United States
Environmental Protection
Agency
EPA-600/R-92-209c
November 1992
<&EPA Research and
Development
PROCEEDINGS: 1991
INTERNATIONAL CONFERENCE ON
MUNICIPAL WASTE COMBUSTION
Volume 3. Sessions 1C, 2C, 3C, 4C, 6C,
7C, 8C, 9A, andlOA/C
Prepared for
Office of Environmental Engineering
and Technology Demonstration
Prepared by
Air and Energy Engineering Research
Laboratory
Research Triangle Park NC 27711
-------
EPA REVIEW NOTICE
This report has been reviewed by the U.S. Environmental Protection Agency, and
approved for publication. Approval does not signify that the contents necessarily
reflect the views and policy of the Agency, nor does mention of trade names or
commercial products constitute endorsement or recommendation for use.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.
>
-------
EPA-600/R-92-209 c
November 1992
PROCEEDINGS:
1991 INTERNATIONAL CONFERENCE ON MUNICIPAL WASTE COMBUSTION
VOLUME 3. Sessions 1C, 2C, 3C, 4C, 6C, 7C, 8C, 9A, 10A, and 10C
Theodore G. Brna, Compiler
U.S. Environmental Protection Agency
Air and Energy Engineering Research Laboratory
Research Triangle Park, NC 27711
Prepared for:
U.S. Environmental Protection Agency
Office of Research and Development
Washington, DC 20460
l.b
-------
Conference Organization
Conference Chairman
Theodore G. Brna, Air and Energy Engineering Research Laboratory (AEERL)
U.S. Environmental Protection Agency (U.S. EPA)
Technical Program Committee
U.S. EPA
James D. Kilgroe, AEERL
Steven J. Levy, Office of Solid Waste
Carlton C. Wiles, Risk Reduction Engineering Laboratory
Air & Waste Management Association
James R. Donnelly, Davy McKee Corporation
Steve Stasko, Air & Waste Management Association
American Society of Mechanical Engineers
Richard S. Magee, Northeast Hazardous Substance Center,
New Jersey Institute of Technology
Robert E. Hall, AEERL, U.S. EPA
Environment Canada
A. Finkelstein, Technology Development Branch
Steven E. Sawell, Wastewater Technology Centre
Exhibition
Len Mafrica, Air & Waste Management Association
Sponsors
Air & Energy Engineering Research Laboratory, U.S. EPA
Risk Reduction Engineering Laboratory, U.S. EPA
Air & Waste Management Association
Participating Organizations
American Society of Mechanical Engineers
Environment Canada
ii
-------
PREFACE
The Second International Conference on Municipal Waste Combustion
(MWC), held in Tampa, Florida during April 16-19,1991, was sponsored by the
Air and Energy Engineering Research Laboratory and the Risk Reduction
Engineering Laboratory of the United States Environmental Protection Agency
and the Air and Waste Management Association. The American Society of
Mechanical Engineers and Environment Canada were participating
organizations.
The Conference program provided an opportunity for exchange of current
research developments on MWC and ash disposal as well as unit operating
experience. The topics discussed included overviews on MWC in the United
States, Canada, and Europe; MWC processes; dry/wet flue gas cleaning
experience; ash characterization, treatment, utilization, and disposal; chlorinated
dioxin/furan control; novel flue gas cleaning technology; environmental
compliance; risk and quality control/quality assurance; municipal waste
management; mercury emission control; sampling and analysis; economic and
social issues; and regulatory effects.
Conference authors came from Australia, Belgium, Bermuda, Canada,
Denmark, Finland, Germany, Italy, Japan, Norway, Sweden, Taiwan, The
Netherlands, the United Kingdom, and the United States of America. Panelists
related U.S. environmental compliance requirements and MWC plant operation
in Session 5 and discussed the international outlook for environmental regulation
in the Conference's concluding session, Session 11. Records of the panel
discussions are not included in these proceedings. The 82 presentations made in
the three concurrent sessions were complemented by the keynote address and
international overviews, all of which are contained in these proceedings.
The proceedings have been compiled in three volumes with papers on
combustion, ash, and flue gas cleaning grouped separately in different volumes.
Papers on related topics have been included in each volume, subject to limiting
each volume to approximately the same size. The papers are listed according to
sessions in which presented, and the sessions contained in each volume are
presented below.
iii
-------
Preface (continued)
Volume 1 contains papers from the following sessions:
• Session P: Opening Plenary Session
• Session 0: International Overviews
• Session 1 A: Combustion I
• Session 2A: Combustion II
• Session 3A: Combustion III
• Session 4A: Fuel Cleaning
• Sessions 6A/6B: Environmental Effects/Risk and Quality
Assurance/Quality Control
• Session 9C: MSW Characteristics and Downstream Effects
• Session 10B: Regulatory Effects
Volume 2 contains papers presented in the following sessions:
• Session IB: Ash Characterization
• Session 2B: Ash Treatment/Utilization I
• Session 3B: Ash Disposal I
• Session 4B: Ash Disposal II
• Sessions 7A/7B: Municipal Waste Management 1/Ash
Treatment/Utilization II
• Sessions 8A/8B: Municipal Waste Management U/Ash Leaching
• Session 9B: Economic and Social Issues
Volume 3 contains papers presented in the following sessions:
• Session 1C: Recent Flue Gas Cleaning Experience I
• Session 2C: Recent Flue Gas Cleaning Experience II
iv
-------
Preface (continued)
Session 3C:
Session 8C:
Session 4C:
Session 6C:
Session 7C:
Session 9A:
Wet and Dry Flue Gas Cleaning Experience
Flue Gas Cleaning: PCDD/PCDF Control
Novel/Emerging Flue Gas Cleaning Technology I
Mercury Control
Novel/Emerging Flue Gas Cleaning Technology II
Sampling and Analysis I
Sessions 10A/IOC Sampling and Analysis n/Flue Gas Cleaning System
Performance
Members of the two panels are listed below:
Session 5 - Impact of Plant Operation on Environmental Compliance
Moderator: Charles O. Velzy, Roy F. Weston, Inc., Valhalla, NY
Panelists: David S. Beachler, Westinghouse Electric Corporation,
Pittsburgh, PA
Francis A. Ferraro, Wheelabrator Technologies, Inc.,
Hampton, NH
Gary Pierce, ABB Resource Recovery Systems,
Windsor, CT
David B. Sussman, Ogden Martin Systems, Inc., Alexandria, VA
Session 11 - Outlook for Environmental Regulation
Moderator: James D. Kilgroe, AEERL, U.S. EPA, Research
Triangle Park, NC
Panelists: A.G. Buekens, Free University of Brussels, Brussels, Belgium
A. Finkelstein, Environment Canada, Ottawa, Ontario
Fred L. Porter, OAQPS, U.S. EPA, Research Triangle Park,
NC
David W. Scott, Warren Spring Laboratory, Stevenage,
Hertfordshire, United Kingdom
v
-------
Intentionally Blank Page
vi
-------
CONTENTS
VOLUME 1
SESSION P: OPENING PLENARY SESSION 1
Opening Remarks Theodore G. Brna
Welcome Martin E. Rivers
Introduction of Keynote Speaker G. Blair Martin
Keynote Address
"Waste Management Issues and Their Impacts on Municipal Waste
Combustion," Stephen A. Lingle 3
SESSION 0: INTERNATIONAL OVERVIEWS 11
Session Co-Chairmen:
Theodore G. Brna
Air and Energy Engineering Research Laboratory (AEERL)
U.S. Environmental Protection Agency (U.S. EPA)
Research Triangle Park, NC
Carlton C. Wiles
Risk Reduction Engineering Laboratory (RREL)
U.S. EPA
Cincinnati, OH
"An Overview of Environment Canada's National Incinerator Testing and
Evaluation Program (NITEP)," Abe Finkelstein 13
"Overview of Recent Standards and Guidelines for Municipal Waste
Combustors in the United States," Fred L. Porter 29
"Municipal Waste Combustion Developments in Europe," A. G.
Buekens and F. De Geyter 37
Preceding page blank
vii
-------
SESSION 1A: COMBUSTION I.
61
Session Co-Chairmen:
Richard S. Magee
Northeast Hazardous Substance Research Center
New Jersey Institute of Technology
Newark, NJ
Robert E. Hall
AEERL
U.S. EPA
Research Triangle Park, NC
"The Effect of Sulfur Compounds on the Formation Mechanism of PCDD
and PCDF in Municipal Waste Combustors," Brian K. Gullett, Kevin R.
Bruce, and Laura O. Beach 63
"Role of Soot in the Transport of Chlorine in Hydrocarbon-Air Diffusion
Flames," S. Venkatesh, K. Saito, J.M. Stencil, V. Majidi, and M. Owens 81
"Control and Simulation of System Behavior of Batch-Fed Solid Waste
Incinerators," Jing T. Kuo and J.S. Wang 101
"Experimental Techniques to Study the Combustion Characteristics of
Two Plastics Commonly Found in Municipal Wastes," Yiannis A.
Levendis and Thomai Panagiotou 119
SESSION 2A: COMBUSTION II 135
Session Co-Chairmen:
Robert E. Hall
AEERL
U.S. EPA
Research Triangle Park, NC
Richard S. Magee
Northeast Hazardous Substance Research Center
New Jersey Institute of Technology
Newark, NJ
"Development of Natural Gas Injection Technology for NOx Reduction
from Municipal Waste Combustors," Hamid A. Abbasi, Mark J. Khinkis,
Craig A. Penterson, Frank Zone, Rob Dunnette, Kunihiro Nakazato,
Patricia A. Duggan, and David G. Linz 137
viii
-------
"Improving Combustion Conditions at RDF Facilities," Joseph J. Bacchi
and Robert J. Allen 153
"Montgomery County South Incinerator Test Project: Formation, Emission,
and Control of Organic Pollutants," James D. Kilgroe, W. Steven Lanier,
and T. Robert von Alten 161
SESSION 3A: COMBUSTION m 177
Session Co-Chairmen:
Robert E. Sommerlad
R-C Environmental Services & Technologies
Branchburg, NJ
James D. Kilgroe
AEERL
U.S. EPA
Research Triangle Park, NC
"Results and Future Plans for the ASME Research Committee's Program
on Research of Fireside Problems in Municipal Waste Combustors,"
Robert E. Sommerlad, Herbert I. Hollander, and Richard W. Bryers 179
"Advanced Pollution Control in Municipal Waste Combustors Using Natural
Gas," W. R. Seeker, G.C. England, R. Lyon, and P. Duggan 197
"Selective Noncatalytic Reduction (SNCR) Performance on Three California
Waste-to-Energy Facilities," Barry L. McDonald, Gary R. Fields, and Mark D.
McDannel 207
SESSION 4A: FUEL CLEANING/PROCESSING 225
Session Co-Chairmen:
Steven J. Levy
Office of Solid Waste (OSW)
U.S. EPA
Washington, DC
Simon Friedrich
Office of Waste Reduction (OWR)
U.S. Department of Energy (DOE)
Washington, DC
ix
-------
"The Effects of Changing Municipal Solid Waste Characteristics on Com-
bustion Fuel Quality," Nicholas S. Artz and Marjorie A. Franklin 227
"Continuous Belt Dryer for the Production of Sludge-Derived Fuel," Klaus
S. Feindler 241
"Partitioning of Elements by Refuse Processing," Gary L. Boley 261
SESSION 6A: ENVIRONMENTAL EFFECTS 277
Co-Chairmen:
Brian K. Gullett
AEERL
U.S. EPA
Research Triangle Park, NC
Hunter F. Taylor
Black & Veatch
Richmond, VA
"Comparison of Potential Greenhouse Gas Emissions from Disposal of
MSW in Sanitary Landfills vs. Waste-to-Energy Facilities," Hunter
F.Taylor 279
"Landfill Gas Issues Affecting the Design and Operation of Waste-to-
Energy Facilities," Eric R. Peterson, Robert B. Gardner, and Paul K.
Foxwell 293
"Case Study: Environmental Review of the Dakota County Resource
Recovery Project," Paul A. Smith and James F. Walsh 305
SESSION 6B: RISK AND QUALITY ASSURANCE/QUALITY
CONTROL 319
Co-Chairmen:
David M. White
Radian Corporation
Research Triangle Park, NC
Robin R. Segall
Entropy Environmentalists, Inc.
Research Triangle Park, NC
-------
"Impact of Risk Assessment on Medical Waste Incinerator Design," James
Hsu and Jean Tilly 321
"A Multi-pathway Health Risk Assessment Methodology for Municipal and
Medical Waste Incinerators," Grant Chin, Narcisco Gonzalez, Lynn Terry,
and Melanie Marty 341
"External QA/QC for Medical Waste Incinerator Emission Testing," R.R.
Segall, S.A. Shanklin, A.L. Cone, and C.E. Riley 355
SESSION 9C MSW CHARACTERISTICS AND DOWNSTREAM
EFFECTS 375
Co-Chairmen:
Simon Friedrich
OWR
U.S. DOE
Washington, DC
Steven J. Levy
OSW
U.S. EPA
Washington, DC
"Methodology for Size and Category Classification of MSW and the
Downstream Effects," J. P. Aittola, W. Hogland, D. Nagelhout, A.J. Poll
and H. Rosvold 377
"Waste-to-Energy Process/Products," Ray D. Gutshall 397
"Chimney for Municipal Waste Combustion Projects," John C. Sowizal and
Richard Wilber 407
SESSION 10B REGULATORY EFFECTS 421
Co-Chairmen:
Robin R. Segall
Entropy Environmentalists, Inc.
Research Triangle Park, NC
David M. White
Radian Corporation
Research Triangle Park, NC
xi
-------
"Environmental Regulation on Municipal Waste Combustion," Scott W.
Clearwater and Martin J. Marchaterre 423
"Municipal Solid Waste Incineration: Emission Testing Experiences in
Minnesota," James F. Idzorek 441
"USEPA Performance Testing and Monitoring Requirements for Municipal
Waste Combustor Air Emissions," Gary McAlister and C.E. Riley 459
xii
-------
CONTENTS
VOLUME 2
SESSION IB: ASH CHARACTERIZATION 1
Session Co-Chairmen:
Carlton C. Wiles
RREL
U.S. EPA
Cincinnati, OH
Steven E. Sawell
Wastewater Technology Centre
Environment Canada
Burlington, Ontario
"An International Perspective on Ash from Municipal Waste Incinerators,"
A.J. Chandler, T.T. Eighmy, J. Hartlen, O. Hjelmar, D. Kosson, S.E. Sawell,
H.A. van der Sloot, and J. Vehlow 3
"Major Findings of the U.S. EPA/CORRE MWC-Ash Study," Haia K.
Roffinan 11
"A Summary of the National Incinerator Testing and Evaluation Program
Ash Characterization and Solidification Studies," S.E. Sawell and T.W.
Constable 9
SESSION 2B: ASH TREATMENT/UTILIZATION 1 57
Session Co-Chairmen:
Steven E. Sawell
Wastewater Technology Centre
Environment Canada
Burlington, Ontario
Carlton C. Wiles
RREL
U.S. EPA
Cincinnati, OH
"The U.S. EPA Program for Evaluation of Treatment and Utilization
Technologies for Municipal Waste Combustion Residues," Carlton C.
Wiles 59
xiii
-------
"A Comparison of Five Solidification/Stabilization Processes for Treatment
of Municipal Waste Combustion Residues, Part I - Physical Testing," Teresa
Holmes, David Kosson, and Carlton Wiles 83
"A Comparison of Five Solidification/Stabilization Processes for Treatment
of Municipal Waste Combustion Residues, Part II - Leaching Properties,"
David S. Kosson, Hans van der Sloot, Teresa T. Holmes, and Carlton C.
Wiles 103
SESSION 3B: ASH DISPOSAL 1 119
Session Co-Chairmen:
David S. Kosson
Department of Chemical and Biochemical Engineering
Rutgers University
Piscataway, NJ
A. John Chandler
A.J. Chandler & Associates
Willowdale, Ontario
"Assessment of the Environmental Impact of MSWI Ash Disposal in
Bermuda," Ole Hjelmar, Jorgen Birkland Andersen, Kirsten Jebjerg
Andersen, Estelle Bjornestad, Erik Aagaard Hansen, Anthony Knap, Kent
Simmons, Clay Cook, Sue Cook, Ross Jones, Alison Murray, and
Tony Riggs 121
"Incinerator Ash Disposal in the Tampa Bay Region," James C.
Andrews, Jr 139
"Compactibility of MSW Ash and Wastewater Sludge Mixtures for
Landfill Codisposal," Jean Benoit and T. Taylor Eighmy 155
SESSION 4B: ASH DISPOSAL II 173
Session Co-Chairmen:
A. John Chandler
A. J. Chandler & Associates
Willowdale, Ontario
David S. Kosson
Department of Chemical and Biochemical Engineering
Rutgers University
Piscataway, NJ
xiv
-------
"An Overview of the Fate of Metals During Combustion at the Mid-
Connecticut Resource Recovery Facility," R. Michael Hartman 175
"Engineering Evaluation of Resource Recovery Residue Utilization Modes,"
Richard W. Goodwin 191
"Ash/Residue Vitrification: Waste Products into Beneficial By-Products,"
Herbert I. Hollander and Howard E. Clark 209
SESSION 7A: MUNICIPAL WASTE MANAGEMENT 1 217
Co-Chairmen:
David B. Sussman
Ogden Martin Systems, Inc.
Alexandria, VA
David Hay
Environment Canada
Conservation and Protection
Ottawa, Ontario
"A Unique Approach to Municipal Waste Management in Chianti, Italy,"
Prakash H. Dhargalkar 219
"Operating a Waste-to-Energy Facility in an Increasingly Stringent
Regulatory Environment," Francis A. Ferraro 231
"Operations Monitoring of Waste-to-Energy Facilities: The Contracting
Communities' Role," Alexander J. Pyatsky and Anthony M. LoRe, Jr 245
"Treatment Options of MSW in Helsinki Area: Combustion or Materials
Recovery and Recycling," J-P. Aittola, P. Korhonen, P. Vanni and, L.
Helminen 253
SESSION 7B: ASH TREATMENT/UTILIZATION II 271
Co-Chairmen:
T.T. Eighmy
Department of Civil Engineering
University of New Hampshire
Durham, NH
XV
-------
Ole Hjelmar
Water Quality Institute
Research Centre
Horsholm, Denmark
"Bottom Ash Reuse as an Aggregate Substitute in Asphaltic Concrete, Part I:
Evaluation of Physical Properties of Bottom Ash," David Gress, Xishun
Zhang, Scott Tarr, Ingrid Pazienza, and Taylor Eighmy 273
"Processed Ash Demonstration Project," Douglas G. Mehan and William
F. Hooper 289
"The Neutralysis System," Robert S. Merdes 305
"Environmental Issues Associated with the Use of MSW Combustor Ash
in Asphalt Paving Mixes," Warren H. Chesner 315
SESSION 8A MUNICIPAL WASTE MANAGEMENT II 333
Co-Chairmen:
David Hay
Environment Canada
Conservation and Protection
Ottawa, Ontario
David B. Sussman
Ogden Martin Systems, Inc.
Alexandria, VA
"Municipal Solid Waste Incineration in the United Kingdom," D.W. Scott 335
"Can a Waste Combustion Plant Improve the City Air Quality?" Antonio
Bonomo 353
"A Case Study of an Intergovernmental Success Story," John S. Hadfield 373
SESSION 8B ASH LEACHING 389
Co-Chairmen:
Ole Hjelmar
Water Quality Institute
Research Centre
Horsholm, Denmark
xvi
-------
T.T. Eighmy
Department of Civil Engineering
University of New Hampshire
Durham, NH
"Analysis of Laboratory and Field Leachate Test Data for Ash from Twelve
Municipal Solid Waste Incinerators," Carol A. Andrews and Mary T.
Hoffman 391
"Leaching Behavior of Residues from Mass Burn and Refuse Derived
Fuel Incinerators," D.W. Sundstrom, H.E. Klei, B.A. Weir, and A.J.
Perna 411
"Chemical Composition of Ash-Monofill Leachates: Actual Field Data," Haia
K. Roffman 427
SESSION 9B ECONOMIC AND SOCIAL ISSUES 447
Co-Chairmen:
Hunter F. Taylor
Black & Veatch
Richmond, VA
Brian K. Gullett
AEERL
U.S. EPA
Research Triangle Park, NC
"Municipal Solid Waste Incinerator Impacts on Residential Property Values
and Sales in Host Committees," Chris Zeiss 449
xvii
-------
Intentionally Blank Page
xviii
-------
CONTENTS
VOLUME 3
SESSION 1C: RECENT DRY FLUE GAS CLEANING EXPERIENCE 1 1
Session Co-Chairmen:
Dennis C. Drehmel
AEERL
U.S. EPA
Research Triangle Park, NC
Jeffrey L. Hahn
Ogden Projects, Inc.
Berkeley, CA
"Overview of Air Pollution Controls for Municipal Waste Combustors,"
James R. Donnelly 3
"Toxic Metal Emissions from MWCs and Their Control," Theodore G. Brna 23
"Determination of Efficiency of Flue Gas Cleaning Systems on Municipal Solid
Waste Incinerators in Denmark," Peter Blinksbjerg 41
Two-and-a-Half Years' Operating Experience at the Warren County Energy
Resource Recovery Facility," Claus Jorgensen, William B. VanHooser, Janine G.
Kelly, and William B. Cook 47
SESSION 2C: RECENT DRY FLUE GAS CLEANING - EXPERIENCE II 69
Session Co-Chairmen:
Jeffrey L. Hahn
Ogden Projects, Inc.
Berkeley, CA
Dennis C. Drehmel
AEERL
U.S. EPA
Research Triangle Park, NC
"Operating Experience and Emission Rates of Scrubber Baghouses, Scrubber
ESP's, and Furnace Injection of Lime at RDF Facilities," J. Michael Smith and
R. Michael Hartman 71
Preceding page blank
-------
"Review of SD/FF versus SD/ESP System Performance for Acid Gas, Particulate,
and CDD/CDF Control," David M. White and Kristina L. Nebel 77
"Reduction of Dioxin and Heavy Metals by Semi-dry Scrubber and Baghouse
from MSW Incinerator," T. Nakao and E. Shibuya 87
SESSION 3C: WET AND DRY FLUE GAS CLEANING EXPERIENCE 99
Session Co-Chairmen:
Marjorie J. Clarke
Environmental Consultant
New York, NY
David S. Beachler
Resource Energy Systems Division
Westinghouse Electric Corporation
Pittsburgh, PA
"Gas Cleaning in Connection with Waste Incineration," Bjorn Lindquist 101
"The Joint EC/EPA Mid-Connecticut Test Program: A Summary," T. G. Brna,
J. D. Kilgroe, and A. Finkelstein 113
"Effective Control of Mercury and Other Pollutants from MWC Facilities by
EDV Technology," Gerwyn Jones 133
SESSION 4C: FLUE GAS CLEANING: PCDD/PCDF Control 149
Session Co-Chairmen:
David S. Beachler
Resource Energy Systems Division
Westinghouse Electric Corporation
Pittsburgh, PA
Marjorie J. Clarke
Environmental Consultant
New York, NY
"Application of DeNOx: Catalysts for the Reduction of Emissions of
PCDD/PCDF and Other PICs from Waste Incineration Facilities by Catalytic
Oxidation," H. Hagenmaier, K-H. Tichaczek, H. Brunner, and G. Mittelbach 151
"Techniques for Dioxin Emission Control," Vladimir Boscak and George
Kotynek 159
XX
-------
"Reduction of Dioxins/Furans Formed in Combustion by DeNovo Synthesis,"
Aaron J. Teller 175
SESSION 6C: NOVEL/EMERGING FLUE GAS CLEANING
TECHNOLOGY 1 189
Co-Chairmen:
James R. Donnelly
Davy McKee Corporation
San Ramon, CA
Karsten S. Felsvang
Niro Atomizer, Inc.
Columbia, MD
"Pilot-Scale Testing of the Ammonia Injection Technology for Simultaneous
Control of PCDD/PCDF, HC1, and NOx Emissions from Municipal Solid Waste
Incinerators," Laszlo Takacs and George L. Moilanen 191
"A Novel Calcium-Based Sorbent for the Removal of Flue Gas HC1 by Dry
Injection," Wojciech Jozewicz , Brian KGullett, and Shiaw C. Tseng 209
"Results of Full Scale Dry Injection Tests at MSW Incinerators Using a New Active
Absorbent," Karsten S. Felsvang and D. Helvind 229
SESSION 7C: MERCURY CONTROL 245
Co-Chairmen:
Karsten S. Felsvang
Niro Atomizer, Inc.
Columbia, MD
James R. Donnelly
Davy McKee Corporation
San Ramon, CA
"Municipal Waste Combustors: A Survey of Mercury Emissions and
Applicable Control Technologies," David M. White, Kristina L. Nebel, and
Michael G. Johnston 247
"Controlling Mercury Emissions from RDF Facilities," Gary Pierce 259
"Mercury Emission Control: Sodium Sulphide Dosing at Hogdalen Plant in
Stockholm," Christer Andersson and Bengt Weimer 275
XXL
-------
"Control of Mercury and Dioxin Emissions from United States and European
Municipal Solid Waste Incinerators by Spray Dryer Absorption Systems,"
B. Brown and K.S. Felsvang
287
SESSION 8C NOVEL/EMERGING FLUE GAS CLEANING
TECHNOLOGY E 319
Co-Chairmen:
Charles B. Sedman
AEERL
U.S. EPA
Research Triangle Park, NC
R. Michael Hartman
ABB Resource Recovery Systems
Windsor, CT
"Retrofit Acid Gas Emissions Controls for Municipal Waste Incineration: An
Application of Dry Sorbent Injection," Jan T. Zmuda and Peter V.
Smith 321
"Pilot Plant Tests to Obtain HC1 and Hg Reduction in Emissions Produced by a
Municipal and Hospital Solid Waste Combustor," A. Magagni and G. Boschi 339
SESSION 9A SAMPLING AND ANALYSIS 1 349
Co-Chairmen:
James D. Kilgroe
AEERL
U.S. EPA
Research Triangle Park, NC
Stellan Markland
Institute of Environmental Chemistry
University of Umea
Umea, Sweden
"Mercury Emission Monitoring on Municipal Waste Combustion," Hartmut
Braun and Andreas Gerig 351
"A New Method for Sampling Halogenated Dioxins and Related Compounds
in Flue Gases," S. Markland, I. Fangmark, and C. Rappe 367
xxii
-------
SESSION 10A
SAMPLING AND ANALYSIS II
373
Co-Chairman:
Stellan Markland
Institute of Environmental Chemistry
University of Umea
Umea, Sweden
James D. Kilgroe
AEERL
U.S. EPA
Research Triangle Park, NC
"Assessment of an On-Line CI-Mass Spectrometer as a Continuous Emission
Monitor for Sewage Sludge Incinerators," K.R. Campbell, D.J. Hallett, R.J.
Resch, J. Villinger, and V. Federer 375
"Sampling and Analysis of Air Toxics from Municipal Waste Combustors,"
Justice Manning, John Chehaske, Paul Fraley, Dave Wetmore, and Eric
Hollins 381
"Development of Source Testing, Analytical and Mutagenicity Bioassay
Procedures for Evaluating Emissions from Municipal and Hospital Waste
Combustors," R.R. Watts, P.M. Lemieux, R.A. Grote, R.W. Lowans, R.W.
Williams, L.R. Brooks, S.H. Warren, D.M. Marini, and J. Lewtas 407
SESSION 10C FLUE GAS CLEANING SYSTEM PERFORMANCE 425
Co-Chairmen:
R. Michael Hartman
ABB Resource Recovery Systems
Windsor, CT
Charles B. Sedman
AEERL
U.S. EPA
Research Triangle Park, NC
"A Review of Activated Carbon Technologies for Reducing MSW Incinerator
Emissions," Marjorie J. Clarke 427
xxiii
-------
"Emissions Control of Polychlorinated Dibenzo-p-Dioxins and Polychlorinated
Dibenzofurans at Municipal Waste Combustors," Shiaw C. Tseng,
Wojciech Jozewicz, and Charles B. Sedman 447
"Design and Operation of Pulse-Jet Fabric Filters for Incineration Air Pollution
Control," William Gregg 465
xxiv
-------
SESSION 1C: RECENT DRY FLUE GAS CLEANING EXPERIENCE I
Co-Chairmen:
Dennis C. Drehmel
Air and Energy Engineering Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC
Jeffrey L. Hahn
Ogden Projects, Inc.
Berkeley, CA
1
-------
Intentionally Blank Page
2
-------
The work described in this paper was not funded by the U.S. Environmental
Protection Agency. The contents do not necessarily reflect the views o1 the
Agency and no official endorsement should be inlerred.
OVERVIEW OF AIR POLLUTION CONTROLS FOR MUNICIPAL WASTE COMBUSTORS
J.R. Donnelly
Davy Environmental
San Ramon, California 94583
ABSTRACT
The growth in incineration of municipal solid waste has lead to concerns of potential harmful
emissions of acid gases, heavy metal and toxic trace organic compounds into the environment. This
has lead to the promulgations of emissions control limits in many countries in Europe, the United
States and Japan.
Several different technologies are currently available and new approaches are emerging for
improved control of specific pollutants of concern. Technology transfer is such that a successful
application of a new technology any where in the world may rapidly lead to applications throughout
the world.
This paper presents an overview of technologies being applied to MWC's for the control of NOx
acid gases, particulate matter, heavy metals and toxic trace organic compounds (PCDD's/PCDF's).
The technologies presented are reviewed as to their state of development and control efficiencies.
Preceding page blank 3
-------
INTRODUCTION
Environmentally sound disposal of municipal and industrial refuse has become a major issue in
the past decade. The volume of waste generated has continued to grow annually and traditional
disposal methods (land filling or ocean dumping) are becoming less acceptable because of cost
and environmental concerns. Incineration of refuse in modern high efficiency combustors is being
employed to a growing fraction of the waste stream to achieve significant reduction in refuse
volume while in many instances achieving energy recovery in the form of steam or electricity.
This increase in incineration has been coupled with an increase in the complexity and efficiency of
air pollution controls to limit incinerator emissions. This paper presents a discussion of the air
pollution controls applied to municipal waste combustors.
REFUSE INCINERATION
Refuse or municipal solid waste includes non hazardous waste generated in households,
institutions, (excluding hospital wastes) commercial and light industrial facilities, agricultural
wastes and sewage sludge. In 1988 the United States generated approximately 180 thousand tons
of refuse. (1)
The refuse per capita generation rate in the U.S. is the highest in the world and has shown an
annual growth over the past several decades. Because of this continual refuse growth,
environmental concerns with land filling or ocean dumping and the lack of availability and cost of
landfills, the public has began looking into alternate ways to handle its refuse. This has included
composting, recycling and incineration. Of these three alternatives, incineration has shown the
ability to achieve the greatest reduction in refuse volume (70 - 90%) and can be used in
conjunction with the other two alternates to achieve even greater reduction in disposal volume.
Incineration is currently being used to treat about 15% of the refuse stream with this percentage
expected to increase to 25-30 % by 1995.(2)
This increase in refuse incineration has lead to increased concern over air pollution from
these incinerators, which in turn has lead to the promulgation of air emissions standards and the
application of modern air pollution controls to limit these emissions. The U.S. Environmental
Protection Agency in response to the public's concerns issued "New Source Performance
Standards and Emissions Guidelines for Existing Facilities" in February 1991. Table 1 summarizes
these proposed standards and guidelines.
4
-------
Table 1. Municipal Waste Combustion Emission Standards. C3)
New Source
Performance Standards
Emission Guidelines For Existing Facilities
Capacity-Tons/day
Unit
Unit
Facility
>250
>250 s 2200
>2200
Particulate Matter-(gr/dscf)
0.015
0.030
0.015
Opacity-%
10
10
10
Organic Emissions-ng/dscm
Total Chlorinated PCDD Plus PCDF
-Mass bum units
-RDF fired units
30
30
125
250
60
60
Acid Gas Control
% Reduction or Emissions-(ppm)
HC1
95 (25)
50 (25)
90 (25)
so2
80(30)
50 (30)
70 (30)
NOx
(180)
None
None
Carbon
Monoxide, ppm
50-150*
50-250*
50-250*
* Range of values reflect differing types of MWC's
In proposing these standards, the EPA recognized differences in facility size, the type of
incineration (mass burn fired versus refuse derived fuel fired) and new sources versus existing
sources. The facility capacity refers to the total burn rate for all refuse combustors at a single
site. EPA selected total particulate matter emissions limits as the way of controlling trace heavy
metal emissions limits. EPA will add emission limits for mercury, cadmium and lead emissions in
the coming year.
Emissions limits are established for the total emissions of the poly-chlorinated dibenzyl-dioxins
(PCDD) plus polychlorinated dibenzyl-furans (PCDF). These compounds were selected as
surrogates for organic emissions because of their potential adverse health effects. In addition
EPA has established carbon monoxide (CO) emission limits as a measure of good combustion
practices which limit the formation of PCDD, PCDF and their key precursors. ¦ The proposed CO
limits vary from 50 to 250 ppm (@ 7%Oz dry gas conditions) depending on the type of
combustion.
Acid gas emissions (HC1 and S02) are based on either a percent reduction or a maximum
stack emission level whichever is the least stringent. Nitrogen Oxides (NOx) emissions levels are
proposed only for large new sources.
5
-------
of MWC facility operating supervisors and site specific training for operating personnel.
In addition to EPA standards, many states and local air pollution control districts have
developed their own sets of emission standards for new sources. Many of these standards (or
permit conditions) are more stringent than the proposed EPA standards.
INCINERATION SOURCES
Several different technologies are employed in refuse combustion. These include mass burn
combustors (modular, traveling crate and rotary combustors) refuse derived fuel-fired combustors
and to a lesser extent fluid-bed combustors.
MASS BURN COMBUSTORS
Mass burn combustors are the predominant type of incinerators currently being employed for
refuse. These are characterized by accepting refuse which has undergone very limited
preprocessing other than removal of large oversized items. Mass burn combustors may be of the
modular (or starved air) type, the traveling grate type or the rotary combustor type. Figure 1
shows a cutaway view of a traveling grate mass burn incinerator.
Figure 1. - Mass Burn Incinerator. (4)
(Courtesy - Joy Technologies Inc.)
6
-------
Typically, refuse is dumped into a tipping bay where it is mixed and oversize objects are
removed by an overhead crane. The overhead crane dumps the mixed refuse into a feed chute
where it is fed into the grate system by a hydraulic ram. The traveling grates move the refuse
through the various zones of the combustion chamber in a tumbling motion. Air is added at
multiple points through the grate to assist in refuse dry out, combustion and burnout. Additional
combustion air may be added above the grate to assist in flue gas mixing and complete
combustion.
Residual ash falls from the grate into a wet quench system and then is removed and sent to a
landfill. Rue gases pass upward to a burnout zone where the temperature is maintained at about
1800°F for 1 to 2 seconds to ensure complete destruction of organic compounds. The flue gas
then passes through the boiler and economizer where its temperature is reduced to 350 - 450°F
prior to entering the flue gas cleaning system.
REFUSE DERIVED FUEL (RFD) COMBUSTORS
RDF fired combustors are designed to burn a fuel which has been pre-processed to produce a
fuel with a more uniform size, composition and heat rate. A variety of combustor designs can be
employed with RDF depending on the degree of preprocessing. The simplest form and most
commonly employed method of RDF processing is shredding of refuse followed by magnetic
separation, to remove ferris metals, and in some cases air classification to remove ash. This type
of fuel is often burned in a spreader stoker combustor or suspension fired over a stoker.
The RDF may be further processed to produce; a densified fuel, by pelletizing; a recovery
prepared RDF in which a larger portion of metals and glass are removed; or a fluff RDF for co-
firing with coal in suspension fired combustors.
RDF combustors range in size from approximately 400 to 1000 TPD capacity. Because of the
nature of the fuel and firing, particulate matter carry over to the air pollution control system is
generally much higher than for a mass burn combustor.
EMISSIONS CHARACTERIZATION
Refuse inciner; ttion has the potential of emitting a wide range of pollutants to the
environment. These potential emissions arise from compounds present in the refuse stream, are
formed as a part of the normal combustion process, or are formed due to incomplete combustion.
Table 2 lists principal potential MWC emissions and the prime source for each.
7
-------
Table 2. Principal MWC Emissions and Sources.
Pollutant
Principal Source
Particulate Matter
Acid Gases
Ash In Waste Stream
Heavy Metals
(Arsenic, Cadmium,
Lead, Mercury)
Organic Compounds
(Dioxins, Furans)
HC1
so2
S03
HF
NO,
CO
Chlorinated Plastic In Waste Stream
Sulfur Compounds In Waste Stream
Oxidation of S02 In Flue Gas
Fluorocarbons In Waste Stream
Air & Fuel Nitrogen Conversion
Incomplete Combustion
Metal Compounds In Waste Stream
Products of Incomplete Combustion
or Contained in Waste Stream
Particulate matter consists primarily of non-combustible inorganic material entrained in the
flue gas. Particulate matter typically ranges in size from less than one micron to about 50
microns. The uncontrolled particulate matter emission rate varies substantially for the different
types of MWC's. Modular incinerators produce the lowest levels of uncontrolled emissions with
RDF fired units having the highest.
Acid gases; hydrogen chloride (HC1), sulfur dioxide (S02) and hydrogen fluoride (HF) are
formed during the combustion of chloride, sulfur and fluoride containing compounds found in the
waste stream. A small fraction (approximately 1 to 5 percent) of the S02 in the flue gas is
oxidized to sulfur trioxide (S03). These gases, in the presence of water or water vapor, react to
form hydrochloric, sulfurous, hydrofluoric or sulfuric acid.
Nitrogen oxides (NO,) are found predominantly in the form of Nitrous Oxide, NO, and are
formed primarily through the conversion of fuel bound nitrogen although some nitrogen in the
combustion air may also be converted. Carbon monoxide (CO) is formed through the incomplete
combustion of organic compounds in the waste stream and is used as an indicator of combustion
conditions.
Heavy metals compounds of concern emitted from MWC's include the oxides and chlorides
of arsenic, cadmium lead and mercury. These compounds are formed from the combustion of
heavy metal containing components »>f the waste stream such as batteries, plastics, paper products
and metal alloys. A number of these compounds have boiling points or sublime at temperatures
below the 1800°F typical of incineration systems and are thus vaporized into the flue 2 gas. As
the flue gas temperature cools, they tend to condense out and are concentrated on fine
particulate matter in the flue gas. For the compounds of mercury and lead, a significant fraction
may remain in the vapor state at typical incinerator exit flue gas temperatures.
8
-------
Organic Emissions are a result of incomplete combustion of compounds found in the waste
stream. The prime organic compounds of concern are polychlorinated dibenzyl-dioxins (PCDD)
and polychlorinated dibenzyl-furans (PCDF). These emissions can arise from; incomplete thermal
destruction of PCDD and PCDF containing materials in the waste stream, incomplete thermal
destruction of other organic compounds that produce PCDD/PCDF precursors and through
chemical reactions that occur at relatively low temperatures downstream of the combustor.
Table 3. shows typical uncontrolled and controlled emissions for a number of pollutants of
concern from refuse incineration. Percent reduction ranges typical of levels being achieved
utilizing best available control technologies are also shown for each pollutant.
Table 3. Typical Refuse Incinerator Uncontrolled and Controlled Emissions.
Pollutant
Uncontrolled
Emissions
Controlled
Emissions
Percent
Reduction
Particulate Matter, gr/dscf
0.5-4.0
0.002-0.015
99.5 +
Acid Gases ppmdv
HC1
so2
HF
NOx
400-100
150-600
10-0
120-300
10-50
5-50
1-2
40-120
90-99+
50-90+
90-95 +
30-65*
Heavy Metals mg/nm3
Arsenic
Cadmium
Lead
Mercury
<0.1-1
1-5
20-100
<0.1-1
<0.01-0.1
<0.01-0.5
<0.1-1
<0.1-0.7
90-99+
90-99+
90-99+
10-90+
Total PCDD/PCDF ng/nm3
20-500
<1-10
80-99
Reference conditions - Dry Gas @ 12%C02
* Reduction associated with non-selective catalytic reduction
Modern refuse incinerator installations achieve very low emissions due to the proper
application and operation of available air pollution control systems. The average incinerator
emissions levels for all pollutants has decreased substantially over the past five years as more
modern installations have been brought into service.
9
-------
AIR POLLUTION CONTROL SYSTEMS
Air pollution control systems for refuse incinerators can be classified by either the pollutant
they control or their operating principals. Often more than one control device will be used in
series to control a number of pollutants. The most common examples of this are the use of an
electrostatic precipitator followed by a wet scrubber or a spray dryer absorption system including
an electrostatic precipitator or fabric filter. The following sections discuss control techniques for
the major stack emissions from municipal waste combustion.
CARBON MONOXIDE CONTROLS
Carbon monoxide emissions are controlled by employing "Good Combustion Practices".
These practices include operational and incinerator design elements for controlling the amount
and distribution of excess air in the flue gas to ensure there is enough oxygen present for
complete combustion. The design of modern efficient combustors is such that there is adequate
turbulence in the flue gas to ensure good mixing, a high temperature zone (greater than 1800°F)
to complete burnout and a long enough residence time at the high temperature (1-2 seconds).
The feed to the combustor is controlled to minimize fuel spikes that lead to fuel rich firing.
The combustor is equipped with adequate instrumentation and combustion air controls to adjust
for rapid changes in fuel conditions. These types of controls can limit CO formation to 150 ppm
or less depending on the combustor design.
Good combustion practices also limit PCDD/PCDF emissions exiting the incinerator. This
is accomplished by maintaining firing conditions that destroy PCDD/PCDF's found in the fuel and
by destroying PCDD/PCDF's precursors that may be formed from the combustion of other
chlorinated and organic compounds.
NITROGEN OXIDES CONTROLS
Nitrogen oxides (NO,) emissions are controlled by limiting their formation in the incinerator
using staged combustion or applying selective non catalytic reduction to reduce the NO, content
in the flue gas. Staged combustion is accomplished by splitting up the introduction of combustion
air into the combustor so that areas of fuel rich and fuel lean firing are established. This will
lower the peak flame temperatures and limit the amount of oxygen available to react with
nitrogen in the air at the peak temperature. Introduction of additional secondary air downstream
in the combustor will ensure complete combustion and minimize CO formation. Generally
staged combustion is effective in reducing NOx formation due to air nitrogen conversion but is not
very effective in limiting conversion of fuel bound nitrogen to NOr
10
-------
NOx present in the flue gas can be reduced by employing either a selective catalytic or non
catalytic reduction process. The selective catalytic reduction (SCR) process utilizes ammonia
injection upstream of a catalytic reactor, at about 600-650°F, to reduce NOx to nitrogen. Selective
catalytic reduction has been applied to a wide range of combustion sources where 80-85 percent
NOx reduction has been demonstrated. However, because for the nature of the compounds found
in refuse incinerator flue gas, the successful application of SCR requires installation downstream
of the acid gas and particulate control systems with subsequent reheat to the reactor operating
temperature. Because of these constraints, only limited SCR applications to refuse incinerator
flue gases have been attempted.
Selectively, non-catalytic reduction (SNCR) reduces flue gas NOx through the reaction with
ammonia in a temperature range of 1700-1900°F. The ammonia may be supplied as anhydrous
ammonia, aqueous ammonia or as urea. At flue gas temperatures above 1900°F, the oxidation of
ammonia to NOx increases and SNCR can actually result in an increase in overall NOr At
temperatures below about 1700° NOx reduction falls off and ammonia break through increases
leading to the potential for a visible ammonium chloride plume.
Ammonia injection, also known as Thermal De-NOD has been applied to many different
combustion sources including mass burn refuse incinerators. NOx reduction levels of up to 65
percent have been demonstrated at an ammonia to NOx ratios of about two with ammonia break
through as low as 5 ppm. This corresponds to an NOx emission level as low as approximately 60
ppm. Thermal De-NOx operates most efficiently under steady state operating conditions.
Changes in fuel feed rate, excess air rate or incinerator load can significantly change flue gas
conditions at the ammonia injection point leading to a major change in control efficiently.
Urea injection has been demonstrated full scale on refuse combustors in the U.S. and
Europe. Urea injection offers the advantage on not requiring a hazardous material for operation.
At the injection temperatures employed (1600-1900°F) the urea quickly breaksdown to form the
active reagent. In some cases, reaction enhancers are added to the urea to expand the effective
temperature window to as low as 1200°F. Tests with urea injection have achieved greater than 65
percent NOx reduction with very low (approximately 5 ppm) ammonia slip. (5)
Acitivated carbon reactors are under development in Germany for the removal of NO,,
PCDD/PCDF, and mercury from incinerator flue gases. These reactors are placed downstream of
a dry scrubbing system in cleaned flue gases at temperatures between 60 and 150°C. Ammonia is
injected upstram of the reactor. Test work has shown 50-75% NOx reductiion and NO, emissions
between 60 and 120 ppm. (<)
PARTICULATE MATTER CONTROLS (7)
Particulate emissions are primarily controlled by electrostatic precipitators (ESP's) or fabric
filters, although wet scrubbers are sometimes used on small incinerators or in series with ESP's
for additional control. ESP's are installed either alone, to control particulate emissions, or after a
11
-------
spray dryer, as a part of an acid gas cleaning system. Fabric filters are typically installed
downstream of a quench tower or spray dryer where the conditions of increased fl ue gas moisture
and lowered temperature aid in protecting filter bags from hot embers.
Electrostatic Precipitators
Electrostatic precipitators collect particulate matter by introducing a strong electrical field in
the flue gas which imparts a charge to the particulates present. These charged particles are then
collected on large plates which have an opposite charge applied to them. The collected
particulate is periodically removed by rapping the collection plates. The agglomerated particles
fall to a hopper where they are removed. Key design parameters for electrostatic precipitators
include, particulate composition, density and resistivity; flue gas temperature and moisture
content; inlet particulate loading and collection efficiency; specific collection area (SCA = square
feet of collecting surface per 1,000 acfm of flue gas) and number of fields; flue gas velocity,
collector plate spacing; rapping frequency and intensity; and transformer rectifier power levels.
Table 4. presents sizing parameters typical for ESP's applied for incinerator particulate
emissions control.
Table 4. Electrostatic Precipitator - Design Parameters.
Particulate Loading, gr/acf 0.5-9
Required Efficiency, % 98-99.9
Number of Fields 3-4
SCA, ft2/1000 acfm 400-550
Average Secondary Voltage, kv 35-55
Average Secondary Current mA/1000ft3 30-50
Gas Velocity, ft/sec. 3.0-3.5
Acid Gas
Particulate Control
Flue Gas Temperature, °F 350-450 230-300
Flue Gas Moisture, % Vol. 8-16 12-20
Ash Resistivity, ohm-om 10®-1012 108-109
The ranges in parameters shown reflect straight ESP particulate control applications and ESP
applications as a part of an acid gas cleaning system. Although the inlet particulate loading to the
ESP is much higher as part of an acid gas cleaning system, the number of fields and specific
collecting area required to achieve a similar outlet emission, don't change significantly. This is
due to lower ash resistivity values and increased flue gas moisture contents which improve the
12
-------
ESP's performance. Incinerators which have had spray dryers retrofitted in front of existing
ESP's, have in most cases been able to maintain the same level of particulate emissions (e.g. 0.01
to 0.015 gr/dscf @ 12% C02)
Weighted wire, rigid frame and rigid electrode type precipitators are employed for incinerator
applications, however, rigid frame and rigid electrode types predominate. This is related to the
corrosive gas conditions and sticky nature of the fly ash being collected. Electrode failures
associated with rigid frame and rigid electrode systems are less frequent than for weighted wire
years. This is especially true where higher rapping forces are needed to dislodge the sticky fly
ash. For rigid frame systems, high alloy (e.g. Incoloy 825) spring wound electrodes are also used
to minimize electrode corrosion problems.
The insulator compartment ventilation system is designed to minimize the effects of the
corrosive nature of the flue gas and fly ash stickiness. A pressurized ventilation system, employing
heated air is recommended to maintain clean insulators and reduce potential electrical tracking
problems.
Fabric Filters
Both reverse air and pulse jet type fabric filters are used for particulate emission control on
refuse incinerators. Each type offers advantages that should be evaluated on a site specific basis.
Both types are capable of achieving particulate emissions on the order of 0.01-0.015 gr/dscf @
12% C02 or lower. Table 5 presents design parameters typical of incinerator fabric filter
applications.
Table, 5. Fabric Filter Parameters Reverse Air.
Reverse Air
Pulse Jet
Operating Temperature, °F
Type Fabric
Fabric Coating
Woven Fiberglass
10% Teflon B or Acid
Resistant
230 -450
Fabric Weight, Oz/Yd2
Bag Diameter, Inches
Net Air to Cloth Ratio
1.5 - 2.0:1
6
4-6
3 - 4
9.5
8
16 or 22
6
3.5 - 4.0:1
4
8- 10
1.5 - 2
Minimum Compartments
Overall Pressure Drop, in W.G.
Estimated Bag Life, Years
The temperature ranges shown represent both operation after a dry quench chamber
13
-------
(350-450°F) and a spray dryer (230-300°F). For these temperatures ranges woven fiberglass is
typically used as the bag material although Nomex fabric is also used. A 10 percent Teflon B
coating is the most commonly specified with acid resistant coating also used.
The bag sizes differ substantially for the two types of filters. Reverse air filters generally
employ eight inch diameter by up to 24 feet long bags. Pulse jet bags are usually six inch
diameter by 12 to 14 feet long. However, some vendors offer a low pressure pulse filter with up
to 24 feet long bags. The biggest differences in operating parameters are in the air to cloth ratio
and system pressure drop. Pulse jet filters generally operate at double the air to cloth ratio as
reverse air filters and nearly double the pressure drop. This results in more frequent bag cleaning
and a substantially shorter bag life.
The main advantages of a pulse jet fabric filter are a lower capital cost and a smaller foot
print. However, because of the shorter bag life and higher pressure drop, the pulse jet filter
generally has a higher total evaluated cost for plants exceeding 15 years of life. A reverse air
filter typically has lower particulate emissions when compared to a pulse jet filter.
The majority of fabric filter applications are as a part of an acid gas cleaning system and
incorporate specific design features for operating after a spray dryer. The flue gas after a spray
dryer has been cooled (240-300°F), has a high moisture content (12-20 percent), is closer to the
dew point (80-160°F) and may have a higher particulate loading. These flue gas conditions can
lead to severe corrosion and baghouse plugging.
Corrosion control is accomplished by: insulation design; control of air in-leakage into the
filter; hopper heating; and, in some instances, coating of the fabric filter internals with an acid
resistant material. Insulation specifications usually require a minimum of four inches with double
lapping on side panels and with the insulation extending into the hopper crotch areas. Air in-
leakage is controlled by good quality control during erection and by minimizing the number of
openings into the filter. Hopper heating is used to maintain the hopper skin temperature at the
flue gas temperature to prevent cold spots and aid in maintaining product flowability.
As part of an acid gas cleaning system, the fabric filter also acts as a reactor to aid in acid gas
absorption, especially for sulfur dioxide. Sulfur dioxide in the flue gas is absorbed by alkaline
material in the filter cake on the bags. Therefore when a bag is freshly cleaned, S02 absorption
decreases. In order to minimize this impact on overall absorption, the number of bags being
simultaneously cleaned should be minimized. This can be accomplished by increasing the number
of compartments. A minimum of six compartments is generally specified for acid gas cleaning
systems.
Wet Scrubbers
Wet scrubbers are typically employed as part of a two stage flue gas cleaning system
downstream of an electrostatic precipitator where they function as a particulate removal polishing
stage and as an acid gas absorber. A venturi scrubber followed by a packed or tray tower is
-------
commonly used, however, other types of wet scrubbers such as charged droplet scrubbers are also
used. Figure 2. shows a typical wet scrubber design used for both particulate and acid gas control.
Figure 2.
1 saturator venturi /1 st scrubbing stage
2 lamellar droplet separator
3 radial-flow scrubber/2nd scrubbing stage
4 lamellar droplet separator
Saturator Venturi with Radial-Flow Scrubber.
(Courtesy Lurgi Corporation)
Typically, water is recycled in the venturi stage to achieve particulate removal. Hydrogen
chloride present in the gas would also be removed in this stage. Additional particulate and acid
gas removal can take place in the second scrubbing stage. Absorption of SOz is enhanced in this
stage by maintaining a recirculating solution pH in the range of about 6.5 to 9 through addition of
caustic (sodium hydroxide). A blow down stream is maintained for each stage to control the
recirculating solution solids content. Typical design parameters for refuse incinerator wet
scrubber applications are presented in Table 6.
Table 6. Wet Scrubber Design Parameters.
Venturi Stage Absorber Stage
Gas Velocity, Ft/Sec
Pressure Drop, Inches W.C.
L/G, Gal/Kacfm
Scrubbing Media
Solution pH
90-150
40-70
10-20
Water
<1-2
6- 10
4 - 8
20 -40
Caustic
6.5 - 9
15
-------
Materials of Construction
High Alloy Steel
(eg. Inconel,
Hastelloy)
FRP
Lined Carbon Steel
The venturi section is subjected to severe corrosive conditions due to the low circulating
solution pH, high hydrochloric acid concentration and the presence of small amounts of sulfuric,
nitric and hydrofluoric acids. The scrubber inlet temperature may be as high as 450°F which
precludes the use of corrosion resistant resins, therefore, high alloy steels are typically specified as
the materials of construction. In cases where the inlet flue gas contains high levels of particulate
matter, the venturi section may be lined with a corrosion resistant materials such as bricks. The
venturi section typically is equipped with a set of emergency quench nozzles to ensure that the
flue gas temperature leaving the venturi is maintained at an acceptable level for the absorber
stage materials of construction.
The absorber stage may be a packed tower, a tray tower or a radial flow tower as shown in
Figure 2. Materials of construction for the absorber are typically fiberglass reinforced plastics
(FRP) although carbon steel vessels lined with rubber or a corrosion resistant resin material, are
also used.
ACID GAS CONTROLS (8)
Control of refuse incinerator acid gas (HC1, SO^ S03, and HF) emissions is achieved by dry
sorbent injection, spray dryer absorption or wet scrubbing. Each of these types of technologies
has been successfully applied to meet existing emissions regulations, however, as emissions
limitations become more stringent, the trend is toward spray dryer absorption and wet scrubbing.
Dry Sorbent Injection. DSI
Dry sorbent injection involves the addition of an alkaline material, usually hydrated lime,
Ca(OH)2 or soda ash, Na2(C03), into the gas stream to react with acid gases present thus
producing a salt which is collected in a particulate collection device. This very simple process can
capture up to 90% of the HC1 present in the flue gas and about 50% of the S02. However,
stoichiometric ratios (equivalents of alkali added per equivalents of acid in the'flue gas) are high,
typically on the order of 2 to 4. Therefore, simple DSI applications are normally limited to small
facilities with moderate emissions control requirements.
The overall acid gas control efficiency of DSI can be improved and reagent consumption
decreased by:
° Increasing flue gas humidity
° Recycling reaction products into the flue gas stream
16
-------
Increasing the flue gas relative humidity can be accomplished by cooling the flue gas using
heat exchangers or by quenching the flue gas using water sprays. Both approaches are
commercially applied however, the use of a quench chamber predominates.
Typically flue gas from the incinerator enters a 3 to 5 second retention time cooling tower
(or dry quench chamber) where water is sprayed into the gas to lower the temperature. The flue
gas temperature leaving the cooling tower is maintained at a temperature high enough to ensure
that all water droplets evaporate (300-350°F). Dry reagent is then mixed with the flue gas via
pneumatic transport systems or eductor Venturis. The reagent reacts with acid gases prior to
removal in a dust collector (typically a fabric filter). A portion of the collected reaction products
in some cases is re-injected to increase acid gas removal and decrease reagent consumption .
Humidification and reagent injection steps can also be carried out together in specially designed
reactors. This type of process can achieve greater that 95% HC1 removal and 90% S02, removal
at stoichiometric ratios between 1 and 2.
Spray Dryer Absorption SPA
Spray dryer absorption has been widely applied for refuse incinerator emissions control and
has been specified as Best Available Control Technology (BACT) in a number of air permits.
The SDA process combines a spray dryer with a dust collector. Reagent addition, flue gas
humidification and some acid gas absorption takes place in the spray dryer. Additional acid gas
absorption and collection of the dry fly ash reaction products mixture takes place in the dust
collector. The SDA process is capable of achieving very high removal efficiencies for all acid
gases (99+%HCl, 95%S02, 99+%S03, 95%HF) as well as for the removal of trace metals and
organic compounds at stoichiometric ratios between 1 and about 1.8. Figure 3. is a simplified flow
diagram for the SDA process.
HEAD TANK
STACK
FLUE GAS
TEMP
OUST COLLECTOR
Figure 3. Spray Dryer Absorption Process.
(Courtesy of Niro Atomizer)
17
-------
Incinerator flue gas enters the spray dryer where it is contacted by a cloud of finely atomized
droplets of reagent (typically a hydrated lime slurry). The flue gas temperature is decreased and
the flue gas humidity is increased as the reagent slurry simultaneously reacts with acid gases
present and evaporates to dryness. In some systems, a portion of the dried product is removed
from the bottom of the spray dryer while in others it is carried over to the duct collector.
Collected reaction products are sometimes recycled to the feed system to reduce reagent
consumption.
Several different spray dryer design concepts have been employed for incinerator SDA
applications. These include; single rotary, multiple rotary and multiple dual fluid nozzle
atomization; down flow, up flow and up flow with a cyclone pre-collector spray dryers; and single
and multiple gas inlets. Flue gas retention times range from 10 to 18 seconds and flue gas
temperatures leaving the spray dryers range from 230°F up to 300°F.
Generally the lower the spray dryer outlet temperature, the more efficient the acid gas
absorption. The minimum reliable operating outlet temperature is a function of the spray dryer
and dust collector design and the composition of the dry fly ash reaction product mixture. The
spray dryer outlet temperature must be maintained high enough to ensure complete reagent
evaporation and the production of a free flowing product. Low outlet temperature operation
requires; efficient reagent atomization, good gas dispersion and mixing, adequate residence time
for drying and design of the dust collector to minimize heat loss and air in-leakage.
The dust collector downstream of the spray dryer may be an electrostatic precipitator, a
reverse-air baghouse or a pulse-jet type baghouse. The selection of a specific type of dust
collector is dependent on site specific factors such as particulate emission limits, overall acid gas
removal requirements and project economics. Each of these dust collection devices offers process
advantages and disadvantages that are evaluated on a site specific basis. Generally where high
acid gas control is required (95+%HCL, 85+%S02) a baghouse is utilized, as it is a better reactor
than an electrostatic precipitator.
Whether a fabric filter or ESP is selected as the dust collector, minimization of heat loss
from the dust collector to avoid corrosion and increased product stickiness is a prime design
consideration. Four methods employed to achieve this, are as follows:
° Insulation, to control heat loss
° Control of air in-leakage, to minimize cold spots
° Hopper heating, to maintain product temperature
° Operating procedures to maintain product flowability and minimize cold areas
The end product from the SDA process is a fine hygroscopic material with a significant
soluble fraction. The end product tends to be stickier than MSW fly ash and more difficult to
convey and store. Major end product constituents include:
° Fly Ash
° Calcium Hydroxide
O
o
Calcium Sulfite
Calcium Sulfate
18
-------
° Calcium Chloride ° Calcium Fluoride
° Calcium Carbonate ° Moisture
The calcium chloride formed at typical spray dryer outlet temperatures is a mixture of mono
and dehydrates (CaCl2H20 and CaCl22H20) and at lower temperatures will absorb moisture until it
reaches the hexahydrate form (CaCl26H20) and melts. Therefore, it is necessary to keep the
product from being exposed to cold and/or moist air. This is accomplished by proper design of
the product conveying and storage systems.
Wet Scrubbing
Wet scrubbing systems are capable of achieving high acid gas removal efficiencies and have
been applied to a large number of installations in Europe. Typical wet scrubbing applications
included two-stage scrubbers located downstream of an electrostatic precipitator. The first stage
is used for HC1 removal and the second for S02 removal. Water is used to capture the HC1 and
either caustic or hydrated lime is used for S02 capture. Figure 2., shows a typical two-stage wet
scrubber while Figure 4. shows a process flow diagram for an application of wet scrubbing with fly
ash treatment.
Electrostatic Scrubber
SO;
HCI
Scrubbing Water
Heavy Metals
Residual Salts
Waste-
water
Scrubbed
Fly Ash
Neutralization ¦ and
Precipitation Stage
Solids
Separation
Figure 4. Wet Scrubbing With Ash Treat,nent.
Courtesy of Lurgi, Inc.)
In this process, the HCI stream from the first scrubbing stage is pumped to a fly ash leaching
tank where it is used to leach out heavy metals from the fly ash collected in the dust collector.
After leaching, residual fly ash solids are either disposed of or used in construction applications.
The heavy metals bearing HCI stream is then treated alone or with the sodium sulfite/sulfate
19
-------
solution from the second scrubber stage in a neutralization/precipitation stage to concentrate the
heavy metals and produce salt-containing wastewater for disposal. When lime is used in the S02
absorption section of the scrubber, the calcium sulfite slurry can be oxidized to calcium sulfate
(Gypsum) for utilization.
Wet scrubbers offer some advantages:
° They are relatively inexpensive to install and require relatively small plot space;
° They are capable of achieving very high removal efficiencies for acid gases
(99+%HC1, 95+S02),
° They are capable of high removal efficiencies for many volatile trace compounds;
° They require the lowest reagent stoichiometrics (1.0 - 1.2) of any of the alternatives
considered.
Wet scrubbers also have some disadvantages:
° They produce a wet effluent which requires additional treatment with complex
effluent treatment systems, economics and space requirements are not as attractive
as the other alternatives.
° Wet scrubbers are more prone to corrosion problems and may require expensive
materials of construction.
° Historically, wet scrubbers have experienced more operating problems and higher
maintenance requirements that the alternatives.
HEAVY METALS CONTROLS
The primary heavy metals of concern from refuse incinerators (arsenic, cadmium, lead, and
mercury) are collected in the particulate control device or in the acid gas control system. The
major fraction of these metals exist as solid particulates at incinerator exit flue gas temperatures
and are collected as particulate matter. However, some arsenic, lead, and mercury compounds
exist in the vapor state at incinerator flue gas exit temperatures and these compounds must be
collected by condensation through cooling of the flue gas. This can be accomplished with either
an SDA or wet scrubbing process.
In the SDA process, the flue gas cooling takes place rapidly in a cloud of finely atomized
droplets. These droplets serve as sites for metals to condense or be absorbed onto. The
condensed metal is then removed with the reaction products in the downstream dust collector.
Collection efficiencies for arsenic and lead at typical SDA system operating temperatures are
greater than 90 percent.
A significant fraction of mercury remains in the vapor phase even at SDA system outlet
temperatures of 250°F. Addition of small amounts of powdered activated carbon or sodium
20
-------
sulfide upstream of the spray dryer have been used to enhance mercury control and greater than
90% capture has been achieved.
Wet scrubbers following a dust collector, operate at saturated flue gas temperatures (150° -
180°F) and can achieve greater than 90% removal of mercury. They can also remove a major
fraction of the other metals which may escape the particulate control device.
PCDD/PCDF CONTROL
PCDD/PCDF emissions are controlled by good combustion practices which inhibit their
formations and by particulate and acid gas controls. Combustion temperature above 1800°F for
greater than two seconds, are specified as a method of destroying PCDD/PCDF found in the
waste stream and their precursors formed from the combustion of other organic and chlorine
containing compounds. However, some PCDD/PCDF compounds may still form downstream of
the incinerator on the surface of fly ash at temperatures from 500° - 700°F.
Control of PCDD/PCDF compounds found in the flue gas leaving the incinerator is achieved
by electrostatic precipitators operating below 450°F or by acid gas control systems. Acid gas
control system achieve a higher PCDD/PCDF capture efficiency because of their reduced outlet
temperatures and the large droplet surface area available for adsorption to take place.
PCDD/PCDF capture efficiencies up to 99% can be achieved and total emissions can be reduced
to less than about 10 ng/nm3
CONCLUSIONS
Modern efficient incinerators are increasingly being used to simultaneously reduce municipal
solid wastes volumes while recovering energy in the form of steam or electricity. As the fraction
os MSW incinerated increases from today's 15% to about 25-30% in 1995, pressures will continue
to ensure that these incinerators environmental impacts are minimized.
EPA has recently established New Source Performance Standards and Emissions Guidelines
for existing facilities to reduce air pollution emissions. These regulations will require installation
of new or more advanced air pollution control systems to meet these standards.
Air pollution control systems available today have demonstrated the ability to reduce flue gas
emissions to below levels currently mandated. A wide range of manufacturers have experience
with guaranteeing the emissions levels required.
The work described in this paper was not funded by the U.S. Environmental
Agency and therefore the contents do not necessarily reflect the views of the
Agency and no official endorsement should be inferred.
21
-------
References
1C -1
(1) EPA Characterization of Municipal Solid Waste in the United States 1960-2000 (update 1988)"
Final Report, USEPA Office of Solid Waste and Emergency Responses, Franklin Associates Ltd.,
March 30, 1988
(2) EPA "Decision-Makers Guide To Solid Waste Management", USEPA Official Solid Waste and
Emergency Responses, EPA/530-500-89-072, November, 1989
(3) USEPA, New Source Performance Standards and Emissions Guidelines for Existing Municipal
Waste Combustors" CFR Part 60 Ca, Federal Register, Volume 56, No. 28, February 11, 1991
("} Joy Energy Systems, Inc. Incineration Brochures, Charlotte
(5) D.G. Jones, etal "Two Stage De NO. Process Test Data from Switzerland's Largest Incineration
Plant". International Conference on Municipal Waste Combustions, Hollywood, April 1989
(6) W. Panknins, "Research into Activated Carbon Technology on Harmful Organic Substances. Heavy
Metals and NO. Control". Air and Waste Management Association 83rd. Annual meeting, Pittsburgh,
June 1990
(7) B. Brown, J.R. Donnelly, T.D. Tarnok, R.J. Triscori, "Dust Collector Design Considerations For
NWS Acid Gas Cleaning Systems". EPA/EPRI 7th. Particulate Symposium, Nashville, March 1988
(8) J.R. Donnelly, "Design Considerations for MSW Incinerator APC Systems Retrofit". Air and
Waste Management Association 83rd. Annual meeting, Pittsburgh, June 1990
22
-------
TOXIC METAL EMISSIONS FROM MWCs AND THEIR CONTROL
Theodore G. Brna
U.S. Environmental Protection Agency
Office of Research and Development
Air and Energy Engineering Research Laboratory
Research Triangle Park, North Carolina 27711
ABSTRACT
Energy recovery in combustion facilities complements the volume reduction of waste for
land disposal, but it also produces air polluting emissions and concentrates some pollutants in
the ash or residue requiring disposal. Air pollutant emissions, such as acid gases, trace
organics, trace heavy metals, and particulate matter, can be controlled by flue gas cleaning
processes. These processes include both dry and wet scrubbing, with dry scrubbing being
preferred for municipal waste combustors in the United States and wet scrubbing being
extensively used in Europe.
Dry and wet acid gas scrubbing processes will be described and their effectiveness in
controlling metal and related pollutant emissions will be presented. The control of metal
emissions by these processes (particularly mercury by supplemental means) will be
emphasized, and the treatment of scrubber wastes prior to disposal will be noted.
This paper has been reviewed in accordance with the
U.S. Environmental Protection Agency's administrative
review policies and approved for presentation and publication.
23
-------
INTRODUCTION
Waste combustion reduces the volume of waste requiring disposal, can thermally destroy
toxic organic compounds present in the municipal solid waste (MSW) feed, and may be
complemented by thermal energy recovery. However, combustion generates air polluting
emissions, such as acid gases, trace organics [products of incomplete combustion including
organics and carbon monoxide (CO)], trace metals, and particulate matter (PM), which depend
on waste composition. Flue gas cleaning (FGC) processes to control these emissions transfer the
reaction products or the pollutants to liquids or solids which may need treatment to remove
components, such as heavy metals, or to stabilize the waste prior to its disposal or beneficial
use.
The objectives here are to discuss potential ways of controlling metal emissions
resulting from waste combustion. While fuel cleaning or materials separation may be a feasible
approach for reducing selected air pollutant emissions, the focus here will be on controlling air
pollutant emissions by FGC processes. Both wet and dry FGC processes will be discussed, but
disposal of process waste will be noted only briefly. Generally, special landfills are used for
waste requiring disposal while reuse is limited.
It should also be noted that Section 129 of the Clean Air Act Amendments of 1990
requires the U.S. Environmental Protection Agency to review and revise the promulgated
standards and guidelines for municipal waste combustors (MWCs) within 1 year.1 The Act
stipulates that numerical emission limits be set for cadmium (Cd), lead (Pb), and mercury
(Hg) by November 15, 1991. Thus, the control of these metals will be stressed. The standards
and guidelines for MWCs are also to be broadened to apply to units of 225 tonnes/day (250
tons/day) or less within 2 years.
METAL EMISSIONS
The combustion of wastes containing metallic compounds usually leads to their thermal
decomposition as temperatures above 700°C (1300°F) are present.2 For example, the heating
of compounds to over 700°C (1300°F) during combustion can lead to reduction processes such
as:
HgS + O2 --> Hg + SO2
HgO + C --> Hg + CO
where the mercuric sulfide (HgS) and mercuric oxide (HgO) in the waste feed react with oxygen
(O2) in combustion air and fuel carbon (C), respectively, to produce elemental Hg, sulfur
dioxide (SO2), and CO. On cooling of the combustion gases during energy recovery or prior to
FGC, some Hg would be oxidized as Hg readily combines with O2 at temperatures of 340 to 600°C
(650 to 1100°F) to form HgO. Similarly, the presence of chlorine (CI2) in the combustion
gases would result in mercuric chloride (HgCl2) being formed as the gases cooled.
Metallic vapors also condense as the flue gases cool downstream of the combustion
process. Condensation concentrates the volatile species on fine particles because of their high
24
-------
surface areas (on a mass basis relative to large particles). Thus, the submicrometer fraction
of the PM is enriched with condensible volatile species.3
Comparing data for MWCs having electrostatic precipitators (ESPs) for PM collection,
Damle et al.4 concluded that the metallic species are collected primarily in the fine
(submicrometer) PM fraction and that efficient collection of this fraction is essential to
controlling the emission of semi-volatile species. They also observed that fine particles are
collected less effectively than large particles in an ESP and that at stack conditions [-120 to
205°C (250 to 400°F)], the metallic species, except Hg, are mostly present on fine particles.
Figure 1 gives saturation curves for several metals and their compounds. Since the
sampling and analytical methods (e.g., EPA's multimetals train and pertinent analytical methods
for metals) determine the concentration of a metallic species regardless of its chemical form,
it is not possible to compare the saturated vapor concentrations for specific metal species with
their measured concentrations. Metal data, except for Hg, predominantly show over 90% of the
metals analyzed to be in the solid phase (i.e., in particulate form) despite the high volatility of
some of their chemical compounds. Elemental arsenic (As), Cd, Hg, and selenium (Se), as well
as the chlorides of As, Hg, Se, and zinc (Zn), may be completely in the vapor phase leaving the
combustor because of their relatively high volatilities.
In contrast to organic compounds, inorganic metal compounds have a low tendency for
adsorption on PM. The high proportion of the total metallic species found in the particulate
phase, based on flue gas sampling at 205°C (400°F) or lower, suggests that most metals were
primarily in their less volatile chemical forms. Mercury, however, even at stack
temperatures of 120 to 150°C (250 to 300°F), was in the vapor phase, since it was collected
predominantly in the back half of the sampling train (i.e., after the PM filter)4
POST- COMBUSTION EMISSION CONTROL PROCESSES
As discussed earlier, air pollutant emissions requiring control include acid gases
[mainly hydrogen chloride (HCI) and SO2], trace organic compounds [polychlorinated
dibenzodioxins and dibenzofurans (CDD/CDF)], trace metals (Cd, Hg, Pb,...), and PM. It has
also been suggested that high PM control provides effective emission control of metals of
concern, except Hg.5 However, Figure 2 suggests that Hg control is affected by inlet PM
concentration (at least for dry scrubbers), with the lower Hg concentrations corresponding to
higher PM concentrations entering dry scrubbing systems.6 It should be noted that the highest
inlet PM concentrations corresponded to refuse-derived fuel (RDF) units (Biddeford and Mid-
Connecticut) and the Quebec City mass burn unit (with a slipstream pilot spray dryer
absorber/fabric filter system), which were believed to have higher C in their flyashes than
other mass burn units represented in Figure 2. The use of acid gas control processes can
complement the control of metal emissions through improved PM control as well as lowering the
gas temperature which enhances condensation of some metal species (e.g., As, Hg, Se) and
adsorption of both condensed metals and organics on particles.
Both wet and dry acid gas control processes are used to remove HCI and SO2. Usually,
HCI control is more of a concern than SO2 when MSW is burned, but SO2 is harder to control
25
-------
than HCI although present at a significantly lower concentration than HCI. For small
incinerators, wet acid gas scrubbers may be more economical than dry scrubbers. Dry
scrubbers permit good acid gas and high PM control and may be less costly than wet scrubbers
for moderate-to-large incinerators with low-to-moderate acid gas concentrations.
A two-stage wet scrubber is shown in Figure 3. In the first stage, the water venturi
primarily effects PM control, accomplishes some HCI removal, and through gas cooling
promotes condensation of vaporized species. The second stage removes SO2 through the use of a
sorbent, such as sodium hydroxide (NaOH), and also removes additional HCI and PM. Except for
small MWCs, primary particulate collection is usually accomplished with an ESP preceding wet
scrubbing. Since the flue gas is saturated with water vapor at 40 to 60°C (105 to 140°F) on
leaving the scrubber, the flue gas may require heating to control corrosion of a metallic stack
and/or preclude a vapor plume at the stack exit. Heating of the cleaned flue gas may also be
required if additional gas treatment is needed after wet scrubbing, such as Hg removal via
absorption filtration and nitrogen oxides (NOx) control by selective catalytic reduction.
If the incinerator operates without energy recovery (i.e., without a boiler or heat
exchangers), the flue gas from the combustor would require cooling from about 985°C
(1800°F) to the acid gas scrubber temperature range. Flue gas cooling may be achieved using a
spray dryer (SD) upstream of a wet acid gas scrubber, a heat exchanger or quench tower ahead
of dry sorbent injection (DSI), or a spray dryer absorber (SDA). Both the DSI and SDA
processes would be followed by a PM collector.
A dry acid gas scrubber [DSI or SDA plus PM collector] differs from a wet one in two
ways: the waste discharged from a dry scrubber is a dry powder, not a liquid slurry, and the
clean flue gas leaving the dry scrubber is not saturated with water vapor. While both dry and
wet acid gas scrubbers require PM control components, PM removal follows acid gas
neutralization in dry scrubbing and precedes it in wet scrubbing.
In both forms of dry scrubbing [DSI or SDA followed by either an ESP or a fabric filter
(FF) for PM collection], calcium-based sorbents are generally used. With DSI, the sorbent,
usually calcium hydroxide [Ca(OH)2], is injected into flue gas which has been cooled to about
150°C (300°F) or less. The circulating bed (Figure 4) permits longer gas/sorbent contact
time for the acid gas neutralization reaction to occur than does sorbent injection into the duct
(Figure 5) where the gas velocity is higher than in the reactor. The solids entrained by the flue
gas are removed in the PM collector, usually a FF (baghouse) because it enables additional acid
gas removal by unreacted sorbent present in the dust cake on the bags.
While acid gas removais are 90 to 95% for HCI and 75 to 80% for SO2 with the DSI/FF
system, metals removal also is high because of the high collection of both total PM mass and fine
particles. If Hg control is not adequate because of insufficient Hg in a collectible form (i.e., Hg
species are not in the solid phase), it can be improved by injecting either sodium sulfide (Na2S)
or activated carbon (C) into the flue gas before Ca(OH)2 addition. The elemental Hg is converted
to HgS or adsorbed on the C particles, both of which are collected as PM. (As noted earlier, the
suspected higher C content of flyashes in the inlet PM concentrations shown in Figure 2 may be
partially responsible for the lower outlet Hg concentrations of the Biddeford, Mid-Connecticut,
26
-------
and Quebec City units.) Figures 4, 5, and 6 show injection locations for these additives for
supplementing Hg control.
The SDA/PM collection system, also referred to as a semi-dry scrubber, is shown in
Figure 6. In this system, a Ca(OH)2 slurry is atomized into the surrounding dirty flue gas. As
the gas cools to the 120 to 150°C (250 to 300°F) range, it evaporates the finely atomized
droplets containing sorbent which reacts with the acid gases. The dry powdery solids formed
contain calcium salts, flyash enriched with metals, and unreacted sorbent. Some of these solids
may be collected in the bottom of the SDA, but most are entrained by the flue gas to the PM
collector (either an ESP or a FF) for removal from the gas. HCI removals exceeding 95% and
SO2 removals over 90% can be attained with lime SDA/PM collection systems. However, dry
systems require higher reagent-to-acid-gas ratios than do wet systems for comparable acid gas
control when attainable.
Activated powdered C is also being applied at several incineration facilities in Europe
primarily to improve CDD/CDF control, but the injected C is also enhancing Hg capture.
Incineration plants in Germany are also using carbon beds after wet scrubbing to improve both
CDD/CDF and Hg control in order to comply with tighter air emission regulations.
Another supplemental Hg control process uses a selenium (Se) filter as shown in Figure
7. This process, which was developed for the metal smelter industry in Sweden, is being
applied to a MWC in Sweden after reheating the gas from a wet acid gas scrubber and just ahead
of the stack.7 The Hg/Se reaction in the filter leads to the retention of the solid-phase product
in the filter. Sizing of the filter permits effective operation for the desired period (~5 years)
before its replacement. Operating limitations include an unsaturated (dry) flue gas at 60°C
(140°F) or less and low particulate loading [preferably below 10 mg/Nm3 (0.005 gr/dscf)] to
avoid premature cleaning of the porous Se-impregnated filter material. Hg concentrations of
about 10 mg/Nm3 (referenced to 10% CO2) at the outlet of the filter were reported.7
In Japan, Hg removals of 30 to 70% by a wet scrubber have led to several approaches
for enhancing Hg capture.8 One method used with caustic-soda-based acid gas scrubbing injects
a liquid chelating agent and cupric chloride into flue gas to absorb atomic Hg in the flue gas.
Atomic Hg removal ranging from 65 to 85% over the increasing pH range of 2 to 12 in the
absorbing solution was reported when the additive was injected into 600°C (1100°F) gas with
1500 ppm HCI. A second method involves sodium hypochlorite addition. This additive reacts
with Hg to form HgCl2 which is soluble in water. Mercury removals of 90 to 95% were
achieved with this method. As reported earlier, this additive had been proposed for NOx control
by wet scrubbing.s To achieve the concurrent removal of both Hg and NOx, bromine has been
added to reduce NOx by 30 to 50%.
EFFLUENT/RESIDUE TREATMENT AND DISPOSAL
The wet acid gas scrubber generates a liquid effluent which may require treatment to
remove metals prior to discharging blowdown to a sanitary sewer or water body, if permitted.
If liquid discharges are not permitted or treatment costs are high, zero effluent discharge can be
27
-------
achieved by a system such as shown in Figure 8. Hot flue gas enters the SD which receives
blowdown after treatment to remove metals by precipitation and pH adjustment. While the SD
functions primarily to evaporate the treated blowdown, it may also remove some acid gases. The
cooled flue gas passes to a PM collector, such as an ESP, before entering the two-stage scrubber.
If high PM and high fine particle (including aerosol) control is required, an ionizing wet
scrubber can supplement wet acid gas scrubbing. If needed, the water-saturated flue gas can be
reheated after leaving the ionizing wet scrubber and before entering the stack as unsaturated
flue gas to address corrosion or vapor plume problems. This hybrid wet/dry scrubber enables
high multipollutant control and discharges dry solids. The dry waste may require treatment to
remove leachable metals, such as Cd and Pb, before the remaining waste is landfilled or reused
where permitted. The concentrated metals may require disposal as hazardous waste.
Wet scrubber effluent treatment processes are used to concentrate and remove metal,
such as Pb and Cd to preclude their leaching in landfills, prior to their disposal as hazardous
wastes in Europe. The use of C beds following wet scrubbing to remove Hg and/or organic
compounds or of the Se filter for Hg removal produces wastes which may need treatment or
disposal of hazardous wastes.
Dry scrubber wastes are frequently landfilled as special wastes in the U.S., with the
excess lime and pozzolanic character of the ash/residue aiding its stability. Monitoring of
landfill leachate is also a usual regulatory requirement.
RESULTS
While Hg control information for the Se filter in Sweden and with wet acid gas scrubbing
in Japan were mentioned earlier, test results, including limited data for Cd and Pb control, are
presented here. High metal (except Hg) removal (>95%) has generally been achieved with both
SDA/ESP and SDA/FF systems. The SDA/FF system has shown higher Hg capture (>90%) than
the SDA/ESP system (35-70%), probably because of better fine particle removal. Tables 1,
2, and 3 show Cd, Hg, and Pb control results for dry acid gas scrubbers with and without
supplemental Hg control processes.
Table 1 suggests good Cd and Pb control because of their low outlet concentrations and
low PM emission, except for Unit 2, Dutchess County. Hg control for St. Croix appears to be
high based on its outlet concentration of 35 ng/dscm and expected inlet concentrations for
MWCs. Hg control at Springfield appears to be slight and nil for Unit 1 at Dutchess County, but
good for Unit 2 at Dutchess County although it had a higher PM emission than Unit 1.
The metals data in Table 2 and 3 show very high removals of Cd and Pb, generally above
98% when dry scrubbers with either FFs or ESPs are used. Vory high Hg control (>95%) has
also been obtained with SDA/FF systems on RDF MWCs. Less effective Hg control has been seen
with dry scrubbers on mass-burn units, possibly because of the apparent lower C in flyash
from these combustors than in RDF units. Although no test data were found to support this
theory, the effectiveness of C injection for improving Hg control has been demonstrated (see
Table 3) and provides qualitative support for this theory. Very high Hg control in the
February, 1989, Mid-Connecticut tests corresponded to loss-on-ignition (LOI) FF residue
values as low as 4%. Assuming that water of hydration and C account for the LOI, these values
28
-------
ranged from 4 to 10% for the Mid-Connecticut tests, while the Hg removal was over 95% for
all 13 runs. Thus, the amount of C needed for high Hg control, if it improves Hg control, is
small. This is consistent with the results of the activated C addition tests (Table 3) which
required small addition rates to improve Hg removal significantly.
Table 2 indicates close agreement between PM removal and Cd and Pb capture for both
ESPs and FFs and less effective Hg control with ESPs than with FFs. The Millbury outlet Hg data
suggest little or no control (assuming atypical Hg inlet concentration between 500 and 1000
ng/Nm3), while the SEMASS (which has five-field ESPs while Millbury has three-field ESPs)
data imply at least moderate control.
Table 3 indicates that both Na2S and C injection result in high Hg capture by dry
scrubbers. Removals exceeding 85% seem feasible when these additives are used.
Table 4 shows Hg removal data for plants with wet scrubbers in Europe. It indicates that
Hg captures of 85% or more were achieved with wet scrubbers. The results from Japan noted
earlier indicated similar Hg removals when additives were used with wet scrubbers.
SUMMARY
Effective PM control is essential to control emissions of toxic metals from MWCs. High
PM removal generally implies good metals control, with Hg being a possible exception.
Additives to flue gas (Na2S solution or activated powdered C) ahead of dry scrubbers improve Hg
control, while C beds following wet scrubbing reduce air pollutant emissions to low levels. A Se
filter for removing Hg from conventionally wet scrubbed flue gas followed by gas reheating
before filtration has been proposed for MWC applications, but no performance data have yet
been reported. Additives to enhance Hg control in wet scrubbers, such as cupric chloride with a
chelating agent and sodium hypochlorite, are being successfully used in Japan. For high
multipollutant emissions control, wet scrubbing is favored on MSW incinerators in Japan and
Europe, while the semi-dry scrubber predominates in the U.S.
REFERENCES
1. FACT SHEET: New Municipal Waste Combustors and FACT SHEET: Existing Municipal
Waste Combustors, U.S. Environmental Protection Agency, Office of Air Quality Planning
and Standards, Research Triangle Park, NC, January 11, 1991.
2. Lindqvist, Oliver, Waste Management and Research, 4:35, 1986.
3. Vogg, H., H. Braun, M. Metzger, and J. Schneider, Waste Management rind Research,
4:65, 1986.
4. Damle, Ashok S., David S. Ensor, and Norman Plaks, "Condensible Emissions from
Municipal Waste Incinerators," Presented at Eighth Symposium on the Transfer and
Utilization of Particulate Control Technology, San Diego, CA, March 20-23, 1990.
5. Air Pollution Standards of Performance for New Stationary Sources; Rule and Proposed
Rules (40CFR Parts 60, 51, and 52), Federal Register, Vol. 54, No. 243, Wednesday,
29
-------
December 20, 1989, pp. 52188-52304.
6. Municipal Waste Combustors - Background Information for Proposed Standards: Post
Combustion Technology Performance, EPA-450/3-89-027c, U. S. Environmental
Protection Agency, Office of Air Quality Planning and Standards, Research Triangle Park,
NC, August 1989.
7. Solbu, E., "Selenium Filter for Mercury Removal from Gases," unpublished paper
(selinfo, 530, May 30, 1990), Boliden Contech AB, S-932 Skelleftehamn, Sweden.
8. Nakazato, Kunihiro, "Latest Technological Experience of the Removal of Mercury in Flue
Gas and the Management of Flyash from MSW Incinerator," In Proceedings, 1990
National Waste Processing Conference, Long Beach, CA, June 3-6, 1990, American
Society of Mechanical Engineers, New York, NY, pp. 163-169.
9. Sedman, C.B. and T. G. Brna, Municipal Waste Combustion Study: Flue Gas Cleaning
Technology, EPA/530-SW-87-021d (NTIS PB87-206108), U. S. Environmental
Protection Agency, Office of Solid Waste, Washington, DC, June 1987.
10. McClanahan, D., A. Licata, and J. Buschmann, "Operating Experience with Three APC
Designs on Municipal Incinerators," Presented at International Conference on Municipal
Waste Combustion, Hollywood, FL, April 11-14, 1989.
11. Brna, Theodore G. and James D. Kilgroe, Journal of Air & Waste Management Association,
4 0 (9): 1 324,1 990.
12. SEMASS Waste-to-Energy Resource Recovery Facility: Compliance Test Report.
Prepared by Eastmount Engineering, Inc., Walpole, MA, for SEMASS Partnership,
Rochester, MA, August 1989.
13. Docket No. A-89-08, Item IV-M-20, U.S. Environmental Protection Agency,
Washington, DC, October 15, 1990.
14. City of Zurich, Switzerland (Waste Disposal Department), "For a Clean Zurich," undated
brochure reporting October 1986 test results for Josefstrasse plant.
30
-------
Temperature,°F
200 400 600 800 1000 1200 1400
104
m
E
Cr>
e
103
V.
c
0
-H
+J
(0
*-l
-P
C
102
O
c
0
0
ZnCl2
(AS2O3) 2
PbCl2
CdCl2
N1CI7
0.5-1 mg/m3
Hg in flue
gas from MW
(uncontrolled)
200
800
400 600
Temperature, °C
Figure 1. Saturation curves for selected metals and compounds
31
-------
1 000
Lo
K3
CT
s?
N
d
£
o
V)
¦o
X
cr>
3
C
o
c
ID
O
c
o
o
cr>
X
3
o
900 -
800
700 -
600 -
500
400 -
300 -
200
100
~
* Mid-Connecticut
+ Quebec City (Slipstream Scrubber)
° Biddeford
A Commerce (1988)
x Commerce (1987)
~ Long Beach
r Marion County
•Individual runs during same test period at Commerce.
0.8
1.2
1.6
2.4
2.8
3.2
3.6
Inlet PM Concentration (gr/dscf
-------
SUPPLEMENTAL
Hg CONTROL OPTION:
SELENIUM
FILTER
CLEANED
FLUE GAS
FLUE GAS
HEATER
WATER
VENTURI
SCRUBBER
MIST
ELIMINATOR
COOLING
ABSORPTION
X ZONE
STACK
PUMP
HYDROCLONE
CAUSTIC
SODA
WASTEWATER
PUMP
INDUCED
DRAFT FAN
TO WASTEWATER
TREATMENT PLANT
Figure 3. Wet flue gas scrubbing system.
1. LIME SILO
2. REACTOR
3. CYCLONE
4. PARTICULATE COLLECTOR
5. STACK
SUPPLEMENTAL Hg 6. WASTE SILO
CONTROL OPTION: Na2S SOLUTION
CLEANED FLUE CAS*
LIME
AIR"0)
WATER
(optional)
RECYCLE
(optional)
AIR £7
FLUE GAS
DRY WASTE
Figure 4. Dry sorbent injection into fluid bed reactor.
33
-------
1. LIME SILO
2. REACTOR
3. CYCLONE
4. PARTICULATE COLLECTOR
5. STACK CLEANED
FLUE GAS
6. WASTE SILO
LIME
AIR -0^
FLUE GAS
SUPPLEMENTAL Hg
CONTROL OPTION:
Na2S SOLUTION
DRY WASTE
Figure 5. Dry sorbent injection process.
SUPPLEMENTAL Hg
CONTROL OPTION:
CARBON POWDER-
cezzd
CLEANED FLUE GAS
1. LIME FEEDER
2. UMESLAKER
3. LIME SLURRY
FEEDTANK
4. SPRAY DRYER ABSORBER
5. PARTICULATE COLLECTOR
6. STACK
L_X
PARTICLE RECYCLE
(optional)
DRY WASTE
Figure 6. Spray dryer absorption (semi-dry) process.
34
-------
HgSe
INCOMING GAS
Figure 7. Selenium filter.7
1. FLUE GAS
2. EXHAUST GAS
3. SPRAY DRYER
4. ELECTROSTATIC PRECIPITATOR
OR FABRIC FILTER
5. GAS-GAS HEAT EXCHANGER
6. VENTURI SCRUBBER
7. NEUTRALIZATION TANK
. SLUDGE TANK
UMESILO
UME SLAKER
SODIUM HYDROXIDE STORAGE
SODIUM HYDROXIDE FEED TANK
DRY WASTE
WATER
DRY AIR
Figure 8. Wet scrubbing process with zero effluent discharge.
-------
TABLE 1. CONTROL OF PARTICULATE MATTER (PM)
AND SELECTED HEAVY METALS WITH DRY LIME
INJECTION/FABRIC FILTER SYSTEMS
Average PM
Concentration3
gr/dscf @ 12% CO?
Average Concentration, ug/dscm
@ 7% O?
Reference
Location and Test Date
outlet
cadmium (Cd)
outlet
lead (Pb)
outlet
mercury (Hg)
outlet
•St. Croix, WI
6/88
0.015b
2.3
18
35
6
•Springfield, MAC
7/88
0.0016
1
21
300
10
•Dutchess County, NY
Unit 1,2/89
Unit 2,2/89
0.0097
0.035
2.72
3.03
38.9
49.1
1080
84.7
6
aMultiply gr/dscf by 2288 to obtain mg/dscm.
''Not measured simultaneously with metal, but is value from 5/88 tests.
CA11 concentrations are referenced to dry gas with 12% CO2. Inlet PM concentration = 0.090 gr/dscf.
-------
TABLE 2. CONTROL OF PARTICULATE MATTER (PM) AND SELECTED HEAVY METALS
WITH LIME SPRAY DRYER ABSORBER (SDA)/FABRIC FILTER (FF) OR
ELECTROSTATIC PRECIPITATOR (ESP) SYSTEMS
Average PM
Concentration3
gr/dscf @ 12% CO?
Average Concentration
Hg/dscm @ 7% O2
cadmium (Cd) lead (Pb) mercury (Hg)
Removal, %
Refs.
Location, Control
System, and Test Date
outlet
inlet
outlet
inlet
outlet
inlet
outlet
PM
Cd
Pb
• Marion County, OR
Unit 1, SDA/FF
9/86
0.0023
1,121
2.6
20,500
19
NAb
239
99.7
98.1
99.9
NA
6
•Biddeford, ME
Unit A, SDA/FF
12/87
0.014
1,114
NDC
27,352
159
389
ND
995
NA
99.4
100
6
•Mid-Conn.^
Unit 11, SDA/FF
7/88
2/89e
0.0040
0.0018
1068
595
NA
ND
37,386
7,513
ND
52
884
646
50
8.7
99.8
99.9
NA
100
100
99.9
94.3
98.7
6
11
•Millbury, MA
Unit 1, SDA/ESP
2/88
Unit 2, SDA/ESP
2/88
0.0018
0.0083
NA
NA
17.7
215
NA
NA
278
330
NA
NA
565f
954
NA
NA
NA
NA
NA
NA
NA
NA
6
• SEMASS
Unit 1, SDA/ESP
3/89
Unit 2, SDA/ESP
3/89
0.008
0.012
NA
NA
9.53
6.74
NA
NA
299
234
NA
NA
59.1
105
99.8
99.6
NA
NA
NA
NA
NA
NA
12
aMultiply gr/dscf by 2288 to obtain mg/dscm.
^Not available or not measured.
cNot detected.
^All concentrations are for dry gas referenced to 12% COj.
eValues shown are averages for normal SDA/FF temperature
operation (performance tests 6,8,12,13, and 14).
'From May 1988 test.
-------
TABLE 3. REMOVAL OF SELECTED METALS USING
ADDITIVES WITH DRY SCRUBBERS
Additive: Na2 S Solution Before Dry Ca(OH>2 Injection
Average Outlet Concentration, HR/Nnr
Reduction, %
Reference
Cadmium
Lead
Mercury
Location
Date
Unit
(Cd)
(Pb)
(Hr)
Cd
R>
Hr
Burnaby Plant3
1989-90
1
NAb
NA
117(3)c
NA
NA
87^
13
Vancouver, British
1989-90
2
NA
NA
127(3)
NA
NA
86d
Columbia, Canada
1989-90
3
NA
NA
155(2)
NA
NA
83d
Hogdalen Plante
1986
1
NA
NA
42(2)
NA
NA
89
13
Bandhagen, Sweden
1987
1
NA
NA
26(3)
NA
NA
88
1988
1
NA
NA
16(3)
NA
NA
89
1989
1
NA
NA
4(1)
NA
NA
98
Additive: Powdered Activated Carbon Before Lime Spray Dryer
Average Outlet Concentration, ug/Nm^
Reduction, %
Reference
Location
Date
Unit
Cadmium
(Cd)
Lead
(Pb)
Mercury
(Hr)
Cd
ft
Hr
Josefstrasse Plant'
Zurich, Switzerland
1986
1
14
7008
42
99
99
91
14
(SDA/ESP)
Range of Data*1
With Additive
Without Additive
NA
NA
1
1
NA
NA
NA
NA
20-50
100-250
NA
NA
NA
NA
85-93
20-70
13
13
Amagerforbraending I
Amager, Denmark
SDA/FF
13
Range of Data'
With Additive
Without Additive
NA
NA
NA
NA
NA
NA
NA
NA
15-25
<100
NA
NA
NA
NA
90-95
30-85
a Concentrations referenced to 7% C>2. e Concentrations referenced to 10% CO2.
k Not available or not measured. ' Concentrations referenced to 11 % 02
c- Number of test runs shown in parentheses. 8 Value for lead + zinc.
d Inlet Hg concentration estimated at 900 ng/Nm^. ^ Concentrations believed referenced to 11% O2.
-------
TABLE 4. MERCURY REMOVAL WITH WET SCRUBBERS PRECEDED BY ESPs13
Average Hg Concentration, jig/dscm @ 7% 02
Removal, %
Location
Unit
inlet
outlet
Lyon-North Plant
1
370(3)a
<53b(3)
>85
Lyon, France
2
177(3)
66(3)
63
Lyon-South Plant
1
550(2)
<65(2)
>88
Lyon, France
2
435(2)
<64(2)
>69
Basel Plant
1
326(5)
19(5)
94
Basel, Switzerland
2
222(1)
<19c(ll)
>91
a Number of test runs shown in parentheses.
b Detection limit of 53 jig/dscm for the Lyon plants.
c Detection limit of 14 fig/dscm for the Basel plant.
-------
Intentionally Blank Page
40
-------
The work described in this paper was not funded by the U.S. Environmental
Protection Agency. The contents do not necessarily reilect the views of the
Agency and no official endorsement should be inferred.
Determination of efficiency of flue gas cleaning systems on Municipal Solid
Waste Incinerators in Denmark.
Peter Blinksbjerg, Chem. Ing.
dk-TEKNIK
Danish Boilers Owners' Association
Gladsaxe M0llevej 15
DK-2860 Soborg
Abstract.
Flue gas cleaning equipment on waste incinerators to removal of
acid gases, is a relative new technology in Denmark. This paper
summarizes the investigations made on four different kind of
equipment.
It is found that in general the plants are operating with accep-
table removal efficiencies for acid gases, heavy metals and
dioxins. To observe the regulation limits set up for hydrogen-
chloride, some of the plants seems to have a rather high lime
consumption.
Introduction.
A guideline from the Danish Environmental Protection Agency is-
sued in 1986 requires a reduction of the emissions from munici-
pal solid waste (MSW) incineration. The limit set for the flue
gas concentration of hydrogenchloride (HCl), is at such a
strict level, that most of the incinerators in Denmark have to
be equipped with flue gas cleaning equipment to remove acid
gases.
This type of equipment has not previously been used in Denmark
and therefore we have carried out a number of investigations on
the first 4 flue gas cleaning systems installed at Municipal
Solid Waste incinerators in Denmark. The idea of the investiga-
tions was primarily to examine the acid gas removal efficiency
when the flue gas cleaning systems were run under normal con-
ditions .
Preceding page blank
-------
In Denmark only a small amount (less than 10%) of the produced
waste are landfilled, some waste are recycled or retrofitted,
and most of the waste are burned in massburners. Therefore all
MSW-incinerators in these investigations are mass-burners. The
MSW-incinerators, that participated in the investigations were
equipped with flue gas cleaning systems as described below.
Plant No. 1, Refa: Dry injection of lime in a simple reactor
followed by an electrostatic precipitator (ESP).
MWC capacity: 3.5 t/h. Manuel regulation of lime consumption.
Plant No. 2, Kara: Injection of a limeslurry resulting in a dry
recidue (semi-dry process) followed by an ESP. Under normal
conditions this plant are run with a certain amount of recyc-
ling of the flue gas cleaning residue to minimize the lime con-
sumption.
MWC capacity: 7.0 t/h. Regulated by the outlet SO2- and
HCl-concentrations.
Plant No. 3, Nordforbraending: Dry injection of lime in a reac-
tor followed by a fabric filter (FF).
MWC capacity: 3.0 t/h. Manuel regulation of lime consumption.
Plant No. 4, Amagerforbraending: Injection of limeslurry in a
Spray Dryer Adsorber followed by a FF.
MWC capacity: 12.0 t/h. Regulated by the HCl-outlet concentra-
tion.
Test programme:
It has to be noted that different measuring programmes have
been carried out at each plant which means that e.g. the con-
ditions the plants were run under during sampling differs from
plant to plant.
Plant No. l: Acid gases (HCl, SO2) and dioxins (PCDD + PCDF).
Plant No. 2: Acid gases (HCl, SO2), dioxins and heavy metals
(Hg, Pb and Cd).
Plant No. 3: Acid gases (HCl, SO2) and heavy metals (Hg, Pb and
Cd)
Plant No. 4: Acid gases (HCl, SO2), dioxins and mercury.
Furthermore, samples have been extracted and analysed at plant
No. 2 with the primary aime of establishing a total massbalance
for heavy metals over the flue gas cleaning system.
42
-------
RESULTS.
In the following, the test results are listed. Furthermore the
emission values are compared to the Danish limits, set for flue
gas concentrations for MWC's.
Acid gases.
Table 1 shows the result from the measurements for acid gases.
It is seen that the regulation limits for emission of HCl and
SO2, set to 100 resp. 300 mg/m3(s,d), is not exceeded in any
case.
The plants with the largest lime consumption are plant No. 1
and 3, which are equipped with the most simple flue gas clea-
ning systems. At the same time they are the smallest, and the-
refore a big lime consumption is not as critical as for the
bigger plants, even though a stochiometric ratio (SR) of 3-4
seems rather high.
Heavy metals.
Table 2 shows the result from measurements for lead, cadmium
and mercury. The good removal efficiency (better than 90%) of
lead and cadmium is due to a good removal efficiency of parti-
culate. Analysis of the inlet measurements showed that most of
the lead and cadmium are carriied by the particulate.
The tests at plant No. 4 were performed with the aime to opti-
mize the mercury absorbtion. It is shown that use of activated
carbon as additive had a positive effect on mercury removal.
Increasing amount of activated carbon resulted in decrease of
mercury emission. A decrease in outlet temperature resulted in
lower mercury emission both with and without additive.
It is seen that the regulation limits for emission of lead,
cadmium and mercury, set to 1.4, 0.1 resp. 0.1 mg/m3(s,d) is
only exceeded in one case, it is the mercury emission from
plant No. 2. It seems to be due to the poorest removal effi-
ciency .
Dioxln (Total Dioxins and Furans).
The removal efficiens were in general good, e.g. plant No. 1:
85% removal, plant No. 4: 98% removal and Plant No. 3: no
tests.
At plant No. 2 there were found a poorer efficiency (6 5% remo-
val). The only condition that differs this plant from the
others are the amount of recycling of the flue gas cleaning
residue. This has not to be regarded as a proof for the
poorer efficiency.
43
-------
As well as for mercury, at plant No. 4, there were performed
tests with the aime to optimize the dioxin absorbtion. It is
seen that an outlet temperature at 127°C the outlet values
seems to be lower than comparable tests with an outlet tempera-
ture at 140°C. The highest removal efficiencies were obtained
with addition of additive (activated carbon), no significant
influence of the amount added could be established.
Massbalance.
At plant No. 2 samples have been extracted in such way that it
has been possible to establish massbalances for lead, cadmium
and mercury. All massflow are determined by flue gas measure-
ments and calculations, because of that it should be acceptable
with a difference in the massbalance within 25%.
The massbalances are shown in table 3. The balances for cadmium
and mercury are established within the acceptable difference.
The reason why the difference in the massbalance for lead is as
big as 88% has not been identified. It is important to notice
that the 3 balances has been established with a negative dif-
ference, which might indicate a systematic but not identified
error.
Conclusions.
It is found that the examined plants, which are the first of
their kind in Denmark, in general are operating with acceptable
removal efficiencies for acid gases, heavy metals and dioxins.
To observe the regulation limits set for different emissions,
some of the plants seems to have a rather high lime consump-
tion. In general it is seen that the limits is not exceeded.
We have succeeded in establishing acceptable meassbalances for
cadmium and mercury at one of the plants. Of some not identi-
fied reasons the massbalance for lead is not acceptable.
References.
Kirsten Kragh Nielsen: Optimization of dioxin removal by semi-
dry flue gas cleaning on full scale incinerator. A/S Niro
Atimozer, Soborg, 1989.
Milj0styrelsen: Dioxinemission from waste combustion (In
Danish, English summary available), 1989. (Some of the results
were presented on DIOXIN '89, Toronto).
44
-------
Table 1: Values for efficiency, emission concentration
and lime consumption (Stochiometric Ratio)
for acid gases.
Plant
No.
Efficiency
%
S02 HC1
Emission
mg/m3(s,d)
S02 HCl
S.R
1
83
—
50
3-4
2
80
96
40
22
0.9
3
47
90
68
26
~3
4
69
96
170
35
1.5
Table 2: Values for efficiency and emission concentration
for heavy metals (Lead, Cadmium and Mercury).
Plant
No.
Efficiency
%
Emission
mq/ms(s,d)
Pb
Cd
ng
Pb
Cd
Hg
1
2
92
90
54
0.9
0.078
0.054
3
>99
>90
44
<0.8
<0.022
0.13
4
-
-
66*
-
-
0.066*
*: Some tests were performed with an additive to the
lime.
Table 3: Massbalance for Lead, Cadmium and Mercury.
Performed at plant No. 2. Unit: g/h.
Pb
Cd
Hg
Inlet
Flue gas
464
33
5.0
Lime
1
~0
~0
Total inlet
465
33
5.0
Outlet
Flue gas
40
2.6
2.1
Residue (from ESP)
836
37
3.4
Total outlet
876
40
5.5
Difference*
-88%
-21%
-10%
*: Difference = (Inlet + Outlet)/Inlet
45
-------
46
-------
The work described in this paper was not funded by the U.S. Environmental
Protection Agency. The contents do not necessarily reflect the views of the
Agency and no official endorsement should be inferred.
TWO-AND-A-HALF YEARS OPERATING EXPERIENCE
AT THE
WARREN COUNTY ENERGY RESOURCE RECOVERY FACILITY
By
Claus Jorgensen and William B. VanHooser
Environmental Elements Corporation
Baltimore, Maryland 21227,
Janine G. Kelly
Blount Energy Resource Corporation
Montgomery, Alabama 36116,
and
William B. Cook
Warren Energy Resource Company
Oxford, New Jersey 07863
ABSTRACT
The Blount Energy Resource Corporation's Municipal Waste Combustion Facility located in Warren County,
New Jersey, consists of two 200 ton/day boiler units equipped with a spray dryer/baghouse system for acid
gas and particulate control. This state-of-the-art resource recovery facility was the first of its kind to be built
and started-up in New Jersey, and it has presently been operating for about two-and-a-half years. Design
features and operational history of the air pollution control system supplied by Environmental Elements
Corporation is described in this paper. Following a successful start-up of the facility the initial stack
compliance and flue gas cleaning system performance tests for hydrogen chloride, hydrogen fluoride, sulfur
dioxide, and particulate were conducted in November 1988. The test demonstrated that the system passed
all emission and performance requirements. Two yearly compliance tests have been performed since this
initial testing, demonstrating continued compliance of the system in terms of acid gas and particulate
emissions. Also, a series of metals stack measurements have recently been performed aiming at evaluating
the potential emissions of mercury and other metals from the facility. The test results are presented and
discussed.
Preceding page blank
47
-------
INTRODUCTION
The present paper describes the air pollution control (APC) system installed at the waste-to-energy
facility located in Warren County, NJ. This state-of-the-art facility was the first of its kind to be built and
started-up in New Jersey. The facility was also the first waste-to-energy plant to be awarded to Blount
Energy Research Corporation (Montgomery, AL) in the United States. The air pollution control system
comprising lime spray dryer and pulse-jet baghouse was awarded by Blount to Environmental Elements
Corporation (EEC). This was EEC's first municipal solid waste (MSW) contract with total responsibility for
acid gas and particulate control. EEC contracted with Komline-Sanderson Engineering Corporation,
Peapack, NJ to supply the rotary atomization and spray drying technology essential for achieving acid gas
control. Total process responsibility resided with EEC. Although EEC had experience for operating MSW
facilities using electrostatic precipitators for particulate control, this was for EEC the first opportunity to use
its established pulse-jet fabric filtration technology on this particular application. The waste-to-energy facility
has now been operated by Blount Energy Resource Company for more than 2V2 years and the APC system
has shown very satisfactory performance. The design and the operating experience of the APC system is
described in the following.
WASTE-TO-ENERGY FACILITY DESIGN
COMBUSTOR/BOILER
The Warren County waste-to-energy facility consists of two (2) 200-ton/day boiler trains equipped with
Widmer+Ernst horizontal grates. Each boiler is sized for the production of 56,000 lb/h of steam which is
used for the generation of 13 MW electricity. The facility uses 2 MW to support its own operation.
AIR POLLUTION CONTROL SYSTEM DESIGN CRITERIA
Each boiler is equipped with an APC system consisting of a lime spray dryer for acid gas removal
followed by a pulse-jet baghouse for particulate control. A flow diagram of the APC system is shown in
Figure 1. It is sized for a flue gas flow at maximum continuous rating (MCR) of 51,000 acfm (320°F), or
153,000 lb/h. The APC system is designed for boiler exit flue gas temperature excursions up to 500°F and
will maintain stack temperatures at 265-285°F under all boiler operating conditions.
The expected acid gas concentrations in the flue gas exiting the boiler are 500-900 ppm hydrogen
chloride (HC1), 40-50 ppm hydrogen fluoride (HF), and 150-450 ppm sulfur dioxide (S02) (all dry volume
corrected to 7% 02). The required removal performance of the APC systems is for HC1 95%, or a stack
concentration of 25-45 ppm , whichever is less stringent; 95% HF removal; and 90% removal of S02, or a
stack concentration of 40 ppm, whichever is less stringent. The required maximum particulate stack emission
is 0.01 gr/dscf (7% Oj).
In addition to these emission guarantees three process performance guarantees were also provided by
EEC. The lime consumption guarantee for the above specified conditions was typically in the range 295-480
lb/h (total for two boiler trains) corresponding to an equivalence ratio (ER) of 1.4.
The equivalence ratio, is here defined as
2 x (lb-mol/h of CaO)
ER =
lb-mol/h of HC1 + lb-mol/h of HF + 2 x (lb-mol/h of S02)
48
-------
A total APC system power consumption guarantee of about 85 kW, and a pressure drop guarantee of
10.0 in.WG were also offered.
As to specific equipment warranties it should be mentioned that a 24-month useful bag life guarantee
and a 6 to 9-month useful life guarantee of the atomizer inserts (protecting against erosion by the abrasive
slurry) were provided.
AIR POLLUTION CONTROL SYSTEM DESIGN FEATURES
The APC system can for clarity be divided into three major components, namely the spray dryer absorber
(SDA) (including the rotary atomizer), the pulse-jet filter (PJF), and the lime storage and slurry preparation
system. The SDA's and the rotary atomizers were designed and fabricated by Komline-Sanderson
Engineering Corporation in Peapack, NJ. The PJF's (including associated ductwork, access, etc.) were
designed and fabricated by EEC in Baltimore, MD. The common lime storage and slurry preparation
equipment was supplied as a complete package by Chemco Equipment company in Pittsburgh, PA.
SPRAY DRYER ABSORBER
The spray dryer vessels are designed as downflow, single-rotary-atomizer units with a 22-ft diameter and
a 28-ft cylindrical height section equipped with a 60° bottom cone. A schematic of the SDA is shown in
Figure 2.
The hot flue gas exiting the boiler is ducted to the roof of the SDA and turned vertically downward
along the centerline of the vessel, surrounding the approximately 18-inch diameter central tube enclosing the
rotary atomizer assembly. Swirl vanes located immediately above the atomizer provide for some rotation of
the flue gas as it exits this annular gas disperser and contacts the cloud of finely atomized slurry droplets.
A uniform distribution of the flue gas around the atomizer as well as an adequate gas velocity and direction
are essential features for achieving satisfactory initial drying conditions. The swirling flow of gas with
entrained droplets continues down through the cylindrical vessel and is inverted in the cone section prior to
the gas exiting via the single duct located in the side of the cone. The excellent conditions for simultaneous
heat and mass transfer in the vessel result in the reaction between alkaline slurry droplets and acid
components in the gas during the evaporative cooling process. The end result is that the cooled and dry flue
gas (typically at a temperature of 265-285°F) with increased moisture content exits from the vessel, and a free-
flowing, dry solid of flyash, salt (mainly calcium chloride and calcium sulfite/sulfate) and unused lime is
generated.
Part of the dry solids is discharged from the spray dryer cone bottom, but most is entrained with the flue
gas to the downstream located particulate collector. As the potential for wet wall deposits as a result of
insufficient slurry drying was an occasional concern a motorized lump breaker was included at the cone
discharge. This device would serve to break up any non-powdery product being discharged and avoid
problems in the conveying system.
A schematic of Komline-Sanderson's rotary atomizer, Model 860 is shown in Figure 3. The rotary
atomizer consists of a water-cooled variable-speed 60-HP-drive motor controlled by a frequency inverter,
eliminating the need for a gear box. A fine mist of oil is carried on a stream of compressed air to the upper
and lower bearings of the rotor assembly. Interlocks protect the assembly against loss of compressed air flow,
low oil level in the mist generator, or loss of water flow to the water jacketed stator. All connections to the
atomizer (oil mist, water, electricity) are through one central terminal stand which is located in the penthouse
on top of the SDA roof. The speed of the atomizer is controlled by the frequency inverter with a
thumbwheel adjustment for variation of the wheel speed. This feature provides for tuning for best system
49
-------
performance and optimization of energy consumption. This adds process and operating flexibility that would
be difficult to achieve with a constant-speed type of atomizer. Typical operating speed for the present process
application is 14-15,000 rpm.
The atomization of the lime milk slurry in the atomizer unit is accomplished by its acceleration in the
light-weight rotating 7.5-inch-diameter wheel (fabricated of titanium) to which the lime slurry is gravity fed
from the head tank. The atomizer wheel is equipped with tungsten carbide inserts (for abrasion resistance)
and is mounted on the "rigid" shaft (see Figure 3).
The atomizers were fitted with vibration monitors to provide an early warning system against imbalance.
However, no significant potential for vibration problems are expected with this atomizer type. This is due
to the unique feature of a "rigid" spindle shaft and a light-weight wheel body which makes the rotating
assembly less prone to vibration than most other types of rotary atomizers. The wheel body is manufactured
of titanium, as mentioned above, and the entire spindle shaft length is only 19 inches, of which less than 4-Vz
inches extends beyond the lower bearing. These two design considerations result in a rotating assembly which
operates well below the first critical speed.
A hoist is provided inside the penthouse for removing or placing the atomizer in service. A spare
atomizer assembly, common for both boiler trains, is also provided. This arrangement allows for easy
replacement of an atomizer unit within a relatively short time period, i.e. approximately 20 minutes.
PULSE-JET FABRIC FILTER
Each boiler train is equipped with a four (4)-compartment PJF downstream of the SDA for particulate
control. A schematic of the PJF is shown in Figure 4. Each compartment houses 210 woven fiberglass bags
with Burlington-373 chemical resistant finish, each bag being 6 inches in diameter and 14 ft long. The PJF
is designed for off-line cleaning, using a gross air-to-cloth ratio of 2.7 ft/min (all four compartments in
service), and a net air-to-cloth ratio of 3.6 ft/min (one compartment in the cleaning mode). Each
compartment consisting of housing with tube plate (separating dirty and clean gas side), pyramidal hopper
and compressed air manifold and distribution system is shop assembled as an individual module
(approximately 12 ft x 12 ft x 40 ft) and shipped to the jobsite for easy erection.
The four baghouse modules are connected to the inlet and outlet manifolds. The EEC baghouse design
incorporates a patented stepped inlet manifold and a patented hopper vaning arrangement which assures
improved flow balance and even dust distribution among compartments and among filter bags within each
compartment (see Figure 4). The uniform gas distribution and low turbulence result in a low mechanical
pressure loss (the mechanical pressure drop is less than 1.0 in.WG at all operating loads). Hopper vaning
assures good gas and dust distribution to all the bags resulting in increased bag life by elimination of
potentially damaging high local velocities and dust loadings.
While many different cleaning modes can be used, the recommended procedure is to clean the bags off-
line. This mode of cleaning eliminates dust reintrainment and promotes longer bag life. Cleaning can be
initiated manually, on a timed cycle basis, or by a preset pressure drop across the PJF.
Each compartment has 15 pulse pipes, each pipe cleaning a row of 14 bags by supplying a pulse of 60-80
psig compressed air from the compartment manifold via a solenoid-activated diaphragm valve. The entire
cleaning sequence for one compartment (15 pipes) is of a 3-minutes duration.
The baghouse is of a walk-in plenum type of design, i.e. each clean-air plenum above the entire
compartment tubesheet is tall enough to accommodate the full bag length (14 ft). This plenum is entered
by a single door and bag removal, inspection and replacement can conveniently be done regardless of weather
conditions. There is no reason to enter the dirty side of the compartment to replace bags. Once the pulse
50
-------
pipe just above the tubesheet is disconnected each bag and cage assembly is inserted through the 6-inch
diameter openings in the tube plate and retained by the top flange of the cage.
The 55°-valley-angle hoppers include electrical heaters (for maintaining up to 400°F inside surface
temperature in the lower section), level detectors, vibrators, as well as poke holes and strike plates
immediately above the 12-inch diameter hopper discharge flange. Rotary airlock valves are included between
each hopper flange and the mechanical conveying system. The conveying system is conservatively designed
(not EEC scope) and is not provided with heat tracing.
A flue gas bypass system is included for use during boiler start-up on No. 2 fuel oil. Also, a separate
hydrated lime injection system consisting of a day bin, feeder, blower and piping to the PJF inlet manifolds,
intended for pre-coating bags prior to garbage burning, was supplied.
LIME STORAGE AND SLAKING SYSTEM
The lime system was designed and shop assembled as a separate by Chemco. This approach allows for
extensive shop testing of the separate components and easy field installation. This system measures 13 ft in
diameter and has a total height of about 55 ft when installed in the field. The relative arrangement of
equipment inside the lime system enclosure is indicated on Figure 1.
The quicklime storage silo is designed for truck delivery, and its 13 ft diameter and 14 ft cylindrical
height provides a capacity of 60 ton, corresponding to an inventory of about 10 days of reagent at the
maximum consumption rate. The silo has a 60° conical hopper equipped with a 5-ft bin activator for
discharge into one slaker unit via a volumetric feeder. The slaker is a Wallace & Tiernan paste type unit with
a capacity of 1000 lb/h, which is about 200% of the maximum process consumption rate. This unit is designed
for slaking at a low water-to-lime ratio resulting in a high temperature (180-200°F) and efficient conversion
of quicklime (CaO) to hydrate [Ca(OH)2]. The slaker always operates at its rated capacity in an on/off type
of mode responding to the process demand (see below). Well-water with no water pre-heating is used for
slaking, and grit is removed immediately downstream of the unit by a 20-mesh Sweco screen discharging into
a dump container via a 5-inch screw conveyor located inside the enclosure. The slaked and screened lime
with a solids content of about 22% is stored in the 2,850 gallon agitated slurry tank providing a total supply
of about 8 hours at maximum rated conditions. Low and high level switches at intermediate levels automati-
cally start and stop the slaker, typically every one to two hours, to ensure that a reasonable inventory of
slaked lime slurry is maintained.
A slurry bleed valve gravity feeds the 22% lime slurry to the lime dilution tank located at grade.
Dilution of the lime slurry to the actual concentration level required by the flue gas cleaning process is done
here, typically to 5-10% solids. This tank has a capacity of about 450 gallons of lime slurry to be fed to the
two spray dryers, representing a slurry retention time of 15-60 minutes.
The diluted lime slurry is pumped from grade to the two head tanks located in the spray dryer
penthouses. One pump for each spray diyer is transferring the lime slurry via 1-inch diameter steel pipes at
typical velocities of 7-9 ft/sec, with an overflow return from the head tank to the diluted lime milk tank at
grade. The 30-gallon head tank provides a small diluted lime inventory of about 2-8 minutes and primarily
serves the purpose of ensuring a constant head to the feed control valve for gravity feeding the lime slurry
to the atomizer.
51
-------
PROCESS INSTRUMENTATION & CONTROL SYSTEM
The operation of the flue gas cleaning system is controlled and monitored by a Bailey Network 90
Distributed Control System. Local I/O panels & PLCs with residing control logic for the APC system were
provided by EEC, and were tied in to the plant computer with terminals including monitors and keyboards
located in the central control room. This provides for operation (start-up, shut-down, adjustment of setpoints,
etc.) of the APC system from a central location with all required process readouts available.
The key operational parameter for the APC system is the flue gas outlet temperature from the SDA
which normally is controlled in the range of 265-285°F. This temperature set-point is used for controlling
the flow of lime slurry to the atomizer. A temperature lower than 265°F may lead to insufficient drying of
the hygroscopic calcium chloride salt formed by the reaction between lime and HC1. Insufficient drying would
result in solids deposits in the spray dryer vessel and/or build-up of moist powder in the baghouse, typically
leading to high fabric pressure drops or product accumulation in the hoppers (moist product has poor
flowability). On the other hand, a temperature higher than about 300°F may lead to insufficient acid gas
control, specifically of the reactive S02, or to an excessive lime consumption rate when aiming at maintaining
acid gas emission compliance.
The actual solids content of lime in the slurry to the atomizer is adjusted by the operator based on S02
data from the plant's Continuous Emission Monitoring (CEM) system. The acid gases HC1 and HF are not
monitored continuously as efficient control of S02 is known to result in even more efficient removal of these
two gases.
Depending on the flue gas flow rate and available exit temperature from the boiler the water evaporation
rate in the spray dryer may vary significantly, typically in the range from 3 to 15 gpm per unit. The rotary
atomizer frequency inverter control system maintains constant wheel speed for all conditions, and changes
in the lime slurry flowrate to the unit therefore results in proportional changes in the power consumption
(kW). However, the specific power consumption (kW/lb slurry) remains constant, resulting in uniform
atomization energy and uniform process performance throughout the operating range.
The control system gathers all relevant operational and emission performance data. A history capability
for the last 26 hours of data exists which is a useful tool for following the most recent trends. All data is
downloaded once a day on permanent storage for future reference.
OPERATIONAL HISTORY
SPECIFIC START-UP ISSUES
Initial start-up of the waste-to-energy boiler Unit #1 occurred on July 3, 1988. Detailed procedures for
start-up of the flue gas cleaning system were developed and instituted. The boiler uses No. 2 fuel oil for
start-up, but the fabric filter could not be by-passed during this initial phase as garbage would be introduced
onto the grates as soon as a satisfactory furnace temperature was reached. To protect the filter bags against
any acid gas attack, hydrated lime was pneumatically injected into the PJF inlet manifold during garbage
combustion at the rate of 200 lb/h per unit from the installed precoat system. This was in addition to the
approximately 2,000 lbs of hydrated lime which had been used for initial precoating. Once the SDA inlet
temperature reached 320°F normal scrubbing using lime slurry spraying was initiated and the hydrated lime
injection stopped.
Boiler Unit #2 was started-up a few weeks later.
52
-------
A couple of upset conditions occurred during the initial weeks of operation which are worth discussing.
During normal garbage combustion on July 12, the spray dryer outlet temperature was inadvertently lowered
to 240°F by water spray only (no lime). This resulted in an increase in the tubesheet differential pressure of
3 to 4 in.WG which could not be reduced by the normal cleaning process of the PJF.
The filter bags were then individually air-lanced by manually inserting a section of copper tubing with
radially drilled holes connected to a compressed air supply into each bag. The entire operation was carried
out one module at a time while the remainder of the baghouse was operating. The PJF operation returned
to normal conditions after completion of this procedure as evidenced by the establishment of the original
pressure drop across the bags.
The second upset condition occurred on July 23. While the boiler was on fuel oil operation, the external
economizer was brought in service for the first time. Lime slurry spray reduced the gas temperature across
the SDA from 340°F to 300°F and the system operated in this mode for 16 hours. For reasons not completely
understood an increase in differential pressure was observed , and again air-lancing was used to return the
filter bags to normal conditions.
Selected filter bags from the facility start-up have been analyzed, including those which had gone through
the described upset conditions. The results of this analysis are presented in Table 1.
It appears that the upset conditions described resulted in a quick loss of the fabric permeability.
However, the permeability could be partly restored by the air-lancing operation (comparison of bags A and
B in Table 1 - bag C's permeability cannot be directly compared as it has been in operation for 2-3 months).
Somewhat surprisingly, the bag which experienced upset conditions, but was not air-lanced (bag A) seems to
have less than a normal amount of attached dust, but a quite high moisture content. All these data seem to
verify the hygroscopic nature of this product and particularly the impact of upset conditions. Also, it seems
important to note that the adverse results of upset conditions on the fabric bags may be reversed such as by
the manual air-lancing procedure described.
ROUTINE OPERATION
During the first year of operation the APC system operated with virtually 100% availability. Since then
the system has shut down on 2 to 3 occasions due to electrical or control problems which appear to have been
related to an incident when the plant was struck by lightning.
The current mode of operation of the APC system which has been selected by the plant is primarily
aiming at maintaining compliance with an adequate margin of safety in S02 emission limits, and maintaining
a fairly constant generation rate of waste product. This mode is characterized by the following two major
points:
• The atomizer runs with a constant slurry density of 10% solids. The typical outlet temperature is
265-285°F. In the event of a spike in the inlet S02 concentration occurring, operators may lower
the SDA outlet temperature in increments of 5°F. This provides additional lime slurry for
scrubbing purposes.
• The baghouse cleaning cycle is operated on a time basis to maintain a fairly constant pressure drop
across each module of 7 to 9 in.WG. This provides a uniform flow and composition of the waste
product.
53
-------
COMPONENT HISTORY
Lime Storage/Slaking System
Although somewhat plagued with problems that appear to be characteristic for these fairly compactly
designed systems, such as high maintenance and dirty working conditions, the system continues to provide
lime slurry at the required capacity with only minor equipment modifications.
Level indicators on the lime silo have had numerous problems, and the plant personnel maintain a visual
inspection of the level of lime through the silo roof hatch. A fresh load of 20-22 tons of quicklime arrives
approximately every three days. Problems with the opening of the silo relief valve during periods of loading
with lime have been traced to improper operation of the exhauster on the bin vent filter.
The possible addition of a spare slaker is by the plant regarded as a desirable luxury for improving the
maintenance situation, even though the existing single slaker is sufficient to meet the present consumption
rate (even at higher than design usage rates). Good slaking operation has been partially attributed to a
decision to switch to and maintain supply of a high-quality pebble lime shortly after start-up. Other areas
of the slaking operation have, however, necessitated improvements. The slaker aspirator now uses air instead
of water and further improvements are being considered relative to ventilation. The slaker is manually
washed out after each slaking cycle to prevent build-ups of lime and grit. The grit screen has been replaced
once in over two years of operation. Currently a 20-mesh screen is used. Additionally, 6"-donuts constructed
of V* rubber tubing are placed on top of screen to prevent build-up of solids.
The density monitors on the slurry feed tank have at times provided unreliable inputs, and operators now
manually take two samples per shift to verify solids content of the slurry.
Slurry Pumps, Valves and Piping System
Pinch valves are used for slurry control and all original valves are still in service. Two slurry pumps have
been replaced as this was determined to be more cost effective than rebuilding the old pumps.
The primary slurry handling problems are associated with pluggage of the lime slurry supply and return
piping. Solids deposits in the piping became an apparent problem V/2 years after initial operation and the
pipes now require close monitoring and cleaning with acetic acid during each outage.
A decision has been made to replace the normal slurry steel piping with heat-traced, impregnated hose
within the near future . It is felt that maintenance will be easier and disassembly of the components will also
be quicker, should replacement be necessary.
A cyclonic separator originally installed in the slurry supply line to the head tank never performed
properly in terms of removing oversize particles and it has been replaced with a dual-basket strainer in the
line between the head tank and rotary atomizer. A cross-over line between the two head tanks has also been
added . This provides for additional redundancy if a sudden pluggage occurs in one of the supply lines, and
also makes on-line maintenance of the individual supply and return lines possible.
Spray Drver Absorber
There have been no deposits on the internal surfaces of the dryer vessels. An inspection after 2Vi years
operation revealed only a slight, uniform coating on the walls. In fact, the lump breaker originally installed
at the hopper discharge has been removed. The dryers routinely operate with an outlet gas temperature as
low as 260°F, and have exhibited no drying problems.
54
-------
Rotary Atomizer
The atomizers have performed well albeit with a few modifications. Each atomizer wheel is water-
flushed daily (on-line) and cleaned in a muriatic acid bath once per week (off-line). The spare wheel is put
in service while the wheel is acid cleaned. The tungsten carbide inserts have performed consistently for one-
year intervals before replacement is required. Likewise, the wheel bottom wear plates are replaced at one-
year intervals. A maintenance agreement exists with Komline-Sanderson to replace all wear parts once each
year and to perform general service on the atomizers.
The vibration detectors experienced interference as a result of external electrical noise and required
adjustment. Early minor lubrication problems have been resolved with a slightly redesigned lubrication
system.
Malfunctions associated with the frequency inverter are believed to be the result of an incident when
the plant was struck by lightning as mentioned earlier. In one case the choke coil, which filters the incoming
current into the DC bus, had been short-circuited and was replaced.
Pulse-Jet Fabric Filter
Initially the PJF used a batch cleaning method based on pressure drop. Once cleaning was initiated,
individual modules were cleaned off-line and returned to service with only a 15-second delay before the next
module was cleaned. The overall cleaning cycle would only be 12-15 minutes. The system now employed
uses a 7-minute time delay between modules and a 15-minute delay between cycles, resulting in an overall
cleaning cycle of 55-60 minutes. This time delay cycle maintains a fairly constant pressure drop of 7-9 in.WG
across the PJF, and results in a quite uniform dry waste conveying rate.
The filter bags on Unit #2 were replaced in November 1990, while replacement of those on Unit #1
took place in March 1991. The original galvanized cages were not replaced. Both replacements were
performed during a scheduled outage, but were not the result of bag failures. It was felt that the bags were
near the end of their useful life, and there was concern about the shelf life of the spare bags which had been
in storage since 1988.
The results of a recent analysis performed on filter bags removed from Unit #1 are listed in Table 2.
The independent testing laboratory concluded that the analysis for strength retention (breaking strength,
Mullen Burst, and flex cycles) show no evidence of either chemical or thermal degradation. There was no
visible evidence of any internal migrated particulate into the inner fabric. Under pulse impact the bags
demonstrated high discharge properties and good flow recovery, even though the residual dust did exhibit
some moisture induced agglomerations. There was evidence of contact with corroded cages, however the
level of damage was minimal. These bags were projected by the test laboratory to have 20 additional months
of continuous service life.
The PJF on Unit #1 experienced one upset condition besides those reported during initial start-up. In
June of 1989 a spike in opacity occurred and it was observed that one of the module hoppers had filled up
with ash to the extent that the bags within that module became buried in dust to a level near the tubesheet.
It was noted that the high opacity was due to holes in a bag near the gas inlet, presumably due to the high
gas velocity caused by the restricted flow conditions.
55
-------
There have been a total of 10 bag failures during 2Vi years of operation on the two boiler trains. Good
service life is believed attributable to a number of factors, such as:
• An excellent preventative maintenance program by plant personnel. Hopper poke holes are
checked once per shift for draft, and corrective action is initiated immediately if a problem exists.
There has been only one incident of an ash-filled hopper, as reported above.
• Good and consistent operating practices: Bags are precoated prior to each cold start, and hopper
heater thermostats are set at 400°F.
• Close adherence to start-up procedures.
• A design using a quite tightly fit between bag and cage. Although the cleaning action may be less
efficient with a tightly fitting bag, long bag life is achieved. It appears that this is accomplished with
only a minor sacrifice in terms of slightly higher pressure drop.
GENERAL DESIGN ISSUES
A number of improvements have been incorporated into EEC's design of SDA systems for future waste-
to-energy plants, largely as a result of operating experiences at Warren Energy Resource Company.
Dilution of the lime sluny in EEC's new design will take place in the head tank. This will provide a
faster response to variation in inlet acid gas conditions. It is also anticipated that solids deposits in the slurry
piping will be minimized, since the majority of the slurry piping will be handling a higher lime slurry solids
content. Also, slurry piping less than 2Vi inches will be avoided or minimized. Dual-basket strainers will
always be included in the slurry lines between the head tank and the atomizer and in the slurry return lines.
The Komline-Sanderson rotary atomizer has incorporated only minimum basic design changes. However,
an enhanced cooling system has been introduced thereby significantly improving motor efficiency and allowing
for a 50% increase in atomization capacity. The bearing and motor winding temperatures have been lowered
through a water purge to the shaft and wheel. Monitoring of the motor winding temperature has been added,
along with a once-through, forced oil lubrication system, and a self-locking wheel-to-shaft assembly. A
program to evaluate the performance of the new atomizer design at the Warren County Facility is planned
for April, 1991.
Evaluation of the original filter bags is currently in progress. An analysis of the periodic fabric test
reports demonstrates good service life of the bags and satisfactory ability to withstand upset conditions. The
chemical resistant finish (Burlington-373) shows promise, although a direct comparison with bags provided
with a typical Acid Resistant or a Teflon B finish has not been made. The collection efficiency of the fabric,
as shown later by the emissions test in Section V, is excellent. Although the particulate generated from a
municipal waste combustor followed by a lime spray dryer is very hygroscopic and appears to have a tendency
to become agglomerated on the fabric, the fabric pressure loss is maintained at a constant level, albeit higher
than most other applications.
TEST RESULTS AND EMISSION PERFORMANCE
PERFORMANCE TESTING OF APC SYSTEM
After successful start-up and initial operation of the waste-to-energy facility in the summer and fall of
1988 Environmental Elements conducted the APC system compliance and performance test program in
56
-------
November of 1988. Clean Air Engineering (Palatine, IL) was contracted to do all field sampling and analysis.
Blount operated the facility during the two days of testing with EEC representatives in an advisory capacity.
All relevant operational data was collected, specifically flow rates, temperatures, pressures, and lime
consumption rates and evaluated together with the analytical results of gas sampling (providing gas flowrates,
composition, and pollutant concentrations), and slurry sampling (providing lime slurry composition). Three
test runs were performed on both boiler Units #1 and #2, and the results averaged for each unit. The most
important test results are presented in Table 3.
The equivalence ratio (ER), or the ratio between lime input and acid gases exiting the boiler, is a key
parameter for the removal performance of the SDA/PJF system (see previous section for the definition of
ER). HC1, the most significant of the acid gases, is not monitored continuously and the actual ER at any
given time is therefore not available to the operator of the facility. A typical operational approach is to
always add a comfortable margin of lime slurry in order to ensure compliance at all acid gas conditions.
About 238 lb/h quicklime was added to Unit #1 during performance testing. As the measured HC1 inlet
concentration was relatively low [about 450 ppmdv (7% OJ] the actual ER exceeded 2 and resulted in
significantly higher S02 removal than required for compliance (in excess of 99%, vs. 90% required). When
testing on Unit #2 a clearly higher average HC1 concentration into the spray dryer was measured. As the
lime input was of the same order of magnitude as during testing on Unit #1 a lower actual ER resulted on
Unit #2. The calculated ER of 1.4 is in line with the lime consumption guaranteed for this facility, and as
Table 1 shows, all acid gas removal efficiencies meet the performance requirements (95% for HC1 and HF,
90% for S02).
The measured particulate stack emissions of 0.0017-0.0031 gr/dscf (7% 02) is comfortably below the
maximum emission requirement of 0.01 gr/dscf. However, the measured pressure drop across the APC system
of 9.8-10.4 in.WG is very close to the guaranteed value of 10 in.WG. It may be noted that the actual gas flow
into the spray dryer appears to be 5-10% in excess of the MCR design value which would explain the slightly
higher than predicted pressure loss.
All test results documented that the APC system had satisfied the performance guarantee which had
been provided by EEC when supplying the system.
YEARLY COMPLIANCE TESTING AND EMISSION RECORDS
Acid gas and particulate emission data have been collected by independent stack sampling companies
for Blount Energy Resource Corporation in 1988, 1989, and 1990 serving to verify continued emission
compliance of the facility. These data are presented in Table 4. All reported data are the average of three
test runs.
The particulate emission data from 1988 and 1989 were all taken when both units were operating with
the originally installed bags and are all well below the contractual requirement of 0.01 gr/dscf (7% 02). Even
though the numbers from 1989 appear slightly higher than from 1988 there are hardly sufficient data to
conclude that any slight deterioration of bags has resulted in higher emissions. Particulate emission
compliance is also verified continuously at the facility by the opacity monitors typically showing between zero
and two percent, as compared with the contractual limit of ten percent opacity.
Very high acid gas removal efficiencies have been achieved in these yearly compliance tests. In excess
of 99% HC1 removal and stack emissions of less than 5 ppmdv (7% Oj) were measured, which is exceeding
the performance requirement of 95% HC1 removal or maximum stack emissions of 25-45 ppmdv (7% Oj).
The S02 ftmoval/emission data has been continuously verified since 1989 by the CEM system. In excess of
90% S02 removal or 40 ppmdv (7% Oj) in the stack is typically demonstrated based on a 3-hr rolling average.
57
-------
Normally, the operator will adjust the lime addition rate to aim at a minimum of 95% S02 removal in
order to maintain some "buffer" capacity in the system should a peak of either HC1 or S02 occur in the boiler
exit gas due to changing chlorine or sulfur content in the garbage being burned. A recent (March 7, 1991)
2-hr history of S02 data from the CEM system on Unit #2 is shown in Figure 5. Typically the S02 is reduced
from about 160 ppm at the boiler exit to less than 40 ppm in the stack. For a significant part of the time the
SO, removal appears to be close to 100%, suggesting that an excess of lime is being introduced. As
mentioned earlier, this mode of operation provides the plant with an adequate safety margin in maintaining
compliance.
Obviously the excess lime results in the formation of a quite alkaline waste product. A high lime content
of the APC waste may also have some beneficial impact on the characteristics of the final waste product
(consisting of a mixture of boiler ash and APC waste). E.g., the leachability of certain metals contained in
the boiler ash or in the APC ash can potentially be lowered by the increased alkalinity of the mixture,
reducing the potential leachability impact of the disposed waste on the environment. This feature seems to
suggest that adjustment and/or optimization of the operating conditions of the APC system downstream of
a boiler burning a quite heterogenous municipal waste may be considered as one of the parameters used for
controlling the final waste disposal options. However, the issue is complicated by the variation between
various metals in the alkalinity reactions and a clear conclusion on the subject has not been reached.
METALS EMISSION TESTING
Sampling and analysis of the stack emissions of metals on a selected boiler train were done in both 1988
and 1989 as part of the yearly compliance verification program. The results of this testing are presented in
Table 5 (all data are the average of three test runs). The boiler permit compliance number for each of the
metals has also been indicated in this table. For all metals existing predominantly in particulate form at the
stack gas temperature of about 265°F, i.e. lead (Pb), arsenic (As), beryllium (Be), cadmium (Cd), chromium
(Cr), and nickel (Ni), the measured emission numbers are well below the level set by the permit requirement.
For many of the analyses the concentrations fall below the maximum detection limit used in these
measurements. These low emissions are in large part due to the efficient particulate collection efficiency of
the fabric filter.
For mercury (Hg), which is known to exist mostly in vapor form even at the relatively low stack gas
temperature of 265°F, the measurement was below the permit level of 0.05 lb/h (the three test runs were in
the range 0.013-0.030 lb/h) when tested in 1988 (see Table 5). However, the three test runs in 1989 showed
an average emission of 0.10 lb/h, mainly due to one single test run showing a emission as high as 0.27 lb/h.
It appeared unclear whether this measurement was a result of a temporary extremely high concentration of
mercury in the garbage being burned during this test period or whether it truly reflected the potential for a
higher-than-predicted general mercury emission level. In order to investigate this issue Blount Energy
Resource Corporation embarked on a program aiming at reducing the amount of mercury-containing waste
being fed to the boiler. Also, a commitment was made to gather more data by testing the stack emission of
mercury at regular 3-month intervals for a one-year period starting in the fall of 1990. The result of the first
test series is presented in Table 6. These data seem to verify that significant variation in the mercury
emission (by a factor of 5) may occur during a time period of only a few hours. A each test run typically
represents the average of a 2-hour period, even larger, but shorter-duration variations might also occur in the
emission level than reflected by these data. This apparent fact certainly emphasizes that caution should be
used when evaluating mercury emission numbers based on a limited number of test data and that averaging
a large number of data is required in order to properly assess the mercury emission level. The average
mercury emission of this first test series is 0.044 lb/h, which is below the permit compliance value of 0.05 lb/h.
When expressed in concentration units a comparison of this average emission of 0.42 mg/Nm3 appear to be
only slightly above the level which is typically seen from similar waste-to-energy facilities equipped with spray
dryer/baghouses (1, 2). It is therefore believed that the above described efforts at Warren Energy Resource
Company can be expected to result in compliance with the existing mercury emission standards for the facility.
58
-------
CONCLUSIONS
• The Warren County waste-to-energy facility consisting of two 200 ton/day boilers has been in successful
commercial operation for about two-and-a-half years.
• The APC system using spray dryer/pulse-jet filter technology was furnished by Environmental Elements
Corporation. It was the company's first reference in the supply of this particular process technology.
• The APC system has demonstrated a very satisfactory operational history in terms of availability and
process performance.
• No major mechanical problems associated with the APC system have been experienced at the waste-to-
energy facility. Most operational problems appear fairly trivial and similar to what has been reported
from other facilities, such as the high maintenance associated with the handling of lime slurry. Good
operational practices and excellent maintenance programs at the facility have also significantly
contributed to this favorable operational record.
• The APC system's guaranteed process performance in terms of acid gas removal (95% HC1 and HF,
90% S02) has been satisfied at the guaranteed lime consumption rates.
• The baghouse has demonstrated particulate emissions well below the guarantee of 0.01 gr/dscf (7% O,).
Total pressure loss across the system is about 10 in.WG.
• Specific environmental concerns relating to the potential emissions of mercury initiated a regular stack
sampling program in 1990 aiming at verifying that compliance with the permit emission limit of 0.05 lb/h
per boiler can be documented.
• The system has also demonstrated compliance with all other metal emission limits, and in many cases
showed emissions an order of magnitude less than the permitted values.
REFERENCES
1) P. C. Siebert: "Estimating Emissions from Municipal Solid Waste Incinerators". Paper presented at
the Society for Risk Analysis 1989 Annual Meeting, San Francisco, CA, October 1989.
2) C. Jorgensen, J. G. Toher, J. L. Hahn, and P. F. Claerbout: "Stack Emission and Dry Scrubber
Performance Testing at the 2362-TPD Indianapolis Resource Recovery Facility". Paper presented at
the IGCI Forum '90: "Air Quality Control Systems for Today and Tomorrow", Baltimore, MD, March
1990.
59
-------
UfcC
SILO
TRUCK
OELTVOY
OF QUICKLIME
PULSE JET
MMOUSC
SCREEN
TO DISPOSE
SLURPf PUMPS
(2+0
APC SYSTEM
WASTE PRODUCT
MIXER
Figure 1. APC System Flow Sheet
60
-------
HEAD TANK
HOIST
HEAD TANK
OVERFLOW RETURN
TERMINAL STAND
LIME MILK
SUPPLY
SPARE ATOMIZER
GAS
ATOMIZER WELL
aow
ATOMIZER IN
OPERATING
POSITION
VANE RING
DISPERSER
CHAMBER
GAS OUTLET DUCT
GAS
FLOW
HOPPER
TO PRODUCT
DISPOSAL
Figure 2. Spray Dryer Absorber
61
-------
13" 0
UPPER BERRING
MAIN HOUSING
STRTTOR
ROTOR
COOLING WATER
JRCKET
DISTRIBUTION
RSSEMBLY
LOWER BEARING
ATOMIZER
WHEEL
Figure 3. Komline-Sanderson Model 860 Rotary Atomizer
62
-------
ACCESS PLATFORM
OUTLET POPPET DAMPERS
BY-PASS
DUCT
OUTLET MANIFOLD
ACCESS DOOR
WALK-IN
PLENUM
COMPRESSED AIR
MANIFOLD
PULSE PIPE
BAG ON CAGE
ACCESS PLATFORM
INLET MANIFOLD
DISTRIBUTION
VANES
ANTI-SNEAK
BAFFLES
INLET DAMPER
TURNING VANES
BY-PASS
POPPET DAMPER
TURNING VANES
GAS FLOW
TV
MODULES
OUT
OUTLET MANIFOLD
MODULES
BY-PASS DUCT
TURNING VANES
GAS FLOW
INLET STEPPED
MANIFOLD
tSJ
MODULES
MODULES
1 it 2 3 ft 4
Figure 4. Puke-Jet Fabric Filter
63
-------
07MhR91 THURSDAY HO .93 BLP.2 iu£
400 ppm
320 ppm
240 ppm
160 ppm
80 ppm
0 ppm
Figure 5. S02 Data From CEM System
. ,-'l '
j-*Vj ¦
•« t—mTI I
- —
: 00 6 • £0 6: 40 7 • 00 7 ¦ C0
40
64
-------
TABLE 1. FILTER BAG TEST DATA (a)
BAG IDENTIFICATION (b)
A
B
C
D
PERMEABILITY (at 0.5"WG):
-As received, cfm/sq.ft
7.52
12 . 60
2.78
20-40
-Vacuumed, cfm/sq.ft
9.60
14.60
7.64
N/A
WEIGHT:
-As received, oz/sq.yd
25.7
22.9
30.3
21.6
-Washed, oz/sq.yd
21.7
22.3
22.3
N/A
MULLEN BURST, psig (avg)
1363
1477
1332
900 (c)
RESIDUAL DUST:
-Moisture content, %
18.5
14 . 0
N/A
-Moisture regain:
at ambient RH, %
22 . 6
16.3
N/A
at 100% RH, %
34.0
25.0
N/A
(a) Grubb Filtration Testing Services, Inc - Report No. 342, Oct.1988
Note: MIT Flex not performed as bags exhibited no physical Wear
(b) Bag A: Experienced low temperature excursions and "water spray
only" operation. Tubesheet pressure drop 8-9 in.wg
Bag B: Experienced same conditions as Bag A, but was air-lanced
and removed from service
Bag C: Experienced same conditions as Bags A and B, but was
returned to normal SDA service
Bag D: Specifications for new bag
(c) Specification lists minimum Mullen burst strength; however, tests
routinely show results in excess of 1200-1400 psig
TABLE 2. FILTER BAG TEST DATA (a)
BAG IDENTIFICATION (b) A B C
PERMEABILITY (at 0.5"WG):
-As received, cfm/sq.ft 2.89 3.22 3.26
-Vacuumed, cfm/sq.ft 7.93 8.67 8.63
WEIGHT:
-As received, oz/sq.yd 27.6 27.1 26.9
-Washed, oz/sq.yd 21.8 22.1 21.9
BREAKING STRENGTH, lbs/inch:
-Warp (length) 460 463 468
-Fill (width) 421 142 145
BREAKING STRENGTH, % loss:
-Warp (length) 38.7 38.9 37.6
-Fill (width) 59.9 59.5 58.7
MULLEN BURST, psig 617 619 622
MULLEN BURST, %loss 38.1 38.1 37 8
FLEX CYCLES (MIT Method):
-Warp (length) 13642 13587 13965
-Fill (width) 4213 4154 4100
FLEX CYCLES, %loss:
-Warp (length) 54.5 54.7 53.5
-Fill (width) 57.9 55.1 59.0
(a) Environmental Consultant Company - Report No.TLN 8764, April 1991
(b) Each bag sampled from different module
65
-------
TABLE 3. RESULTS OF PERFORMANCE TESTS, NOVEMBER 1988
BOILER UNIT:
#1
#2
GUARANTEE:
SPRAY DRYER INLET CONDITIONS:
Flue gas flow rate, ACFM
Temperature, deg.F
HC1 cone, ppmdv (7%02)
HF cone, ppmdv (7%02)
S02 cone, ppmdv (7%02)
STACK CONDITIONS:
Flue gas flow rate, ACFM
Temperature, deg.F
HC1 cone, ppmdv (7%02)
HF cone, ppmdv (7%02)
S02 cone, ppmdv (7%02)
QUICKLIME FEED RATE, lb/h
EQUIVALENCE RATIO (ER)
ACID GAS PERFORMANCE:
HC1 removal, %
HF removal, %
S02 removal, %
PARTICULATE EMISSION:
-gr/dscf (7%02)
PRESSURE LOSS:
-Total system, in.WG
60,100
366
454
1.2
146
57,900
264
21
0.09
0.48
238
2.1
95.4
93
99.7
0.0017
9.8
62,100
365
873
1.8
149
59,800
262
29
0. 08
14
215
1.4
96.6
96
90.4
0.0031
10.4
1.4
95
95
90
0.01
10.0
TABLE 4. RESULTS OF YEARLY COMPLIANCE TESTS, 1988-1990
YEAR
ii
1
1
1
M 1
1
io 1
ii
1
1
1
1
1
1 CO
1
I 00
II
1
1
1
1
M |
1
10 1
ii
H
00 1
1
io 1
1
1
1
1
1
ii
1
1
1
H 1
1
10 1
1
10 1
ii
0
BOILER UNIT
#1
#2
#1
#2
#1
#2
HC1 INLET, ppmdv
(7%02)
840
718
466
182
(a)
(a)
HC1 STACK, ppmdv
(7%02)
2.8
1.8
3.3
1.4
HC1-REMOVAL, %
99.7
99.7
99.3
99.3
—
—
S02 INLET, ppmdv
(7%02)
170
199
(b)
(b)
(b)
(b)
S02 STACK, ppmdv
(7%02)
1.2
0.23
-
-
-
S02-REMOVAL, %
99.3
99.9
—
—
—
—
PARTICULATES, gr/dscf(7%02)
0.0036
0.0018
0.0048
0.0040
(a)
(a)
(a) Results not yet available
(b) Compliance verified by CEM data
66
-------
TABLE 5. RESULTS OF METAL EMISSIONS TESTS, 1988-1989 (a)
YEAR 1988 1989
COMPLIANCE
BOILER UNIT #2 #2 REQUIREMENT:
MERCURY (Hg)
LEAD (Pb)
ARSENIC (As)
BERYLIUM (Be)
CADMIUM' (Cd)
CHROMIUM (Cr)
NICKEL (Ni)
0. 022
0.101
0.05
<
0.0003
<
0.003
0.25
0.000087
<
0.0003
0.002
<
0.00015
<
0.0003
0.0004
<
0.0003
0.00040
0 . 03
<
0.0005
<
0.0019
0.01
<
0.0003
<
0. 001
0 . 004
(a) All data in LB/H per boiler
TABLE 6. RESULTS OF MERCURY EMISSIONS TESTS, 1990 (a)
test
mg/Nm3
LB/H
RUN #1
0.178
0.0185
RUN #2
0.893
0.0959
RUN #3
0.244
0.0427
RUN #4
0.417
0.0430
RUN #5
0.267
0.0293
RUN #6
0.495
0.0525
AVERAGE 0.416 0.0444
(a) All tests done on boiler unit #1
67
-------
Intentionally Blank Page
68
-------
SESSION 2C:
RECENT DRY FLUE GAS CLEANING EXPERIENCE II
Co-Chairmen:
Jeffrey L. Hahn
Ogden Projects, Inc.
Berkeley, CA
Dennis C. Drehmel
Air and Energy Engineering Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC
Preceding page blank
69
-------
Intentionally Blank Page
70
-------
The work described in this paper was not funded by the U.S. Environmental
Protection Agency. The contents do not necessarily reflect the views oi the
Agency and no official endorsement should be inferred.
OPERATING EXPERIENCE AND EMISSION RATES OF
SCRUBBER BAGHOUSES, SCRUBBER ESP's AND FURNACE
INJECTION OF LIME AT RDF FACILITIES
J. Michael Smith
R. Michael Hartman
ABB Resource Recovery Systems
Windsor CT 06095-0500
ABSTRACT
This paper will focus on operational, economic and environmental
considerations of different air pollution control devices at three ABB
Resource Recovery Systems (RRS) refuse derived fuel (RDF) facilities
(Mid-Connecticut Facility in Hartford, CT, The Greater Detroit Resource
Recovery Facility, the H-POWER Facility in Honolulu, HI.).
The air pollution control devices are: (1) a dry scrubber absorber
followed by a fabric filter baghouse, (2) a dry scrubber absorber followed
by a five field ESP, (3) lime injection to the furnace followed by a hot
electrostatic precipitator.
Evaluating these three types of control devices is important because
of new EPA emission guideline regulations that mandate restricted
particulate, dioxin, and acid gas emission limits on municipal waste
combustion facilities (MWC) greater than 250 tons per day combustion
capacity. As a result of these new regulations, many existing MWC's will
be faced with requirements to upgrade their air pollution control systems.
I. INTRODUCTION
This paper will summarize the operational experiences of three
different air pollution control systems at three ABB Resource Recovery
Preceding page blank
-------
Systems designed and operated RDF facilities in the United States.
The differing air pollution control systems are largely the result of
differences in the air permitting process these facilities experienced, as
will be explained later.
The three RDF facilities that will be compared are the Mid-Connecticut
Resource Recovery Facility in Hartford Connecticut, the Greater Detroit
Resource Recovery Facility in Detroit Michigan, and the H-POWER Facility in
Honolulu Hawaii.
The summary of the key design features of these three facilities is
shown in Table 1. The steam generators in each are waterwall spreader
stokers with split traveling grates. Each uses three levels of tangential
overfire air and an undergrate air system that is split into five
compartments. The steam generator design is the basic Combustion
Engineering VU-40 design. Mid-Connecticut is designed to be able to burn
either coal or RDF, while the other two facilities are designed to burn
only RDF, with oil as the startup and auxiliary fuel.
Permit applications to construct the Mid-Connecticut and Detroit
facilities were submitted in the 1983-1984 time period. At that time,
there were no Federal emission standards except for particulates. The
permitting process was essentially controlled at the state level and was
undergoing rapid change with no uniformity of emission standards. In
Detroit, the state air permit was issued first in November 1984. The
principal focus of pollution control in that permit was on good combustion
conditions for control of organics and good control of particulates and
metals by use of a high efficiency five-field hot ESP. There was no
requirement in the Detroit permit for acid gas removal, although there were
emission limits for SO^ and HC1 (see Table 2 for a listing of relevant air
emission limits). There was a dioxin/furan limit, but there was no system
included for control for dioxin other than good combustion conditions.
The Mid-Connecticut Facility, however, was required by the state
permit process, when that air permit was issued in 1985, to add a
scrubber/baghouse for control of acid gases as well as for dioxin and furan
emissions.
The H-POWER project permit was issued in 1987 after the EPA
Administrator issued a landmark "Remand Decision" which, for the first
time, had the effect nationally of requiring scrubbers on all new municipal
waste combustors. By this time, it was becoming better known in the
scientific community that dry scrubber absorbers had a beneficial impact
for a wide range of toxic metal and organic pollutants and was, in fact, a
proven technology. Table 2 summarizes some relevant aspects of the H-POWER
air permit for comparison to those of the Detroit and Mid-Connecticut
facilities. This table does not contain all of the permit limits for these
facilities, only the relevant ones that depend on an air pollution control
system. Because work had already started on fabrication of an ESP for
72
-------
H-POWER when the 1987 EPA Remand Decision came in, the facility retained
its ESP's after the dry scrubber, although it was necessary to add an extra
field to the ESP's.
All of these three facilities are now in operation and have had at
least one set of air emission stack tests. These data will be summarized
when the paper is presented. Because this paper is focused on the effect
the air pollution control system has on conditions after the flue gas
leaves the boiler, the pollutants that will be focused on will be
particulates, acid gases (SC^ and HC1), volatile trace metals (Pb, Hg, Cd,
As) and dioxin/furan emissions after the particulate control device. The
paper to be given at the conference will also discuss the mercury and HC1
emission exceedance problems at the Detroit facility and the actions taken
to install a furnace lime injection system.
To complete the paper, the operational experience for each of these
back-end pollution control systems will be discussed. The discussion will
identify and compare differences in labor needs, maintenance costs, lime
usage, permit compliance experience and other relevant features.
73
-------
TABLE 1
KEY DESIGN FEATURES
Mid-Conn. Detroit
MSW Throughput
Design Capacity
Auxiliary/
Startup Fuel
Steam Generation
Rate and # of
Units
Energy Generation
Air Pollution
Control System
per Unit
2,000 TPD
Coal
Natural Gas
3 Units --
231,000 lbs/hr
each
68 MWe
Dry Scrubber
absorber with 5
compartment bag-
houses with 168
bags in each
compartment,
plus time,
temperature and
turbulence on
combustion
3,300 TPD
Oil
H-POWER
2,160 TPD
Oil
3 Units --
360,000 lbs/hr
each
65 MWe
Currently, lime
furnace injec-
tion with hot
ESP. Dry
scrubber absorber
and baghouse
in process of
being retro-
fitted, plus time,
temperature and
turbulence on
combustion
2 Units --
250,000 lb/hr
each
57 MWe
Dry scrubber
absorber with
5 field ESP,
plus time,
temperature
and turbulence
on combustion
74
-------
TABLE 2
RELEVANT PERMIT LIMITS
POLLUTANT
MID-CONNECTICUT
DETROIT
H-POWER
PCDD/PCDF
1.95 ng/Nm
2,3,7,8-TCDD Toxic Eq .
@ 12% CO.
0.0043 lbs/hr
7.767 ng/Nm
@ 12% C0o
None
PARTICULATE
0.015 gr/dscf
@ 12% CO.
0.019 gr/dscf
(a 12% CO.
0.015 gr/dscf
HC1
90% Control
and 50 ppmv
@ 12% C0o
294 lbs/hr
315 ppmv
@ 12% CO.
None
SO,
0.32 lbs/10 BTU
110 ppmv
@ 12% CO.
457.1 lbs/hr
282 ppmv
(a 12% CO.
30 ppmdv
@ 12% CO.
Pb
Hg
700 ug/m
(a 12% CO
240 ug/m
@ 12% CO
1.37 lbs/hr,
2333 ug/m
0.07 lbs/hr
120 ug/m
0.0028 lbs
per ton RDF
260 ug/m
0.0022 lbs/
ton RDF ,
203 ug/m
Cd
None
0.085 lbs/hr
143 ug/m
None
75
-------
Intentionally Blank Page
76
-------
REVIEW OF SD/FF VERSUS SD/ESP SYSTEM PERFORMANCE
FOR ACID CAS. PARTICULATE. AND CDD/CDF CONTROL
by: David M. White and Kristina L. Nebel
Radian Corporation
Research Triangle Park, NC 27709
ABSTRACT
Over the past five years, a number of new municipal waste combustors
(MWC's) in the U.S. have installed spray dryer (SD) systems followed by either
a fabric filter (FF) or an electrostatic precipitator (ESP). Initial
performance data indicated that SD/FF systems were capable of achieving lower
emission levels of sulfur dioxide (S02) , hydrogen chloride (HCl), particulate
matter (PM), and chlorinated dioxins and furans (CDD/CDF) than SD/ESP systems.
Review of additional data from several recent SD/ESP systems suggests that
these differences are smaller than the initial data indicated. This paper
reviews the available emissions data from several existing SD/FF and SD/ESP
systems and compares the emission reduction potential of both types of
systems. This paper is of relevance to existing MWC's that may be considering
retrofit of a SD system upstream of an existing ESP.
This paper has been reviewed in accordance with
the U.S. Environmental Protection Agency's peer
and administrative review policies and approved
for presentation and publication.
Preceding page blank
-------
BACKGROUND
The first SD/FF system on an MWC in the U.S. began operating in Marion
County, Oregon in 1986, and the first SD/ESP system began operating in 1988 at
Millbury, Massachusetts. Based on the anticipated emissions control
performance of SD/FF and SD/ESP systems and on the number of permits issued
for such systems, the U.S. Environmental Protection Agency (EPA) issued
operational guidance to permit writers on June 26, 1987, indicating that a SD
followed by either a FF or an ESP was "best available control technology"
(BACT) for MWC's under EPA's prevention of significant deterioration (PSD)
program. Also in 1987, EPA initiated development of emission standards for
new and existing MWC's. These standards were proposed in December 1989 (54 FR
52251) and were promulgated in February 1991 (56 FR 5488). This paper reviews
acid gas, PM, and CDD/CDF emissions data associated with these two control
technologies.
REVIEW OF AVAILABLE EMISSIONS DATA
As of the end of 1990, there were an estimated 66 MWC units in the U.S.
operating with either SD/FF or SD/ESP systems. These units account for a
combined waste combustion capacity of 39,000 tons/day (tpd). As shown in
Table 1, these include 48 units equipped with SD/FF systems and 18 units with
SD/ESP systems. The caustic used by all these units to control acid gas
emissions is a lime-based (i.e., calcium) slurry. The information presented
in this paper is based on compliance test and continuous emission monitor
(CEM) data collected from the MWC's noted by asterisks (*) in the four right-
hand columns of Table 1.
ACID GASES
The primary acid gases emitted by MWC's are S02 and HC1. Of these two
acids, HC1 has a stronger affinity for reaction with lime at typical SD
operating temperatures (250-300°F). As a result, removal efficiencies for HC1
across the SD/FF or SD/ESP system are always higher than for S02. Typical S02
and HC1 emissions and reduction efficiencies measured during compliance
testing are presented in Table 2. The data presented in Table 2 typically
reflects the arithmetic average of measurements made during three separate
test runs, each of which generally lasted three hours or less. Note that both
SD/FF and SD/ESP systems achieved stack HC1 emissions of less than 30 ppmv1
and greater than 97 percent HC1 reductions during compliance testing. S02
emissions, however, showed greater variability, with stack concentrations
ranging up to 70 ppmv and emission reductions on a plant-wide basis of 73 to
97 percent. In most cases, the S02 control efficiency is higher for SD/FF
systems than for SD/ESP systems.
Compliance test data indicate the level of emission reduction that a
control system is capable of achieving. To ensure that a control system is
operated to continuously achieve high levels of emissions reduction, however,
it is necessary to continuously monitor system performance. Both S02 and HC1
CEM systems are commercially available. However, HC1 monitors are more
expensive than SO, monitors and do not have the level of demonstrated
reliability that nas been achieved with S02 monitors. Based on the
limitations on HC1 CEM systems and on the ability to achieve high levels of
HC1 removal even at relatively low levels of S02 control, EPA based the
demonstration of continuous acid gas emission reductions on the use of S02
monitors alone.
xAll emissions concentrations reported in this paper are at either 7% 02
or 12% C02 and are on a dry-basis at standard temperature (68°F).
78
-------
To confirm the SO, reduction capability of SD/FF and SD/ESP systems
applied to MWC's, EPA obtained SO^ CEM data from four commercial MWC's. Three
of the facilities were equipped with SD/FF systems (York County, Pennsylvania;
Stanislaus County, California; and Bridgeport, Connecticut) and one was
equipped with an SD/ESP system (Millbury, Massachusetts).2 The data
indicated that short-term (e.g., hourly) concentrations of S02 measured before
and after the SD/FF or SD/ESP can be much higher than weekly or monthly
averages due to the heterogeneity of municipal waste and short-term
variability in control system performance. The median (i.e., 50th
percentile), minimum, and maximum hourly average inlet and stack S02 levels
measured at each of these facilities are listed in Table 3 (inlet S02 levels
were not monitored at Stanislaus County). Note that the maximum inlet S02
levels at the three MWC's with data are over three times the median value, and
that the maximum outlet concentrations are 4 to 70 times higher than the
median values. The higher median outlet level at Bridgeport, compared to
Stanislaus County and York County, reflects Bridgeport's less stringent S02
emission limit.
As shown in Figure 1 (based on inlet data collected at the Bridgeport
MWC), the hourly average SD inlet S02 data from all of the facilities were
found to be lognormally distributed. Analysis of hourly average S02 levels
measured after the control system found that stack emissions of S02 were also
lognormally distributed. Because of the lognormality of the S02 CEM data, a
feometric mean is statistically appropriate for estimating expected levels for
onger time periods. Analysis of geometric means derived from inlet and stack
S02 CEM data, for averaging periods of 3 hours to 7 days, found that a 24-hour
averaging period was sufficient to account for most of the short-term "spikes"
in S02, while being short enough to require operating personnel to carefully
monitor control system performance.
Analysis of the CEM data from York County and Millbury indicate that
carefully operated SD/FF systems are capable of continuously achieving greater
than 80 percent S02 reductions and that carefully operated SD/ESP systems can
continuously achieve greater than 70 percent S02 reductions. To continuously
achieve these levels of S02 reduction, however, it will be necessary to
operate with a system SO, control setpoint that will result in annual average
S02 reductions of over 90 percent for SD/FF systems and over 80 percent for
SD/ESP systems. The higher level of S02 reduction by SD/FF systems reflects
the additional S02 reduction achieved by the filter cake on the FF bags.
PARTICULATE MATTER
Table 4 summarizes PM compliance test data collected during individual
runs at several representative SD/FF and SD/ESP systems. In most instances,
the PM emission rate measured during individual runs from both types of
systems were less than 0.015 gr/dscf. The only exceptions were during single
runs at Biddeford and Marion County, both of which are equipped with SD/FF
systems, and at SEMASS and Millbury, both of which are equipped with SD/ESP
systems. Except for one of the eight compliance tests conducted at Marion
County, PM emission rates based on the average of three individual runs were
less than 0.015 gr/dscf in all cases.
Based on review of the data from individual runs, elevated PM emissions
during a compliance test can occur either as the result of higher emissions
during all runs or during a single run. For example, above average PM
emissions were measured during each of the runs at Biddeford and during each
of the runs associated with the above compliance test at Marion County. The
2HCl CEM data were also collected at the Millbury MWC. These data
indicated HCl reductions were almost always greater than 97 percent, except
during periods when high inlet concentrations of HCl or S02 caused HCl
reductions to decrease to 90-97 percent (based on an 8-hour averaging period).
79
-------
attributable to the relatively high design net air-to-cloth ratio (5.2
acfm/ft2) of the unit's pulse-jet FF. The higher PM emission rate (averaging
0.016 gr/dscf) during one of eight Marion County tests may indicate a bag leak
during this test. One the other hand, the higher PM emission rate (0.020
gr/dscf) measured during the single run at Millbury may have been due to an
upset condition in the ESP or combustor during this run. During 20 other test
runs conducted at Millbury, the next highest PM emission rate was 0.005
gr/dscf.
At the other end of the spectrum, PM emissions of less than 0.003
gr/dscf have been measured at SD/FF systems and at several new SD/ESP systems
(e.g., the 5-field ESP in Honolulu and 4-field ESP in West Palm Beach) that
are equipped with large ESP's able to collect fine PM and to minimize PM
reentrainment and bypass. Based on these data, the PM control performance of
SD/FF and SD/ESP units depends on system design and operation. As a result,
there is no clear distinction in the ability of recently built SD/FF and
SD/ESP systems to control total PM.
DIOXINS AND FURANS
Table 5 summarizes the CDD/CDF data collected from several SD/FF and
SD/ESP systems. Both SD/FF and SD/ESP systems have demonstrated the ability
to achieve CDD/CDF emissions of less than 30 ng/dscm, with several facilities
achieving emissions of less than 10 ng/dscm. However, two of the MWC's
equipped with SD/ESP systems (Millbury and SEMASS) have experienced relatively
large variations in CDD/CDF emissions between individual test runs. For
example, the SEMASS MWC had CDD/CDF emissions between 5 and 18 ng/dscm during
five of six runs, but measured 907 ng/dscm on the sixth run. At Millbury,
five of the six runs were between 40 and 70 ng/dscm, while the sixth run was
103 ng/dscm. Based on available data, it is not possible to determine whether
the higher CDD/CDF level measured during these individual runs are caused by
an upset condition in the combustor (causing higher CDD/CDF levels at the
combustor exit) or in the SD/ESP system (resulting in reduced CDD/CDF
collection efficiency). Similar large variations in stack concentrations of
CDD/CDF have not been identified at MWC's using SD/FF systems, however. This
suggests that the filter cake on FF bags may enhance the ability of SD/FF
systems to handle upset operating conditions in the combustor and SD system.
Based on CDD/CDF levels measured at the combustor outlet at Commerce,
Marion County, Mid-Connecticut, and the Quebec City pilot-scale tests, SD/FF
systems have been able to reduce CDD/CDF emissions by more than 95 percent,
and are frequently in excess of 99 percent. The only MWC with a SD/ESP that
has simultaneously measured both inlet and outlet CDD/CDF levels is Millbury.
During testing at Millbury, CDD/CDF reductions were between roughly 50 and 70
percent, with an average of approximately 65 percent. Based on these data, it
appeared that SD/ESP systems were less efficient at CDD/CDF collection than
SD/FF systems. However, subsequent testing conducted at several RDF-fired
MWC's with SD/ESP systems having 4 or 5 fields (versus Millbury which has 3
fields) measured lower CDD/CDF emissions, suggesting that removal efficiencies
from properly designed and carefully operated SD/ESP units can approach levels
achieved by SD/FF systems.
CONCLUSIONS
The emissions control performance of both SD/FF and SD/ESP systems
depends on proper design and operation. Based on compliance test data
collected from recently built MWC's with SD/FF or SD/ESP systems, the PM and
CDD/CDF control performance of both systems can be comparable. However,
SD/ESP systems appear to be more sensitive to upsets in operating conditions
than SD/FF systems, and may experience noticeable increases in PM and CDD/CDF
emissions during these upsets. The failure to detect similar "spikes" in PM
and CDD/CDF emissions from SD/FF systems may reflect the ability of the FF
80
-------
filter cake to moderate short-term duration increases in the two pollutants.
Because most of the reaction between HC1 and sorbent occurs before the
flue gas enters the PM control device, HC1 removal efficiencies by SD/FF and
SD/ESP systems appear to be similar. However, a significant fraction of S02
removal occurs in the PM control device. Because of the S02 removal by the
filter cake in a FF, the S02 emissions from a SD/FF are roughly one-half those
associated with a SD/ESP.
REFERENCES
U.S. Environmental Protection Agency. Municipal Waste Combustors --
Background Information for Proposed Standards: Post-Combustion Technology
Performance. Research Triangle Park, North Carolina. August, 1989. EPA-
450/3-89-27c.
U.S. Environmental Protection Agency. Municipal Waste Combustors --
Background Information for Promulgated Standards and Guidelines: Summary of
Public Comments and Responses. Appendices A and C. Research Triangle Park,
North Carolina. November 1990. EPA-450/3-91-004.
Individual compliance test reports from Hempstead, Portland, Charleston, West
Palm Beach, and Honolulu.
Sussman, D. B. (Ogden Martin Systems). Testimony Before the National Air
Pollution Control Techniques Advisory Committee. Research Triangle Park,
North Carolina. January 31, 1991.
81
-------
TABLE 1. EXISTING U.S. MWC'S EQUIPPED WITH SD/FF OR SD/ESP SYSTEMS
Capacity -Compliance Test Data- S02
Control No. of Per Unit Startup Acid CEM
Location State Type Units (tpd) Date Gases PM CDD/CDF Data
Marion County
OR
SD/FF
2
275
1986
*
*
*
Biddeford
ME
SD/FF
2
350
1987
*
*
*
Commerce
CA
SD/FF
1
300
1987
*
*
*
Bridgeport
CT
SD/FF
3
750
1988
*
Bristol
CT
SD/FF
2
325
1988
*
Mid-Connecticut
CT
SD/FF
3
667
1988
*
*
*
Indianapolis
IN
SD/FF
3
787
1988
*
*
*
Long Beach (SERRF)
CA
SD/FF
3
460
1988
*
*
*
Penobscot
ME
SD/FF
2
360
1988
*
*
*
Stanislaus County
CA
SD/FF
2
400
1988
*
*
*
Babylon
NY
SD/FF
2
375
1989
*
*
*
Hennepin County
MN
SD/FF
2
600
1989
Broward County (North)
FL
SD/FF
3
750
1990
Broward County (South)
FL
SD/FF
3
750
1990
Fairfax County
VA
SD/FF
4
750
1990
*
Gloucester County
NJ
SD/FF
1
575
1990
*
Hempstead
NY
SD/FF
3
773
1990
*
*
*
Huntsville
AL
SD/FF
2
345
1990
*
Kent County
MI
SD/FF
2
312
1990
*
York County
PA
SD/FF
3
448
1990
48
26714
Millbury
MA
SD/ESP
2
750
1988
*
*
*
Portland
ME
SD/ESP
2
250
1988
*
*
*
Charleston
SC
SD/ESP
2
350
1989
*
*
*
Haverhill
MA
SD/ESP
3
550
1989
*
SEMASS
MA
SD/ESP
2
950
1989
*
*
*
West Palm Beach
FL
SD/ESP
2
1000
1989
*
*
Essex County
NJ
SD/ESP
3
750
1990
Honolulu
HI
SD/ESP
2
900
1990
*
*
*
18
12300
-------
TABLE 2. SELECTED ACID GAS COMPLIANCE TEST DATA
Stack Cone.
Control (ppmv) % Reduction
Location Type S02 HC1 S02 HC1
Marion County
Biddeford
Commerce
Mid-Connecticut
Stanislaus County
Millbury
Portland
SEMASS
West Palm Beach
SD/FF
31
SD/FF
22
SD/FF
3
SD/FF
32
SD/FF
4
SD/ESP
58
SD/ESP
40
SD/ESP
61
SD/ESP
33
18
85
97
6
76
99
6
97
99
12
83
97
2
90
15
76
98
87
65
16
73
97
TABLE 3. SUMMARY S02 CEM STATISTICS
Location
Bridgeport
Stanislaus County
York County
Millbury
Control
Sampling
1-Hour
Average
(ppmv)
Type
Location
Median
Minimum
Maximum
SD/FF
Inlet
155
50
617
Outlet
33
11
286
SD/FF
Outlet
5
0
169
SD/FF
Inlet
102
3
381
Outlet
9
1
377
SD/ESP
Inlet
173
55
547
Outlet
35
2
227
83
-------
TABLE 4. SELECTED PM COMPLIANCE TEST DATA
Control Sampling Run Average (gr/dscf)
Location Type Location Average Minimum Maximum
Biddeford
Bridgeport
Marion County
Mid-Connecticut
Stanislaus County
Honolulu
Millbury
SEMASS
West Palm Beach
SD/FF
Outlet
0.0140
0.0095
0.0190
SD/FF
Outlet
0.0022
0.0004
0.0130
SD/FF
Outlet
0.0073
0.0030
0.0211
SD/FF
Outlet
0.0030
0.0017
0.0059
SD/FF
Outlet
0.0039
0.0011
0.0086
SD/ESP
Outlet
0.0016
0.0007
0.0027
SD/ESP
Outlet
0.0045
0.0004
0.0201
SD/ESP
Outlet
0.0100
0.0070
0.0170
SD/ESP
Outlet
0.0013
0.0003
0.0028
TABLE 5. SELECTED CDD/CDF COMPLIANCE TEST DATA
Control Sampling Run Average (ng/dscm)
Location Type Location Average Minimum Maximum
Babylon
SD/FF
Outlet
21.9
12.6
27.2
Biddeford
SD/FF
Outlet
4.4
3.5
5.2
Mid-Connecticut
SD/FF
Outlet
0.7
NDa
1.4
Stanislaus Co. #1
SD/FF
Outlet
6.3
4.6
8.9
Stanislaus Co. #2
SD/FF
Outlet
6.5
5.0
8.5
Honolulu #1
SD/ESP
Outlet
9.9
4.1
16.7
Honolulu #2
SD/ESP
Outlet
2.9
1.5
3.8
Millbury
SD/ESP
Outlet
59.2
40.4
103.0
SEMASS #1
SD/ESP
Outlet
9.3
5.1
13.6
SEMASS #2
SD/ESP
Outlet
311.0
6.6
907.0
aNot detected
84
-------
400-1
45 75 105 135 165 195 228 255 205 315 345 375 408 438 465 495 525 555 585 615
Inlat 80S Concentration (ppa)
300-1
3.97 4.12 4.28 4.43 4.88 4.73 4.88 5.03 5.18 5.33 5.48 8.83 5.78 5.83 8.08 8.23 8.38
Natural Log of Xnlot 808 Concentration
Figure 1. Distribution of Hourly Average inlet S02
Concentrations at Bridgeport
85
-------
Intentionally Blank Page
-------
The work described in this paper was not funded by the U.S. Environmental
Protection Agency. The contents do not necessarily reflect the views of the
Agency and no official endorsement should be inferred.
REDUCTION OF DIOXIN AND HEAVY METALS
BY SEMI-DRY SCRUBBER AND BAGHOUSE FROM MSV INCINERATOR
by : E. SHIBUYA, T. NAKAO
NKK CORPORATION
2-1 Suehiro-cho, Tsuruini-ku, Yokohama, 230, Japan
ABSTRACT
This paper presents the result of investigations about PC-DDs/PCDFs and
heavy metals discharged from MS'- incineration plants.
As to the PC-DDs/PCDFs, inlet tenperature of the baghouse appeared to
be most important, parameter. To control the PC-DDs/PCDFs, lo,;er tenperature
is better. And dust load on the fabric filter is important paraneter too.
The heavier dust load on the fabric filter, the better the removal of the
PCDDs/PC-DFs.
As to the heavy metals, Pb, Cn, Zn, are removed (¦•ell at the baghouse
and the removal of them depends on the dust load on the fabric filter same
asPCDDs and PCDFs. Hot-ever, it is difficult to remove Hg at baghouse in
spite of lov temperature and heavy dust load on the fabric filter.
INTRODUCTION
The problem of dioxin discharged from refuse incinerators is attracting
attention on a global scale. The refuse incineration rate in Japan is high
at 72 percent in 1989. The development of technologies for controlling
dioxins is a matter of utmost urgency.
Vith this background, research is under i-ay on tvo items, namely,
technologies for controlling the generation of dioxins in the incineratior
and technologies for removing dioxins. This paper presents removal
efficiency of dioxins andheavy metals by the semi-dry scrubber and baghouse.
Preceding page blank
-------
EXPERIMENT CONDITIONS
The dioxins and heavy netal contents were investigated at the inlet and
outlet of a serai-dry scrubber and baghouse installed in stoker type
incinerator. Two refuse incineration plants A and B were selected as
objectives of the investigation.
(1) Refuse incineration plant A is made up of tvo continuous operating
incinerators each having an incinerating capacity of 150 t/d. This
plant is equipped with semi-dry scrubbers and electric precipitators.
One of the two incinerators is so arranged that a fullscale baghouse,
exhaust gases flowing is 10000 Nra:/H ,is installed in parallel along
ESP flow .
(2) I'iaste incineration plant B is made up of one continuous operating
incinerator vith an incinerating capacity of 150 t/d. This plant is
equipped with semi-dry scrubbers and baghouse.
A simplified flow is shown in Fig.l.
PCDDs/PCDFs ANALYTICAL METHODS
Flue gas samples were taken from the duct isokinetically using the 5-
train-method modified I'S EPA with the dust collector settled outside of the
duct. The sample preparation consist of natrix-specific extraction (Soxhlet
for solid samples after 2N-HC1 treatment of particulate, liquid-liquid for
aqueous samples by toluene), acid-base partitioning followed by silicagei
and basic alumina column chromatography. 1:C;:-2,3,7,8-TCDD, :~C:;-2,3,7,8-
TCDF, i;,C —-0C-DD, ": C ± - - 0 C D F were added to samples after the extraction as
the recoverey standards during the analytical procedure. PCDDs/PCDFs isomer
specific analysis was performed on SI'1 by HRGC-/MRMS with a DB-5 and SP- 2331
fused-silica capillary columns.
For quantification of PCDDs/PCDFs, the two most abundant ions were
monitored, and the peaks in sample chromatograms where the> isotopic
ratios of each twoions are closed to that in native 2,3,7,8-substituted
PCDDs/PCDFs standard, were identified and quantified as PCDDs/PCDFs by
comparison ofpeak area.
RESULTS AND CONSIDERATIONS
DIOXINS
Bag filter temperature and dioxins
It has been reported by Yamagishi...(1),(2)that there is a close
relationship between the bagfilter inlet temperature and dioxins. The
relationship of baghouse inlet temperature to outlet PCDDs/PCDFsf2,3,7,8-
aa
-------
TCDDeq, Internatirial TEF) is sho'- n in Fig. 2. As can be seen in Fig. 2, the
lover the baghouse inlet temperature, the lover the dioxins . The relation
ship between dioxins removal rate and baghouse inlet temperature is shovn in
Fig.3.
Dioxin removal rate = —— x 100 {%)
DXNin
DXNin PCDD/PC-DF at inlet of bag filter (Hg/Nm3)
DXNoutl PCDD/PCDF at outlet of bag filter (Hg/Nm3)
As can be seen in Fig.3, the lover the baghouse inlet temperature,the
higher the removal rate. The reason is given as the lover the temperature,
the higher the susceptibility of dioxins to adsorb on the dust layer.
The relationship of PCDDs/PCDFs(2,3,7,8-TCDDeq,Internatinal TEF) in fly
ash to baghouseinlet temperature is shovn in Fig.4. The figure shovs that
the lover the temperature, the higher the dioxins concentration. This
indicates that the lover the temperature, dioxins contained in the exhoust
gas is easier to remove and it is further considered that the shift into the
dust is also made easier. Several papers (i),(2) have already reported that
production of dioxins takes place in dust collectors as the dust collector
temperature rises to the 300 C level and the dioxins concentration rises.
Accordingly, Fig.4 vill also shov that the dioxin concentration rises vhen
the baghuse inlet temperature is raised near the 300 C.
Dust load and dioxins
Dust load is indicated by the amount of dust layer accumulated on the
baghouse and expressed by the following equation.
1 |
Dust load = C x 10"- x 0 x -t— x x T ,, , M
A bO (kg/cm-)
C Dust concentration (g./Nnr)
Q I Exhaust gas volume (Nm:/h)
A ' Area of filter cloth (m~)
T Back vash interval (min)
The comparatively large particles contained in dust descends in the
baghuse by gravitation. Therefore, the dust load obtained by the equation
given above does not express the accurate amount of dust caught by the
fablic filter, but it is capable of indicating the extent of dust layer. The
relationship of dioxins removal to dust loadis sho^n in Fig.5. The figure
shovs that the larger the accumulation of dust load, that is, thenore dust
accumulated on the fablic filter, the higher the dioxins removal rate.
89
-------
The relationship between dust load and PCDD/PCDF(2,3,7,8-TCDDeq,Inter
natinal TEF) in fly ash is shown in Fig.6. It can be seen that the higher
the accumulation of dust load, the higher dioxin concentration in dust.
This indicates that the dust layer has become thicker and has raised the
possibility of catching and adsorbing particles of dioxins or gaseous
dioxins in exhaust gas.
HEAVY METALS
Pb, Zn, and Cd
The relationship of Pd, Zn, and Cd removal efficiency to baghouse inlet
temperature is shown in Fig.7 to 9. The removal efficiency mentioned here
is referred to as the extent of removal of heavy metals at the inlet and
outlet of semi-dry scrubber. The lower the baghouse inlet temperature,the
removal efficiency of Pb and Zn tends become higher. However, Cd, irrespec
tive of temperature, indicates a high removal efficiency of notless than 90
percent.
The relationship between the removal efficiency of Pb, Zn, and Cd
and dust load is shown in Fig. 10 to 12. As the figures show, the higher
the dust load, the higher the removal efficiency. However, the removal
efficiency of Cd, irrespective of dust load, is greater than 90 percent.
Hfi
The relationship between Hg removal efficiency, baghouse inlet
temperature, and dust load is shown in Fig.13 and 14. The figures show large
fluctuations in Hg removal efficiency values. The burning of refuse in
incinerators causes the Hg concentration in exhaust gas to fluctuate sharply,
but the fluctuations may have been amplified since the measurements were
taken in this experiment in batches instead of using a continuous meter.
CONCLUSION
The experimental results indicated that the following items i«ere found
to be important in inhibiting the generation of dioxins and heavy metals.
1) Maintaining the exhaust gas temperature at the inlet of baghouse as low
as possible.
2) Maintaining the dust load as high as possible.
However, removal efficiency of not less than 90 percent can be
obtained for Cd without observing the conditions of 1) and 2) given above.
Since there are sharp fluctuations in the Hg data, definite conclusions
could not be obtained from the experiment.
90
-------
The i-'ork described in this paper vas not funded by the U.S.
Envirnmental Protection Agency and therefore the contents do
not necessarily reflect the vie^s of the Agency and no
official endorsement should be inferred.
REFERENCES
1. Yamagishi,M.,Shibuya,E.,"Conyrol of PCDD/PCDF Emissions from Waste to
Energy Plant",9th International Symposium on Chlorinated Dioxins and
Related Compound,1989
2. Yamagishi,MShibuya,E"Simultaneous Control of Dioxins and NOx in
Municipal Waste Incinerator",10th International Symposium on Chlorinated
Dioxins and Related Compound,1990
91
-------
Gas cooling
system
Gas cleaning system
PLANT A
150t/24h X 2units
LCV=1500~1900kCal/kg
-» IDF
PLANT B
150t/24h Xlunit
LCV=1500~2500kCal/kg
Waste heat
boiler
Stack
IDF
ESP
Stack
Baghouse
Baghouse
Waste heat
boiler
Ca(0H)2 slurry atomizing
seni-dry scrubber
Ca(0H)2 slurry atomizing
seni-dry scrubber
Fig.l Plant flow diagram
Cx_
CjJ
C^i
CO
PLANT A |
- PLANT B I
o
o
o
O
•
k
o
o
•
o
o
o
0
120 140 160 180 200 220
Dust collecter inlet temperature CC)
Fig.2 Dioxins VS dust collecter inlet temp.
92
-------
100
o
CD
X
o
(50)
PLANT « PLANT 1
(100)
180
200
220
120
140
160
Dust collecter inlet temperature (°C)
|: Inlet dioiins < 20t>nj/Nn3
Fig.3 Dioxins removal VS dust collecter inlet temp.
1 1
PLANT A PLANT B I
1
0
c
o
o
»
•
O °
• •
120 HO 160 180 200 220
Dust collecter Inlet temperature (°C)
Fig.4 Dioxins in fly ash VS dust collecter inlet temp.
93
-------
~ ~
~
*
*
/
*
PLANT A
o
PLANT B
•
Dust load (kg/m2)
|; Inlet dioxins < 200nj/Kn3
Fig.5 Dioxins removal VS dust load
PLAKT A
0
PLAKT B
•
o
O O
o
•
o
o
•
•••
•
0.1 0.15 0.2 0.25 0.3
Dust load (kg/m2)
Fig.6 Dioxins in fly ash VS dust load
94
-------
100
"=> 60
40
20
o
o
0
o
PLANT k PLANT B
o •
1 1
0
120 140 160 180 200 220
Dust collecter inlet temperature CO
I •' Inlel lejd < 5o{/Nr3
Fig.7 Lead removal VS dust collecter inlet temp.
100
60
40
20
> f-
» o
o
~
4-
o
o
PLANT k PLANT fi
o •
T_—pi
0
120 140 160 1 HO 200 220
Dust collecter inlet temperature TO
I : Inlet zinc < i5og/Ni>3
Fig.8 Zinc removal VS dust collecter inlet temp.
95
-------
80
« 60
EE
CD
40
'O
20
0
120 140 160 180 200 220
Dust col teeter inlet temperature (°C)
Fig.9 Cadmium removal VS dust collecter inlet temp.
0
c
o
o
PLANT A PLAKT B
o •
, r,-r-P.
0 0.05 0.1 0.15 0.2 0.25 0.3
Dust load (kg/m2)
I • Inlet leid < 5ng/Nn3
Fig.10 Lead removal VS dust load
96
-------
o
o
*
~
o
PLANT A PLANT
O •
1 "I 11
8
0 0.05 0.! 0.15 0.2 0.25 0.3
Dust load (kg/m2)
I; Inlet zinc < I5»j/>ln3
Fig.11 Zinc removal VS dust load
<0»
o
o
PLANT A PLANT B
O •
r™i 7
0 0.05 0.1 0.15 0.2 0.25 0.3
Dust load (kg/m2)
Fig.12 Cadmium removal VS dust load
97
-------
o
PLANT A I
°
PLANT 8 I
2J
i
o
o
o
•
•
•
•
o
i
A
•
•
0" * ¦ 1 ¦ 1 ¦ - 1 * t"D 1
120 140 160 180 200 220
Dust collecter inlet temperature CC)
Fig.13 Mercury removal VS dust collecter inlet temp.
PLANT A PLAMT B
O •
o
» o
o
o •
o
5
•
{
* ..
•
3
0 0.05 0.1 0.15 0.2 0.25 0.3
Dust load (ks/m2)
Fig.14 Mercury removal VS dust load
98
-------
SESSION 3C:
WET AND DRY FLUE GAS CLEANING EXPERIENCE
Co-Chairmen:
Marjorie J. Clarke
Environmental Consultant
New York, NY
David S. Beachler
Resource Energy Systems Division
Westinghouse Electric Corporation
Pittsburgh, PA
99
-------
Intentionally Blank Page
100
-------
The work described in this paper was not funded by the U.S. Environmental
Protection Agency. The contents do not necessarily reflect the views oi the
Agency and no official endorsement should be inferred.
GAS CLEANING IN CONNECTION
WITH WASTE INCINERATION
by: Bjorn Lindquist
BOLIDEN CONTECH AB
Skellefted, Sweden
ABSTRACT
The Boliden process for gas treatment after waste incineration is based upon general
experience in treating smelter and roaster gases and industrial waste waters at the Boliden
smelters in Sweden, and is adapted to the specific conditions by waste incineration.
The Boliden technique is worked out to give:
- low emissions to air and water
- products that can be easily handled, stored or utilised
- high reliability
- minimum service demand
- easy operation and maintenance
- low consumption of energy and chemicals
These demands are fulfilled in the Boliden wet gas treatment process which consists of
gas washing in a scrubber, separating the fine dust and aerosols in the wet electrostatic
precipitators, and eliminating mercury in a selenium filter.
Preceding page blank
-------
The fly ash which is collected in the scrubber is leached in the scrubber liquid and
separated in a filter, and the water bleed is purified in a waste water treatment plant.
BASIC CONDITIONS
The Boliden process for gas treatment after waste incineration is based upon general
experience in treating smelter and roaster gases and industrial waste waters at the Boliden
smelters in Sweden. The process has been introduced at other sites. The system is adapted to
the specific conditions encountered when gases emanate from such an inhomogenous source
as waste incineration.
This process, wet gas treatment according to the Boliden system, consists of units
designed in order to meet the environmental demands for dust, hydrochloric acid and other
acidic compounds, dioxins and mercury in the gas leaving the plant; for heavy metals and
other contaminants in the effluent water; and for composition and stability of the solid waste
products.
Our technique is worked out to give:
- low emissions to air and water
- products that can be easily handled, stored or utilised
- high reliability
- minimum service demand
- easy operation and maintenance
- low consumption of energy and chemicals
These demands are fulfilled in our wet gas treatment process which consists of gas
washing in a scrubber, separating the fine dust and aerosols in the wet electrostatic
precipitators, and eliminating mercury in a selenium filter. The fly ash which is collected in the
scrubber is leached in the scrubber liquid and separated in a filter, and the water bleed is
purified in a waste water treatment plant.
We deliver complete turn-key plants including design, fabrication, delivery, erection,
commissioning, start-up, and initial operation of the plant, including documentation and
education of operators and maintenance personnel. In the plant there are systems for cooling
and purification of the incoming waste gas; for handling of chemicals and waste products, for
water and effluent water; electricity; process control; insulation and surface treatment.
PROCESS DESCRIPTION
A flow sheet of the process is shown in figure 1.
The gases from the incineration furnace pass a waste heat boiler and an economizer
before they enter the scrubber through the central gas inlet nozzle.
102
-------
In the scrubber most of the solid impurities and the acidic compounds are removed from
the gas. The pH of the scrubber liquid is automatically controlled by addition of limestone
from silo into the pump tank.
Since the dust load is very high, scrubber liquid is bled from the system in order to
avoid too high slurry concentrations. This bleed goes to a reaction tank where it is mixed with
fly ash collected in the boiler and economizer. The soluble compounds are dissolved in the
acidic liquid, and then the solids are removed in a filter. The leached fly ash, which is returned
to the furnace, has a low content of zinc, cadmium and other noxious metals. A part of the
filtrate is taken as a bleed to the water treatment, while the rest is returned to the scrubber.
The gases from the scrubber continue to the Editube wet electrostatic precipitators,
where aerosol and very fine dust particles, containing noxious matters such as hydrochloric
acid, dioxins and other chlorinated organic compounds, are removed from the gas. The
precipitator consists of a GRP inlet chamber and a stainless steel tube bundle with negatively
charged needle point emission electrodes concentric in each tube. The dust and aerosol
particles are charged by corona emissions and collected at the tube surfaces.
The precipitator also acts like a gas cooler, and the low steel temperature together with
the dilution of the impurities by the condensate forming inside the precipitator tubes, ensures
that metal corrosion is kept at a minimum. The clean gas leaves the precipitator by way of the
upper chamber and the gas outlet nozzle. The dirty condensate collects in the bottom cone and
drains out through the condensate nozzle. The Editube precipitator may be integrated with a
heat recovery system: cooled by having a closed circuit that in its turn is cooled by sea water
or cooled with water which is cooled by air in cooling towers.
After reheating the gas in order to avoid condensation, elemental mercury is removed in
a selenium filter, and the purified gas is led to the stack.
The water bleed is treated in two steps in a water treatment plant. In the first step,
sodium sulphide is added and the heavy metals are precipitated as sulphides and removed from
the system in a centrifuge. Either sodium sulphide solution may be delivered in sufficient
quantities from a nearby pulp mill to a storage tank, or the solution is prepared from a solid
product. The sulphide sludge may be processed for recovery of its metal values.
In the second step ferrous hydroxide is precipitated by increasing the pH to between 7
and 8, and this precipitation removes the rest of the contaminants not 100% removed to the
first step where the iron goes into solution again, and therefore only the iron losses from the
system have to be compensated.
103
-------
The offgas will have the following maximum concentrations based upon a dry gas with
10% C02:
TABLE 1. EMISSION TO AIR
Dust content
10 mg/Nm3
HC1
10 mg/Nm3
Hg total
10 Jig/Nm3
Dioxins
0.1 ng/Nm3
The purified gases are reheated in order to avoid condensation in the selenium filter and
in the stack.
The waste water will be treated in a multi-stage process. The effluent water will contain
the following maximum concentrations:
TABLE 2.
EMISSION TO WATER
Zn
0.1 mg/1
Pb
0.01 mg/1
Hg
2 M-g/1
F
30 mg/1
CaCl2
10-100 g/1
PH
7-9
REMOVAL OF FINE DUST, MIST AND DIOXINS
Removal of these impurities is done in the Editube wet electrostatic precipitator.
MATERIAL CHOICE
The standard materal chosen in the Editube wet electrostatic precipitator is the stainless
steel Avesta 254 SMO which has excellent corrosion properties in chloride containing acidic
solutions. If corrosion tests indicate that other metallic materials (for instance Sandvik Sanicro
28 or Titanium grade 2) may be more suitable, we will deliver the units accordingly.
BASIC DESIGN
In order to get a low-cost, easy transportable precipitator, it was chosen to produce it as
a modular-shaped unit. If a higher gas capacity is needed that can be handled in one unit, the
required number of units are put in parallel.
104
-------
Each precipitator unit consists of gas inlet and outlet chambers, put together by a shell-
and-tube bundle, see figure 2. Each chamber has a gas nozzle fitted with built-in shut-off
damper. The lower (inlet) chamber is also equipped with a perforated plate in order to get a
uniform distribution of the gas into the tubes. This plate also acts as a floor for personnel
entering the precipitator.
The gas passes 90 grounded tubes (in the standard unit) which act as collecting surfaces
for the impurities to be separated. The tube bundles are made of stainless steel or titanium,
while the chambers may be in other materials as glass-reinforced plastics. In the case that the
chambers are in other materials than the tube bundles, the parts are fitted with flanges and
bolted together, otherwise the construction is all-welded.
The tube bundle is surrounded by a shell of the same material, and when cooling water
is connected at the shell side, the unit is also acting as a shell-and-tube heat exchanger.
Besides that this gives the unit a gas cooling capacity, it also has the advantage that the metal
surface in contact with the gas is kept at a low temperature and is wetted with a condensate
film which dilutes and continuously washes away the impurities that are collected on the
surface, and this keeps the corrosion rate at a minimum. As an extra safety, the unit is
equipped with an automatic flushing system. The flushing intervals can be adjusted by a timer.
Concentric in each circular tube there are negatively charged emission electrodes of a
particular design which creates higher current density than conventional discharge electrodes.
The electrodes are made of the same material as the tube bundle.
The emission electrodes are positioned by upper and lower frame-works. The upper
frame is supported by beams which are introduced into the upper chamber through gas-tight,
electrically heated, insulator compartments.
The high-voltage current to the electrodes comes from a rectifier-transformer unit placed
at the top of the precipitator. The local control cabinet is equipped with a kV-panel, a mA-
panel, a spark-counter and an operating-timer.
CAPACITY
The Editube wet electrostatic precipitator is designed as a standard unit, and is capable
of cleaning gas flows between 20-40 000 Nm3/h on the gas composition. Only one
purification step is necessary, while conventional precipitators operate two in series for the
same duty. The clean gas has a dust content of 10 mg/Nm3 or lower.
The cooling capacity of the unit is about 0.8 MV.
DELIVERY
The Editube wet electrostatic precipitator is delivered by road or rail to the customer as a
complete, workshop-fabricated and tested unit which is ready for operation after only a few
days for erection and connection to the gas, cooling water, flushing water and electric
systems.
105
-------
DIOXIN REMOVAL EFFICIENCY
Measurements with dioxin removal in the Editube wet electrostatic precipitator has given
the following results:
TABLE 3. DIOXIN REMOVAL EFFICIENCY
Inlet conc.
Oudet conc.
Separation
ng/Nm3 TCDD-eqv.
(N)
%
46.23
0.07
99.85
56.0
0.17
99.7
19.7
0.06
99.7
The over-all aerosol removal efficiency was 98.9%, which surprisingly is lower than
the dioxin removal efficiency.
More than 98% of the aerosol particles which remain in the gas after treatment have
diameters >15 fim. The removal efficiency of the particles >15 fim is only about 75% in the
wet electrostatic precipitator, while there is an almost complete removal of the smaller
particles. However, the concentration of dioxins is lower on the >15 fim particles than on the
veiy small particles, which explains why the dioxin removal is higher than the general particle
removal.
If all particles >15 |im are already removed from the gas in an efficient scrubber system
or in a vane separator, this gives a total aerosol removal efficiency, and consequently dioxin
removal efficiency, which is much higher than in the table above. It is therefore possible to
guarantee a maximum dioxin content of 0.1 ng/Nm3 in the gas after treatment.
MERCURY REMOVAL AND RECOVERY
The growing use of mercuiy-based batteries, e.g. in personal electronic equipment, will
inevitable increase the residual Hg content of municipal wastes. This is catered for in the
Boliden plant, which incorporates a reactive filter first developed for and now widely used in
metallurgical smelting operations. This consists of a cylindrical shell containing graded porous
material impregnated with selenium, which has a strong affinity for mercury, see figure 3. In
waste-gas cleaning practise, gas from the electrostatic precipitator - reheated to avoid dewpoint
effects - passes through the selenium filters before being discharged to the stack. The filter
cartridges have lines of several decades in municipal systems and, when spent, are replaced on
a favourable exchange basis.
106
-------
ASH TREATMENT
Most of the flue dust from the furnace off-gas is retained in the waste-heat
boiler/precipitator/economizer system which forms part of a normal waste-incineration plant.
Any dust still entrained in the gas leaving the economizer is removed as it passes through the
scrubber and wet electrostatic precipitator.
It is not possible to recycle the generated flue-dust directly to the incineration furnace,
because it contains such heavy metals as zinc and cadmium; or, for environmental reasons, is
disposal of the dust to land-fill a viable alternative.
Boliden has therefore developed a system whereby the dust is leached of heavy metals,
leaving only the residue to be recycled to the furnace. Leach residue is separated out by a band
filter. The filtrate is treated with sodium sulphide, and sludge with a high heavy-metal content
is formed which is removed by a centrifuge. This sludge, which represents only 2% by
weight of the flue dust, can be re-utilized in a metal-processing plant such as a zinc or copper
smelter. As a result there is no flue dust remaining to be disposed of.
UTILITIES DEMANDS
The following approximate figures will apply for a plant treating 50 000 Nm3/h:
TABLE 4. UTILITIES DEMANDS
Electric power
300 kW
Limestone 1-100 (im
400 tons per year
Sodium sulphide (60%)
100 tons per year
Polymer
0.8 tons per year
Iron sulphate (18% Fe)
2 tons per year
Sodium hydroxide (50%)
30 m3 per year
Sodium hydroxide (100%)
20 tons per year
OPERATING ECONOMICS
The Boliden system has such a high flexibility to take care of pollutants that no previous
sorting of the domestic waste is necessary. Also, as the amount of dust formed is so small
compared with other systems, the costs for storage of flue ash are more or less eliminated.
The surplus of energy generated from the incineration make the total plant in many cases
profitable at the same time as the world's most stringent environmental demands can be
fulfilled.
107
-------
REFERENCE LIST FOR BOLIDEN GAS CLEANING PROCESSES (JAN. 1991)
Client
Location
Product
Process
Capacity (gas) Year
compl./
status
Kemira
Kemi*
Kemira
Kemi*
Boliden
Mineral
Boliden
Mineral
Boliden
Mineral
Preussag
New
Consort
Boliden
Mineral
Nuova
Samim
Ronte
Adle
Helsingborg
Sweden
Helsingborg
Sweden
Skelleftehamn
Sweden
Skelleftehamn
Sweden
Skelleftehamn
Sweden
Nordenham
Germany
Barberton
RSA
Skelleftehamn
Sweden
Arzberg Germany
Bilbao
Spain
Sulphuric acid
Sulphuric acid
Liquid sulphur
dioxide
Liquid sulphur
dioxide
Copper, lead,
sulphuric acid,
liquid SO2
Lead
Gold
Sulphuric acid
Power plant
Porto Vesme Lead
Italy
Sulphuric acid
Scandust
Scharin Ursviken
Landskrona Alloy metals
Sweden from dust
Thiosulphate
Mercury Removal
Wet Arsenic
Separation
50 000 Nm3/h 1969
50 000 Nm3/h 1971
Mercury Separation 60 000 Nm3/h 1972
in Selenium Filters
Mercury Separation 100 000 Nm3/h 1976
in Selenium Filters
Wet Gas Treatment 300 000 Nm3/h 1978
Thiosulphate
Mercury Removal
Dry Arsenic
Separation
Wet Electrostatic
Precipitators
Wet Arsenic Sep./
Mercury Removal
Processes
Wet Arsenic
Separation
Wet Arsenic
Separation
Mercury Separation
in Selenium Filters
Wood chip Wet Electrostatic
30 000 Nm3/h 1980
10 000 Nm3/h 1982
100 000 Nm3/h 1986-
-87
2x3 000 Nm3/h 1987
16 000 Nm3/h 1987
60 000 Nm^/h 1987
20 000 Nm3/h 1987
70 000 Nm3/h 1988
108
-------
Client Location Product Process
Capacity (gas) Year
compl./
status
UnitexAB Sweden
boiler
Precipitators
Hoechst Frankfurt/Main Power plant
Germany
Cominco Trail
Canada
Enami
Norsk
Hydro
Norsk
Hydro
Santiago
Chile
Porsgrunn
Norway
Porsgrunn
Norway
Chematur Karlskoga
Internat. Sweden
Outo-
kumpu
Haijavalta
Finland
Lead
Copper
Chlorine
Magnesium
Municipal Waste
Incineration
Sulphuric acid
Wet Arsenic Sep./
Mercury Removal
Processes
Wet Arsenic Sep./
Cadmium Separation
Wet Arsenic Sep./
Thiosulphate
Mercury Removal
Wet Electrostatic
Precipitators
Dioxine Removal
by Wet Electrostatic
Precipitators
Flue Gas Treatment
Thiosulphate Mercury
Removal
1 500 Nm3/h 1989
16 000 Nm3/h 1990
88 000 Nm3/h 1990
2 300 Nm3/h 1990
70 000 Nm3/h 1990
35 000 Nm3/h 1991
69 000 Nm3/h 1990
* Boliden Kemi before 1989
The work described in this paper was not fundet y the U.S. Environmental Protection Agency
and therefore the contents do not necessarily reflect the views of the Agency and no official
endorsement should be inferred.
109
-------
>
Figure 1. Flow-sheet; Treatment of gases from waste incineration
110
-------
Figure 2. Selenium Filter; Hg + Se HgSe
111
-------
EDITUBE Wet Electrostatic Precipitator
112
-------
THE IOINT EC/EPA MID-CONNECTICUT TEST PROGRAM: A SUMMARY
T. G. Brna and J. D. Kilgroe
U. S. Environmental Protection Agency
Air and Energy Engineering Research Laboratory
Research Triangle Park, NC 27711, USA
A. Finkelstein
Environment Canada
Technology Development Branch
Ottawa, Ontario
Canada, K1A OH3
ABSTRACT
In early 1989, Environment Canada and the U.S. Environmental Protection
Agency sponsored a comprehensive test program on a refuse-derived fuel (RDF)
unit of the Mid-Connecticut facility in Hartford. The program, conducted in
cooperation with the Connecticut Resource Recovery Authority, the facility's
owner, and ABB Resource Recovery Systems, the operator, included
characterization and performance test phases. The results of the characterization
tests were used in defining both the combustion and flue gas cleaning system
operating conditions for the performance tests.
The results of the performance tests are emphasized here and will be
summarized in three parts. First, the combustion tests results will be addressed and
related to good combustion practice for RDF combustors. Then, the performance of
the lime spray dryer absorber/fabric filter system in controlling acid gas (hydrogen
chloride, sulfur dioxide), trace organic (polychlorinated dibenzo-p-dioxin and
polychlorinated dibenzofuran), trace metal (arsenic, cadmium, chromium, lead, and
mercury), and particulate matter emissions will be discussed. Finally, the results of
ash/residue analyses will be presented.
This paper has been reviewed in accordance with the administrative review policies
of the U.S. Environmental Protection Agency and approved for presentation and
publication.
113
-------
INTRODUCTION
The joint Environment Canada (EC)/U.S. Environmental Protection Agency (EPA)
program to evaluate the combustion and air pollution control systems for a modern
refuse-derived fuel (RDF) combustor was performed on Unit 11 at the Mid-
Connecticut facility (Mid-Conn) in Hartford. The test program, conducted with the
cooperation and assistance of the owner, Connecticut Resource Recovery Authority
(CRRA), and the operator, ABB Resource Recovery Systems (ABBRRS) - - formerly
Combustion Engineering, Inc. (CE) - - involved two phases: characterization and
performance. Since the results of the 28 characterization tests were reported
earlier,1'2 the emphasis here will be on the 13 valid performance tests made during
February 14-March 1, 1989. These results will be discussed in three parts:
combustion, flue gas cleaning, and ash/residue, with emphasis on these topics in
the order listed.
A major objective of the test program was to study the impact of combustor
operation on the control of organic emissions. Another objective was determining
the effect of flue gas system variables (temperature and sorbent/acid gas ratio) on the
control of acid gases as well as organic, trace metal, and particulate matter emissions.
Since the air pollutants from combustion gases are converted to solids by dry flue
gas cleaning or are in particulate carryover, the third test program objective
discussed here concerns organics and metals contained in the fabric filter
ash/residue.
TEST FACILITY DESCRIPTION
The facility contains a processing plant and a RDF power plant.1 The power plant
contains three CE steam generating units, each consisting of a RDF spreader stoker, a
natural circulation welded-wall boiler, a superheater, an economizer, and a tubular
combustion air preheater. All tests were conducted on Unit 11, which is designed to
produce 105 tonnes/hr (231,000 lb/hr) of steam at full load. Figure 1 provides a
schematic of the unit.
The fuel burning system for each unit consists of a RDF injection system, a traveling
grate stoker, and a combustion air system (see Figure 2). RDF is pneumatically
injected through four ports in the front face of each combustor. The lighter fraction
burns "in suspension" and the heavier falls onto the stoker where combustion is
completed. Underfire air is provided at controlled rates to 10 zones under the grate.
There are two separate overfire air (OFA) systems: a tangential system and a wall
system. The tangential system has four tangential overfire air (TOFA) windbox
assemblies in each furnace corner. Each TOFA assembly contains three elevations of
nozzles which can be manually adjusted in the horizontal plane. The wall system
contains one row of OFA ports on the front wall and two rows on the rear wall.
114
-------
The flue gas cleaning (FGC) system consists of a lime-based spray dryer absorber
(SDA) followed by a reverse-air-cleaned fabric filter (FF) or baghouse. This system is
capable of controlling the gas temperature at the SDA outlet and the sulfur dioxide
(SO2) concentration at the FF outlet. The SDA outlet temperature is controlled by
the lime slurry flow rate to the SDA, while the SO2 removal rate is controlled by
adjusting the lime concentration in the feed. The baghouse has 12 compartments,
each with 168 Teflon-coated glass fiber bags.
TEST CONDITIONS, MEASUREMENT METHODS, AND TEST RESULTS
All performance tests were run at slightly de-rated load conditions because of
unusually wet RDF being fired and insufficient combustion air fan capacity.2
Combustion and FGC process conditions for the performance tests were based on the
results of 28 characterization tests conducted in January 1989.2'3 During the
performance tests, a computerized data acquisition system continuously recorded
combustion and FGC process conditions. Continuous emission monitors (CEMs)
measured the concentration of oxygen (O2), carbon monoxide (CO), carbon dioxide
(CO2), total hydrocarbons (THCs), hydrogen chloride (HC1), and SO2 at the SDA inlet
and FF outlet. Nitrogen oxides (NOx) were monitored at the SDA inlet. CEMs also
measured the concentration of CO, HC1, and SO2 at the "mid-point" between the
SDA and FF. Modified Method 5 (MM-5) sampling trains were used to collect
organic samples at the SDA inlet and FF outlet during all tests. Organic samples
were also taken at the air heater inlet during four tests (PT07, PT08, PT09, and PT10).
Method 5 (M-5) sampling trains collected total particulate samples at the SDA inlet
and FF outlet. All sampling and analysis were done in accordance with protocols
approved by EC and EPA. Figure 1 indicates sampling locations.
The duration of each test was from 4 to 6 hours to ensure that sufficient volumes of
samples had passed through the MM-5 sampling trains. Combustion and FGC
process conditions were set and the test was begun after stable operating conditions
were obtained.
All polychlorinated dibenzo-p-dioxin and dibenzofuran (PCDD/PCDF), CO, NOx,
SO2, HC1, and other air pollutant data presented here have been corrected to 12%
CO2. All PCDD/PCDF data are given in nanograms per standard cubic meter [25°C,
101.3 kPa(77°F, 1 atm)], denoted by ng/Sm3.
Fourteen performance tests were conducted; however, since performance test 1
(PT01) did not meet all sampling requirements, its results are not given.
Combustion and FGC process conditions were varied independently. The
combustion tests were structured to evaluate the effects of good and poor
combustion conditions on organic concentrations at the SDA inlet.
The primary combustion test variables were boiler steam load, underfire-to-overfire
air ratio, and OFA distribution. During testing, the criterion for judging good or
115
-------
poor combustion conditions was the CO concentration at the SDA inlet. The effects
of load were evaluated by conducting tests at low (L), intermediate (I), normal (N),
and high (H) boiler steam flow rates. Underfire-to-overfire air ratios, which
influence the relative amount of RDF burned in suspension and the entrainment of
particulate matter in the flue gases (particulate matter carryover), were controlled by
changing the number of levels of OFA. Distributional mixing effects were evaluated
by changes in the underfire-to-overfire air ratio and by using rear-wall overfire air
(ROFA) in combination with different levels of TOFA.
Combustion test conditions, in order of increasing load, and the resultant average
CO and PCDD/PCDF concentrations are summarized in Table 1. The CO values are
averages based on measured values at the SDA inlet and FF outlet. The NOx and
PCDD/PCDF values are from the SDA inlet.
DISCUSSION
Combustion Tests
Test results were evaluated to assess the effects of combustion conditions on furnace
emission of organics, NOx, and metals using concentrations of these air pollutants
at the SDA inlet.
Although organic compounds (such as PCDD/PCDF) may be in the waste feed, it is
unlikely that they will pass through the combustor undestroyed.4 They may also be
formed in high temperature regions of the furnace from the thermal decomposition
products which are incompletely oxidized due to insufficient combustion air,
mixing, temperature, or residence time. They may also originate from catalytic
reactions on the surface of flyash downstream of the combustion chamber.
Multiple regression analyses show that PCDD, PCDF, chlorophenol (CP),
chlorobenzene (CB), and polynuclear aromatic hydrocarbon (PAH) concentrations at
the SDA inlet can best be correlated with two or more of the following easily
monitored furnace/flue gas properties:
a) CO concentration
b) THC concentration
c) NOx concentration
d) Moisture (H2O) concentration
e) Temperature in the furnace or at the economizer outlet
For example, the best correlation for PCDD concentrations at the SDA inlet (R2=0.9)*
is based on the CO, NOx , and H2O concentrations at this location.
* R2is the correlation coefficient with R2=1.0 indicating a perfect (exact) correlation.
116
-------
Multiple regression analyses provide correlations which indicate that the combustor
operating variables (steam load, combustion air flows, RDF moisture content, etc.)
can be used to control PCDD, PCDF, CP, CB, and PAH concentrations at the SDA
inlet. These operating variables control combustion in the furnace by impacting the
fundamental combustion parameters: time, temperature, air/fuel ratio, and mixing.
Multiple regression analyses based on easily monitored flue gas properties provide
for better correlations of organics (R2=0.8 to 0.98) than combustor operating variables
(R2=0.6 to 0.8). Optimum control systems for limiting the furnace emission of
organics will probably require both flue gas property measurements and combustion
operating variables as inputs.
Single parameter regression analyses using average test values show that either CO
or THC tracks the furnace destruction of organics. Moderate correlations between
CO (or THC) and PCDD/PCDF at the SDA inlet [R2=0.70 (or 0.68)] were obtained
when the entire data set (13 tests) was considered (see Figures 3 and 4). An excellent
correlation (R2=0.95) occurred for CO versus PCDD/PCDF under poor combustion
(CO>200 ppm). No statistically significant correlation was found between CO and
PCDD/PCDF for tests with good combustion conditions (CO<200 ppm). Tests with
average THC concentrations greater than 7 ppm provided excellent correlations
between THC and PCDD/PCDF emissions, but tests with less than 7 ppm showed no
significant correlation.
Single parameter correlations between average CO (or THC) concentrations and CP,
CB, or PAH concentrations at the SDA inlet provided better correlations [R2= 0.88
(0.92), 0.83 (0.87), and 0.83 (0.85), respectively] than those for PCCD or PCDF. There
were no significant correlations between average CO (or THC) and PCB at the SDA
inlet. Thus, CO or THC concentration at the SDA inlet appears to be a good
indicator for the furnace emission of most trace organics of concern.
Previous field tests have shown a strong positive correlation between the amount of
PM entrained in flue gas (PM carryover) and emissions of PCDD/PCDF.5 The Mid-
Conn data show a fair correlation (R2=0.61) between PCDD/PCDF and PM
concentrations at the SDA inlet for good combustion conditions (CO<200ppm); yet,
no significant correlation between these variables was seen for all combustion
conditions. Possibly for good combustion, the emission rate of PM (as PM is
postulated to provide reaction sites where PCDD/PCDF are formed) is the principal
variable affecting the furnace PCDD/PCDF emission rate, while for poor combustion
the effects of other parameters obscure the relationship between PM and
PCDD/PCDF.
PCDD/PCDF and other chloro-organic compounds can be formed downstream of
the furnace by de novo synthesis reactions on the surface of flyash.6'7'8 The amount
formed is believed to be proportional to the amount of flyash and the time
individual particles with reaction sites exist in the temperature range of about 450 to
117
-------
150°C (750 to 300°F). Mid-Conn has an air heater just upstream of each SDA with
flue gas in the upper end of this range. Thus, an increase in PCDD/PCDF
concentrations across the air heater due to de novo synthesis was expected. The flue
gas temperature at the economizer outlet (which approximates that at the air heater
inlet) typically ranges from 343 to 388°C (650 to 730°F), while the temperature at the
outlet of the air heater ranges from 177 to 210°C (350 to 410°F). To evaluate the
hypothesis that PCDD/PCDF form on particulate deposits within the air heater,
PCDD/PCDF measurements were made during four test runs just upstream of the
air heater inlet simultaneously with those at the SDA inlet (air heater outlet).
Contrary to expectations, a comparison of measurements at the inlet and outlet of
the air heater indicated a reduction of PCDD/PCDF concentrations during three of
four runs. No explanation for these unexpected results is presently available.
Average NOx emissions ranged from 149 to 193 ppm. Generally, NOx emissions
increase with increased combustion temperatures and improved mixing.
Accordingly, low CO emissions corresponded with high NOx emissions.
Conversely, the lowest NOx emissions were associated with high concentrations of
CO, THC, and organics at the SDA inlet. An evaluation of 30-second emission
averages from the CEM data indicates that, to obtain a NOx emission less than 180
ppm (the new U.S. New Source Performance Standard requirement for large
municipal waste combustors), the Mid-Connecticut units would have to operate at a
CO emission concentration of 71 ppm or higher.
There were no apparent correlations between combustion conditions and the
concentration of metals in flyash at the SDA inlet.
Flue Gas Cleaning System Tests
As noted earlier, Unit 11 has a lime SDA/FF system for acid gas and PM control.
Lime is slaked and the resultant calcium hydroxide slurry is diluted with water from
the facility's surface water runoff/coal pile drainage pond, or Connecticut River
water when pond water is unavailable, prior to being fed to the SDA's atomizer.
During the performance tests, dilution water was mainly from the pond. As
reported earlier,2 the SDA outlet temperature was used as a surrogate for the
approach to saturation temperature since it was directly measurable, and the SO2
concentration at the FF outlet served as the on-line indicator of sorbent-to-acid gas
ratio (or lime stoichiometry). Because the plant's measuring instrument for slurry
density was inoperative and dilution water composition varied during the test
program, the lime stoichiometry could only be estimated; thus, it is not
quantitatively noted here. When stoichiometry is used here, it refers to lime
stoichiometry; high, moderate (or medium), and low stoichiometry correspond to
low, medium, and high FF outlet SO2 concentration, respectively.
The HC1 and SO2 concentrations were monitored continuously, and their averaged
values along with removals for each test run are shown in Table 2. Note that the
data for PT02 and PT05, PT03 and PT11, and PT12, PT13, and PT14 have been
118
-------
grouped, since the FGC system set points (SDA gas outlet temperature and FF outlet
SO2 concentration) were the same. Table 2 shows that the HC1 and SO2 values at the
SDA inlet averaged 445 and 185 ppm, respectively, over all tests, with the individual
test average values being ±10% of these averages, except that for PT08 when the HC1
concentration was about 20% higher.
As expected the HC1 and SO2 removals increased with decreasing SDA outlet gas
temperature (approach to saturation temperature) and increasing lime
stoichiometry. The HC1 removal averaged 95% or more and the SO2 removal over
83% when the lime stoichiometry was high (low FF outlet SO2 concentration), such
as during normal FGC system operation, and showed only a slight decrease with
temperature as the average SDA outlet temperature increased from 122 to 166°C (252
to 330°F). The HC1 removal was above 92% for all temperature and stoichiometry
combinations tested, except for high temperature (171°C or 339°F) and high
stoichiometry when it was about 77%. The SO2 removal was more sensitive to the
change in stoichiometry than to temperature, being 83 to 95% for high stoichiometry
and ranging from 62 to 76% for medium stoichiometry over the tested SDA outlet
temperature range [122 to 171°C (252 to 339°F)]. At low stoichiometry, the SO2
removal unexpectedly rose from 30 to 32% as the SDA outlet gas temperature
increased from 122 to 142°C (252 to 287°F); however, the SO2 removal was -6% at
171°C (339°F) (i.e., the SO2 concentration at the FF outlet was greater than at the SDA
inlet), suggesting that SO2 was being desorbed in the filter cake. This suspected
desorption was similar to that observed in the Mid-Conn characterization test
series,2 is consistent with pilot study findings,9 and is likely due to HCl's being more
reactive than SO2 with the calcium-based compounds in the filter cake. These
results suggest that "normal" operation of the SDA/FF system [140°C (285°F),
moderate stoichiometry] would be expected to yield 95% or more HC1 removal and
about 70% SO2 removal.
Table 3 provides inlet concentrations and removal efficiencies for selected organics.
With the exception of PT09 (highest SDA outlet temperature and low
stoichiometry), the PCDD removal was at least 99.8% for SDA inlet concentrations
ranging from 95 to 397 ng/Sm3. For PT09, the inlet PCDD concentration was 71
ng/Sm3 and the removal was 99.2%. Since the SDA outlet temperature,
stoichiometry, and inlet PCDD concentration may affect PCDD removal, the specific
reason for the relatively low removal during PT09 is not evident. All three
variables had values suggesting low control (i.e., high temperature, low
stoichiometry, and low inlet concentration). The PCDF removal was 99.9% or
greater for all tests, although the SDA inlet PCDF concentration changed almost
three-fold from 341 to 1007 ng/Sm3. CP control was usually slightly better than CB
control, with CP control generally being 97% or more and CB control being 95% or
more. The CP inlet concentration ranged from 11,329 to 62,938 ng/Sm3, more than
double that for CB. The PAH removal generally increased with inlet concentration,
ranging from 58.6 to 97.7% as the concentration rose from 6,289 to 88,626 ng/Sm3.
The lowest PAH removals in the FGC system correspond to the lowest inlet
119
-------
concentrations and occurred during good combustion (low CO). However, the PAH
(as well as other organic) emissions varied with flue gas cleaning system variables.
Generally, organics removal was high (over 95%), except PAH removal which was
usually moderately high (over 92%), for all tests and conditions so that changes in
the FGC process variables had little effect over the range evaluated. For example,
the removal of combined PCDD and PCDF was 99.7% or higher for all tests.
As shown in Table 4, the PM concentration in the flue gas at the SDA inlet ranged
from 3,274 to 4,949 mg/Sm3, while the outlet concentration ranged from 2.68 to 7.62
mg/Sm3. The corresponding PM removal was 99.8% or more. Because of the high
PM removals, the effect of process changes could not be reliably distinguished.
Table 4 also presents selected metals control data, with complete removal of As and
Cd being indicated for all tests. The removal of Pb followed very closely the
removal of PM, usually being above 99%, despite a five-fold ratio of the maximum
to minimum inlet concentration for the tests. Surprisingly, Hg control was over
96% for all the tests. The Cr removal was high and paralleled that of Hg.
While high C content in flyash, which is characteristic of RDF combustors, may be a
factor in the observed high control of Hg emissions, conclusive test data were not
obtained on C content during the test program. However, the loss-on-ignition data
in Table 5 range from 4.26 to 10.45% and suggest high C content, since these values
are believed representative of the C plus water of hydration content of the fabric
filter ash. Since the flyash from highly efficient mass burn combustors is from 1 to
2%, the values in Table 5 suggest C values greater than this range even for water of
hydration contents as high as 50%. However, PM loading may also be a factor
affecting Hg control, since condensible metals are believed to enrich particles,
especially small ones, because of the relatively high surface area on a mass basis.
Other variables considered in studying the Hg removal were flue gas temperatures
and lime stoichiometry. Hg removal decreased slightly with increasing
temperature, probably due to decreasing condensation or adsorption on PM as the
temperature rose. Using outlet SO2 concentration as a surrogate for lime
stoichiometry (note that the average inlet SO2 values varied by less than 7% from
185 ppm, the average of all test averages), Hg removal decreased with increasing
stoichiometry. Possibly a reaction between mercuric chloride and (excess) calcium
hydroxide in the filter cake led to the liberation of Hg2+ ions and the increased Hg
emission. Thus, operation at high lime stoichiometry to ensure high acid gas
removal can be accompanied by increased Hg emission relative to low or moderate
stoichiometry.
Ash/Residue Results
During the test program, RDF and ash/residue samples were collected. RDF was
collected from the conveyor serving the combustor. Ash/residue were collected
from the economizer, fabric filter, and grate (siftings). In addition, the ash from the
120
-------
grate's surface was collected, after quenching, as bottom ash, and the flyash at the
SDA inlet and FF outlet was collected using sampling trains. The distribution of
metals in the ash streams is discussed elsewhere,10 as are the physical and chemical
properties of the noted ash/residue.11 The results to be discussed here concern the
metals and organics collected in the FF ash (or residue).
While the FGC system data indicate complete removal of As and Cd, the
concentrations of these metals in Table 5 suggest otherwise. This premise is based
on the lime slurry's (slaked lime plus dilution water) not being a significant source
of these elements. The Pb concentration in the FF ash was the highest of the metals
shown in Table 5 and was expected as it had the highest flue gas concentrations at
the SDA inlet and was removed very effectively (generally, over 99%) in the FGC
system. While the SDA inlet concentrations and removals of Cr and Hg were quite
similar, the Cr transferred to the FF was 4 to 20 times as great. Possibly the high
volatility of Hg during collection and storage of the samples prior to their analyses
may have affected the concentrations of Hg in the FF ash.
Considering the minimum and maximum PM emission rates of 2.68 (PT06) and 7.62
mg/Sm3 (PT04), the calculated Hg transfer rates to the FF ash for these runs are 44.57
and 66.48 g/hr, respectively. The Hg removal rates, based on flue gas sampling, are
90.32 and 92.98 g/hr, respectively, for the tests. Thus, the calculated Hg transfer rates
to the FF ash for these tests are about 50 to 70% of those determined from flue gas
data. For a less volatile metal, such as Pb, the calculated rates are 4,538 and 4,727
g/hr for PT06 and PT04, while removals in the FGC system are 1,120 and 1,549 g/hr,
respectively, about 30% of the calculated values. These examples suggest that the
input-output method is, at best, a crude check on gas sampling and analytical
measurements.
Comparison of PCDD removal rates by the FGC system with values observed in the
FF ash yields results similar to those noted for the metals above. For example, the
PCDD removal rate by the FGC system ranged from 10.12 (PT09) to 57.57 mg/hr
(average of PT02 and PT05), while the corresponding FF ash gained PCDD at the
calculated rates of 147.5 and 41.18 mg/hr for these tests. Thus, the calculated PCDD
transfer rate for PT09 was almost 15 times that of the FGC system, while for
combined PT02 and PT05, the calculated rate reasonably approximates the PCDD
removal from the flue gas, about 30% below the flue gas removal rate. Similar
results would be expected from the input-out analyses for other organics.
SUMMARY
The comprehensive 1989 test program on an RDF municipal waste combustion unit
at the Mid-Conn facility obtained combustion and FGC performance data useful for
the design and operation of similar systems. The jointly sponsored EC/EPA effort
also analyzed ash/residue for organic and metal contents as well as generation rates.
121
-------
Multiple regression analyses show that organic concentrations at the spray dryer
inlet can be best correlated by two or more of the following easily monitored flue gas
properties: CO, THC, NOx , moisture, and temperature at the economizer outlet (or
in the furnace). Similar analyses indicate that combustion operating variables
(steam load, air flows, RDF moisture content, etc.) can be used to control organic
emissions, including PCDD, PCDF, CP, CB, and PAH emissions. For tracking the
furnace destruction of organics, CO or THC is useful. CO is an excellent indicator of
PCDD/PCDF for poor combustion (CO>200ppm), but not for good combustion
(CO<200 ppm). PCDD/PCDF formation via de novo synthesis was not observed as
expected in the air heater where the temperature supported their net production.
The FGC system demonstrated high removals of acid gases, organics, metals, and
PM. HC1 removals above 95% and SO2 removals over 90% were achieved with high
lime slurry flows (high lime stoichiometry) and flue gas temperatures at the SDA
outlet as high as 166°C (330°F). PCDD and PCDF removals exceeded 99% under all
test conditions and appeared to be independent of combustion conditions. Lime
stoichiometry had a greater effect on acid gas removal than did SDA outlet
temperature. Metal removals were very high: 100% for As and Cd, 98.5% or more
for Pb, 96.9% or greater for Hg, and 96.4% or more for Cr for all tests. PM control was
99.8% or greater, and the removal of Pb, Hg, and Cr was only slightly less.
Generally, the organics in the FF ash were higher for poor combustion than for good
combustion. The PCDD/PCDF ranged from 74 to 509 ng/g for the tests. Pb was the
predominant metal in the ash, while As was the least prominent. Both were
similarly ranked in the flue gas at the SDA inlet.
References
1. Boley, G. L., Startup and Operations of the Mid-Connecticut Resource
Recovery Project. Paper presented at the International Conference on Municipal
Waste Combustion, Hollywood, Florida, April 1989.
2. Brna, T.G. and R.K. Klicius, Characterization of a Lime Spray Dryer
Absorber/Baghouse on a Refuse-Derived Fuel Combustor. Paper presented at the
International Conference on Municipal Waste Combustion, Hollywood, Florida,
April 1989.
3. Kilgroe, J.D. and A. Finkelstein, Combustion Characterization of RDF
Incinerator Technology: a Joint Environment Canada - United Staes Environmental
Protection Agency Project. Paper presented at the International Conference on
Municipal Waste Combustion, Hollywood, Florida, April 1989.
4. Miller, H., S. Marklund, and C. Rappe, "Correlation of Incineration
Parameters for the Destruction of Poly chlorinated Dibenzo-p-dioxins,"
Chemosphere, 18, 7/8, (1989): 1485-1494.
122
-------
5. Schindler, P.J. and L.P. Nelson, "Municipal Waste Combustion Assessment:
Technical Basis for Good Combustion Practice," EPA-600/ 8-89-063 (NTIS PB90-
154949), August 1989.
6. Stieglitz, L. and H. Vogg, Formation and Decomposition of Polychlorinated
Dibenzo-dioxins and -furans in Municipal Waste, Report KfK4379, Laboratorium fiir
Isotopentechnik, Institut fiir Heisse Chemie, Kernforschungszentrum Karlsruhe,
Februar 1988.
7. Hagenmaier, H., H. Brunner, R. Haag, and M. Kraft, "Catalytic Effects of Fly
Ash from Waste Incineration Facilities on the Formation and Decomposition of
Polychlorinated Dibenzo-p-dioxins and Polychlorinated Dibenzofurans," Environ.
Sci. Technol. 21 (1987): 1080-1084.
8. Gullett, B.K., K.R. Bruce, and L.O. Beach, "Formation of Chlorinated Organics
During Solid Waste Combustion," Waste Management and Research, 8, (1990): 203-
214.
9. Chang, J.C.S., T.G. Brna, and C.B. Sedman, Pilot Evaluation of Sorbents for
Silmutaneous Removal of HC1 and SO2 from MSW Incinerator Flue Gas by Dry
Injection Process. Paper presented at the International Conference on Municipal
Waste Combustion, Hollywood, Florida, April 1989.
10. Hartman, R.M., An Overview of the Fate of Metals During Combustion at the
Mid-Connecticut RDF Facility. Paper to be presented at the Second International
Conference on Municipal Waste Combustion, Tampa, Florida, April 1991.
11. Sawell, S.E., T.W. Constable, and R.K. Klicius, A Summary of the National
Incinerator Testing and Evaluation Program: Ash Solidification and
Characterization Studies. Paper to be presented at the Second International
Conference on Municipal Waste Combustion, Tampa, Florida, April 1991.
123
-------
ECONOMIZER
AIR HEATER
SAMPLING
(ORGANICS)
SUPERHEATER
INLET SAMPLING AND CEMs
OUTLET SAMPLING
AND CEMs
MIDPOINT CEMs
SPRAY
DRYER
ABSORBER
TUBULAR
AIR HEATER
RDF DISTRIBUTORS
FABRIC FILTER (BAGHOUSE
FRONT
OVERFIRE AIR
OVERFIRE AIR FAN
GRATE
COMBUSTOR
STACK
CEMs = CONTINUOUS EMISSION MONITORS
INDUCEO DRAFT Fan
Figure 1. RDF traveling grate stoker boiler with spray dryer absorber/fabric filter (Combustion Engineering).
-------
tangential
OVERFIRE AIR
RDF
DISTRIBUTORS
GRATE
SURFACE
WATER SEAL
IDLER SHAFT
DRIVE SHAFT
SIFTING
SCREW
CONVEYOR
UNDERGRATE
AIR
COMPARTMENT
Figure 2. RDF traveling grate stoker boiler (Combustion Engineering).
125
-------
2000
1500
1000
500
•
0
1
1
0
0
n° 0
•
O Good operation
OO
0 Poor operation
0
200
400
CO (ppm)
600
soo
1000
R2 = 0.70
Values corrected to 12% CO2
Figure 3. Dependence of uncontrolled PCDD/PCDF on CO emissions.
a
c
Uh
61
Q
Q
ki
2000
1500
1000
500
A
A.
A A
1 0
R2 = 0.68
^ Good operation
A Poor operation
20 30 40
THC (ppm)
50
60
Values corrected to 12% CO2
Figure 4. Dependence of uncontrolled PCDD/PCDF on THC at SDA inlet.
126
-------
TABLE 1. COMBUSTION CONDITIONS AND RESULTS AT SDA INLET
Test No. Load
Comb.
Overfire Air
CO
NOx
(PT) 1000 ke/hr
Cond.a
TOFAb
ROFAc
OFAd
EEm
13 71 (L)
G
2
nil
47
158
157
14 74 (L)
G
2
nil
49
70
177
PCDD/PCDF
ng/Sm^e
599
428
10
87 (I)
G
2
nil
52
77
186
667
02
88 (I)
G
2
nil
52
108
184
946
05
84(1)
P
1
65
38
903
149
1861
09
95 (N)
G
2
65
51
92
188
449
08
96 (N)
G
2
65
48
89
193
1162
11
96 (N)
G
2
65
52
68
175
536
07
101 (N)
P
3
nil
51
387
172
1003
04
98 (N)
P
3
nil
54
214
172
774
03
99 (N)
P
1
65
44
432
160
1008
12
117(H)
G
2
65
53
116
180
282
06
118 (H)
P
2
nil
57
397
157
1202
a Good (G) or poor (P) combustion conditions
b Number of levels of TOFA
c Pressure in ROFA plenum, mm Hg
d OFA as a percentage of total combustion air
e Standard conditions: 25°C, 101.3 kPa
-------
TABLE 2. FLUE GAS CLEANING (FGC) SYSTEM PERFORMANCE: ACID GASES
Concentrations, ppm
Inlet Outlet Removal, %
Test No. FGC Cond.
(PT)a
Temp./SO?b
HC1
SO2
HC1
SO?
HC1
SO?
02,05
L/H
470
173
20
121
95.7
30.1
03,11
H/L
416
187
20
17
95.2
90.9
04
H/M
471
186
31
44
93.4
76.3
06
M/L
404
192
10
32
97.5
83.3
07
L/L
399
183
8
17
98.0
90.7
08
M/H
538
184
41
126
92.4
31.5
09
H/H
432
178
98
189
77.3
-6.2C
10
L/M
429
194
19
74
95.6
61.9
12,13,14
M/M
444
187
18
59
95.9
68.4
a Values are averaged for multiple runs.
b High temperatures (H) ranged from 166 to 171 °C (330 to 339°F), medium temperatures (M) from 141 to 142°C (285 to 287°F), and low (L)
temperatures from 122 to 124°C (252 to 255°F) for the spray dryer outlet gas. Fabric filter SO2 outlet concentrations were above 100 ppm for
high (H) concentration, between 21 and 100 ppm for medium (M) concentration, and 20 ppm or less for low (L) concentration. All concentra-
tions are referenced to 12% CO2 in dry gas [25°C (77°F), 101.3 kPa (1 atm)].
c Desorption of SO2 in the filter cake is suspected for low lime stoichiometry and relatively high HC1 concentration.
-------
TABLE 3. FLUE GAS CLEANING (FGC) SYSTEM PERFORMANCE: ORGANICS
Test No.
FGC Cond.
Inlet Concentrations,
ng/Sm3a
Removal, %
(PT)b
Temp./SO?c
PCDD
PCDF
CB
CP
PAH
PCDD
PCDF
CB
QP
PAH
02,05
L/H
397
1,007
10,860
62,938
60,176
• 99.9
99.9
96.2
97.4
92.0
03,11
H/L
161
611
6,159
20,798
46,976
99.8
100d
95.2
99.1
92.2
04
H/M
151
623
5,964
16,964
25,519
99.8
99.9
98.4
99.0
92.2
06
M/L
317
885
9,403
41,588
88,626
99.9
100d
94.3
96.9
97.7
07
L/L
207
796
7,074
25,168
51,774
99.9
100d
98.5
99.1
97.3
08
M/H
211
951
7,071
20,226
10,259
99.9
100d
98.4
99.1
76.7
09
H/H
71
378
4,848
11,329
32,421
99.2
99.9
97.7
96.5
92.5
10
L/M
243
424
6,170
16,198
6,289
99.9
100^
99.3
99.9
58.6
12,13,14
M/M
95
341
4,647
14,419
7,747
99.6
100d
99.1
99.4
63.2
a Organics are: polychlorinated dibenzo-p-dioxins (PCDD) and dibenzofurans (PCDD), chlorobenzenes (CB), chlorophenols (CP), and
polynuclear aromatic hydrocarbons (PAH),
b Values are averaged for multiple runs.
cHigh temperatures (H) ranged from 166 to 171 °C (330 to 339°F), medium temperatures (M) from 141 to 142°C (285 to 287°F), and low
temperatures (L) from 122 to 124°C (252 to 255°F) for the spray dryer outlet gas. Fabric filter outlet SO2 concentrations were above 100 ppm for
high (H) concentration, between 21 and 100 ppm for moderate (M) concentration, and 20 ppm or less for low (L) concentration. All concentrations
are referenced to 12% CO2 in dry gas [25°C (77°F), 101.3 kPa (1 atm)].
^Value is based on rounding off to three significant figures.
-------
TABLE 4. FLUE GAS CLEANING (FGC) SYSTEM PERFORMANCE:
PARTICULATE MATTER AND SELECTED METALS
Test No. FGC Cond. Particulate Matter
(PT)a Temp./S02b (PM) mg/Sm3 Inlet Concentration, [ig/Sm3 Removal, %
Inlet
Outlet
As
Cd
Cr
Pb
Eg
PM
Asc
Cdc
Cr
Pb
Hg
02,05
L/H
4,949
4.83
250
548
859
13,472
680
99.9
100
100
98.3
99.7
99.0
03,11
H/L
4,313
5.60
214
594
579
11,479
622
99.9
100
100
98.6
99.6
96.8
04
H/M
3,274
7.62
168
536
538
10,050
614
99.8
100
100
98.1
99.6
97.8
06
M/L
3,308
2.68
194
437
353
7,229
583
99.9
100
100
97.7
99.5
98.0
07
L/L
4,230
4.39
176
515
520
5,877
584
99.9
100
100
98.5
99.5
98.7
08
M/H
4,745
3.88
224
832
862
4,649
646
99.9
100
100
96.4
99.1
99.3
09
H/H
3,894
5.79
196
668
1,491
2,592
644
99.9
100
100
99.3
98.5
97.8
10
L/M
4,531
4.09
210
599
871
4,770
718
99.9
100
100
99.0
99.1
98.8
12,13,14
M/M
3,433
5.46
219
569
949
8,563
668
99.8
100
100
98.2
99.3
98.6
a Values are averaged for multiple runs.
b High temperatures (H) ranged from 166 to 171 °C (330 to 339°F), medium temperatures from 141 to 142°C (285 to 287°F), low temperatures (L)
from 122 to 124°C (252 to 255°F) for the spray dryer outlet gas. Fabric filter SO2 outlet concentrations were above 100 ppm for high (H)
concentration, between 21 and 100 ppm for medium (M) concentration, and 20 ppm or less for low concentration. All concentrations are referenced
to 12% CO2 in dry gas [25°C (77°F), 101.3 kPa (1 atm)].
c All outlet concentrations were nondetectable and assigned zero values for calculating removal.
-------
TABLE 5. FABRIC FILTER ASH CONTENT: ORGANICS AND SELECTED METALS
Test No. FGC Cond. Ash Rate LOIc
(PT)a Temp./SQ2b kg/hr Concentration, ng/g Concentration, |ig/g %
PCDD
PCDF
CB
CP
PAH
As
Cd
Cr
Pb
Hg
02,05d
L/H
429
96
71
1,085
2,870
9,437
15
70
264
1,987
25
6.52
03,lle
H/L
2,140
49
100
704
2,225
1,087
18
97
240
2,405
30
4.50
04
H/M
1,385
84
172
1,059
3,320
1,806
20
96
179
3,413
48
8.15
06
M/L
1,239
227
282
1,684
6,095
7,431
19
96
154
3,666
36
10.45
07
L/L
550
154
271
941
4,997
1,992
17
90
147
3,051
37
9.97
08
M/H
434
62
96
729
1,636
2,905
22
62
210
2,439
25
5.00
09
H/H
1,317
112
222
1,266
4,336
4,780
21
119
287
4,545
37
9.30
10
L/M
1,166
27
47
684
1,924
1,402
19
87
274
2,352
27
4.26
12,13,14f
M/M
707
102
111
1,218
1,832
4,093
19
118
207
2,812
39
8.89
a Values are averaged for multiple runs.
b High temperatures (H) ranged from 166 to 171 °C (330 to 339°F), medium temperatures (M) from 141 to 142°C (285 to 287°F), and low
temperatures (L) from 122 to 124°C (252 to 255°F) for the spray dryer outlet gas. Fabric Filter SO2 outlet concentrations were above 100 ppm for
high(H) concentration, between 21 and 100 ppm for medium (M) concentration, and 20 ppm or less for low(L) concentration. All concentrations
are referenced to 12% CO2 in dry gas [25°C (77°F), 101.3 kPa (1 atm)].
c LOI is loss on ignition,
d Data shown are based solely on PT05.
e Data shown are based solely on PT11.
f Data shown are based on PT12 and PT14.
-------
Intentionally Blank Page
132
-------
The work described in this paper was not funded by the U.S. Environmental
Protection Agency. The contents do not necessarily reflect the views o1 the
Agency and no official endorsement should be inferred.
EFFECTIVE CONTROL OF MERCURY AND OTHER POLLUTANTS
FROM MWC FACILITIES BY EDV TECHNOLOGY
Gerwyn Jones, C.Eng.
Belco Technologies Corporation
7 Entin Road
Parsippany, NJ 07054
ABSTRACT
The paper describes with several performance case histories operating
experience of EDV gas cleaning technology for controlling particulate matter,
heavy metals including mercury, acid gases and PCDD/PCDF emissions from both new
and retrofitted MWC and waste water sludge incinerator facilities. Recent
developments realized in EDV technology aimed at recovering new materials from
waste streams are also reported. In addition, a status update of supplementary
pollutant controls demonstrated on a 330 TPD MWC is presented.
Introduction
Thermal processing of municipal and hazardous wastes are two of the most
closely and severely environmentally regulated operations in the U.S. today. As
a consequence, these industries have become prime examples of community and
commercial services that are dependent on the performance of pollution control
technologies. In this environment, municipal agencies and operator managements
need to re-evaluate the influence of pollution and pollution control on plant
economics. In a growing number of situations there are compelling economic
reasons why the control of pollution should be recognized as a legitimate part
of the production process. As the influence of pollution control increases so
should the priority allocated to assessing technology and equipment. A global
cost-benefit perspective is therefore becoming an essential complementary
evaluation process to the best available control technology approach.
Concerns over the fate of heavy metals contained in wastes processed by
municipal solid waste combustion (MWC) continue to impact on air pollution
control technology and residue management. Residues from current BACT MWC air
pollution control systems operating in the U.S. contain heavy metals, excess lime
and exhibit poor recycle properties. Alternate air pollution control technology
extensively used in other countries, e.g. EDV, provide means for the controlled
extraction and recovery of heavy metals and the production of inert residue
suitable for commercial use.
Achieving effective control of pollution is dependent on understanding the
nature of the problem. The formation mechanisms, the influence of combustion
variables, the profile of individual pollutants, adaptability to future
regulations are some of the parameters that should be addressed in the technology
selection process. Impact of the growing application of continuous emission
monitors for an ever widening range of pollutants and shorter averaging times
will likely fuel reassessment of air pollution technologies applied to many
thermal processes.
Preceding page blank
-------
EDV Technology
EDV technology has been developed to provide a versatile range of gas
cleaning systems to suit a wide variety of applications and performance
requirements. Emissions of particulate matter, PM 2.5, acid gases, PCDD/PCDF,
heavy metals, can be effectively and efficiently controlled by EDV gas cleaning
systems.
The EDV concept consists of a sequence of fundamental functions namely
saturation, expansion, condensation, ionization and filtration.
Saturation of the gas takes place in an open spray tower by means of
proprietary LAB-G nozzles. For high temperature applications, quenching of the
gas is a necessary prerequisite and the tower is designed accordingly. The
turbulent contact produced by the LAB-G nozzles between the scrubbing liquor and
the gas achieves a large degree of particulate removal as well as establishing
the saturation condition. For applications where gas absorption of acid gases
is required, this turbulent contact zone is used.
On leaving the spray tower, the saturated gas enters a tubular module(s).
As the gas traverses the converging/diverging section, adiabatic expansion takes
place, which is accompanied by a change of energy state.
The adiabatic expansion of a gas saturated with water vapor produces a
state of supersaturation. Particulate matter suspended in the gas acts as
condensation nuclei and a film of water is formed on the surface of the dust
particles. The encapsulated fine dust particles achieve growth in both size and
mass which will significantly improve the effectiveness of the subsequent
particulate removal mechanisms. The presence of the water film also ensures that
the particle is capable of receiving and retaining an electric charge regardless
of the dielectric nature of the core material.
The divergent pressure recovery section of the module is designed to
minimize pressure loss and ensures suitable ionization conditions. An electrode,
mounted on the axial centerline of the module, creates a corona within the gas
flow path. As the water-encapsulated particles traverse the ionizing zone, they
become negatively charged. The electrical energy expanded in ionization is
nominal in comparison with the energy levels used in conventional electrostatic
precipitation devices.
The filtering device developed for the EDV concept provides for a near
absolute removal of particles/mist from the flue gas. Two separate contact
mechanisms are integrated in the design, namely physical impaction and
electrostatic attraction. A centrally mounted hollow cone spray nozzle is
positioned adjacent to, and facing, the pressure recovery exit section of the
module. The proximity of this nozzle to the polarizing tip of the electrode
ensures that the water curtain created by the nozzle is positively charged under
the influence of the electric field generated by the electrode. Since the
filtering nozzle and feed pipe are earthed, there is a migration of electrons
from the water curtain which results in the subsequently formed filtering spray
curtain to retain a positive electrical charge.
134
-------
The charged (negative) particles/mist are removed from the gas stream by
the filtering spray curtain.
Collision capture by the large filtering spray droplets is boosted by the
electrostatic attraction effect. The ELECTRO-DYNAMIC FILTRATION approximates a
total removal of particles/mist from the gas stream which affords important
benefits to plant operation.
Since the virtual elimination of dust bearing droplets from the gas stream
can be assured, the hydraulic circuit of the EDV can be engineered to operate at
a high suspended/dissolved solids level with the result that the associated
materials recovery/effluent treatment plant is of a much reduced size and cost.
Mercury Capture
Mercury emissions from MWC have been studied by laboratory synthesis and
a measurement program of over 4000 hours using a continuous emission monitor
(CEM). Table 1 illustrates the primary findings of this research. Figure 1
shows the influence of temperature on mercuric chloride formation. The
concentration of hydrogen chloride present in the flue gas further influences the
speciation of mercury. Extensive test data from multiple MWC and other
incineration processes show mercuric chloride as the dominant species present in
the flue gas. Thus a controlled washout of this water soluble compound is
readily achieved in the EDV system.
Figure 2 shows a CEM trace of total mercury from a MWC that serves a
community that practices source separation of batteries. Correlation between the
mercury CEM and standard measurement methodology is shown in Figure 3.
EDV performance for the control of MWC mercury emissions shown in Table 2
(Basle, Switzerland) were made at the flue gas adiabatic saturation temperature
of 56/57"C (133/135*F). Subcooling of the flue gas to 35"C (94'F) improved
capture beyond the detection limits of the test method (see Table 3) . The
stability of the captured mercury in the solid EDV cake residue is illustrated
by lixiviation results by Swiss, Japanese and U.S. (TCLP) test methods shown in
Table 4.
Typical EDV performance on other heavy metals is illustrated in Table 5.
PCDD-PCDF Suppression - Capture - Destruction
The majority of MWC facilities in the USA are of modern design and employ
good combustion practices. As a consequence, the generation of PCDD/PCDF
precursors is curtailed which results in a comparatively low level of PCDD/PCDF
formation. The thermal designs of MWC facilities built in the 1950s, 60s and 70s
frequently support formation of substantial quantities of PCDD/PCDF precursors.
Incinerator-heat recovery boiler-electrostatic precipitator is typically
the process train of many of these earlier facilities. Temperatures in the range
of 400'- 650"F are usual for flue gas exiting the heat recovery boiler. This
temperature range and the presence of catalysts in the flyash have been found to
promote the rapid formation of PCDD/PCDF in MWC flue gases.
135
-------
A diagrammatic illustration of the effect of temperature-residence time on
PCDD/PCDF formation is shown in Figure 4. The residence time is measured from
the time that the flue gas exits the heat recovery boiler. The 220'C (428'F)
inclined line represents the rapid rate of formation of PCDD/PCDF during passage
of the flue gas through the electrostatic precipitator. After the facility had
been retrofitted with an EDV gas cleaning system, the rate of PCDD/PCDF formation
collapsed as represented by the line 61"C. When an EDV system is used as a
stand-alone gas cleaning system, PCDD/PCDF formation follows the "dash 61 'C"
line, i.e. maximum suppression, minimum formation.
Condensation, and hence capture, of PCDD/PCDF is temperature dependent.
This is illustrated in Figure 5. EDV systems typically operate in a temperature
range of 55-65"C (131-149'F) as against 132-160'C (270-320'F) in spray dryer-
fabric filter systems which can be important for retrofits of earlier generation
(pre-1984) of incinerators.
EDV technology has been developed to achieve +99% capture of PCDD/PCDF.
To alleviate possible future concerns over deposits of contaminated residues, the
destruction of PCDD/PCDF has been integrated into EDV technology. Test data from
commercial MWC shows that the destruction of captured PCDD/PCDF achieved in
Phase I of the development program is already at a 90% level.
CEM
The introduction of continuous emission monitoring of pollutants and the
steady reduction in averaging time imposes a new threshold of regulatory
responsibility onto incinerator plant operators. Variable feedstocks and
feedback control systems typically employed by spray-dryer-fabric filter systems
aggravate an already sensitive operational situation.
Figures 6,7, and 8 show simultaneous inlet/outlet CEM strip chart records
of S02, HC1 and NOX across an EDV system installed on a MWC. The heavy line
trace in each chart highlights the specified pollutant. The buffering capacity
of the EDV system for acid gases is well illustrated in their CEM strip chart
records.
The MWC at which the NOX strip chart (Figure 8) was recorded operates a
THERMAL DeNOX system for controlling the emissions of nitrogen oxides. The EDV
controls the ammonia slip and ammonium compounds produced. In addition, the EDV
produces a further significant reduction in NOX emissions. This supplementary
DeNOX capability has been complemented by the development of a wet DeNOX concept.
The performance and long term efficiency of this wet DeNOX as an integral feature
of an EDV system has been demonstrated on a 330 TPD MWC. The performance of the
wet EDV achieved by DeNOX concept is shown in Figure 9.
136
-------
Conclusion
In order to achieve maximum environmental and economic benefits from
municipal services, a comprehensive perspective is an essential first step.
An integrated approach that embraces source separation, recycling of
marketable materials and the thermal processing of residuals is applicable to the
disposal of municipal solid wastes. Thermal treatment ought to be viewed not
only for energy recovery but as an intermediate stage to the recovery of
construction materials and heavy metals from ash residues. Realization of such
objectives requires that the selection of technologies, especially gas cleaning
technology, be made on site-specific global environmental-economic criteria
rather than on institutionalized procedures. Environmental and economic
objectives have been demonstrated to be compatible and mutually beneficial.
137
-------
100
400 500 600 700 800 900
200 300
100
figure l - MERCURIC CHLORIDE FORMATION
2 HOUR CEM RECORD - MERCURY
Time
FIGURE 2 EDV PERFORMANCE
138
-------
Correlation between standard method
and continuous measurement of mercury
EDV Inlet
EDV Outlet
400
Standard Method microgram/Nm3
FIGURE 3 EDV PERFORMANCE
220"C
6I"C
dash 61 °C
Residence l ime
FIGURE 4 I'CDD - I'CDF
139
-------
Vapor Pressure - Temperature
25 50 75 Mil) 125
Temperature °C
FIGURE 5 - PCDD - PCDF
12 HOUR CEM RECORD - S02 ppm
I I I I ||j,
°°°S888888
V
I.
;1
/I
Mvv/t ^ 'u y Wv'VN
" OUTLET
FIGURE 6 - EDV PERFORMANCE
-------
12 HOUR CEM RECORD - HCL ppm
2000
100
INLET
OUTLET
figure 7 - EDV PERFORMANCE
12 HOUR CEM RECORD - NOX ppm
200
INLET
888888888
i i i i i
i i i i '
i i
i i
200
.-I.-, . i i •, .
I - •> V'' V ' ' v
A.'<"
'-"'V V '
FIGURE 8
OUTLET
EDV PERFORMANCE
Reproduced from
best available copy. Si&W
-------
240
220
200
180
1 60
140
120
100
80
60
40
20
0
V"
A,
A.
A,
fx
V
*A
/
\
V
A
w
J
\
/
-V" Ct If »
•>rrv'
-\r '
-V-
-
-------
TABLE 1 MERCURY CAPTURE
1. Mercury reacts with hydrogen chloride (HC1)
to form mercuric chloride (HgCl;) at temperatures
above 300°C.
2. Higher temperatures and higher HC1 concentrations
result in higher production rate of mercuric
chloride.
3. Mercuric chloride is water soluble.
4. Oxidation of mercuric chloride in low pH solution
promotes formation of stable compounds.
5. EDV absorption zone operates below pH2.
6. Metallic (atomic) mercury reacts with mercuric
chloride to form insoluble mercurous chloride
(Hg2Cl:).
7. Lower flue gas temperature promotes higher
mercury removal.
8. EDV system operates below 66°C (150°F).
143
-------
TABLE 2 EDV PERFORMANCE DATA - BASEL
Test Run
1
3
4
Unit No.
1
1
->
->
Boiler Outipt
Gas flow
Nm3/h wet
84800
84800
89100
76800
Gas Temp. °C
182
180
183
181
HC1
1170
1800
550
740
SO,
110
150
130
150
Hg
0.193
0.128
0.166
0.126
EDV Outlpr
Gas Temp. "C
57
56
57
57
Particulate
0.9
0.5
0.6
1.5
HC1
<1 DL
<1 DL
<1 DL
<1 DL
SO,
<11 DL
<13 DL
<24 DL
<13 DL
Hg
0.012
0.013
0.005
0.016
Note: 1. Values in mg/Nrrr dry corrected 11% 0:.
2. DL - detection limit for SO, determined by sampling duration.
144
-------
TABLE 3 SAMPLE BLANKS FOR MERCURY
AT BASEL MSW INCINERATION PLANT (SWITZERLAND)
MSW TRAIN #
Date
Test Run no.
03/20/90 04/04/90 04/06/90 03/20/90 04/06/90
7 A 13A 23A 7B 23B
Hg sampled in micrograms
Volume sampled in N1 dry (2 hrs)
Hg sample blank
concentration mg/Nm3 dry
1.8
200
0.009
2.0
200
0.010
0.6
200
0.003
0 . 3
200
0.002
0.5
200
0.003
-------
TABLE 4
TABLE 4 EDV PERFORMANCE
MERCURY LIXIYIATION TEST DATA
CH J USA
BUS 88 KK 13 TCLP
Residue mg/kg 126 126 126
Leachate
Concentration mg/1 0.002 0.0005 0.002
Regulatory Limit
(maximum) mg/1 0.01 0.005 0.2
TABLE 5 EDV PERFORMANCE DATA
HEAVY METALS
5/18/88 5/20/88
Inlet Outlet Inlet Outlet
Zn 0.406 0.004 0.42 0.007
Cd
0.0084
<0.001
0.0038
<0.001
Ni
0.009
0.002
0.074
<0.005
Pb
0.21
<0.001
0.144
0.005
Cu
0.154
<0.001
0.116
<0.001
Cr
0.098
<0.001
0.108
<0.0017
As
0.004
<0.001
0.006
<0.001
Data as mg/Nm3
to 7% 02 dry.
< means under detection limit
146
-------
TABLE 6 PCDD - PCDF CAPTURE
1. PCDD - PCDF formation is temperature dependent.
2. PCDD - PCDF condensation is strongly temperature
dependent.
3. The presence of carbon in scrubbing liquid
accentuates PCDD - PCDF capture.
4. PCDD - PCDF are destroyed in the scrubbing liquid.
147
-------
Intentionally Blank Page
148
-------
SESSION 4C:
FLUE GAS CLEANING: PCDD/PCDF CONTROL
Co-Chairmen:
David S. Beachler
Resource Energy Systems Division
Westinghouse Electric Corporation
Pittsburgh, PA
Marjorie J. Clarke
Environmental Consultant
New York, NY
Preceding page blank
-------
Intentionally Blank Page
150
-------
The work described in this paper was not funded by the U.S. Environmental
Protection Agency. The contents do not necessarily reflect the views of the
Agency and no official endorsement should be inferred.
APPLICATION OF DeNOx-CATALYSTS FOR THE REDUCTION OF
PCDD/PCDF AND OTHER PICfi FROM WASTF. INCINERATION FACILITIES
BY CATALYTIC OXIDATION
H.Hagenmaier, K.-H. Tichaczek, H. Brunner, G. Mittelbach*
Institute of Organic Chemistry, University of Tubingen,
D-7400 Tubingen, FRG
^Deutsche Babcock Anlagen AG, D-4150 Krefeld 11, FRG
ABSTRACT
In laboratory experiments it was demonstrated that PCDD/PCDF
can be decomposed catalytically in the presence of oxygen in the
temperature range of 150 - 500°C. Several commercially available
oxidation catalysts and other catalysts on the basis of metals
and metal oxides were tested. It was found that catalysts produced
for the selective catalytic reduction of nitrogen oxides in the
presence of ammonia (SCR-catalysts) exibit some unexpected
properties as oxidation catalysts for the decomposition of
organohalogen compounds in general and of PCDD/PCDF in particu-
lar. While most recognized oxidation catalysts did not exhibit
any activity below 300°C under the conditions of our laboratory
experiments, SCR catalysts on the basis of Ti02 showed under
otherwise identical conditions high activity as oxidation
catalysts for organochlorine compounds at temperatures of 250 -
350°C. Under appropriate conditions, organochlorine compounds
could be decomposed by catalytic oxidation without PCDD/PCDF
formation. In pilot plant studies at municipal waste incinerators
we were able to demonstrate that catalysts originally developed
for selective reduction of nitrogen oxide are capable of reducing
the PCDD/PCDF emissions to levels below 0.1 ng TCDD-equivalents
(I-TEQ) per m3. Other products of incomplete combustion (PICs)
are reduced simultaneously. Prerequisites for this PCDD/PCDF
reduction are residence times in the range of 0.3 to 0.7 seconds
and (either the absence of ammonia or) ammonia concentrations of
less than 10 ppm. When run in the temperature range of 250 -
350°C, these catalysts can be used either solely as oxidation
Preceding page blank m
-------
catalysts to lower the PCDD/PCDF emission, or as combined
catalysts to also simultaneously reduce nitrogen oxides and
oxidise halogenated compounds. In the latter case the catalyst
volume has to be increased accordingly.
Since SCR catalysts are already in use in power stations and
municipal waste combustors (MWCs) and have proven their resi-
stance against catalyst poisoning, this technique for the
reduction of PCDD/PCDF emissions from MWCs can be easily applied
on a full technical scale. This full scale applicability has
recently been demonstrated by measurments at the municipal waste
incinerator at Vienna, Austria.
INTRODUCTION
At the 6th International Dioxin Symposium at Fukouka we
reported upon the catalytic effects of filter ash on the
decomposition of PCDD/PCDF /1,2/. Subsequently we investigated
the source of this catalytic activity in filter ash. We found
that every metal and metal oxide tested exhibited some catalytic
activity in the decomposition of PCDD/PCDF in the temperature
range of 250 to 350°C. In extensive laboratory tests we evaluated
the conditions under which not only PCDD/PCDF could be decomposed
by oxidation catalysts in the presence of oxygen, but also
organochlorine compounds in general without the formation of
PCDD/PCDF /3/. In order to apply these findings to the reduction
of PCDD/PCDF emissions from waste incineration facilities
effective PCDD/PCDF decomposition in the temperature range of
250 to 350°C appeared to be essential since this catalytic
oxidation is best installed as the final gas cleaning device. We
found in our laboratory tests that SCR catalysts of the titanium
oxide, ferric oxide and zeolithe type in combination with the
usual additions of other metal oxides are also very good oxidation
catalysts for organochlorine compounds when applied in the
absence of ammonia or at low ammonia concentrations. Some of
these catalysts exhibit this property very efficiently at the
desired temperature range of 250 to 350°C.
We therefore tested such catalysts in pilot plant studies at
municipal waste incinerators.
152
-------
LABORATORY EXPERIMENTS
With the model compounds hexachlorobenzene, 2,4,8-trichlo-
rodibenzofuran and tetrachloroethene, the efficiency of decom-
position with oxidation catalysts was tested in laboratory
experiments . It could be demonstrated that in the temperature
range of 200-500°C and space velocities of about 2000 h-1 these
compounds can be successfully decomposed in the presence of oxygen
(air) and water without the formation of PCDD/PCDF /3/. We could
also demonstrate that PCDD/PCDF are much more readily decomposed
than hexachlorobenzene or terachloroethene.
Tetrachloroethene therefore proved to be a very good model
compound in laboratory tests of the temperature dependence of
the decomposition efficiency of oxidation catalysts for organo-
chlorine compounds. While most of the commercially available
oxidation catalysts tested show only limited efficiency below
350°C, the SCR catalysts on titanium oxide basis with dotations
of vanadium oxide exhibit good efficiencies at temperatures as
low as 250°C. This is shown in Figure 1. Here three different
"low dust" deNOx catalysts are compared with one "high dust"
deNOx catalyst. While at high temperature (>400°C) there is little
difference in the decomposition efficiency of the four catalysts,
distinct differences are apparent at temperatures below 300°C.
100 T
90 -
80 -
% decomp.
70 -
60 --
50 -
225 275 325 375 425 T (°C)
Figure 1: Temperature dependence of hexachloroethene decom-
position by catalytic oxidation on deNOx catalysts. Kal 1 to Kat
3 "low dust"-type on Ti02 basis. Kat 4 "high dust"-type on Ti02
basis. Air/20% H2O with 300 ppm hexachloroethene; space velocities
2000 h-1.
O—S—
D kat 1
•X- kat 2
A kat 3
O kat 4
153
-------
It has been shown that there also exists a good correlation
between the SO2 oxidation rates of individual catalysts and their
efficiency in PCDD/PCDF decomposition /4/. Therefore the SO2 to
SO3 conversion factor of the individual catalyst is also a good
indicator for its efficiency in PCDD/PCDF decomposition.
PILOT PLANT STUDIES
The laboratory experiments suggested that oxidation catalysts
of the SCR type could be successfully used for the reduction of
stack gas emissions of PCDD/PCDF and related compounds in waste
incineration plants. Thus at three different waste incinerators
pilot facilities, originally installed for the test of SCR
catalysts with regard to NOx reduction, were used to test the
reduction of PCDD/PCDF and related compounds. A side stream of
the off-gas in the range of 100 to 250 m3/ h was diverted through
the catalytic reactor after preheating to the desired test
temperature. The catalyst volume was divided into various parts,
so that the influence of the catalyst volume could be evaluated
in addition to temperature, catalyst configuration and NH3-ad-
dition to the flue gas. In case of small catalyst volumes (high
space velocities) and addition of NH3 just the denox reaction
took place while the decomposition of PCDD/PCDF was suppressed.
In the absence of NH3 or at NH3 concentrations of a few ppm, under
otherwise identical conditions, the PCDD/PCDF reduction was >95%
/ 4 / .
This means that in order to obtain the neccessary reduction
in PCDD/PCDF emissions by catalytic oxidation in combination
with NOx reduction one has to increase the volume of the catalytic
reactor to obtain sufficient catalytic volume for the oxidation
reaction.
Representative results for different starting concentrations
of PCDD/PCDF are shown in detail in Table 1. The test Illb shows
that even under very low starting concentrations a reduction of
PCDD/PCDF concentration was obtained. This demonstrates that
under the conditions of catalytic oxidation in the temperature
range of 250 to 350°C no relevant PCDD/PCDF de novo synthesis
had occurred.
154
-------
Examples from
Test II
Test Ilia
Test lllb
in
out
in
out
in
out
ng/cbm
ng/cbm
ng/cbm
ng/cbm
ng/cbm
ng/cbm
Tetrachlordibenzodioxins
23.9
0.19
0.07
0.03
0.023
n.d.
Pentachlordibenzodioxins
25.2
0.37
1.09
0.05
0.029
n.d.
Hexachlordibenzodioxins
27.2
0.55
3.66
0.06
0.068
0.032
Heptachlordibenzodioxins
18.4
0.44
2.68
0.09
0.056
0.038
Octachlordibenzodioxin
8.3
0.41
1.20
0.06
0.113
0.031
Total PCDD
103.0
1.96
8.70
0.29
0.289
0.101
Tetrachlordibenzofurans
28.8
0.36
1.65
0.62
0.353
0.087
Pentachlordibenzofurans
20.4
0.53
5.08
0.45
0.384
0.074
Hexachlordibenzofurans
9.6
0.35
6.21
0.16
0.503
0.033
Heptachlordibenzofurans
5.1
0.24
3.09
0.14
0.258
0.013
Octachlordibenzofuran
0.5
0.10
0.69
n.d.
0.041
n.d.
Total PCDF
64.4
1.58
16.72
1.37
1.539
0.207
2,3,7,8-Tetrachlordibenzodioxin
0.17
n.d.
n.d.
n.d.
n.d.
n.d.
1,2,3,7,8-Pentachlordibenzodioxin
0.47
0.005
0.039
n.d.
n.d.
n.d.
1,2,3,4,7,8-Hexachlordibenzodioxin
0.58
0.011
0.039
n.d.
0.007
0.004
1,2,3,6,7,8-Hexachlordibenzodioxin
1.24
0.036
0.184
0.002
0.008
0.006
1,2,3,7,8,9-Hexachlordibenzodioxin
0.68
0.025
0.123
0.002
0.006
n.d.
1,2,3,4,6,7,8-Heptachlordibenzodioxin
9.67
0.250
0.878
0.048
0.029
0.051
2,3,7,8-T etrachlordibenzofuran
0.63
0.018
0.106
0.013
0.012
0.006
1,2,3,7,8-Pentachlordibenzofuran
1.53
0.035
0.167
0.033
0.052
0.007
2,3,4,7,8-Pentachlordibenzofuran
1.17
0.023
0.233
0.024
0.038
0.005
1,2,3,4,7,8-Hexachlordibenzofuran
0.97
0.037
0.495
0.014
0.073
0.005
1,2,3,6,7,8-Hexachlordibenzofuran
1.09
0.041
0.800
0.030
0.089
0.005
1,2,3,7,8,9-Hexachlordibenzofuran
0.06
0.004
0.038
0.003
n.d.
n.d.
2,3,4,6,7,8-Hexachlordibenzofuran
0.73
0.030
0.953
0.013
0.053
0.003
1,2,3,4,6,7,8-Heptachlordibenzofuran
3.96
0.210
1.820
0.090
0.204
0.013
1,2,3,4,7,8,9-Heptachlordibenzofuran
0.13
0.006
0.370
0.008
0.016
n.d.
l/TEQ (NATO/CCMS)
1.81
0.041
0.451
0.023
0.049
0.006
% reduction in l/TEQ
97.70
94.90
86.90
Table 1. PCDD/PCDF-analyses of waste incinerator off-gases
before (in) and after (out) catalytic oxidation. Examples from
test series with a range of initial concentrations.
In Figure 2 residual PCDD/PCDF concentrations after catalytic
oxidation from a range of initial concentrations in relation to
the catalyst volume are shown. In the concentration range of
interest the relative reduction in PCDD/PCDF concentration under
comparable conditions was slightly dependent on the initial
concentration. This means that in order to obtain a final
concentration of <0.1 ng TEQ/m3, the surface of the catalytic
reactor has to be adjusted according to the expected initial
concentration range. Further optimization appears possible by
155
-------
adjusting the catalyst configuration and/or the reaction tempe-
rature .
Under comparable conditions no substantial difference in the
degree of PCDD/PCDF reduction could be detected between new
catalysts and those which had been used for 6000 h under both
denox conditions and oxidation conditions. This already indi-
cates that the reduction of PCDD/PCDF is due to catalytic
oxidation and not due to adsorptive effects. Analyses of catalyst
material after use for more than 6000 h in the pilot facility
did also not show any accumulation of PCDD/PCDF by adsorption.
PCBs, chlorobenzenes and chlorophenols were shown to be
reduced by catalytic oxidation to the same extent as reported
here for PCDD/PCDF.
100
a
u
a.
v. 0.001 -i
go 0,5 1 1,5 2 2,5 3
ratio of catalyst surface to gas flow (rel.)
Figure 2. Residual concentrations after catalytic oxidation
in relation to the outer catalyst surface. The inlet concentra-
tions of PCDD/PCDF are the values at catalyst volume zero, the
outlet concentrations are correlated to different catalyst sizes.
156
-------
Recently a full scale deNOx-unit at the MWC Spittelau/Vienna
was tested with regard to its capacity in lowering PCDD/PCDF
emissions /6/. With a catalyst volume designed for deNOx function
only, Reduction of PCDD/PCDF of 90 to 95% could be obtained
under deNOx conditons. With increased catalyst volume further
reduction can be expected.
CONCLUSIONS
Laboratory tests and pilot studies at waste incinerators have
shown that catalysts of the SCR type, especially those on Ti02
basis, are also excellent oxidation catalysts for the decompo-
sition of PCDD/PCDF and related compounds.
PCDD/PCDF levels <0.1 ng TEQ/m3 in stack gas were obtained in
pilot plant studies using SCR catalysts.
Simultaneous reduction of NOx and PCDD/PCDF requires additio-
nal catalyst volume either before or after the denox reaction.
No de novo synthesis of PCDD/PCDF and no accumulation of
PCDD/PCDF in the catalytic reactor was observed under the
conditions employed.
REFERENCES
1. H. Hagenmaier, M. Kraft, H. Brunner, R. Haag, 6th
International Symposium on Chlorinated Dioxins and Related
Compounds, Fukuoka 198 6
2. H. Hagenmaier, M. Kraft, H. Brunner, R. Haag, Environ. Sci.
Technol. 21, 1080-1084 (1987)
3. H. Hagenmaier, VDI-Berichte 730, 239-254 (1989)
4. H. Hagenmaier, G. Mittelbach, VGB Kraftwerkstechnik,
70, 491-493 (1990)
5. H. Hagenamier, K.-H. Tichaczek, H. Brunner, G. Mittelbach,
10th International Symposium on Chlorinated Dioxins and
Related Compounds, Bayreuth 1990
6. R. Boos, K. Scheidl, T. Prey, F. Wurst, R.P. Kuna,
Chemosphere, accepted for publication
157
-------
Intentionally Blank Page
158
-------
The work described in this paper was not funded by the U.S. Environmental
Protection Agency. The contents do not necessarily reflect the views of the
Agency and no official endorsement should be inferred.
TECHNIQUES FOR DIOXIN EMISSION CONTROL
by
Vladimir Boscak, Vfclund Miljoteknik A/S,
Copenhagen, Denmark
George Kdtynek, V®lund USA Ltd.,
Oak Brook, Illinois
ABSTRACT
Only several years ago it was generally thought that PCDD/F were either
formed and/or not destroyed in the combustion of municipal solid waste
(MSW). Today it is generally accepted that they are formed by the de-novo
synthesis downstream from the combustion chamber at a temperature of about
300°C.
Fly ash, specifically copper, acts as a catalyst for the formation of
free chlorine which chlorinates the aromatic structures through a Deacon
lite process. The transition of unburned carbon through a mesqphase may play
an important role in the formation of precursors.
The de-novo synthesis of PCDD/F hamologues and isomers can be best
postulated as the unimolecular reactions involving an activated complex.
Once the FCDDs/Fs are formed, they are partitioned between gas, liquid, and
solid phases depending on the spatial and temporal temperature profile in
the flue gas.
A correlation between PCDD/F emissions from MSW incinerators and flue
gas temperatures was quantified frcm data in the Danish Ministry of Environ-
ment study. An important practical implication of this correlation is that
exit flue gas temperature can serve as a surrogate for estimating PCDD/F
emissions.
An understanding of chemico-physical behavior of PCDD/F in MSW combus-
tion allows the optimization of operating conditions to minimize their emis-
sions.
Preceding page blank 159
-------
Hie techniques for controlling PCDD/F emissions can be divided into
three categories:
o Minimization of precursors for PCDD/F formation
o Prevention or reduction of PCDD/F formation
o Post formation emission control and destruction
Velund MiljeJteknik &/S has implemented several mitigative techniques
for the reduction of PCDD/F emissions in the design of MSW combustion
plants.
INERDmCITON
Ever since the discovery of emission of polychlorinated dibenzodioxins
and dibenzodifurans, PCDD/F, commonly referred to as dioxins, frcm the waste
incineration facilities in the 1970s, they have been a hot topic of discus-
sion. Both the scientific community and the general public are still in-
volved in the controversy over their environmental impact, and specifically
their health risk. The wide implementation of solid waste incineration, in
spite of the landfill shortage situation, has been frequently opposed by the
general public. One of the major arguments against incineration has been the
health risk associated with the emissions of dioxins.
It has been discovered in recent years that the formation/emission of
dioxins is a general phenomenon associated with combustion. Dioxins have
been detected in the exhausts from automobiles, power plants, petroleum
refineries, pulp and paper mills, etc. They are also generated in the every-
day human activities like grilling of meats, dry cleaning, and burning logs
in the fire places.
DIOXIN REGULATIONS
The future trends in the regulation of MSW incinerator emissions will
be dictated by health risk standards. Dioxins, specifically TCDD, are the
most toxic chemicals known to mankind. Until a couple of years ago, dioxin
formation and/or not destruction was thought to take place in the combustion
zone and therefore, remedial actions revolved around the three Ts of combus-
tion.
Consequently, in the past, several European countries adopted regula-
tions specifying furnace operating conditions such as temperature, residence
time, and turbulence, while at the same time no specific emissions limits
were mandated. New, however, most European regulations call for a dioxin
l.'init of 0.1 ng/m based on the toxic equivalents (TCDD).
Recent U.S. regulations for the dioxin emissions from MSW combustion
plants distinguish between new and existing facilities and their capacity.
160
-------
3
For new facilities total dioxin/furans emission limits are 75 ng/m for
installations under 250 tons/day capacity and 5-30 ng/m for larger units.
For the existing facilities, the regulations limit total TCDD/F emis-
sions to 500 ng/m for plants under 250_tons/day, 125 ng/m for plants
between 250-2200 tons/day and 5-30 ng/m for the larger facilities.
FORMATION OF DIOXINS IN WASTE INCINERATION
The formation/emission of dioxins in the waste incineration has been
extensively studied, resulting in a better understanding of their formation
and reduced emissions.
In recent years it has been shewn conclusively that dioxins are
destroyed in the combustion zone but are formed downstream from the furnace
at a temperature of about 300°C. The exact mechanism(s) of formation are not
yet fully understood, but it appears that at least two mechanisms pertain.
Chlorine donors and ring donors react in at least two different ways, and
the dioxin products reach equilibrium, resulting in predictable isomer
ratios.
COPPER CATALYTIC ACTION
Copper present on the fly ash particles frcan waste combustion plays an
important role in a dioxin formation. An important step is the formation of
free chlorine through a Deacon-like reaction.
The overall reaction in the Deacon process is:
4 HC1 + 02 —> 2 Cl2 + 21^0 - 116.8 kT (350°)
This reaction is catalyzed by CuCl_/CuCl and proceeds in three consecutive
steps.
The chlorine radical CI" formed through cleavage of chlorine is essen-
tial for chlorination of aromatic structures in dioxin formation.
The oxidative cyclization of the aromatic structures is also catalyzed
by copper.
The catalytic activity of copper in PCDD/F formation indicates one way
for their reduction. Vogg (1) and Naikwadi (2) have demonstrated through
laboratory tests with fly ash that a reduction in PCDD/F formation can be
accomplished by the inhibition of fly ash catalytic activity. Vogg demon-
strated about 8 5% reduction in the dioxin formation when ammonia was present
in the flue gas. Without the presence of ammonia, dioxin concentration was
6,197 ng/g in fly ash while in the presence of 300 mg NH_/m only 914 ng/g
were measured in the fly ash.
161
-------
Naikwadi shewed that both triethyl amine and CaO inhibit dioxin forma-
tion. Of more practical significance is the six fold reduction accenplished
when 2% CaO is mixed with fly ash.
Vikelsoe (3) also shewed the beneficial effect of CaO in the full scale
incineration tests, but to a lesser degree.
An obvious, but impractical technique for dioxin reduction would be the
elimination of copper from wastes.
THE POLE OF CARBON IN DIOXIN PQEMATTON
The combustion of refuse takes place within luminous diffusion flames,
luminosity resulting frcnt the radiation of carbon in the yellcw spectral
region. The carbon structures formed in the flames consist of a number of
roughly spherical particles strung together lite pearls an a necklace. As
the thermal transformation of carbon proceeds, the lower molecular weight
components lite benzene or phenol are vaporized while the higher molecular
weight aromatic structures polymerize forming a pitch.
The basic organic molecules like benzene and phenol are readily chemi-
sorbed on the fly ash metallic surfaces. Ordered structures of adsorbed
benzene have been observed on several metallic surfaces. The optimal struc-
ture of benzene adsorbed on rhodium is shewn in the top middle of Figure 1
(4). It shews that four benzene molecules form a rectangular unit cell, but
the molecules are distorted and shew losses of seme vibrational frequencies.
FORMATION of dioxins through an activated complex
The formation of dioxins and furans (PCDD/F) in refuse incineration
from aromatic structures, chlorine, oxygen, and hydrogen under copper cata-
lytic action through an activated carrplex was proposed by Boscak (5).
Figure 1 illustrates the transition of aduct benzene molecules into an
activated complex where chlorination and oxidative cyclization takes place.
The energy of an activated complex is distributed among all its inter-
nal degrees of freedom. Some degrees of vibrational freedom are replaced by
translational motion along the reaction coordinates. This part of energy
gives rise to the transition of the activated complex either in the direc-
tion toward the products (PCDD/F) or back to the aduct molecules.
If dioxins are indeed formed frcm the aromatic structures oil metallic
surfaces, the obvious technique for their reduction is the complete combus-
tion of carbon in the combustion zone.
Good mixing of refuse with primary cctribustion air and introduction of
secondary air into the flame zone to create a turbulent premixed flame is an
important technique for the reduction of soot (particulate carbon) forma-
tion.
162
-------
DEACON PROCESS
4 HCl + O-
<-
> 2 Cl2+ 2H20
3 3
U (J
3
(J
ON
Lo
GAS
Cu1
i +
Cu
I, IOUID
MELT
Cli-' f
SOL I D
£ul±—
Q i
M
0
+
D 4 '
F-*
M
4-
r
Cu
ff | c,/
i '¦ v /
+ ^ +
Cu Cu
ACTIVATED COMPLEX
Figure 1 Formation of PCDD/F through an activated complex
Cu
Cu
OPTIMAL STRUCTURE
OF BENZENE ON
METALLIC SURFACE (4)
PCDFs
PCDDs
Reproduced from
best available copy.
-------
The presence of calcium, barium, strontium, and lithium salts has a
strong inhibiting effect on soot formation. On the other hand, presence of
sodium, potassium, and cesium salts increases soot formation.
OTHER CCMEUSTTON CONDITIONS
Oxygen concentration in the flue gas plays an important role in FCDD/F
formation. The increase of 0_ content (high excess air) increases FCDD/F
formation drastically with an especially pronounced formation of hepta and
octa congeners.
The presence of water vapor is also important. While the "dry" condi-
tions result in the formation of higher hcmologues, the "moist" conditions
produce a maximum of the hexa isomers with increased penta and tetra iso-
mers.
In general, good combustion i.e. high temperature, flame oxidative
state, lew CD content, etc. results in a trend toward higher chlorinated
homilogues. With poor combustion, lew temperature, and high CD content more
lew chlorinated hcmologues are formed.
PHASE EQUILIBRIA
Once the dioxins are formed they are partitioned between the vapor,
liquid, or solid phase depending on the spatial and temporal tenperature
profile within the flue gas stream.
The fugacity calculations by Schramm et al. (6) shewed that the emis-
sion of PCDD/F from incinerators depends on the partition of these canpounds
between stack gas (gas phase) and fly ash (solid phase). The calculations
shewed that the separation of fly ash at lew temperatures should decrease
the total emissions up to a factor of one hundred. Increased fly ash con-
centrations would also enforce the separation effect, whereas lew particle
concentrations in the gas phase lead to high relative amounts of PCDD/F in
the gas phase.
PCDD/F present in the gas phase can, upon temperature reduction, either
condense on the fly ash or undergo homogeneous nucleation.
Since the aerosols formed through the homogeneous nucleation of FCDDs/
Fs would be in the sutamicron range, they would be very difficult to collect
in the particulate control equipment.
The presence of PCDD/F aerosols in the flue gas and the problems with
their collection in the sampling train have been reported by Fangmark et al.
(7). The result of the study shewed that an aerosol filter has to be intro-
duced after the condenser to collect particle-bound PCDD/F efficiently.
The PCDD/F condensed on the fly ash are readily collected in the
particulate control equipment and can be subsequently destroyed via thermal
treatment.
164
-------
THE DANISH STUDY ON DIOXIN EMISSIONS
An important study on dioxin emissions from the waste incineration
plants was carried out in Denmark frcm 1986 until 1989. The Danish Ministry
of the Environment issued the report on the study shewing that emissions are
much lower than expected, and provided the data correlating the emissions
with the operating parameters (8). Several correlations between dioxin
emission and operating parameters were seen. It should especially be noticed
that the emissions of dioxins increase with decreasing combustion tempera-
ture, increasing exit gas temperature, and increasing content of carbon
monoxide.
One interesting tut unexplained finding was that dioxin emissions
varied significantly at the same plant when measured at different times.
In order to explain the variation in dioxin emissions at various times
an attempt was made by the author to correlate the incinerator load and the
exit gas temperature since the incinerator load was one of the test para-
meters. This correlation for the two plants where tests were conducted on
several occasions is shewn in Figure 2.
It can be seen that the incinerator load and the exit gas temperature
do not have a singular functioned relationship. The correlation is notice-
able only if one correlates the data for each test period.
OQRRELATION BETWEEN EXIT GAS TEMPERATURE AND PCDD/F EMISSIONS
The significance of the different exit gas terrperature regimes during
different testing periods is that it clearly correlates with the PCDD/F
emissions: the higher the exit gas terrperature the higher are the PCDD/F
emissions.
The correlation between exit gas terrperature and PCDD/F emissions at
100% incinerator loads for all 5 tests is shewn in Figure 3 (9). The plot
shews a clear exponential dependency between exit gas terrperature and PCDD/F
emissions. While the total PCDD/F emissions are about 100 ng/m at a flue
gas terrperature of approximately 200 °C it increases to about 1000 ng/m at a
flue gas terrperature of approximately 240°C.
The simple explanation of this correlation is the vapor phase - solid
phase partition of PCDD/F which depends on the terrperature.
An important practical implication of this correlation is that the exit
gas terrperature can serve as a surrogate for PCDD/F emissions with several
advantages including simplicity and measurement in real time.
The most important and easiest technique to implement for dioxin
emissions control is to keep exit gas terrperature lew. This can be accom-
plished by several means like use of an economizer in the boilers, cooling
165
-------
Figure 2 Correlation between exit gas
temperature and load
Exit
temp
260- -
FENC? NOP D
TEST 2 +
TEST 3
RENO SYD
TEST 1 ~
TEST 2 O
160-"
1 1 0
130 * Load
Figure 3 Correlation between PCDD/F emissions
and exit gas temperature
*
-+-
-+-
200 2 20 2 4C cC
Exit gas tenperature
166
-------
of the flue gas either by heat exchangers or water injection, wet or semi-
dry flue gas treatment etc. Most of the modern waste-to-energy plants
designed by Volund have lew exit gas temperatures, because they are equipped
with semi-dry flue gas treatment systems.
THE ROLE OF BOILER CLEANING INTERVALS
The different functional relationships between boiler load and exit gas
temperature can be easily explained by the boiler cleaning intervals.
As the time after a manual boiler cleaning elapses, more and more fly ash
accumulates on the boiler tubes. The fly ash and scale on the boiler tubes
serve as an insulation reducing the heat transfer rate from the bulk gas to
the boiler tubes. The consequence is a higher exit gas temperature and
corresponding higher PCDD/F emissions.
The discussion with plant personnel at the Reno-Nord facility confirmed
that in the tests with higher emissions more time had elapsed between the
manual boiler cleaning and the testing.
An interesting phenomenon regarding dioxin emissions was observed
during an incinerator test in Denmark (3). After a manual boiler cleaning,
the dioxin emissions rose dramatically, but returned to normal after two
days. The researchers speculated that possibly the cleaning had temporarily
changed the chemical properties of the boiler surfaces, thereby introducing
enhanced catalytic dioxin generating activity or previously formed dioxins
were exposed.
High PCDD/F emissions during boiler cleaning by steam soot blowing was
reported by Vogg (4). They concluded that the high emissions were probably
due to the increased moisture content in the flue gas.
One technique for reducing of dioxin emissions is therefore to optimize
the cleaning of boilers. Boilers should be cleaned continuously with as
little steam as possible when steam soot blowing is used. Switching to
cleaning with compressed air or mechanical rappers should also be con-
sidered. Most modern waste-to-energy plants designed by Volund use either
steel shots or mechanical rappers to clean the boiler tubes as required.
DIOXIN EMISSION TESTS AT AMAGERFORBRSNDING
The Danish Ministry of Energy sponsored a study on the influence of
flue gas cleaning system operating parameters on the removal of dioxins
(10). The investigation was carried cut on the fourth incineration line at
Amagerforbraending supplied by Vfolund Miljcteknik as shewn in Figure 4. This
line uses a combination of grates and a rotary kiln for the complete com-
bustion of refuse. The boiler is cleaned by a mechanical rapping system vised
in the recent Volund installations. The Volund-Limar flue gas cleaning
system consists of a spray dryer absorber with a rotary atomizer followed by
a pulse-jet fabric filter. The process flew diagram of the Volund-Limar
system is shewn in Figure 5. The system treats 70,000-95,000 Nm /h of flue
167
-------
m
Ji i
3. Boiler
4. Spray Absorber
5. Fabric Filter
6. Ash and Clinker Transport
Figure 4 Sideview of the fourth incineration line at Amagerforbraending
Reproduced trom
best available copy.
-------
Water
Reactor
Boiler
,-M—
Container
Wnshwatrr
t ank
M i x i n q
tank
Semi-dry absorption system V0lund-Limar
Figure 5
-------
gas at a spray dryer absorber temperature of 140"C. The system operates in a
single pass mode and uses a suspension of hydrated lime as the absorbent.
The test program was designed to investigate the effect of the spray
dryer absorber outlet temperature as well as the use of activated carbon and
active lime on dioxin removal.
Results of FCDD/F emissions and calculated Nordic toxic equivalents for
the different tests are shewn in Table 1 as reported by Nielsen et al. (10).
Reduction of total PCDD/F is higher than 98% and for Nordic toxic equiva-
lents higher than 99.5% in all cases. Outlet values are in all cases signi-
ficantly belcw the 0.1 ng Nordic toxic equivalents/Nm at 10% 0_. Neither
addition of activated carbon nor of active hydrated lime seems to have any
effect on PCDD/F removal.
Figure 6 shows the correlation between exit gas temperature and dioxin
emission for three plants including Amagerforbrasnding. Good exponential
dependency of dioxin emissions on temperature is noticeable through the four
orders of magnitude of dioxin concentrations.
The conclusion of this study was that using a spray dryer absorber
followed by a fabric filter requires no modification in normal-operation to
secure an emission level of 0.1 ng Nordic toxic equivalents/Nm at 10% 0 .
Even during start-up conditions, where an inlet level of 50 ng Nordic toxic
equivalents/Nm was measured, the outlet level was belcw the 0.1 ng limit.
SUMMARY OF TECHNIQUES FOR DIOXIN EMISSIONS CONTROL
The techniques for PCDD/F emissions control can be divided into three
categories:
1. Minimization of precursors for PCDD/F formation
Good combustion is essential for the minimization of reactive aromatic
species which generate the precursors such as benzene, phenol etc. for
PCDD/F formation. The keys to good combustion are the three Ts of
combustion with the mixing of fuel and combustion air (both primary and
secondary) being the most important. Reduction of copper in refuse is
beneficial though that may not be practical. A low oxygen content in
the flue gas will result in a reduction of PCDD/F formation.
2. Prevention or reduction of PCDD/F formation
The inhibition of copper catalytic activity in PCDD/F formation can be
accampl.shed in the presence of ammonia, lime, etc. The continuous
cleaning of boiler tubes and avoidance of steam use will reduce PCDD/F
formation. The removal of fly ash frcm flue gas at temperatures above
300°C will reduce fly ash (copper) catalytic action.
170
-------
TABLE 1: SPRAY DRYER ABSORBER (SDA) AND PULSE JET FABRIC FILTER (PJF)
OPERATING PARAMETERS AND RESULTS (10)
SDA Temperature
140 °C
127 0 C
Additive
Carbon kg/h
Active lime kg/h
4.5* 0.5 1.5 4.5
14
1.5 4.5
14
SDA. PCDD+PCDF
in
PJFout PCDD+PCDF
% Rem PCDD/PCDF
132 2170 283 276 201 278
2.1 3.2 1.2 2.4 1.1 3.5
Average (2.25)
98.4 99.9 99.6 99.2 99.5 98.8
254 154 154 307
1.3 0.37 0.65 2.8
Average (1.3)
99.5 99.8 99.6 99.1
SDA. N.t.e. TCCD
in
PJFout N,t-e- TCCD
% Rem. N.t.e. TCDD
2.8 50 4.8 8.3 4.0 7.6
.0076 .050 .0075 .045 .035 .015
99.7 99.9 99.8 99.5 99.1 99.8
7.7 5.0 4.5 4.9
.0047 N.D. .002 .043
99.9 =100 =100 99.1
PFJ .
out
N.t.e. max. TCDD
.027 .079 .022 .060 .052 .033
.020 .016 .026 .056
3
All numbers in ng/Nm dry at 10% O
N.t.e. = Nordic toxic equivalents
N.D. = Non detectable
* = Simulated start-up
-------
1000
PCDD/F
ng/m
Reproduced from
best available copy.
1 ' ' '
'
i ; 1 i i i
1 • /¦ 1
1 i : i 1; | ,
1 ¦ 1 ¦ 1 i 1 • I
[ -1 i il : ¦ ' !
1 J
•t 1
1
. ,J„ ; 1
100 140 180 220 2 6 0°C
Figure 6 Correlation between exit gas temperature
and PCDD/F emissions
£ Reno-Nord and Syd plants
| Amager plant
172
-------
3. Post-formation emission control and destruction
Lew exit flue gas exit temperature and high degree of particulate (fly
ash with condensed PCDD/F) removal is essential for prevention of
releases of PCDD/F to the atmosphere. The exit gas temperature can
serve as a surrogate for PCDD/F emissions measurement. Adsorption of
PCDD/F on filter cake or an activated carbon bed will also reduce the
emissions. PCDD/F adsorbed on fly ash can be destroyed by returning or
alternative thermal or catalytic treatments.
The implementation of fully automatic computer operated incineration
line controls, continuous mechanical boiler cleaning and use of a semi-
dry flue gas treatment system (consisting of spray dryer absorber and
fabric filter) at Amagerforbraending (supplied by Volund Miljateknik)
has demonstrated that dioxin emissions can be kept below the 0.1 ng
Nordic toxic equivalents/Nm .
173
-------
references
1. Vogg, H., Metzger, M., Stieglitz, L., Recent Findings on the Formation
and Decomposition of PCDD/PCDF in Municipal Solid Waste Incineration,
Waste Management & Research, 5, 239, 1987.
2. Naikwadi, K.P., Karasek, F.W., Prevention of PCE® Formation in MSW
Incinerators by Inhibition of Catalytis Activity of Fly Ash Produced,
Chemosphere, Vol. 19, Nos. 1-6, pp. 299-304, 1989.
3. Vikelsoe, J., et al., Significance of Chlorine Sources for the Genera-
tion of Dioxins During Incineration of MSW, Dioxin '90, Vol. 3, pp.
193-196, Bayreuth, September 1990.
4. Vogg, H., et al., Chemical Process Aspects in Dioxin Reduction in Waste
Combustion Process (in German), VGB Kraftwerkstechnik, No. 8, August
1989.
5. Boscak, V., Role of Temperature in Formation and Emission of PCDD/F,
Dioxin '90, Vol. 3, pp. 35-38, Bayreuth, September 1990.
6. Schramm, K-W., Lenoir, D., Hutzinger, 0., Fugacity Calculations of
Vapor - Flyash Partition in Polyhalogenated Dioxins and Furans, Chemo-
sphere, Vol. 20, No. 5, pp. 563-568, 1990.
7. Fangmark, I., Wikstrcm, L-E., Marklund, S., Rappe, C., Studies on
Sampling Methods for PCDDs and PCDFs in Stack Emission, Chemosphere,
Vol. 20., Nos. 10-12, pp. 1333-1340, 1990.
8. Milj©styrelsen (Danish Ministry of Environment), Dioxins Emission from
Waste Incineration Plants, (in Danish), Miljcprojekt No. 117, 1989.
9. Boscak, V., Effect of Temperature on PCEO/PCDF Formation/Emission in
MSW Incineration, Presented at the Symposium on Municipal Solid Waste
Management: Making Decisions in the Face of Uncertainty, Toronto,
Canada, April 25-27, 1990.
10. Nielsen, K.K., Felsvang, K., Blinksbjerg, P., Holm, T.S., Meeting the
0.1 ng Toxic Equivalent Limit Using Spray Dryer Absorber and Fabric
Filter for Flue Gas Cleaning on Incinerators, Dioxin *90, Vol. 3, pp.
135-138, Bayreuth, September 1990.
174
-------
The work described in this paper was not funded by the U.S. Environmental
Protection Agency. The contents do not necessarily reflect the views of the
Agency and no official endorsement should be inferred.
REDUCTION OF DIOXINS-FURANS
FORMED IN COMBUSTION AND DENOVO SYNTHESIS
Aaron J. Teller
Research-Cottrell, Inc.
P.O. Box 1500
Somerville, NJ
Tel: 908/685-4063
Fax: 908/685-4834
The acceptance that Denovo synthesis of dioxins and furans occurs on the surface of flyash
particles in the temperature range of 200°C-400°C, establishes that post combustion
formation of dioxins must be addressed. Dioxin-furan emission reduction can be achieved
by addition of synthesis inhibitors, adsorbers, catalytic destruction, or utilization of the
inherent capability of silica-alumina for chemisorption of the dioxins and furans within
appropriate temperature ranges.
The chemisorption procedure, as applied to commercial incinerator operation adds no
additional equipment or capital and operating costs to the system. The equilibria rather
kinetics controls the absorption. Thus, wide variation in inlets are accepted by the system
with no externally induced system response required.
Experience at the Commerce, California waste to energy system established reductions of
total dioxin and furan cogeners to 1.9 ng/Nm3 from inlet concentrations ranging from 28 to
735 ng/Nm3, within the existing acid gas-particulate-heavy metal removal system, using only
temperature control to affect the removal by chemisorption.
The increase in dioxin emissions present in the flue gas after the boiler occurred during a
period of 18 months. During this period there was no change in the combustion conditions
of time-temperature and carbon monoxide outlet concentrations. The significant increase
in emissions dioxin-furan cogeners is attributed to Denovo synthesis on the boiler tubes.
The system consists of a Quench Reactor™-Dry Venturi™-and Baghouse. The baghouse
operating temperature is maintained at 115-135°C to optimize the chemisorptive removal
mechanism. The baghouse cake thickness is maintained in the range of 5 to 15 mm, at a
pressure drop of about 100 mm aq., with a 6 hour cleaning cycle. The achievement of the
thick fixed bed in the baghouse, resulting from the Dry Venturi™ action, provides for
equilibrium control of the adsorption of the dioxins and furans as well as acid gas reductions
of 99%, heavy metal reduction of 99.5%, and particulate emissions of less than 0.004
gr/dscf.
175
-------
INTRODUCTION
The concern with emissions of dioxins and furans from municipal waste incineration
processes has intensified, resulting in the emergence of regulations in Europe, Japan, and
North America. The general target is 0.1 to 0.2 ng/Nm3 toxic equivalent in the stack gas.
The mode of control of these emissions has gone through an evolutionary period with
the present consensus developing that dioxins and furans are formed both in the combustion
zone and in the gas cooling zone. Thus, control of the emissions of dioxins and furans
require more than control of combustion conditions.
In the 1970's and 1980's, theoretical and laboratory studies, such as conducted by
Rordorf (8), Bozeka (1), and Tsang (16), indication that both dioxin and furan emissions
from the combustion process could be limited to close to zero if combustion temperatures
were maintained in the range of 1000°C for 1 to 2 seconds. The ASME Solid Waste
Committee and many regulatory agencies recommended this practice.
The achievement of this condition on a continuous basis was found to be difficult
because of the heterogeneity of the waste fed to the incinerator. Clarke (2) reported high
dioxin and furan emissions, up to 700 ng/Nm3 where a furnace operated at an average
temperature of 1065°C.
A phenomenon apparently creating dioxins and furans downstream of the combustion
zone became evident at the PEI tests on an incinerator provided with an afterburner. In this
case, the dioxin-furan emission from the combustion zone were essentially zero, but the
stack emissions contained significant quantities of dioxins and furans (3).
It was proposed by Teller (12,14) that formation of the dioxins and furans were most
probably occurring in the window of 200°C to 500°C in the boiler region, catalyzed by flyash
and that effective control would be achieved by treatment of the gas downstream of the
boiler.
The confirmation of dioxin-furan formation downstream of the combustion zone was
presented by Hiraoka (5, 6), Naikwadi (7), and Stieglitz (10) in 1990.
The data presented by Hiraoka (5) indicated significant formation of dioxins in the
gas/air heat exchanger operating in the temperature range 270°C to 400°C. Data presented
by the authors are as follows:
176
-------
Cone, in ng/Nm3
T°C T4CDD P5CDD H6CDD H7CDD Q£DD
Inlet Gas/Air 400 13.1 <0.25 21.2 50.5 19.1
Heat Exchange
Inlet of EP 270 29.6 33.7 84.3 132 141
A further study by Hiroaka et.al. (6) indicated that formation of dioxins in an
electrostatic precipitator represented by the concentration of toxic equivalent 2, 3, 7, 8
TCDD, as a function of the operating temperature of the precipitator, with a ten-fold
increase from 250°C to 290°C operating temperature.
Naikwadi (7) indicated that the temperature range for formation of the dioxins and
furans, post combustion was 200°C to 400°C and that flyash appeared to be the catalyst.
Stieglitz (10) indicated that the denovo synthesis of dioxins and furans from
preformed aromatic structures in a flyash environment.
The synthesis of dioxins and furans, catalyzed by flyash, from aromatic structures, that
are more refractory than the dioxins, formed during the thermal destruction of dioxins in
the furnace, is another potential component of the denovo synthesis.
The now confirmed formation of dioxins and furans downstream of the combustion
zone requires reduction of these compounds in a flue gas emission control system.
TECHNOLOGY
In anticipation of the confirmation of the denovo synthesis of dioxins and furans
downstream of the combustion zone, a technology for the recovery of these compounds
including all congeners, was proposed in 1984 (13). It established the pseudo vapor pressure
of the dioxin congeners in contact with flyash as 10"7 times the true vapor pressure proposed
by Schroy (9), Figure 1. It further established the degree of chemisorption on flyash as a
function of Temperature (13). The process, integrated with the Quench Reactor-Dry
Venturi-Baghouse system, with specific temperature control requirements, now generally
referred to as the RC/Teller system was patented in 1985 (12).
The process is now operational at MWS Resource Recovery systems in the U.S.,
Japan, and Italy. The data included are those from the Commerce, California facility.
The system (Figure 2), consists of the upflow Quench Reactor, Dry Venturi, ad
Baghouse (17). Inasmuch as the thickness of the cake providing for fixed bed adsorption
177
-------
and contact with the resident flyash is critical in the effectiveness of the chemisorptive
process, the contribution of the Dry Venturi is significant.
The Dry Venturi action changes the characteristic of the baghouse cake, permitting
a buildup of thickness to the range of 5mm with no significant increase in pressure drop.
Thus a cake buildup of 4 to 8 hours between cleaning, at a maximum pressure drop of 4-5
in. w.g., is a normal operating condition.
The efficiency of recovery of acid gas as a measured and that of TCDD as calculated
for 120°C-130°C operation, based on HCD and S02 data approaches 96% in the baghouse
as a function of the moving wave front number (Figure 3).
SYSTEM DATA
The system performance, TABLE I, was essentially constant over the 15-month
period occurring between tests. The outlet levels and efficiencies of recoveries for
particulate, acid gases, and non-volatile heavy metals were the same within the
reproducibility of sampling and analysis.
The significant changes were improvement of recovery of mercury and the radical
change in the inlet levels of dioxins and furans (Tables II, III, and IV). The total
PCDD/PCDF at the boiler exit increased from 28.5 ng/Nm3 to 739 ng/Nm3. The CO level
remained essentially the same.
If, indeed, CO concentrations are an indicator of the PCDD formed in the
combustion chamber, the relationship observed by several investigators, then the significant
increase is reflection of denovo synthesis.
It is believed that this increase is reflective of the seasoning of boiler tubes with the
continuous deposition of flyash. Stieglitz (10) and others have demonstrated this
phenomenon in controlled tests.
Significantly the performance of the emission control system produced final exhausts
essentially the same as when the inlet PCDD/PCDF levels were 25 times the levels
observed soon after startup. This behavior implies confirmation of the theoretical
projections (13) that the vapor phase concentrations of the PCDD/PCDF were temperature
related provided adequate thickness of the adsorbent were available.
Even where the toxic equivalents for inlet and outlet PCDD/PCDF concentrations
were calculated (TABLE IV), the exit toxic equivalents were <0.113 ng/Nm3 (12% C02)
for an inlet of 42.26 ng/Nm3 for an inlet of 2.15 ng/Nm3.
178
-------
The only significant change is an increase in removal efficiencies as the inlet
concentration increased.
The data further confirm the occurrence of denovo synthesis of dioxins and furans
downstream of the combustion zone, and that these compounds in the vapor phase can be
reduced by chemisorption on flyash and other silica-alumina compounds utilizing
temperature control and adequate depth of adsorbent.
179
-------
TABLE I
SYSTEM PERFORMANCE
COMMERCE REFUSE TO ENERGY FACILITY
1987 1988
Cone, at Percent Cone. At Percent
Stack Remova1 Stack Remova1
TOTAL PARTICULATE
9.8
99.9
11.4
99.8
mg/Nm3 (12% C02)
HC1 (ppm at 3% 02)
11.4
99.0
9.4
98.9
S02 (ppm at 3% 02)
1.56
99.5
1.6
98.3
HF (ppm at 3% 02)
0.05
99.7
1.2
98.8
TOTAL PCDD/PCDF
<0.79
>97. 3
<1.21
>99.7
ng/Nm3 (12% C02)
TOTAL PAH
<0.15*
>99.4
mg/Nm3 (12% C02)
<0.47**
METALS mg/Nm3 (12%
co2)
Arsenic
<2.0
>98.1
<0.16
>99.8
Beryllium
<0.7
>90.0
<0.19
>97.2
Cadmium
<3.4
>99.9
2.0
99.9
Chromium
<0.7
>99.9
2.4
99.9
Lead
3.1
>99.9
2.0
>99.9
Mercury
270
0-20
41
91.3
Nickel
<28.0
>99.6
6.3
99.9
* excluding naphthalene
** including naphthalene
180
-------
TABLE II
PCDD MISSIONS FROM COMMERCE
RESPONSE TO ENERGY FACILITY
Concentration ng/Nm3 at 12% C02
(< is below detectable limit)
Boiler
Boiler Exhaust
Exhaust (Test 8DF) Stack Efficiency
(<50 PPM COl Stack (<50 PPM CO! (Test 8-DF) 1987 1988
2,3,7,8 TCDD
Total TCDD
<0.097
0.865
<0.003
0.112
1.16
91.90
<0.007
0.063
"96.9
88.1
>99.4
99.9
1.2.3.7.8 PCDD
Total
0.097
0.448
<0.003
0.051
4.31
100.77
<0.014
0.142
>96.9
88.6
>99.9
99.9
123478 H CDD
123678
123789
Total HXCDD
0.078
0.124
0.124
1.126
<0.003
<0.011
<0.011
0.102
2.52
4.58
13.49
73.58
<0.004
<0.005
<0.012
0.026
>99.6
>91.1
>91.1
92.0
>99.8
>99.9
>99.9
>99.9
1234678 H CDD
Total HpCDD
1.067
2.160
<0.058
0.062
13.53
29.73
<0.012
0.055
>94.6
97.1
>99.9
99.8
OCDD
3.470
0.153
12.43
<0.037
95.6
>99.9
Total PCDD
8.200
<0.314
308.4(1>
<0.308(1>
>96.2
>99.9
Total PCDD
326.3<2>
<3.04(2>
>99.0
(1) Only run, stable operation, absent of interferences, and less than 50 PPM
(2) Average of 3 runs — 1 unstable operation of incinerator
1 interference in analysis
181
-------
TABLE III
PCDF EMISSIONS FROM COMMERCE
REFUSE TO ENERGY FACILITY
Concentration ng/NM3 at 12% C02
(< is below detectable limit)
Boiler
Boiler
Exhaust
Exhaust
(Test 8DF)
Stack
Efficiencv
/<50 Dom CO)
Stack
(<50 PPM CO)
(Test 8-DF)
1987
1988
2,3,7,8 TCDD
<0.59
<0.028
13.78
<0.033
96.6
>99.8
Total TCDF
11.5
0.227
252.62
0.666
98.0
99.7
12378 PCDF
0.78
<0.003
4.66
<0.024
>99.6
>99.5
23478 PCDF
0.51
<0.015
15.04
<0.033
>97.1
>99.8
Total PCDF
2.83
0.059
70.18
0.163
98.0
99.8
123478 H CDD
0.64
<0.03
28.75
<0.003
95.3
>99.9
123678 H CDF
0.37
<0.012
8.10
<0.010
>96.8
>99.9
234678 H* CDF
0.03
<0.015
15.49
<0.007
50.0
>99.9
123789 H CDF
<0.006
<0.001
3.39
<0.005
"84.0
>99.9
Total HXCDF
2.91
0.077
75.94
0.040
97.4
>99.9
1234678 H CDF
<0.0006
<0.001
18.53
<0.007
"84.0
>99.9
1234789 ITCDF
0.156
<0.001
2.06
<0.009
99.4
>99.6
Total HpCF
2.18
0.086
28.42
0.007
96.1
99.9
OCDF
0.88
0.032
3.36
<0.025
96.4
>99.3
Total PCDF
20.3
<0.480
430.6(1>
0.901(1>
97.6
99.8
Total PCDD
1219.5(2>
7.65(2>
99.4
(1) Only run, stable, absent of interferences, and less than 50 PPM CO.
(2) Average of 3 runs — 1 unstable operation of incinerator
1 interferences in analysis
182
-------
TABLE IV
COMPARISON OF PCDD/PCDF TOXIC EQUIVALENTS
THREE MONTHS AFTER STARTUP — EIGHTEEN MONTHS AFTER STARTUP
Concentration in ng/Nm3 (12% C02)
Toxic Wt,
Factor
1987
Boiler
Exhaust
Stack
Boiler
Exhaust
1988
Stack
2378 TCDD
12378 PCDD
123478 HXCDD
123678 HXCDD
123789 HXCDD
1234678 HXCDD
OCDD
1.00
1.00
0.03
0.03
0.03
0.03
<0.097
0.097
0.002
0.004
0.004
0.032
<0.003
0.003
<0.003
<0.0003
<0.0003
<0.0024
1.16
4.31
0,
0,
0,
0,
077
137
405
406
<0.007
<0.014
<0.0001
<0.0002
<0.0004
<0.0004
TOTAL PCDD
0.2361
<0.011
6.495
<0.022
2378 TCDF
12378 PCDF
23478 PCDF
123478 HXCDF
123 678 HXCDF
234678 HXCDF
123789 HXCDF
1234678 HXCDF
1234789 HXCDF
OCDF
1,
1,
1,
0,
0,
0,
00
00
00
03
03
03
0. 03
0.03
0.03
0.00
0.59
0.78
0.59
0.019
011
001
000
000
005
0.
0.
0.
0.
0.
<0.028
<0.003
0.027
0.0
001
0016
0006
000
0.0016
0.0001
0.0
13 .78
4.66
15.04
0.163
0.243
0.465
0.102
0.556
0. 062
0.0
<0.033
0.024
<0.033
<0.0001
<0.0003
<0.0002
<0.0002
<0.0002
<0.0003
0.0
TOTAL PCDF
1.916
<0.063
35.77
<0.091
-------
FIGURE 1: PSEUDO-VAPOR PRESSURE OF 4 CDD
280 EMISSIONS FROM COMBUSTION PROCESSES
lO'1
10
M l0''
*
10*
i
10°
10"®
10'
10
'¦•11
OAT*i SCMAOY (2*1
VERTICAL w:NCS - MNQC
OF tXP. OATA
_ — - AN TO IN| CO »T
SCHROY ti«l
I i » 34.S7083 - t4S0-3,
. ^..1 IPot T,»
t*903.<39
KJ
! I I
I I I I I II I I I i I
I i ! I
iL
i *
1 ' 1
i
i III
KCAL «/
-------
FIGURE 2: TELLER SEMI-DRY SYSTEM
OUTLET GAS
DRY VENTURI
QUENCH
REACTOR
BAGHOUSE
AIR
FLUE GAS
SOLIDS
WATER
TESISORB
SOLIDS
LIME
AIR
11
-------
FIGURE 3: MOVING WAVE FRONT - EFFICIENCY
OF RECOVERY - BAGHOIJSE CAKE ADSORPTION
oicxin rsocvehy by chemisorption 283
a
»i
*1
- ii
fO
¦
ifl'l
O |tf TCN OMTM • MCI
186
-------
BIBLIOGRAPHY
Bozeka, G. G., Proc. Nat. Waste Proc. Conf., (1976), ASME, 215
Clarke, M., Waste Age, November 1987, 102-117
Concord Testing Corp., "National Incinerator Testing and Evaluation Program, Env.
Canada Rept. EPS 3/WP/l, (1985)
Duwel, U. et. al., Dioxin '89, September 1989, Toronto, Canada (in press)
Hiraoka, M. et. al., Dioxin '90, Volume 3, 71-75
Hiraoka, M. et. al., Dioxin '90, Volume 3, 81-85
Naikwadi, et. a., J. Chromet, in press, (1990)
Rordorf, B. F., Chemosphere 14 (6-7), 885, (1985)
Schroy, J. et. al., "Aquatic Toxicology Hazard Assessment", ASTM Special Tech. Pub.,
891, (1985), 409-21
Stieglitz, L., et. al., Dioxin 90, Volume 3, 173
Teller, J. A., USP 3,957,464 (1976), 4,293,524 (1981), 4,319,890 (1982), 4,375,455
(1983)
Teller, A. J, USP 4,502,346 (1985)
Teller, A. J., and Loubert, Proc. 76 Annual Meeting APCA (1984)
Teller, A.J., and Hsieh, J. Y., Resource Recovery, (1988), I, 9-14
Teller, A. J., Proc. Int. Conf. on Combustion, Toronto, (1988), 271-91, "Emissions
from Combustion Process", Clement and Kagel, Lewis Press (1990), 271-91.
Tsang, W., DOE/CE/30790 - IT, (1986)
187
-------
Intentionally Blank Page
188
-------
SESSION 6C:
NOVEL/EMERGING FLUE GAS CLEANING TECHNOLOGY I
Co-Chairmen:
James R. Donnelly
Davy McKee Corporation
San Ramon, CA
Karsten S. Felsvang
Niro Atomizer, Inc.
Columbia, MD
Preceding page blank
189
-------
Intentionally Blank Page
190
-------
The work described in this paper was not funded by the U.S. Environmental
Protection Agency. The contents do not necessarily reflect the views oi the
Agency and no official endorsement should be inferred.
PTTnrr-.qrATF TR^TTNG OF THE AMMONIA INJECTION TECHNOLOGY FDR SIMULTANEOUS CONTROL
OF PCDD/PCDF. HCl AND NOx EMISSIONS FROM MUNICIPAL SOUP WASTE INCINERATORS
by: Laszlo Takacs
Occidental Chemical Corporation
Niagara Falls, New York 14302
George L. Moilanen
Sierra Environmental Engineering, Inc.
Costa Mesa, California 92626
ABSTRACT
This paper describes the pilot-scale phase of a project for the development
of a technology to simultaneously control emissions of PCDD/PCDF
(polychlorinated dibenzo dioxins/polychlorinated dibenzo furans), acid gases
(hydrochloric acid (HCl) and sulfur oxides (SO )) and nitrogen oxides (NO ) from
municipal solid waste (MSW) incinerators. Thextechnology utilizes ammonia (NH )
as the control agent for all species of concern.
Previous work with a bench scale test system indicated that ammonia
was effective both in suppressing low-temperature (<800 °F) PCDD/PCDF formation
and in controlling HCl emissions when combined with a suitable particulate
removal device. This technology promises to have many advantages over currently
used alternative technologies. Mainly, this Ammonia Injection Technology (AIT)
inhibits PCDD/PCDF formation rather than removing it after it formed.
Therefore, AIT mitigates both air pollution and solid waste disposal concerns.
In this article, the theoretical basis for controlling all species, the
previous experimental work and the currently ongoing pilot-scale ammonia injec-
tion experiment are described. In this current phase of the project, a pilot-
scale economizer is used to simulate the behavior of the actual MSW economizer
and to demonstrate the effectiveness of ammonia in suppressing PCDD/PCDF
formation. A pilot-scale particulate control device is used to condense and to
remove ammonium chloride and other ammonium salts in order to reduce HCl and SO
emissions.
After successful completion of the current phase of the pilot-scale study,
additional project phases are planned. During the subsequent phases, the
Preceding page blank 191
-------
effectiveness of ammonia in suppressing PCDD/PCDF formation at higher
temperature (>800 °F), will be investigated and the optional NOx reduction
capability utilizing the demonstrated THERMAL DeNOx technology will be
incorporated into AIT.
The work described in this paper was not funded by the U.S.
Environmental Protection Agency and therefore the contents do not
necessarily reflect the views of the Agency and no official
endorsement should be inferred.
INTRODUCTICW
THE TEST FACILITY
Occidental Chemical Corporation (OxyChem) operates one of the world's
largest Refuse Derived Fuel (RDF) burning municipal solid waste incinerators
in Niagara Falls, New York (Figure 1). This facility has a capacity of
burning 2200 tons of RDF per day. The two boilers generate up to 270,000
pounds per hour steam (750 °F and 1250 psig). The steam is supplied to the
adjacent chemical complex and/or to the two 25 MW turbine/generators. The
basic features of each unit includes waterwall furnace enclosure, air-sweep
stoker feed system, convective superheater, boiler bank, and economizer
section. The OxyChem incinerators are equipped with highly efficient dry
electrostatic precipitators for particulate control. They do not have acid
gas control. As with other units losing RDF technology, the Occidental
incinerators have relatively high PCDD/PCDF emissions' '. Occidental has been
investigating ways to retrofit its units to control these compounds.
THE CONCEPT OF MULTI-SPECIES CONTROL VIA AMMONIA INJECTION
PCDD/PCDF and HC1 acid gas are two important species emitted from MSW
combustion. In addition, emissions of NO and SO are also of concern.
Curing Occidental's investigation of control options the concept of
multi-species emission control utilizing a single reagent was considered. It
was conceptualized that ammonia had the potential to simultaneously control
emissions of several pollutants of concern: PCDD/PCDF, HC1 and NO - and even
SOx, to some degree.
This concept was based on various degrees of previous knowledge. In
1987, Vogg and Stieglitz published results of laboratory furnace treatment of
fly ash from MSW combustion in an air stream shewing that NH^ addition to the
air stream suppressed PCDD/PCDF formation in the ash at the optimal
temperature of 572°F* .
In addition, it is known that in the presence of gas phase HC1, NH_ will
form NH4C1, which will condense out as a solid particle as the temperature
falls. Also, the reduction of NO via gas phase NH^ injection around the
optimum temperature of 1750 °F via^ the THERMAL DeNOx reaction is well proven
and commercialized.
192
-------
This multi-species control concept was first researched and studied "on
paper". It was then concluded that there was opportunity for successful
application of the Ammonia Injection Technology. However, several issues and
unknowns were identified which were deemed to be resolvable only by
experimental means.
OVERVIEW OF THE PROJECT
Based upon the favorable feedback from the "paper study", OxyChem initi-
ated a four-stage project to develop the Ammonia Injection Technology. First,
available background information and theories would be thoroughly researched.
Second, a pilot-scale test program would be undertaken to verify that
PCDD/PCDF is indeed formed at the lower temperature regions (<800 °F) of the
incinerator and to investigate the degree of PCDD/PCDF formation suppression
by ammonia. Also in this stage, the formation and capture of NH.C1 would be
demonstrated. Third, the entire AIT would be piloted. This woula include the
simulation of the equipment down-stream of the superheater, the investigation
of the PCDD/PCDF formation/suppression by ammonia on a relatively larger
scale, and the incorporation of the THERMAL DeNO process. Also in the third
stage, the treatment of the generated waste waiter and solid waste would be
studied. Finally, in the fourth stage, the full scale AIT would be installed
on the Occidental facility. It was decided at the onset of the project that
each subsequent stage would be executed only if the previous stage proved to
be successful. The project was initiated in 1989 and now it is in its third
stage.
BACKGROUND
THEORETICAL BASIS OF THE AMMONIA INJECTION TECHNOLOGY
PCDD/PCDF Control
The key to the Ammonia Injection Technology is the possibility that
ammonia could prevent the formation of PCDD/PCDF. It is generally thought
that PCDD/PCDF is formed either directly in the combustion chamber as a result
of incomplete combustion (primary formation), or in the equipment downstream
from the combustion chamber from unburned hydrocarbon precursors and
chlorine/HCl (secondary formation). In reality, PCDD/PCDF is probably
formed in both the combustion and the cooler downstream zones. According
to prevailing theories' ' , the secondary formation of PCDD/PCDF is most
prevalent in the 550-650 °F temperature range.
There are actually several theories as to why ammonia would prevent the
formation of PCDD/PCDF. We suggest that the ammonia in the combustion gases
competes with the hydrocarbon precursors present for the available chlorine
(chloride). Since ammonia is more reactive with chloride than hydrocarbons
are, ammonium chloride is more likely to form than chlorinated organics such
as PCDD/PCDF. Others theorize that PCDD/PCDF is formed from chloride and
hydrocarbon precursors by a catalytic reaction talcing place on the surface of
193
-------
the fly ash ^ ' K Heavy metals, namely copper, may act as catalysts. It
is thought that the catalytic action can be slowed down, or stopped entirely,
by poisoning the catalyst. Various catalyst poisons have, been tried and
certain organic amines have been found to be effective * . We can then
hypothesize that whatever an organic amine does in this regard, ammonia should
react similarly. At a high injection temperature, the amines dissociate into
NH_ radicals and subsequently NH ions is formed.. At the proper temperature,
ammonia forms the same radical. Yet others' ' further hold that an
electrophilic mechanism is likely to be responsible for the chlorination of
the organics. According to this theory, the fly ash surface functions as a
stoichiometric oxidant, and not a catalyst, in promoting the chlorination of
aromatics by HC1. Reaction of HC1 with iron (3+) on the fly ash produces
surface-bound iron(3+) chloride species that are the actual chlorinating
agents. We again suggest that by tying up the HC1 with ammonia, the
chlorination of the aromatics can be minimized.
Therefore, if either (or all) of the above theories is true, the result
is a reduction in PCDD/PCDF formation.
HC1 Control
As for HC1 control, it is well known that at the proper temperature
ammonia and hydrochloric acid readily react to form ammonium chloride. The
pure NH4C1 would de-sublimate at 635 °F. However, in boiler operation,
ammonium chloride is not likely to be in the solid phase until the flue gas
temperature is much lower, perhaps after the particulate matter control
device. If it condenses in the stack, or in the atmosphere outside the stack,
it creates an unacceptable dense white cloud. Thus the challenge, regarding
HC1 control with ammonia, is to find the proper reaction temperature, to
condense the vapor to a solid, and then capture the resulting sub-micron
particulates in the appropriate control device. SO would behave with ammonia
similarly, although the reaction would be somewhat xslower (and less complete)
and the resulting ammonium salt would be more difficult to handle.
NO Control
X
NO control is not a primary concern in this project. However, when
controlling PCDD/PCDF and acid gases with ammonia, the control of NO would be
an economically achievable added ber^fit. The control of NO witn ammonia,
either by the "THERMAL DeNOx" system^ ', or by the "Selective Catalytic Reduc-
tion" systenr , is well-known technology. Therefore, it is not detailed
here. However, it is included in the proposed technology in order to take
full advantage of the use of ammonia.
SUMMARY OF BENCH-SCALE EXPERIMENT
Bench-scale test system description
In order to validate the above theories, and to establish the feasibility
of ammonia injection as a potential emission control technology, a series of
194
-------
bench-scale experiments were conducted at Occidental Chemical Corporation's
MSW facility.
In the bench-scale experiment, a portion of the ash-containing flue gas
from the boiler outlet was diverted isokinetically into the experimental
apparatus (Figure 2). The system consisted of an 8 foot long (3M I.D.) sample
probe inserted into the boiler outlet just before the economizer. The
approximately 800 °F flue gas sidestream was pulled (via a jet pump) through
an insulated, 10 feet long (3" I.D.) sample section. The sample section had
six ports along its length which were used to sample and/or to monitor
emissions or to inject ammonia (port # 3). The sidestream of flue gas was
sampled to determine both the initial background PCDD/PCDF level (port #1) and
the HCl concentration (port #2). A standard EPA train was used to sample the
background PCDD/PCDF concentration.
A portion of the sidestream was diverted isokinetically into a PCDD/PCDF
"reactor" consisting of thimble filters inside an isothermal oven (Figure 3).
Downstream of the reactor, the back half of a standard EPA PCDD/PCDF sampling
train was used to measure the PCDD/PCDF content of the filtered flue gas. The
ash captured in the temperature-controlled thimble filters was also analyzed
for PCDD/PCDF.
In selected experiments, ammonia was injected upstream of the PCDD/PCDF
reactor. The flue gas exiting the reactor was sent to an NH.Cl condensation
system. This condensation system consisted of a condenser/filter assembly in
a constant-temperature oven, followed by a standard HCl sampling train (Figure
4).
Bench-scale experiment design
Experiments on the bench-scale apparatus had three primary objectives:
first, the verification of the secondary PCDD/PCDF formation at approximately
572 °F; second, the determination of the extent of ammonia suppression of this
formation; third, the confirmation of the removal of HCl via formation and
condensation of NH.Cl.
4
A factorial experimental design was developed to investigate the effects
of the key parameters. These included: ammonia stoichiometry, PCDD/PCDF
reactor temperature, reactor residence time and ammonium chloride condensation
temperature.
Bench-scale test results
This phase of the project was developed in 1989 and camLeted in 1990.
The bench-scale test results are detailed in another paper ' and are only
briefly summarized here. The bench-scale experiments demonstrated that:
a. Substantial amounts of PCDD/PCDF formed, relative to background levels,
in the "reactor" at 572 °F. (The average ash residence time in the reac-
tor, which was equal to 1/2 the sampling time, varied between 30 and 60
minutes.)
195
-------
b. Ammonia injection virtually completely suppressed the PCDD/PCDF forma-
tion. In the presence of ammonia, the total PCDD and total PCDF concen-
trations in the reactor remained at approximately background levels.
c. Greater than 97 percent reduction in HC1 emissions was achieved with the
NH4C1 condensation/removal system.
d. Excess ammonia concentration, over the stoichiometric amount, had no
effect on NH.C1 condensation and removal efficiency (in the range 260 °F
to 180 °F). 4
PII0T-SCALE PROJECT
PILOT-SCALE PROJECT OBJECTIVES
Due to the encouraging results of the bench-scale tests, an expanded
project at the pilot-scale size was approved and is currently in progress.
This stage of the project itself is executed in three phases.
Phase 1
The objective of Phase 1 is to confirm the mechanism of PCDD/PCDF suppres-
sion with ammonia as found in the bench-scale tests. In addition, more de-
tailed and expanded tests, focused on the NH4C1 condensation temperature window
and the NH.C1 particle capture alternatives, are also targeted. For this, a
pilot-scale test system simulating the PCDD/PCDF formation characteristics of
the actual (full-size) economizer and dry electrostatic precipitator (ESP), a
pilot-scale ammonia storage, supply, and injection system were designed and
constructed.
Phase 2
Utilizing the experimental data obtained in Phase 1, Phase 2 expands the
pilot system to test candidate NH.C1 particulate control equipment and to
upgrade the pilot system allowing enhanced and long term operation. In this
phase, the issues of liquid and solid waste treatment, handling and disposal is
also addressed.
Phase 3
Hie objective of Phase 3 is to further expand the pilot system to investi-
gate PCDD/PCDF suppression at higher temperatures (in the combustion chamber)
and to incorporate the NO control technology. In this Phase, the entire system
will be optimized as to tile number and locations of the ammonia injection ports,
and the quantities of ammonia injected at those ports. Finally, the waste
handling system will be refined and also optimized.
196
-------
DESCRIPTION OF PIIOT-SCALE TEST SYSTEM
Phase 1 Pilot-scale test system description
Table 1 presents a comparison of the key design values between the actual
MSW system, the previous bench-scale system and the current pilot-scale system.
TABLE 1. COMPARISON OF THE KEY DESIGN PARAMETERS FOR THE
ACTUAL MSW, THE PILOT-SCALE AND THE BENCH-SCALE UNITS
Parameter
Units
Actual
Economizer
Pilot-scale
Economizer
Bench-Scale
Test System
Economizer
Inlet Gas
Flowrate
DSCFM
WACFM
200,000
540,000
4,000
10,800
20
54
Economizer
Inlet Gas
Velocity
ft/s
22.5
22.5
22.5
Economizer
Area
ft2
400
8
0.04
Flue gas
HCl concen-
tration on
dry basis
ppm 0
7% 02
ppm 0
14% 02
640
360
640
360
640
360
Stoichio-
metric NH3
flowrate
lb/hr
200
4
0.02
Figure 5 presents a schematic of the Phase 1 experimental system. A fue
gas stream of approximately 4000 DSCFM at approximately 800 °F will be
withdrawn isokinetically from the boiler outlet duct through an inserted 7 foot
long (3' I.D.) "probe". An isolation damper is installed between the probe and
pilot system just outside the boiler duct wall. Next to the damper, a long
horizontal section of insulated 3' I.D. duct is installed. The background
PCDD/PCDF sample station is located on this section of the duct.
The flue gas will pass through additional insulated ducting to the ammonia
injection station. The ammonia injection station has provisions for the
continuous introduction and thorough mixing of ammonia over a wide range of
stoichiometric ratios.
197
-------
Down stream of the ammonia injection station the hot flue gas continues
into the pilot-scale economizer, where the gas will be cooled to approximately
500 °F or lcwer. At the outlet of the pilot-scale economizer, the gas stream
will pass through a dry ESP simulator (not energized) and then be drawn through
a cyclone separator to remove large-size particulate matter. Finally, after
passing through the fan, the stream is re-injected into the main flue gas stream
at the inlet of the MSW's dry electrostatic precipitator unit.
Phase 2 Pilot-scale test system description
In the second phase of the pilot-scale experiments the economizer and dry
ESP simulators from the Phase 1 experiments will be re-used and combined with
additional equipment for NH.C1 condensation and removal. Figure 6 presents the
Phase 2 equipment. The gas stream at 500 °F or less from the pilot-scale
economizer outlet will be directed into a transition duct, where provision will
be made for secondary ammonia injection (as required). Next, the stream enters
the flue gas precooler or quench chamber. Following the quench chamber, the gas
passes into a pilot-scale particulate control device (PCD). Finally, through
the ID fan, the stream is exhausted to the atmosphere through a short stack.
Various PCDs are being considered for investigation: a baghouse, a wet
electrostatic precipitator (WESP) and scrubber/demisters of different designs.
The device(s) for actual testing will be selected after completion of Phase 1.
Depending on the PCD employed, solid waste or wastewater in various
quantities will be generated. Provisions will be made to treat and properly
dispose of this waste. If the ammonia salts are collected as solid material,
they might be disposed of as solid waste. If the ammonia salts are collected in
solution, the options for treatment are wide open. These options include the
simple disposal of the waste water with the ash (as a wetting agent). If the
water balance dictates, the waste water might be evaporated (the vapors
condensed and recycled) and the ammonium salt crystallized. Alternatively, the
waste water can be reacted with caustic soda or soda ash. The resulting sodium
chloride solution might be recycled to the chlorine/caustic soda manufacturing
process of the adjacent chemical complex. The waste water can also be reacted
with lime. The calcium sulfite/sulfate would be separated and disposed of as
solid waste. The calcium chloride solution would be discharged. In both the
caustic and lime treatment cptions, the ammonia would be purged from the waste
water and recycled to the ammonia injection ports, resulting in a closed ammonia
cycle. It is expected that a wet option (WESP or scrubber/demister) would also
be beneficial in controlling the emissions of certain heavy metals, mainly of
mercury.
IXiring Phase 2, the pilot-scale process control and monitoring system will
also be upgraded to ensure continuous operation for extended periods.
Phase 3 Pilot-scale test system description
In the Phase 3 experiments, the equipment from Phases 1 and 2 (pilot scale
economizer, pilot-scale PCD and primary and secondary ammonia injection systems)
will be re-used and combined with equipment for the determination of
198
-------
high-ternperature PCDD/PCDF formation/suppression and for the THERMAL DeNOx
process. This expanded system is presented in Figure 7.
A gas stream at a temperature of 1800 °F, or higher, will be withdrawn
from the incinerator furnace and sent through a pre-cooler into an insulated
duct where tertiary ammonia injection will take place. The stream then would
enter into a post-cooler, where the gas temperature will be reduced to appr-
oximately 800 °F. The temperature drop will be adjusted to give the temperature
downstream of the ammonia injection point appropriate for the THERMAL DeNQx
reaction. The outlet of the post-cooler will be connected to the inlet duct of
the pilot-scale economizer previously described.
CURRENT PROJECT STATUS
Phase 1 Pilot-scale project experimental design
The Phase 1 tests will serve two purposes. First, they will demonstrate
PCDD/PCDF formation in the pilot-scale economizer and its suppression by ammonia
injection over a wide range of process conditions. This will include the
evaluation of the extent of PCDD/PCDF formation suppression at very low ammonia
stoichiometries, where NH.C1 formation is relatively minor. Second, they will
be used to make a preliminary determination of the conditions under which NH4C1
condensation and removal occurs, such that the appropriate particulate control
device can be selected for subsequent testing. To make this preliminary
determination, some additional ammonium chloride condensation experiments,
similar to those conducted in the bench-scale project in early 1990, will be
performed.
As in the bench-scale test series, a factorial experimental design has been
developed. Variables include: ammonia stoichiametry, pilot scale economizer/dry
ESP simulator temperature, ammonium chloride condenser temperature, and
pilot-scale economi zer/dry ESP simulator residence time. Sampling and
measurement requirements include: background PCDD/PCDF, inlet HCl concentration,
inlet and outlet ammonia concentrations, pilot-scale test system outlet
PCDD/PCDF, as well as flue gas flowrate, moisture content, oxygen, OCL percent
and 00 concentrations. The PCDD/PCDF analysis will utilize High Resolution Gas
chramatography/High Resolution Mass Spectrography (HRGC/HRMS) with
quantification of the tetra - octa PCDD/PCDF isomers. Standard EPA methods will
be used for sampling and analyses.
Phase 1 Pilot-scale project work in progress
Phase 1 began in the fall of 1990. The pilot-scale system design,
engineering, equipment procurement and installation have been completed.
Testing is currently underway. Phase 1 is scheduled to be completed by May
1991. If Phase 1 results are positive, then Phase 2 will commence immediately
and be completed by November 1991.
As issues became positively resolved, Occidental will continue to expand
the scale and scope of this project, up to and including a full scale
demonstration of the Ammonia Injection Technology.
199
-------
SUMMARY
The Ammonia Injection Technology offers an opportunity to simultaneously
control PCDD/PCDF, acid gases and NO emissions from MSW incinerators. Since
the process uses a gaseous control mJclia, it is relatively simple and requires
less space to install than technologies handling solids, such as the lime
injection system. The main advantage of the AIT is, however, that it
suppresses the formation of PCEO/PCDF, as opposed to control systems which
transfer them to another media.
Occidental Chemical Corporation committed significant resources to
develop this technology. A series of bench-scale experiments completed on its
MSW unit were very encouraging. Based upon these results, a pilot-scale
system has been designed and installed to develop the AIT further. Testing on
this system is currently ongoing and is scheduled for completion by mid 1992.
It is intended that, after successful completion of the development
project, the technology will be installed at QxyChem's Niagara Falls, NY MSW
facility.
REFERENCES
1. Kilgroe, J.D., and Johnston, M.G. EPA Assessment of technologies for
controlling emissions from municipal waste combustion. Solid Waste
& Power. Ill No. 6: 18, 1989.
2. Vogg, H., Metzger, M., and Stieglitz, L. Recent findings on the
formation and decomposition of PCEO/PCDF in municipal solid waste
incineration. Waste Management & Research. 5: 285, 1987.
3. Dickson, L., and Karasek, F. Mechanism of formation of
polychlorinated dibenzo-p-dioxins produced on municipal incinerator
flyash from reactions of chlorinated phenols. Journal of
Chromatography. 389: 127, 1987.
4. Stieglitz, L., and Vogg, H. On formation conditions of PCEO/PCDF in
fly ash from municipal waste incinerators. Chemosohere 16: 1917,
1987.
5. Hagenmaier, H., Kraft, M., Brunner, H., and Haag, R. Catalytic
effects of fly ash from waste incineration facilities on the
formation and decomposition of PCEO's and PCDF's. Environ. Sci.
Technology. 21: 1080, 1987.
6. Hagenmaier, H., Brunner, H., Haag, R., and Kraft, M. Copper-catalyzed
dechlorination/hydrogenation of PCEO's, PCDF's and other chlorinated
aromatic compounds. Environ. Sci. Technology. 21: 1085, 1987.
200
-------
7. Hoffman, R. V., Eiceman, G. A., Long, Y., Collins, M. C., and La, M.
Mechanism of chlorination of aromatic compounds adsorbed on the
surface of fly ash from municipal incinerators. Environ. Sci.
Technology. 24: 1635, 1990.
8. Lyon, R. K. THERMAL DeNOx. Environ. Sci. Technology. 21: 231, 1987.
9. Herriander, B. SCR DeNOx at the Munich South Waste Incinerator.
Paper presented at 83rd Annual Meeting & Exhibition, Air & Waste
Management Association, Pittsburgh, Pennsylvania, June 24-29, 1990.
10. Takacs, L., and Moilanen, G. L. Simultaneous control of PCDD/PCDF,
HC1 and NQ^ emissions from municipal solid waste incinerators with
ammonia injection. J. of Air & Waste Management Association. Accepted
for publication in 1990.
201
-------
to
o
to
To Farrows
SNwMKiFwl
Proportioner
Stacks
Electricity
ToPqww
Truck Waight
Scales
\W Shredded Fuel
torage
Steam To
Plants
4000 TONS
Presorting
lUrfeinea
2 Generators
2 Surge
Bins
3 Shredders
2 Electrostatic
Precipitators
2 Boilers
Multi-Flams
Wit Ash
System
2 Fly Ash
Silos
Figure 1. Schematic
of QxyChem's MSW Facility
-------
¦£>0IUEP CTUTLET OUt-T
P\TCTT TUC€
WITH TMEPMO^SPUPlS
u?usi a" i.d pep&e
rrzj |/ / / / / Z| | / / / /|
e>' UPKkSr PIPE WITH R9R7^
(^" I.D., IkHoULATEC?) /
eSA^
-------
FROM
PCPF3T * 4-
N
~\
L
1 1
1 1
1 1
1 1
1 1
1 1
I 1
1 1
V /
1 1
1 1
\ /
v /
v _ X
N3
O
THIME>LE
FILTERS
OVEN AT 5»-|2®F
Figure 3.
TO eTAMDARp H<^l
<5>A|-*fpL_IN<5? TRAM
TO MM^I <=OWC3ENe>*T!OH
TO eoAMC^RO MMo,
¦5>AMPLIM^ TftAIM
TC (SvV«S>-EMT MOOUuE
COMPEWeATE KUOCBOUT
TRAP
PE«oT OF IMPIKJCSER
TRAIN
Ae> im
EA£K
-------
COVJ&TMJT
TEMPER ATUIQE
^OMTAIMEp
BO°F=
220°F
260>
(iOMDEM^ER
PIUX&9* I PIUTER # a
"ft? "efTAMOJSD
H£l
IMPMS.EP TPAIM
Figure 4. Ammonium Chloride Condensation System
-------
FWOM
EfeTIL-EH
OJTLET-
tPUCT
(—SCWF)
(WSDOl
I HC.L
B»^cK*Mi<^_isiea
<=>TaTI ei«SM3>>TiS>W
w
^e>M
ASH
l 1 tpKbcieere
I I <&AMpuiM<^
Figure 5. Fhase l Pilot-scale Experimental System
-------
£AUU<5
€>TArk?sj
(JH-
—|NHj| «3AMl0UIMTRUHEHTXTk?KJ
MHS
l_.
oisfcoTY
Me>.^U|TA.Tle?N
C<3WTIkJU<3Ue
¦e>AMPUkica
rpFn
IpcppI
(opt)
A*5H * H*0 + MH4CL-
PteAMfUKJ<3i
Figure 6. Hiase 2 Pilot-scale Experimental System
-------
©
-MM,
TO TAUK
MOT*
&ACK<3«<9UUC>
c3
<=>Ta.TIM
N>
o
00
|EgPP~l
Ihcl. 1
Piee
eu?0\^.&a.
-©
t^M&LR>TK7M
FC3CM
eia<5>c"F)
AMMtfUIA
IMJGCTIC7M
eTATICU
©
¦SAi-iCLlfJcS
o
Ah^UIKJ<3j
e>i©£«eT e
•ev»+-1|fUM<3i
TC> PHM€
n ^fr"
-------
A NOVEL CALCIUM-BASED SORBENT FOR THE REMOVAL OF FLUE GAS HCI
BY DRY INJECTION
by: Wojciech Jozewicz
Acurex Corporation
Environmental Systems Division
Research Triangle Park, NC 27709
Brian K. Gullett
U.S. Environmental Protection Agency
Air and Energy Engineering Research Laboratory
Research Triangle Park, NC 27711
Shiaw C. Tseng
Acurex Corporation
Environmental Systems Division
Research Triangle Park, NC 27709
ABSTRACT
A novel calcium-based sorbent for HCI control in municipal waste combustor flue gas was
produced by slurrying waste bottle glass with Ca(OH)2. The reactivity of the sorbent with HCI was
measured in a fixed bed reactor simulating conditions for dry sorbent injection at economizer
temperatures (-500 °C, 1,000 ppm HCI, and 3 s contact time). The novel sorbent has achieved
reactivity levels over three times that of its base Ca(OH)2 under similar conditions.
The sorbent is produced by slurrying crushed waste bottle glass with Ca(OH)2 at elevated
temperatures for extended periods of time. The glass-to-Ca(OH)2 weight ratio (GCWR) was varied
from 1:1 to 4:1, slurrying temperature from 25 to 90 °C, and slurrying time from 0.25 to 5 h. The
median particle size of glass used for the production of sorbent varied from 1.3 to 84.5 jim. The dried
sorbent had a surface area up to 119 m2/g, depending on the GCWR, the slurrying conditions, and the
size of substrate glass particles. X-ray diffraction analyses of the sorbent revealed the formation of
calcium silicate hydrate.
The reactivity of selected glass/Ca(OH)2 sorbents was tested from 200 to 700 °C in the fixed
bed reactor. Based on findings from the current program, some implications for large-scale process
configurations for HCI/S02 control are discussed.
This paper has been reviewed in accordance with the U.S. Environmental Protection Agency's
peer and administrative review policies and approved for presentation and publication. Reference in
this paper to any specific commercial product is to facilitate understanding and does not imply
endorsement or favoring by the U.S. Environmental Protection Agency.
For presentation at the Second International Conference on Municipal Waste Combustion, Tampa, FL,
April 16-19, 1991.
209
-------
INTRODUCTION
Municipal solid waste (MSW) incineration produces numerous pollutants in the form of
particulate matter, gases [hydrogen chloride (HCI), hydrogen fluoride (HF), sulfur dioxide (S02)], volatile
organic compounds (VOCs), and heavy metals. Wet, semi-dry, and dry processes are considered for
the removal of acid gases from MSW incineration. Dry sorbent injection offers the possibility of a
technically simple and low cost alternative. Typically, the dry calcium-based sorbent is injected into the
incinerator, economizer, or downstream flue gas duct (1,2) regions where it reacts with hot combustion,
post-combustion, or cooled post-combustion gases, respectively. In the process of dry sorbent injection
into the duct, the flue gas is humidified to a close approach to adiabatic saturation, either upstream or
downstream of the sorbent injection point. Humidification can significantly increase reactivity of a dry
sorbent with S02 (3). However, in the presence of HCI, drying difficulties may be encountered due to
the strong water retaining properties of absorbed HCI and calcium chloride (CaCI2) (4). These wet
solids, with a decreased rate of drying, can cause wet wall deposits and operational problems with the
particulate control device. To decrease the likelihood of the above adverse effects of CaCI2, the
approach to adiabatic saturation is often widened (5), decreasing the reactivity of the sorbent with S02.
Two contradicting trends described above (minimized approach to saturation for S02 removal
and maximized approach to saturation for system reliability) indicate the need for a very reactive
sorbent for simultaneous control of HCI and S02 from MSW incinerators.
This paper describes the bench scale development and testing under economizer conditions of
such a novel, calcium-based sorbent for HCI removal. Based on findings from the current program and
previous experience (3), some implications for the control of HCI/S02 from MSW incinerators are
discussed.
EXPERIMENTAL
REACTOR
Sorbent reactivity tests with HCI were run in a fixed bed, short time reactor (STR) presented in
Figure 1. In this reactor, sorbent dispersed on a quartz wool bed is rapidly moved into the process gas
stream and then extracted after a preset exposure time. To activate the sample bed holder, the start
button is pushed, which simultaneously activates the air cylinder and starts the timer. This results in
positioning the sample bed holder under the reactive process gas. The flow of gas in the air cylinder is
reversed which drives the bed holder out from under the reactive process gas after the preset time
expires. The sample is loaded into the STR via the sample loading port in the slider assembly and
dispersed into a thin layer of quartz wool. Process gas, preheated to the desired temperature, enters
the slider assembly via the process gas inlet and, after contacting the sorbent dispersed on the quartz
wool, exits to the hood through the process gas outlet. A more detailed description of the reactor is
given elsewhere (6).
The STR was operated at temperatures ranging from 200 to 700 °C, and most often at 500 °C.
The residence time of the sorbent in the process gas stream was 3 s. The volumetric gas flow rate
was 22 L/min (at standard temperature and pressure). The composition of the process gas was 5
volume percent oxygen (02), 1,000 ppm HCI, and the balance nitrogen (N2). The concentration of HCI
was measured with a Thermoelectron HCI analyzer.
210
-------
SORBENTS
Calcium Hydroxide
Reagent grade calcium hydroxide [Ca(OH)2] from Fisher (Fisher C-97) was selected as a
baseline sorbent. It was determined by a thermogravimetric analyzer (TGA) to be 93 percent Ca(OH)2
(the remaining 7 percent was surface water, calcium carbonate, and impurities). Three other
commercially available Ca(OH)2 sorbents were used: Linwood, Marblehead, and Tenn-Luttrell.
Structural properties of commercial Ca(OH)2 and reagent grade Ca(OH)2 are given in Table 1.
Glass/Ca(OH)o
Novel sorbents were produced during the course of this work with the intended application for
HCI removal from a simulated MSW incinerator flue gas. These novel sorbents were produced by
slurrying crushed waste glass with reagent grade Ca(OH)2. Previously (7), sorbents that were several
times as reactive with S02 under duct injection conditions as reagent grade Ca(OH)2 were produced by
slurrying fly ash from coal-fired power plants with Ca(OH)2- The increased reactivity of these sorbents
was attributed to dramatically increased surface area and moisture carrying capabilities. Both
properties were developed following a reaction between silica digested from fly ash and calcium. In
the course of this reaction, amorphous, hydrated calcium silicates were formed. Crushed waste glass
of a clear grade was used as a source of silica (as opposed to colored grade glass). The glass-
to-Ca(OH)2 weight ratio (GCWR) was varied from 1:1 to 4:1. Slurrying time varied from 0.25 to 5 h and
slurrying temperatures from 25 to 90 °C. The water-to-solids ratio during slurrying was 15:1, which
corresponded to an approximately 6 percent solids slurry. Following slurrying, solids were vacuum
filtered and microwave dried for 6 min to rapidly stop the reaction between the silica and calcium.
Waste glass was crushed to a desired fineness using a bench scale attritor from Union
Process. By varying the intensity and time of crushing, glass substrate of divergent structural
properties was produced. The properties of ground glass are given in Table 2. As can been seen from
Table 2, both surface area and pore volume of crushed glass substrate increased as median particle
size decreased. Ground glass substrate, with a nominal particle size of 2 ^m, is shown in a scanning
electron microscope (SEM) photomicrograph in Figure 2. Irregularly shaped particles with smooth
surfaces can be observed.
Analytical Techniques
Sorbents produced for this study were characterized rather extensively using a variety of
analytical techniques. Specific surface area, pore volume, and median pore size were determined by
N2 adsorption/desorption (Micromeritics ASAP 2400). The amount of Ca(OH)2 in a sorbent sample was
detected by TGA (Perkin Elmer TGA7) and defined based on the TGA weight loss between 375 and
545 °C for a heating rate of 10 °C/min. Median particle size measurement was based on x-ray
sedimentation (Micromeritics Sedigraph). Qualitative assessment of phases produced as a result of
slurrying of crushed glass and Ca(OH)2 was done by x-ray diffraction (Siemens D-500). Spectra were
identified by computer comparative analysis with the Joirt Committee for Powder Diffraction Spectra
(JCPDS) spectral file. The morphology of samples was characterized by SEM (Amray SEM).
Conversion following the exposure of solids to HCI was determined by Ca++ analysis by atomic
absorption spectrophotometry (Perkin Elmer AA) and Cr analysis by ion chromatography (Dionex IC).
Conversion percentage is described here as a ratio of half the number of Cr moles to the number of
Ca++ moles multiplied by 100.
211
-------
RESULTS
SORBENT STRUCTURE DEVELOPMENT
Time/Temperature of Sorbent Preparation
Crushed glass substrate with a nominal median particle size of 2 jim was used during this part
of the work. The combined effect of slurrying time/temperature on the development of sorbent structure
was investigated first at a GCWR of 3:1. Slurrying time was varied from 0.25 to 5 h and temperatures
tested were 25, 60, and 90 °C. The results are shown in Figures 3 and 4 giving surface areas and
pore volumes, respectively. Both surface area and pore volume of the sorbent increased with longer
slurrying time. The rate of increase was higher for higher temperatures. For any given slurrying time,
the surface area and pore volume increased with increasing temperature. Only limited structural
enhancement took place during slurrying at 25 °C. Short slurrying times of 0.25 and 0.5 h at 90 °C
were tested because of the high rate of structural enhancement at this temperature. Following 1 h
slurrying at 90 °C, the sorbent developed 33 m2/g surface area. It took 5 h slurrying at 60 °C for
sorbent to develop 32 m2/g surface area.
GCWR
In the next series of experiments, the effect of varying GCWR on the development of structure
was tested. The results for GCWR from 1:1 to 4:1 are given in Figures 5 and 6 for surface area and
pore volume, respectively. Slurrying conditions of 90 °C for 1 h and of 90 °C for 5 h were selected to
represent what was thought to be an early and an advanced sorbent development stage, respectively.
For 90 °C, 1 h slurrying conditions, the surface area and pore volume developed by the sorbent were
practically independent of the GCWR used for its production. Following 90 °C, 5 h slurrying conditions,
surface area and pore volume decreased somewhat as the GCWR increased. This is contrary to
trends observed earlier (8,9) for slurrying fly ash from coal-fired power plants with Ca(OH)2, when the
sorbent's surface area increased with increasing fly-ash-to-Ca(OH)2 weight ratio. However, glass-due
to its amorphous structure-would be expected to dissolve considerably higher amounts of silica per unit
of mass than was the case with fly ash. As a result of this overly increased supply, the excess silica
may blind the reactive surface of calcium silicates and/or result in formation of calcium-lean calcium
silicates. Calcium-lean calcium silicates would be expected to be less reactive than calcium-rich
samples (10). It is not clear at this point why the structural development was independent of GCWS at
the 90 °C, 1 h conditions tested. Perhaps the supply of dissolved silica was too low for the extensive
formation of calcium silicates following slurrying at 90 °C for 1 h.
Glass Fineness
Following the above structure development tests, testing of crushed glass substrates of varying
degrees of fineness was done. The nominal sizes produced varied from 1 to 100 jim. Properties of
sorbents produced by slurrying these substrates at GCWR of 1:1 at 90 °C for 1 h are shown in Figures
7 and 8, giving surface area and pore volume, respectively. Both surface area and pore volume of
sorbents increased with increasing fineness (decreasing median particle size) of glass substrate used to
produce them. A surface area as high as 119 m2/g was developed following) slurrying of 1 jim
(nominal) glass substrate. Structural properties of all glass/Ca(OH)2 sorbents produced are given in
Table 3.
212
-------
CONVERSION
CafOHU Source
Conversion of sorbents with HCi was measured in the STR operated at 500 °C (occasionally
from 200 to 700 °C) with a 3 s residence time. The concentration of HCI was 1,000 ppm during all
tests. Conversion of reagent grade Ca(OH)2 and of Ca(OH)2 from three commercial sources was
measured first and it was approximately 14 percent, regardless of Ca(OH)2 source. The initial
glass/Ca(OH)2 sorbent produced at a GCWR of 3:1, and slurried at 90 °C for 3 h reached a conversion
of 39 percent. Lowering the slurrying temperature to 25 °C and slurrying glass and Ca(OH)2 at the
GCWR rate of 3:1 for 3 h produced sorbent which yielded 33 percent conversion when exposed to HCI.
This comparison is shown in Figure 9 and clearly demonstrates superior conversion levels achieved by
glass/Ca(OH)2 sorbents, compared with Ca(OH)2.
Time/Temperature of Sorbent Preparation
Following the above initial experiments, the combined effect of slurrying time/temperature on
sorbent conversion with HCI was investigated. Crushed glass substrate of nominal size 2 jim was used
and the GCWR was set at 3:1. Three slurrying temperatures were tested and the results are shown in
Figure 10. The STR conditions were 500 °C, 1,000 ppm HCI, and 3 s residence time. There was no
significant effect of slurrying time on conversion with HCI for 25 °C slurrying temperature. Conversion
of sorbents prepared at 90 °C decreased sharply with increases in the slurrying time, whereas
conversion of sorbents prepared at 60 °C initially increased slightly (1 to 3 h) and then decreased with
longer slurrying time (3 to 5 h).
GCWR
Conversion of sorbents prepared by slurrying crushed glass substrate with Ca(OH)2 at GCWR
ranging from 1:1 to 4:1 is shown in Figure 11. The STR conditions were 500 °C, 1,000 ppm HCI, and
3 s residence time. There was no significant effect of GCWR (except for GCWR of 3:1; it is not clear
what caused this aparent anomaly) on the conversion of sorbents slurried at 90 °C for 5 h. Conversion
of sorbents slurried at 90 °C for 1 h was increased with increasing GCWR from 1:1 to 3:1. No increase
of conversion was measured following the increase of GCWR from 3:1 to 4:1.
GLASS/Ca(OH)2 INJECTION PROCESS
Experimental results suggest the potential application of the glass/Ca(OH)2 sorbent in a sorbent
injection into the furnace or, more likely, into the economizer region where a large fraction of HCI would
be removed from the flue gas. Since the preferred reaction temperature would allow HCI capture well
upstream of the particulate collector, this should allow a number of options to be explored, including
operation of the particulate collector at lower temperatures for better S02, metals, and VOC capture.
The results of preliminary tests investigating the temperature window for glass/Ca(OH)., sorbent
injection are shown in Figure 12. Sorbent produced by slurrying crushed waste glass with Ca(OH)2 at
GCWR of 3:1 and at 90 °C for 1 h was used. The conditions in the STR were 1000 ppm HCI and 3 s
residence time. Temperature was varied from 200 to 700 °C. Based on results presented in Figure 12,
the recommended reaction temperature would be approximately 500 °C. Due to the STR temperature
limitations, the actual quenched temperature regime in an MSW incinerator has not been tested.
However, based on data in Figure 12, the conversion would decrease dramatically for reaction
temperatures significantly higher than 500 °C. This suggests that sorbent injection in a quenched
213
-------
system at temperatures in excess of 500 °C, but below the melting point of the probable CaCI2 product
(782 °C), would be optimum. It appears, therefore, that the economizer would be a desired location for
the injection.
To facilitate operation of the particle control device, coarse solids should be used during the
primary injection. However, the reactivity of sorbents is known to decrease with increasing particle
size. To test the effect of sorbent particle size (sorbent particle size was approximated here by crushed
glass nominal size), glass was crushed to varying degrees of fineness. The nominal median particle
size was varied from 1 to 100 |im and complete structural data were given in Table 2. Crushed glass
was slurried with Ca(OH)2 at a GCWR of 1:1 at 90 °C for 1 h. Dried sorbents were exposed to HCI in
the STR operated at 500 °C, 3 s residence time, and with 1,000 ppm HCI. Results presented in Figure
13 indicate decreasing conversion with increasing particle size of crushed glass substrate. Increasing
the nominal particle size of crushed glass substrate from 1 to 100 |xm resulted in conversion decreases
from 23 to 17 percent. The conversion measured for sorbent produced with glass of a nominal size of
50 |xm was 21 percent. While these results are preliminary, they indicate a potential tradeoff between
the coarseness of glass substrate used for production of the sorbent and sorbent reactivity. It appears
that crushed glass as coarse as 50 or 100 iim could be used for sorbent injection.
DISCUSSION
Glass/Ca(OH)2 sorbents were produced, their structural properties analyzed, and reactivity
tested for the removal of HCI from MSW incinerator flue gas. All glass/Ca(OH)2 sorbents tested were
more reactive (yielded higher conversion) than Ca(OH)2 alone. When exposed to HCI under conditions
representative of those encountered during dry sorbent injection into the economizer (11): 500 °C, 3 s
residence time, and 1000 ppm HCI.
Sorbents produced during the course of work described here were characterized by a variety of
analytical techniques in order to better understand how structural properties relate to sorbent reactivity
with S02. Surface area was found not to be a unique indicator of reactivity. For example, comparison
of Figures 3 and 10 indicates that, for a sorbent produced at a GCWR of 3:1 by slurrying at 25 °C, both
conversion and surface area were weakly affected by increasing the slurrying time. SEM examination
of morphology indicated that particles with two different types of surface features were formed. Figure
14 is an SEM photomicrograph of a glass/Ca(OH)2 sorbent produced by slurrying at 25 °C for 1 h at a
GCWR of 3:1. Particles with smooth surfaces typical for glass substrate (see Figure 2) can be seen,
indicating the lack of extensive reaction between glass and Ca(OH)2.
Figure 15 is an SEM photomicrograph of sorbent produced at a GCWR of 3:1, slurried at 90 °C
for 1 h. This sorbent had considerably higher surface area than the one slurried at 25 °C (33 versus
15 m2/g, respectively). The microscope field, shown in Figure 15, reveals particles coated with
amorphous-looking deposits without clearly defined morphological surface features. Increasing the
90 °C slurrying time to 5 h resulted in sorbent with a surface area of 67 m2/g. However, despite the
increased surface area, the conversion decreased compared with sorbent prepared by 1 h slurrying at
90 °C. As can be seen in Figure 16, sorbent slurried at 90 °C for 5 h had particles coated with
deposits. The appearance of these deposits is different from that observed in Figure 15. Small,
nm-sized ridges are visible and they may indicate higher crystallinity of sorbent produced at 90 °C, 5 h
than that of sorbent produced at 90 °C, 1 h. Previously (12), sorbents with high surface area, but with
crystalline morphology, were shown to be less reactive with S02 than ones with amorphous
morphology.
214
-------
X-ray diffraction (XRD) studies indicated decreased amounts of Ca(OH)2 in sorbents that
developed increased surface area. Formation of hydrated calcium silicates was detected by XRD
(JCPDS File No. 33-306). However, the XRD results are not fully conclusive.
Decreasing amounts of Ca(OH)2 in samples which developed increased surface area were
detected by a TGA method that measures the TGA weight loss corresponding to temperature increases
from 375 to 545 °C [the decomposition temperature of Ca(OH)2], This weight loss would be 24.3 and
12.15 percent for pure Ca(OH)2 and glass/Ca(OH)2 sorbent prepared at a GCWR of 1:1, respectively.
A family of glass/Ca(OH)2 sorbents prepared by slurrying at 90 °C for 1 h at a GCWR of 1:1, but using
glass substrate of varying fineness, was examined and results are shown in Table 4. The amount of
Ca(OH)2—as detected by TGA—decreased with decreasing particle size of a glass substrate. Sorbent
reactivity with HCI increased with decreasing amounts of Ca(OH)2 in the sample. Also shown in Table
4 are TGA weight losses for the temperature ranges from ambient to 125 °C, 125 to 375 °C, and 375 to
545 °C. The first corresponds to the surface moisture, the second may be attributed to hydrate
decomposition, and the third may be attributed to calcium silicate hydrate decomposition (13). Both the
amount of surface water and of calcium silicate hydrates increased with decreasing particle size of
glass substrate used for the production of a sorbent. Conversion with HCI (Figure 13) could be
correlated with TGA weight losses described above. However, it should be emphasized that this
correlation was not valid for the whole family of sorbents produced throughout this study. A series of
results presented in Table 4 for sorbents produced at a fixed GCWR, the same slurrying conditions, yet
having dissimilar structural properties and reactivity with HCI, were used here only to demonstrate
structure-reactivity trends. It should be pointed out that the glass/Ca(OH)2 sorbents used for primary
injection (~500 °C) will be subjected to decomposition/sintering in this temperature region. The effect of
decomposition on structural properties of glass/Ca(OH)2 sorbents is unknown. One of the products of
decomposition will be water vapor, which is known to promote sintering (14). Correlations need to be
developed between the amount of calcium silicate hydrate produced, its crystallinity, the amount of
surface water, the decomposition/sintering properties, and the sorbents' reactivity.
CONCLUSIONS
Novel sorbents produced by slurrying crushed waste glass with Ca(OH)2 were demonstrated to
be up to three times as reactive with HCI as Ca(OH)2 under conditions typical for economizer injection
of dry sorbent for the control of HCI from MSW incineration. The structure and reactivity of sorbents
varied with slurrying conditions (GCWR, temperature, time). The amount of Ca(OH)2 detected by TGA
in prepared sorbents was lower than theoretically expected and SEM examination indicated changes of
sorbent morphology as a result of slurrying.
A novel process was described for the removal of MSW incinerator HCI with glass/Ca(OH)2
sorbents. Laboratory tests indicated approximately 500 °C as a preferred reaction temperature and
potential for use of coarse (50-100 p.m) glass substrate for the production of the sorbent.
ACKNOWLEDGEM ENTS
The authors wish to recognize the assistance of Frank E. Briden and George R. Gillis of the
U.S. Environmental Protection Agency and Monsie M. Gillis and Wojciech Kozlowski of Acurex
Corporation. This work was supported by the U.S. Environmental Protection Agency's Air and Energy
Engineering Research Laboratory (EPA Contract 68-DO-0141) located in Research Triangle Park, NC,
USA.
215
-------
REFERENCES
1. Sedman, C.B. and Brna, T.G. Municipal Waste Combustion Study: Flue Gas Cleaning
Technology. EPA/530-SW-87-021d (NTIS PB87-206108), 1987.
2. Clarke, M.J. Emission Control Technologies for Resource Recovery. Jn: Proceedings of the 79th
Annual Meeting of the Air Pollution Control Association. Air Pollution Control Association,
Minneapolis, MN, 1986. p. 86-50.11.
3. Jozewicz, W., Chang, J.C.S., and Sedman, C.B. Bench-Scale Evaluation of Calcium Sorbents for
Acid Gas Emission Control. Env. Proa. 9. 3:137,1990.
4. Chang, J.C.S., Brna, T.G., and Sedman, C.B. Pilot Evaluation of Sorbents for Simultaneous
Removal of HCI and S02 from MSW Incinerator Flue Gas by Dry Injection Process. Presented at
the International Conference on Municipal Waste Combustion, Hollywood, FL, April 11-14,1989.
5. Klingspor, J.S., Roberts, D.L., and Jefcoat, I.A. Acid Gas Emissions: Results of a Spray Dry
Scrubbing Pilot Plant Study, Presented at the International Conference on Municipal Waste
Combustion, Hollywood, FL, April 11-14,1989.
6. Gullett, B.K., Bruce, K.R., and Machilek, R.M. Apparatus for Short Time Measurements in a Fixed-
Bed Gas/Solid Reactor. Rev. Sci. Instr. 61, 2:904,1990.
7. Jozewicz, W. and Rochelle, G.T. Fly Ash Recycle in Dry Scrubbing. Env. Proa. 5, 4:219,1986.
8. Jozewicz, W., Jorgensen, C., Chang, J.C.S., Sedman, C.B., and Brna, T.G. Development and Pilot
Plant Evaluation of Silica-Enhanced Lime Sorbents for Dry Flue Gas Desulfurization. J. of APCA.
38, 6:796, 1988.
9. Jozewicz, W., Chang, J.C.S., Sedman, C.B., and Brna, T.G. Characterization of Advanced
Sorbents for Dry S02 Control. React. Solids. 6:243,1988.
10. Jozewicz, W. and Chang, J.C.S. Evaluation of FGD Dry Injection Sorbents and Additives. Vol. 1.
Development of High Reactivity Sorbents. EPA-600/7-89-006a (NTIS PB89-208920), 1989.
11. Bortz, S.J., Roman, V.P., Yang, R.J., and Offen, G.R. Dry Hydroxide Injection at Economizer
Temperatures for Improved S02 Control, hr Proceedings: 1986 Joint Symposium on Dry S02
and Simultaneous SO-j/NO^ Control Technologies. Vol. 2. EPA-600/9-86-029b (NTIS PB 87-
120457), 1986, p. 31-1.
12. Jozewicz, W., Chang, J.C.S., Sedman, C.B., and Brna, T.G. Silica-Enhanced Sorbents for Dry
Injection Removal of S02 from Flue Gas. J. of APCA. 38, 8:1027,1988.
13. Taylor, H.E.W. The Chemistry of Cements. Academic Press, London, Chapter 22,1964.
14. Borgwardt, R.H. Calcium Oxide Sintering in Atmospheres Containing Water and Carbon Dioxide.
Ind. Eng. Chem. Res. 28:493,1989.
216
-------
PROCESS GAS INLET
SAMPLE LOADING PORT
\j SAMPLE BED HOLDER
QUARTZ WOOL WITH DISPERSED SAMPLE
H) AIR CYLINDER
tSTAINLESS STEEL MESH
SLIDER ASSEMBLY
PROCESS GAS OUTLET
Figure 1. Schematic and cutaway of the short time reactor (STR).
Figure 2. SEM photomicrograph of a crushed glass substrate. Nominal size 2 nm (x5,000).
217
-------
60 C
60
(75 20
2 3 4
Slurrying Time, h
Figure 3. Combined effect of slurrying time/temperature on the development of surface area by
glass/Ca(OH)2 sorbents (GCWR=3:1, nominal glass size 2 urn).
Q)
C5
E
o
©
E
O
>
©
k.
o
a.
0.3
0.2 -
0.1
0.0
—25°C
" 60°C
a 90°C
* *
¦
a a
¦
i
i
2 3 4 5
Slurrying Time, h
Figure 4. Combined effect of slurrying time/temperature on the development of pore volume by
glass/Ca(OH)2 sorbents (GCWR=3:1, nominal glass size 2 nm).
21«
-------
to" 60
• 5h
CD
B 40
1:1 2:1 3:1 4:1
GCWR, g Glass/g Ca(OH)2
Figure 5. The effect of GCWR and slurrying time on the development of surface area by
glass/Ca(OH)2 sorbents slurried at 90 °C (nominal glass size 2 |im).
• 5h
1:1 2:1 3:1 4:1
GCWR, g Glass/g Ca(OH)2
Figure 6. The effect of GCWR and slurrying time on the development of pore volume by glass/Ca(OH}2
sorbents slurried at 90 °C (nominal glass size 2 |im).
219
-------
120
100 -
D>
CVJ
E
CO
CD
80
< 60 -
0
o
CO
•t
D
U)
40
20
0.1 1 10 100
Glass Median Particle Size, ym
Figure 7. The effect of crushed glass size on the development of surface area by glass/Ca(OH)2
sorbents (90 °C, 1 h slurrying time, GCWR=1:1).
_a>
CO
E
o
CD
E
O
>
CD
k—
o
Q.
0.1 1 10 100
Glass Median Particle Size, fim
Figure 8. The effect of crushed glass size on the development of pore volume by glass/Ca(OH)2
sorbents (90 °C, 1 h slurrying time, GCWR=1:1).
220
-------
X
+
+
(0
O
w
O
E
x
CM
O
w
O
E
c
0
w
1
c
o
O
~ Ca(0H)2
El Glass/Ca(OH)2
£
Sorbent Type
Figure 9. Comparison of STR conversion of Ca(OH)2 and glass/Ca(OH)2 sorbents
(500 °C,1,000 ppm HCI, 3 s).
60
% 40
x
+
+
(0
O
v>
O
E
x
CM
O
w
o
E
« 20
c
0
'co
1
c
o
O
~
ft
« ¦
~A
1
~
¦ 25°C
-
• 60°C
~ 90°C
I 1
t
1 3 5
Slurrying Time, h
Figure 10. Combined effect of slurrying time/temperature on the reactivity of glass/Ca(OH)2 sorbents
with HCI (sorbents: GCWR=3:1, nominal glass size 2 n.m; reactor: 500 °C, 1000 ppm HCI,
3 s).
221
-------
8 80
x
+
+
(0
O
c
o
O
60
40
20
1:1 2:1 3:1 4:1
GCWR, g Glass/g Ca(0H)2
Figure 11. The effect of GCWR on the reactivity of glass/Ca(OH)2 sorbents with HCI (sorbents:
90 °C, nominal glass size 2 nm; reactor: 500 °C, 1000 ppm HCI, 3 s).
x.
+
+
(0
O
w
O
E
x
CM
T*~-
o
V)
O
E
c
o
c
o
O
200 300 400 500 600 700
Reaction Temperature, °C
Figure 12. The effect of STR temperature on reactivity of glass/Ca(OH)2 sorbent (sorbent: 90 °C, 1 h
slurrying time, GCWR=3:1; reactor: 1000 ppm HCI, 3 s).
222
-------
X
+
+
(0
O
«
o
E
-------
Figure 15. SEM photomicrograph of a sorbent prepared by slurrying glass/Ca(OH)2 at
GCWR=3:1,90 °C, 1 h (x 5,000).
Figure 16. SEM photomicrograph of a sorbent prepared by slurrying glass/Ca(OH)2 at GCWR-3:1,
90 °C, 5 h (x 5,000).
Reproduced (rom
best available copy.
224
-------
TABLE 1. STRUCTURAL PROPERTIES OF Ca(OH)2 SORBENTS
Ca(OH)2 Source
Surface
Median
Ca(OH)2
Area
Pore Volume
Particle Size
Content
[m2/g]
[cm3/g]
[nmf
[percent]6
Reagent Grade
15.4
0.081
5.6
93
Linwood
13.7
0.066
3.5
87
Marblehead
18.6
0.082
4.3
86
Tenn-Luttreli
18.9
0.102
2.6
90
a by gravity sedimentation
b by TGA, weight loss between 375 and 545 °C
TABLE 2. STRUCTURAL PROPERTIES OF GROUND GLASS
Nominal Size Median Particle Surface Area Pore Volume Median Pore
[|im] Size [|im] [m2/g] [cm3/g] Size [A]
1
1.4
25.0
0.104
166
2
2.1
9.8
0.034
138
10
10.3
1.2
0.003
115
50
47.7
0.5
0.001
107
100
84.6
0.3
0.001
136
225
-------
TABLE 3. STRUCTURAL PROPERTIES OF GLASS/Ca(OH)2 SORBENTS
(Clear Glass and Reagent Grade Ca(OH)2, 6 Percent Solids Slurry)
Glass
Slurrying
Surface
Pore
Nominal Size
GCWR
Time
Temperature
Area
Volume
[jim]
[g Glass/g Ca(OH)2]
[h]
[°C]
[m2/g]
[cm3/g]
2
3:1
1
25
15
0.071
2
3:1
3
25
16
0.068
2
3:1
5
25
20
0.083
2
3:1
1
60
19
0.086
2
3:1
3
60
31
0.150
2
3:1
5
60
32
0.165
2
3:1
1
90
33
0.130
2
3:1
3
90
45
0.175
2
3:1
5
90
67
0.249
2
1:1
1
90
30
0.126
2
2:1
1
90
32
0.159
2
4:1
1
90
29
0.132
2
1:1
5
90
74
0.247
2
2:1
5
90
76
0.290
2
4:1
5
90
88
0.350
2
3:1
0.25
90
16
0.069
2
3:1
0.5
90
19
0.086
1
1:1
1
90
119
0.505
10
1:1
1
90
25
0.114
50
1:1
1
90
12
0.066
100
1:1
1
90
12
0.076
1
1:1
5
90
102
0.448
10
1:1
5
90
40
0.179
50
1:1
5
90
34
0.124
100
1:1
5
90
20
0.088
226
-------
TABLE 4. CHARACTERIZATION OF A SERIES OF GLASS/Ca(OH)2 SORBENTS
(All Slurried at GCWR=1:1, 90 °C, 1 h)
Glass Nominal
TGA Weight Loss [percent]3
Conversion
Size
with HCIb
lum]
ambient-125 °C
125-375 °C
375-545 °C
[percent]
1
4.31
5.27
7.47
23
2
1.85
2.19
8.60
21
10
0.71
1.11
9.35
19
50
0.44
0.64
10.32
21
100
0.43
0.50
10.28
17
a10 °C/min heating rate
bln STR, 500 °C, 3 s, 1000 ppm HCI (see also Figure 13)
227
-------
228
-------
The work described in this paper was not funded by the U.S. Environmental
Protection Agency. The contents do not necessarily reflect the views of the
Agency and no official endorsement should be inferred.
RESULTS OF FULL SCALE DRY INJECTION TESTS AT MSW INCINERATORS USING A NEW
ACTIVE ABSORBENT
by: Karsten S. Felsvang, Niro Atomizer Inc., Columbia, MD.
0. Helvind, A/S Scansorb, Logstor, Denmark
ABSTRACT
Worldwide incineration of municipal solid waste (MSW) has been utilized
to reduce the volume of waste to be disposed of. Increasing environmental
concerns over the potential air pollution impacts have led to emission
limits for pollutants such as HC1, S02, particulate, and more recently also
for mercury and dioxins. For a certain size of incinerators, dry sorbent
injection is the preferred technology for air pollution control.
This paper describes the development of a new active sorbent, Scansorb,
which is particularly suited for use in dry injection processes. The new
sorbent is a lime based product with adjustable properties. Scansorb can be
produced with a specific surface area of 30 to 100 m2/g.
Pilot plant development work has shown that a considerable reduction in
the absorbent quantity can be achieved when Scansorb is used instead of
commercial hydrated lime. Full scale tests performed at four different MSW
incinerators have confirmed the viability of the new active absorbent. The
full scale tests have demonstrated that more than 50# S02 removal can be
achieved with Scansorb at quantities much less than with commercial hydrated
lime.
The potential savings in absorbent and disposal costs might be a major
incentive for MSW operators to use Scansorb for air pollution control.
Preceding page blank
-------
INTRODUCTION
For many years, incineration of municipal solid waste (MSW) has been
utilized in Europe to reduce the volume of waste to be disposed of, while at
the same time producing energy for district heating or in some instances
electricity. In the 1960's, concerns were raised over the potential air
pollution impacts from MSW incineration. Hence, emission limits were
promulgated for pollutants such as HC1 and particulate emission.
In the last two decades, technology has been developed to control HC1
and other pollutants as the emission regulations became more and more
stringent. These technologies include wet scrubbers, spray dryer absorbers,
and dry sorbent injection. Based on state-of-the-art scrubbing technology,
the European Economic Community (EEC) has promulgated emission requirements
for MSW plants in Europe (Table 1). Emission limitations are set for
particulate, HC1, S02, as well as trace metals.
Competition among European countries has led to more and more stringent
emission requirements (Table 2). Pollutants such as S02, NOx, mercury, and
dioxins are now being regulated (1). As a result of the strict competition
for more stringent emission limits, the costs of cleaning up the flue gases
from MSW incinerators are sky rocketing.
In the U.S. new standards have been proposed (2) which regulate the HC1
and S02 emission from new and existing facilities according to size (Table
3). Spray dryer absorption is recommended for most stringent requirement
(95% HC1 reduction, and 852 S02 reduction), whereas, dry sorbent injection
is recommended for the less stringent requirement (80# HC1, and 50% S02).
SCANSORB - A NEW ACTIVE ABSORBENT
In the early eighties, Niro Atomizer developed and patented a lime
based dry active sorbent for use in dry injection processes. However, it
was not until later that the industry showed interest for dry injection
processes.
Niro Atomizer continued the development of lime based dry sorbents with
the purpose to:
1. Save Capital Investment
2. Save Operating and Disposal Costs
The basic idea was to develop dry active absorbents with such a high
reactivity, that the required HC1 and S02 removal could be achieved by a
simple dry injection into the duct work ahead of the existing electrostatic
230
-------
TABLE 1. EMISSION LIMITS (MG/tM3) FOR MSW INCINERATORS ACCORDING TO EEC
DIRECTIVE (JUNE 1990)
MSW Capacity
< 72 TPD >72 TPD
Particulate
100
30
HC1
100
50
so2
300
300
Cd + Hg
0.2
0.2
Ni + As
0.5
0.5
Pb + Cr + Cu + Mn
5
5
TABLE 2. EMISSION LIMITS (MG/M43) FROM EUROPEAN MSW INCINERATORS
(NOVEMBER, 1990)
Austria
Germany
Holland
Italy
Switzerland
Swede
Particulate
15
10
5
30
50
50
HC1
10
10
10
50
30
200
S02
50
50
40
300
500
-
NQx
100
100
70
-
500
-
Hg
0.05
0.1
0.05
0.1
0.1
0.08
Dlaxlns -
Equivalents
(ng/Nn3)
0.1
0.1
0.1
-
-
0.1
Note: TPD denotes Tons of
waste per day.
TABLE 3. PROPOSED EMISSION STANDARD FOR MSW INCINERATORS IN THE U.S.
New facilities
Existing facilities
HC1 Reduction 95% or 25 ppm* None*
801 or 25 ppmb 50%d
95%*
SO, Reduction 85% or 30 ppn None
50% or 30 ppn 50%
85%
a - >250 TPD
b - <250 TPD
c - <250 TPD
d - >250 to 2200 TPD
e - >2200 TPD
231
-------
]recipitator or baghouse. In this case, capital investment would be
minimized to an absorbent receiving silo and some simple injection and
distribution equipment. Further, it was the purpose to accomplish this
emission reduction with absorbent quantities much less than by using
commercial hydrated lime, whereby disposal costs could be reduced. Disposal
costs in Europe range from $20.00 to $70.00/ton and in the U.S. disposal
costs can be as high as $130.00/ton.
As described in this paper, a new lime based dry active absorbent,
called Scansorb, was successfully developed. A production facility for the
production of 10,000 ton/yr. of Scansorb has been built in Denmark. This
facility which starts production in April 1991. will cover the need for dry
sorbent in the northern part of Germany, Denmark, and the southern part of
Sweden. The production process is proprietory (patent pending) and cannot
be disclosed at this point. However, the product is lime based with certain
additives added, which do not impair the waste disposal properties of the
product.
RESULTS OF PILOT PLANT DEVELOPMENT
The pilot plant development work showed that, by carefully adjusting
the parameters in the production process, Scansorb can be produced with a
surface area (BET) ranging from 25 to 120 m2/g. In the initial development
phase, Scansorb products with different surface area were produced and used
in dry injection pilot tests to compare their performance.
Not suprisingly an increase in HC1 and S02 absorption was achieved by
increasing specific surface areas. Figure 1 shows the correlation between
specific surface area and HC1 and S02 absorption respectively. The results
are from a 1500 Nm3/h pilot plant, where the Scansorb was injected into the
ductwork upstream of a pulse jet fabric filter. The temperature was 170° C
and the stoichiometry was approximately 1.2.
The improved performance of the Scansorb products were further
demonstrated in pilot plant tests, in which the Scansorb performance was
compared with commercial hydrated lime with a surface area of 15 m2/g.
Figure 2 and 3 show the comparable HC1 and S02 efficiency of Scansorb 100,
Scansorb 35. and commercial hydrated lime.
The influence of higher specific surface area on HC1 and S02 absorption
has been verified by other investigators (3)*
232
-------
ABSORBENT, BET SURFACE AREA CMa/G>
Figure 1. Pilot plant results with Scansorb
0.0
I
3.0
90—
SO—
70—
60—
a.
g 50-
in
CD
<
„40"
o
1*1
* 30—
20-
10—
0 <
2.0
1.0
ABSORBENT, EQUIVALENCE RATIO
\
3.0
1.0 2.0
ABSORBENT, EQUIVALENCE RATIO
• Commercial Hydra tod Lins X Scansorb 35 ^Scansorb 100 • Commsrcial Hycira tod Lims X Scansorb 35 ~ Scansorb 100
Figuro 2. Pilot plont rosults with Scansorb and H^dratod Linio Figuro 3. Pilot plant results with Scansorb and Hydratod Lira*
233
-------
FULL SCALE TEST RESULTS
DESCRIPTION OF MSW FACILITIES
Full scale dry injection tests with Scansorb were carried out at four
different MSW incinerators in Denmark and Sweden. Table 4 indicates the
location and capacity of the incinerators. They are all traveling grate
incinerators, two of them equipped with electrostatic precipitators as the
only air pollution control equipment, one plant has installed a dry
injection system consisting of a reactor in combination with an
electrostatic precipitator, and the last plant has a reactor followed by a
pulse jet fabric filter. The last two installations currently use
commercial hydrated lime as absorbent. Figure 4 shows the Sonderborg MSW
incinerator equipped with electrostatic precipitators. Figure 5 is a close-
up of the ducts into which Scansorb was injected. Figure 6 shows the big
bag with Scansorb and the variable speed screw conveyor used for injection
of the product. Characteristics of the Scansorb and commercial hydrated
lime used is shown in table 5-
A typical test which lasted up to four hours, was carried out as
follows. Absorbent injection was started and the absorbent rate was
measured by the weight loss of the big bag. HC1, S02, 02, and moisture
content, were measured continuously before the injection point and after the
filter. Before the tests, the delta-T over a pitot-tube installed in the
stack, was correlated with the incinerator flue gas flow. Furthermore, the
flue gas temperature and in some cases the stack opacity was continuously
monitored. All measurements were continuously fed to a data logging system,
which allowed an on-line calculation of removal efficiency and absorbent
usage.
The incinerators were burning a typical municipal solid waste. The
pollutant concentrations that were experienced during the Scansorb dry
injection tests described below are shown in table 6.
TESTS WITH ELECTROSTATIC PRECIPITATOR AS DUST COLLECTOR
The incinerator at Holding is equipped with a small two field
electrostatic precipitator as dust collector. Scansorb and hydrated lime
were injected five meters upstream of the ESP giving a maximum reaction time
in the duct of 1/2 second. The temperature of the flue gas was in average
190 C and the moisture content of the flue gas was only approximately 10$.
Figure 7 and 8 show the achieved HC1 and S02 removal efficiencies using
Scansorb 30, Scansorb 100, and commercial hydrated lime. It can be seen
that for a given HC1 and S02 removal efficiency, a considerable reduction of
the quantity of absorbent is achieved when using Scansorb instead of
commercial hydrated lime.
234
-------
Reproduced from
best available copy.
Figure 5. Dry Injection
Set-up
235
-------
Figure 6. Big bag with Scansorb and screw feeder for absorbent
flow control.
236
-------
TABLE 4. MSW PLANT DATA
Name and plant location
Capacity (TPD)
Air pollution control
Koldlng, Dermatic
Sonderborg, Denmark
Nykobing, Denmark
Sysav, Sweden
2 x 100
2 x 120
2 x 100
2 x 300
ESP
ESP
Reactor + ESP
Reactor + FF
* ESP - Electrostatic precipitator
FT - Fabric filter
TABLE S. CHARACTERISTICS OF USED ABSORBENTS
Product
BET - Surface Area
(mJ/g)
Bulk Density
(g/on3)
Commercial Hydrate
Scansorb 35
Scansorb 100
12-18
27-43
54-93
0.60
0.37
0.25
TABLE 6. POLLUTANT CONCENTRATION DURING FULL-SCALE DRY INJECTION TESTS
Range
Average
HCL, ng/Nn'
S02, mg/Nm3
HjO, Vol. X
02, Vol. X
500-1200
200-500
8-17
10-15
800
300
14
13
4 6 T
ABSORBENT, KG/H/IOOO Nm3(DRY,10% Oa) ABSORBENT, KG/H/IOOO Nm3(DRY,10% Oj)
• Commercial Hydrat«d Lim« X Scansorb 35 ~Scorisorb 100 • Comm^rciol Hydrot«d Lin* X Scorttorb 35 ^Scorisorb 100
Figure 7, HCI Removal - Kolding Figure 8. SOj Removal - Kolding
237
-------
This difference in absorbent quantity was again verified at the
Nykobing incinerator. This incinerator is equipped with a dry injection
system consisting of a reactor and a downstream electrostatic precipitator.
Residence time in the reactor is approximately 2 seconds. Figure 9 and 10
show the results of the full scale dry injection tests with Scansorb 35 and
commercial hydrated lime. The tests were performed at 170° C at a moisture
content of 15%-
Moisture content of flue gas has a significant influence on the
performance of dry injection systems. This was verified at yet another full
scale test carried out at the Sonderborg incinerator in Denmark. This
incinerator is equipped with a two field large electrostatic precipitator
and one of the units has a rather long duct upstream the ESP. This allows
for 1 - 1.5 seconds residence time in the duct before entering the ESP.
The moisture content in the flue gas is to a great extent a function of
how well the operators run the incinerator. The better they can control the
excess air, the higher the moisture content in the flue gas. Figure 11
shows the correlation between moisture content and oxygen content of the
flue gas as developed at the Sonderborg incinerator. Rules for good
combustion practice specifies 10% oxygen in the flue gas, which will
increase the moisture content to a reasonable level. Figure 12 and 13 are
showing the influence of the flue gas moisture content on the performance of
Scansorb 100. By increasing the moisture' content from 10 to 17%t
considerable savings in the absorbent usage can be achieved.
RESULTS OF TESTS USING FABRIC FILTER AS DUST COLLECTOR
The reaction and contact time achieved by dry injection into a duct
upstream of an electrostatic precipitator is not very high. In contrary, a
fabric filter allows for a more intimate contact between the flue gas and
the collected active absorbent. This was verified by dry injection tests
performed at the Kolding incinerator where a pilot fabric filter was
installed. The results (figure 14, 15) show a considerable reduction in
absorbent usage for a given removal efficiency. Furthermore, the figures
show the influence of the flue gas temperature on the consumption.
The improved performance by using a fabric filter as dust collector
were also confirmed by dry injection tests carried out at the Sysav
incinerator in Sweden. This incinerator is equipped with a dry injection
system incorporating a reactor and downstream fabric filter. The
temperature of the flue gas was 140' C and the moisture content was 16%.
Results of the tests are shown in figure 16 and 17.
238
-------
X HCl ABSORPTION
^MW*UIQ)*g9<0O
oooooooooo
' ' » I I ' ' I I '
>
GO
U1
o
T>
CD
m
z
to
OJ
~
IA
Q
s
I
\
8 o
i «
~ a
3 „
o
o
o
z
3
o
7)
<
O
X
~
(/>
0
0
X S0a ABSORPTION
- W w ^ i/i m -g
t U f H t
>
CD
U1
O
X)
OJ
m
z
*
CD
S
x
\
o
o
o
z
3
o
X)
<
o
X
CO
c
->
ft
o
o
¦)
->
ft
o
D
-------
X HCl ABSORPTION
oooooooooo
¦ ' ' I * ¦ ' I ' '
>
CD
U)
O
73
m
m
z
*
o
\
X
\
o
o
o
z
3
o
73
O
X
X S0a ABSORPTION
0000000000
1 I I I I I I I I I
>
m
U)
o
73
W
m
z
*
o
\
z
\
o
o
o
z
3
o
73
<
o
X
(O
c
¦)
X
o
73
ft
3
0
<
0
A
0
3
<0
•V
<—
o
o
o
a
X HCL ABSORPTION
^'^''J^COCDflDCDQXOUXDIOIDO
>irjioiooMiO)(sorjiOicoo
•ooooooooooooooo
I 1 I I I I I I I M I 1 I
>
CD
0)
o
73
w
m
z
A
0
\
X
\
o
o
o
z
3
o
X
ro
o"
(J)
o
(D
c
¦)
U)
o
30
ft
3
0
<
0
A
o
D
(0
*0
0
z
>
CD
0)
o
73
W
m
z
0
\
X
\
o
o
o
z
3
o
X
KJ
o"
Ol
o
% S02 ABSORPTION
-*NU)*l/10)Nja><00
oooooooooo
o o
u
o o o o
-------
MERCURY REMOVAL
Mercury removal efficiency was measured at the Sysav incinerator. As
it appears from table 7. an average of 82% mercury removal was achieved by
Scansorb injection. Table 8 shows mercury removal efficiency measured at
the Holding plant. As this plant was operating at a much higher flue gas
temperature and is using an electrostatic precipitator, only a marginal
mercury removal was achieved. Although the mercury removal during these
tests were marginal, later development has shown that a significant mercury
removal can be achieved by modification of the Scansorb product.
CONCLUSION
The full scale test program has successfully demonstrated the
viability of the new active absorbent Scansorb. The required outlet
emission of S02 and HC1 on existing, as well as new incinerators, can be met
with a simple dry injection system using Scansorb as absorbent.
In table 9 the required amount of Scansorb for 50% S02 removal is
summarized. It appears that when commercially hydrated lime is used the
quantity required to achieve 50% S02 removal is increased with 60 to more
than 100% compared with Scansorb. Also, it is significant that the
quantities of absorbent required for 50% S02 removal when an electrostatic
precipitator is used, is considerably higher than when a fabric filter is
used as a dust collector. The potential savings in absorbent and disposal
costs might be a major incentive for MSW operators to add a new fabric
filter or rebuild the existing electrostatic precipitator into a pulse jet
fabric filter.
241
-------
TABLE 7. MERCURY REMOVAL BY SCANSORB INJECTION AT SYSAV, SWEDEN
Mercury Concentrations (ug/fin3)
Test # Inlet Outlet X Removal
3A 15 4 73
3B 26 2 92
4A 15 2 87
4B 8 2 75
Average: 82%
TABLE 8. MERCURY REMOVAL BY SCANSORB INJECTION AT KOLDING, DENMARK
Mercury Concentrations (ug/fin3)
Test # Inlet Outlet % Removal
3A 116 97 17%
3B 44 45 0
TABLE 9. ABSORBENT USAGE (Kg/1000 t*n3) FOR 5C% S02 REMOVAL
Plant Temp. X 7H 0 ESPb R + ESP FF R + FF
Koldlng 190 10 4.3(7.0)a
Nykoblng 170 15 4.2(7.5)
Sonderborg 155 15 4.8
17 3.0
Koldlng 150 14 1.9(4.8)
170 2.8
Sysav 140 16 1.9(3.8)
a - Scansorb (Hydrated lime)
b - R = Reactor
ESP = Electrostatic precipitator
FF = Fabric filter
242
-------
REFERENCES
1. Hackl , A. Stand und Tendenzen der Rauchgasreinigung nach der
Abfallverbrennung. Paper presented at the Recycle Congress Berlin,
Berlin, Germany. November, 1990.
2. Kilgroe, J.D. U.S. regulatory, research, and legislative activities
related to MWC facilities. Paper presented at the International
Conference on Municipal Waste Combustion, Hollywood, Florida. April 11-
14, 1989.
3. Stumpf, Th. Kalkhydrat grosser Oberflache - Einsatz und
Versuchsergebisse bei Rauchgasreinigungsanlagen. Paper presented at
Fach Seminar - Rauchgasreinigungsanlagen fur Kleinere Anlagen,
Ostbayerisches Technologie - Transfer - Institut, Regensburg, Germany.
Juni 1987.
243
-------
244
-------
SESSION 7C: MERCURY CONTROL
Co-Chairmen:
Karsten S. Felsvang
Niro Atomizer, Inc.
Columbia, MD
James R. Donnelly
Davy McKee Corporation
San Ramon, CA
Preceding page blank
245
-------
246
-------
MUNICIPAL WASTE COMBUSTORS:
A SURVEY OF MERCURY EMISSIONS
AND APPLICABLE CONTROL TECHNOLOGIES
by: David M. White and Kristina L. Nebel
Radian Corporation
Research Triangle Park, NC 27709
Michael G. Johnston
U. S. Environmental Protection Agency
Research Triangle Park, NC 27711
ABSTRACT
Measured mercury emissions from municipal waste combustors (MWC's) are
highly variable. Measured emission reductions from MWC's with acid gas
controls indicate that factors other than air pollution control device (APCD)
operating conditions play a significant role in determining potential emission
reductions. This paper reviews mercury emissions data from MWC's located in
the U. S. that are equipped with spray dryer/electrostatic precipitators
(SD/ESP's) or SD/fabric filters (SD/FF's). Key operating variables examined
include combustor type, APCD, flue gas temperature, and organics emission
levels. The performance of add-on mercury control technologies (sodium
sulfide [Na2S] injection, activated carbon injection) and wet scrubbing are
also discussed.
This paper has been reviewed in accordance with
the U.S. Environmental Protection Agency's peer
and administrative review policies and approved
for presentation and publication.
Preceding page blank 247
-------
INTRODUCTION
As part of Section 129 of the Clean Air Act (CAA) Amendments of 1990,
the U. S. Environmental Protection Agency (EPA) must revise the MWC standards
and guidelines that were promulgated on February 11, 1991. These revisions
must include, among other things, the addition of mercury emission limits for
MWC's.
This paper provides information on mercury emissions from MWC's and
examines how mercury removal relates to various parameters such as APCD inlet
temperature and levels of carbon in the fly ash. Further, this paper
discusses applicable mercury control techniques that could be used to reduce
mercury emissions.
EXISTING MERCURY DATA
Table 1 presents mercury emissions data from 25 MWC units (located at 16
plants) that are equipped with either SD/ESP's or SD/FF's (1). Available
inlet and outlet emissions data, along with combustor type, APCD type, APCD
inlet temperature, and inlet dioxin/furan (CDD/CDF) levels are listed. The
APCD inlet temperature is important because mercury exists in a vaporous form
and does not effectively condense onto particulate matter (PM) at temperatures
greater than roughly 300°F. Inlet CDD/CDF levels serve as a surrogate for
estimating residual carbon in the fly ash, which may enhance mercury removal
due to the adsorption of mercury onto carbon.
UNCONTROLLED EMISSIONS
Uncontrolled mercury levels from eight MWC's (four of which are listed
in Table 1) that range from roughly 200 to 1,400 micrograms per dry standard
cubic meter (/ig/dscm) at 7 percent 0, are reported (1). Based on these
data, there is no clear distinction between inlet mercury levels at mass burn
plants and at refuse-derived fuel (RDF) plants.
CONTROLLED EMISSIONS
For the 12 SD/ESP-equipped MWC's listed in Table 1, outlet mercury
emissions range from 5 to 950 /ig/dscm, with the lower emission rates occurring
at RDF plants. Due to the suspension feeding of fuel into the combustor, RDF
units are believed to have higher PM loadings and carbon contents at the
combustor exit than many mass burn units. The data support the theory that
increased levels of carbon in the fly ash enhance mercury removal, since
mercury adsorbs onto carbon. Inlet APCD temperatures during all of the tests
at these units were less than approximately 300°F.
The 13 SD/FF-equipped MWC's listed in Table 1 have outlet mercury
emissions levels ranging from below detection to 570 /ig/dscm. Flue gas
temperatures entering the APCD's were less than 300°F at all of the
facilities. Additional data provided by Ogden Martin Systems indicate outlet
mercury emissions for individual runs from SD/FF systems that are higher than
those listed in Table 1 (2). Average outlet emissions, however, are similar
to those listed in Table 1. As with the SD/ESP data, the lowest mercury
outlet levels and highest removal efficiencies occurred at the two RDF plants,
Biddeford and Mid-Connecticut. The reported inlet CDD/CDF levels at these two
plants were 903 nanograms per dscm (ng/dscm) and 436 ng/dscm, respectively.
*All emissions results reported in this paper are at 7 percent 02 and are
on a dry-basis at standard temperature (68°F) .
248
-------
The highest outlet mercury level for the SD/FF-equipped MWC's listed in
Table 1 was during the 1987 testing at the Commerce MWC. During this testing,
inlet CDD/CDF levels were very low, averaging 28 ng/dscm. During subsequent
testing the following year, average mercury outlet levels were approximately
40 jig/dscm and 70 ^ig/dscm. The inlet CDD/CDF levels during the 1988 testing
were much higher than during the 1987 testing, averaging roughly 450 ng/dscm
and 780 ng/dscm. This data further supports the theory that increased levels
of carbon in the fly ash enhance mercury removal.
It should be noted that the Hempstead facility, which is a mass burn
MWC, also had low mercury emissions -- 25 ^ig/dscm or less. Inlet CDD/CDF
levels were not reported.
Although not as apparent with the SD/ESP and SD/FF data since all of the
systems operated at flue gas temperatures less than 300°F, reduced temperature
entering the APCD also enhances the adsorption of mercury onto PM. As
discussed previously, the condensation temperature of mercury is around 300°F.
Therefore, at temperatures at or below 300 F, mercury can condense onto PM.
Data from the Quebec City pilot plant which was tested with a dry sorbent
injection/FF (DSI/FF) system demonstrated this trend (1). With an APCD inlet
temperature of 400°F and an inlet CDD/CDF level of roughly 1600 ng/dscm,
Quebec City reported an average outlet mercury emission level over 600
jig/dscm, corresponding to essentially no mercury removal. At APCD inlet
temperatures between 230°F and 285°F and inlet CDD/CDF levels between
approximately 900 and 2400 ng/dscm, average outlet mercury levels were less
than 40 ^ig/dscm and removal efficiencies were between 91 and 97 percent.
Based on the available mercury emissions data, it can be observed that
mechanisms of mercury removal are related to carbon levels in the fly ash,
reduced temperature entering the APCD, and effective collection of PM.
MERCURY CONTROL TECHNIQUES
Supplemental mercury control techniques include the injection of Na,S or
activated carbon into the combustion flue gas, and wet scrubbing. None of
these technologies are currently in use at MWC's in the U. S., but have been
applied to MWC's in Canada and Europe. Brief discussions of these
technologies are presented below.
SODIUM SULFIDE INJECTION (1)
Background
Sodium sulfide is a crystalline solid that dissolves in water. The
resulting Na2S solution is sprayed into the flue gas prior to the acid gas
control device. The specific reactions of Na2S and mercury are not totally
clear, but appear to be:
Hg (gas) + Na2S-H20 ---> HgS (solid) + NaOH, and
HgCl2 (gas) + Na2S-H20 ---> HgS (solid) + NaCl-H20
The effect of (a) flue gas temperature or (b) lime or ammonia injection for
acid gas or N0X control on these mercury reactions i?i uncertain.
Sodium sulfide is currently used for mercury control by MWC's in Avesta,
Koping, and Hogdalen, Sweden; Kempten, Germany; and Burnaby, British Columbia.
Injection of Na2S has been used at the Hogdalen MWC since 1986. The other
three plants in Europe began Na2S injection in 1989. The Burnaby MWC began
testing of Na,S in 1989 and began continuous operation with a temporary system
in December 1989.
249
-------
All of these facilities use DSI/FF systems supplied by Flakt for acid
gas and PM control. Injection of Na2S occurs prior to the DSI system at flue
gas temperatures of 260-300°F. Hogdalen reduces flue gas temperatures prior
to Na2S injection with a heat exchanger which provides hot water for district
space heating. The Burnaby MWC uses a water quench tower for flue gas
cooling. Flue gas temperatures at the stack normally range from 260-300°F.
Flakt reports that Na2S feed rates vary from 0.1 to 1 lb/ton (0.05 to
0.5 kg/Mg) of MSW, depending on site-specific conditions such as the amount of
mercury in the flue gas, the level of control required, and the amount of
carbon present in the fly ash. If a plant has little carbon in its fly ash,
it may be necessary to increase the amount of Na2S injected.
Performance
Mercury control performance data with Na2S injection are shown in
Table 2. The data has been compiled from information provided by Flakt, the
Burnaby MWC facility owner (the Greater Vancouver Regional District (GVRD)),
and from trip reports to the Hogdalen and Burnaby MWC's.
Mercury levels prior to Na2S injection at the Burnaby MWC (400-1400
pg/dscm) are higher than general inlet values reported at European MWC's (55-
560 pg/dscm) (1). The objective of the initial testing conducted at the
Burnaby MWC was to evaluate key system parameters. During these tests, 2 to 7
lb/hr (1 to 3 kg/hr) of Na2S was fed as 10-15 percent concentration solutions
and achieved mercury reductions of 50-65 percent. Subsequent tests conducted
at a feed rate of 4 to 13 lb/hr (2 to 6 kg/hr) of Na2S and a solution
concentration of 2-4 percent achieved average mercury reductions of 86 percent
and outlet mercury concentrations between 84 and 103 pg/dscm. Testing
conducted at a Na,S feed rate of 9 lb/hr (4 kg/hr) and a solution
concentration of 2 percent achieved average outlet mercury concentrations
between 117 and 155 pg/dscm; inlet mercury concentrations were not measured
during these tests, therefore percent reductions could not be calculated. The
improved mercury reduction at lower Na^S concentrations is believed to be the
result of improved atomization and mixing when feeding higher volumes of low
concentration solution versus lower volumes of high concentration solutions.
Test results from the European facilities show mercury emissions ranging
from 40 to 70 pg/dscm, with average levels of approximately 60 pg/dscm. Inlet
mercury levels measured at Hogdalen indicate mercury removal efficiencies
greater than 85 percent. The outlet levels at Kempten of less than 56 pg/dscm
correspond to an estimated removal efficiency between 65 and 90 percent.
All of the existing MWC's using Na2S injection are equipped with DSI/FF
systems. As a result, some uncertainty exists regarding the applicability of
Na2S injection to other APCD configurations, such as spray drying. Flakt
indicated that they do not believe this will be a problem, but do not have any
actual operating experience with application of Na,S injection to spray drying
systems. Flakt did indicate, however, that it would probably be necessary to
have separate Na2S and calcium sorbent feed and injection systems to avoid CaS
scaling of the sorbent feed line.
Cost Estimates
Available cost data for Na2S injection systems are based on estimates
from Flakt, information provided for the Burnaby plant, and chemical costs
from PPG. The Burnaby MWC operator estimated capital costs for a Na,S system
for the Burnaby plant, which has a capacity of 800 TPD, at $150,000-250,000.
The chemical costs for the sodium sulfide, as quoted by Flakt, range from
$0.10-0.50/ton of MSW. This cost is dependent upon the uncontrolled mercury
level and the level of reduction required. The chemical cost reported for the
Burnaby MWC is $0.50/ton of MSW, and the chemical cost (without shipping)
reported by PPG is $0.30/ton of MSW, both of which are consistent with Flakt's
estimate. Based on this information, annualized costs (based on a capital
250
-------
recovery factor of 0.1315 and 8,000 hours of operation per year) for Burnaby
are estimated at $0.50-1.00/ton of MSW.
ACTIVATED CARBON INJECTION
Another mercury control technology used in Europe is the injection of
powdered activated carbon prior to the APCD. This technology has been used
commercially on two MWC's, one located in Zurich, Switzerland, and one in
Geiselbullach, Germany. Additionally, activated carbon injection has been
used during a test program at an MWC in Amager, Denmark.
The Zurich MWC is equipped with an SD/ESP system, the Geiselbullach MWC
is equipped with a DSI/FF system, and the Amager MWC is equipped with a SD/FF
system. At the Zurich plant, powdered activated carbon is injected into the
flue gas, ahead of the SD, at a rate of 2 to 3 kg/hr (4 to 7 lb/hr). The
temperature entering the SD is between 430 and 540°F. Test results from the
Zurich plant show an increase in mercury reduction from 70 percent without
activated carbon injection, to 90 percent with activated carbon injection (3).
At the Geiselbullach plant, solids recirculation is used to improve
sorbent and additive utilization, and a heat exchanger is used to cool the
flue gas prior to its entering the DSI. A powdered mixture of calcium
hydroxide [Ca(0H)2] and coke is then injected into the upward-flowing cooled
and humidified flue gas. Test data and operational features are not available
for the Geiselbullach plant (3).
The Amager MWC operates similarly to the Zurich plant. Testing was
conducted with temperatures at the SD exit of 284°F and at 260°F. As shown in
Table 3, results from the testing with activated carbon injection at the
higher temperature indicate outlet mercury levels between 23 and 77 ^g/dscm,
corresponding to removal efficiencies between 82 and 95 percent. Without
activated carbon injection, outlet mercury emissions were between 67 and 195
^g/dscm, with removal efficiencies between 15 and 65 percent. The highest
removal efficiencies when using activated carbon occurred with increased
additive levels (70 mg/dscm vs. 7 mg/dscm) (4).
Testing at the lower APCD inlet temperature shows greater control of
mercury, especially when activated carbon injection was not used. With
activated carbon, outlet mercury levels ranged from 6 to 24 ^g/dscm (88 to 97
percent removal), and without activated carbon the outlet levels were between
30 and 53 ^g/dscm (72 to 92 percent removal) (4).
No cost data are available for activated carbon injection. However, the
costs are expected to be low, similar to Na2S injection. One of the criteria
for selecting the additive used at the Zurich MWC was its low cost (3).
WET SCRUBBING
Wet scrubbing is a form of acid gas control that has primarily been used
in Europe and Japan. Wet scrubbing involves passing the flue gas through an
ESP to reduce PM, followed by an absorber where flue gas is contacted with an
alkaline solution to saturate the gas stream and reduces flue gas temperatures
to as low as 130°F. The alkaline solution, typically containing calcium
hydroxide [Ca(OH),] , reacts with the acid gas to form salts, which are
generally insoluble and may be removed by sequential clarifying, thickening,
and vacuum filtering. The dewatered salts or sludges are then landfilled.
The wet scrubbing technology results in increased mercury reduction
relative to SD/FF control without supplemental mercury control due to the
lower operating temperature that increases mercury condensation. As a result,
the collection of mercury improves and mercury can be reduced by up to 90
percent. Disadvantages to wet scrubbing, however, include the quantity of
water required and potential difficulties with waste handling. Further,
control of organics may be less than with dry acid gas controls.
251
-------
Test results from two MWC's located in France and one MWC located in
Switzerland are available (5). The two French plants, Lyon-Nord and Lyon-Sud
began commercial operation in 1989 and 1990, respectively, and they are
equipped with ESP's followed by wet scrubbers. Mercury emissions results from
these plants are shown in Table 4. Average mercury outlet emissions at Lyon-
Nord were under 50 g/dscm for Unit 1 and 62 /ig/dscm for Unit 2. Average
removal efficiencies were greater than 82 percent for Unit 1 and 62 percent
for Unit 2. At Lyon-Sud, average mercury outlet emissions were less than
approximately 60 fj,g/dscm for both units, and average removal efficiencies were
greater than 86 percent.
The Basel, Switzerland MWC, which was equipped with only ESP's, was
retrofitted with wet scrubbing systems in 1989. Mercury outlet emissions,
listed in Table 4, ranged from 16 /ig/dscm to 20 /ig/dscm at Unit 1, and from
less than 13 fj.g/dscm to 34 ^g/dscm at Unit 2. This corresponds to average
removal efficiencies between 90 and 96 percent for Unit 1, and between 82 and
96 percent for Unit 2.
REFERENCES
1. White, D. M. and K. L. Nebel (Radian Corporation). Summary of
Information Related to Mercury Emission Rates and Control Technologies
Applied to Municipal Waste Combustors. Prepared for U. S. Environmental
Protection Agency, Research Triangle Park, North Carolina.
September 1990.
2. Sussman, D. B. (Ogden Martin Systems). Testimony Before the National
Air Pollution Control Techniques Advisory Committee. Research Triangle
Park, North Carolina. January 31, 1991.
3. Memorandum from T. G. Brna, U. S. Environmental Protection Agency,
CRB/ORD, to W. H. Stevenson, U. S. Environmental Protection Agency,
SDB/OAQPS. Mercury Emission Control from Municipal Waste Combustors
Using Activated Carbon Injection in Flue Gas. August 20, 1990.
4. Felsvang, K. S., Sander Holm, T., and Brown, B. Control of Mercury and
Dioxin Emissions from European MSW Incinerators by Spray Dryer
Absorption Systems using Rotary Atomizers. Paper presented at 1990
Summer National Meeting, American Institute of Chemical Engineers, San
Diego, California. August 19-22, 1990.
5. Belco Technologies Corporation. Presentation of Performance Test Data
to U. S. Environmental Protection Agency on Municipal Waste Combustors.
August 1990.
252
-------
TABLE 1. EXISTING MERCURY EMISSIONS DATA FROM SD/ESP'S AND SD/FF'S
MWC RUN COMBUSTOR APCD APCD INLET INLET Hg OUTLET Hg REMOVAL
NAME NUMBERS TYPE TYPE TEMP. CDD/CDFa AVERAGEa AVERAGEa EFFICIENCY
(F) (ng/dscm) (ug/dscm) (ug/dscm) (%)
Charleston, Units A & B
1-3
MB/WW
SD/ESP
—
—
b
—
723
—
Honolulu, Unit 1
1-3
RDF
SD/ESP
300
—
c
—
5
—
Honolulu, Unit 2
1-3
RDF
SD/ESP
293
—
c
—
7
—
Mi 11 bury. Unit 1
1-6
MB/WW
SD/ESP
249
—
—
565
—
Mi 11 bury, Unit 2
1-3
MB/WW
SD/ESP
240
170
d
—
954
—
Portland, Unit 1 (12/89)
4-6
MB/WW
SD/ESP
308
—
e
—
550
—
Portland, Unit 2 (12/89)
1-3
MB/WW
SD/ESP
285
--
e
—
382
--
SEMASS, Unit 1
1-3
RDF
SD/ESP
287
—
f
—
59
—
SEMASS, Unit 2
2-4
RDF
SD/ESP
293
—
f
—
105
—
West Palm Beach, Unit 1
3 tests
RDF
SD/ESP
275
—
—
56
West Palm Beach, Unit 2
3 tests
RDF
SD/ESP
278
—
—
23
—
Babylon, Unit 2
1-3
MB/WW
SD/FF
331
—
—
451
—
Biddeford
1-3
RDF
SD/FF
278
903
389
ND g
>99
Commerce (1987)
11,13,14
MB/WW
SD/FF
270
h
28
450
570
-26.7
Commerce (1988)
3,5,9
MB/WW
SD/FF
290
h
446
d
453
39
91.4
Commerce (1988)
13,16,18,29
MB/WW
SD/FF
290
h
783
d
261
68
74.0
Hempstead, Unit 1 (9/89)
1-3
MB/WW
SD/FF
310
h
—
—
9
—
Hempstead, Unit 2 (9/89)
1-3
MB/WW
SD/FF
310
h
—
—
25
—
Hempstead, Unit 3 (10/89)
1-3
MB/WW
SD/FF
310
h
—
—
25
—
Indianapolis, Unit 1
1-3
MB/WW
SD/FF
307
—
—
283
—
Long Beach
1-3
MB/WW
SD/FF
298
305
—
180
—
Marion County
4-6
MB/WW
SD/FF
272
43
d
—
239
—
Mid-Connecticut (7/88)
1-3
RDF
SD/FF
276
1019
1008
—
Mid-Connecticut (7/88)
1-3 i
RDF
SD/FF
284
—
884
50
94.3
Mid-Connecticut (2/89)
12-14
RDF
SD/FF
—
436
668
9
98.7
Quebec City - Pilot
7-8
MB/WW
SD/FF
282
1764
187
10
94.7
Quebec City - Pilot
9-10
MB/WW
SD/FF
284
2157
360
19
94.7
Stanislaus County, Unit 1
14,16,19
MB/WW
SD/FF
295
--
—
499
--
Stanislaus County, Unit 2
38,40,42
MB/WW
SD/FF
290
—
—
462
—
-------
M/E - Modular/Excess Air
M/S - Modular Starved Air
MB/R - Mass Burn/Refractory
MB/WW - Mass Burn/Waterwall
RDF - Refuse Derived Fuel
a Results reported at 7% 02.
b Outlet CDD/CDF values: Unit B - 44.2 ng/dscm (average of three runs conducted during same test campaign),
c Outlet CDD/CDF values: Unit 1 - 6.3 ng/dscm; Unit 2 - 3.8 ng/dscm.
d Inlet CDD/CDF samples collected during separate runs from Hg, but during same test campaign and at similar
operating conditions,
e Outlet CDD/CDF values: Unit 1 - 36.5 ng/dscm; Unit 2 - 43.6 ng/dscm.
f Outlet CDD/CDF values: Unit 1 - 9.3 ng/dscm; Unit 2 - 12.3 ng/dscm.
g Not detected.
h Temperatures not reported for Hg runs; temperatures estimated based on other runs during same test campaign,
i Additional inlet and outlet mercury samples were collected by Method 101A. Not measured simultaneously with
other metals.
254
-------
TABLE 2. MERCURY EMISSIONS DATA WITH Na2S INJECTION
MWC
WASTE
APCD
Na2S
TESTED Hg
EMISSIONSa
PLANT
TPD
TYPE
FEED RATE
INLET
OUTLET
(kg/hr)
(ug/dscm)
(ug/dscm)
Burnaby
265
DSI/FF
Run 1 =1.0
1465
570
Unit ?, 3/89
Run 2 = 2.0
993
407
(10% Na2S)
Run 3 = 2.0
1151
393
AVG 1203
AVG
457
Unit 1, 4/89
All runs =
1423
670
(15% Na2S)
3.0
1443
750
1205
473
AVG 1357
AVG
632
Unit ?, 8/89
Run 1 = 2.5
406
98
(2-4% Na2S)
Run 2 = 6.0
775
91
Run 3 = 2.0
670
84
Run 4 = 3.0
793
101
Run 5 = 6.0
661
103
AVG 661
AVG
95
Unit 1, 12/89
All runs =
NR
138
(2% Na2S)
4.0
NR
67
NR
146
AVG
117
Unit 2. 12/89
All runs =
NR
149
(2% Na2S)
4.0
NR
115
NR
118
AVG
127
Unit 3, 12/89
All runs =
NR
152
(2% Na2S)
4.0
NR
159
AVG
155
Hogdalen - 3 Units
400
DSI/FF
1.24
NR
70
Unit 3, 8/86-9/86
1.05
370
40
1.4
497
61
0.89
NR
55
AVG
57
Kempten (Germany)
210
DSI/FF
NR
NR
<56
REDUCTION
EFFICIENCY
(%)
AVG 62
AV6 53
AVG 86
NA
NA
NA
89
88
65-90 b
NR - Not reported; NA - Not applicable
a Results reported at 12% C02 (assummed to be equal to 7% 02).
b Based on general inlet values at German MSW plants ranging from 170-560 ug/dscm.
255
-------
TABLE 3. MERCURY EMISSIONS DATA FROM THE AMAGER MWC
SD OUTLET INLET OUTLET REMOVAL
ADDITIVE TEMPERATURE MERCURY MERCURY EFFICIENCY
(mg/dscm) (F) (ug/dscm) (ug/dscm) (%)
284
203
229
219
202
165
154
195
86
74
67
24
15
61
64
59
284
378
227
58
40
85
82
20
284
214
248
336
31
35
36
86
86
89
70
284
1516
318
77
23
95
93
260
421
196
163
189
32
48
30
54
92
76
82
71
23
83
260
260
201
198
220
24
6
7
88
97
97
*A11 concentrations are at 7 percent 02.
256
-------
TABLE 4. MERCURY EMISSIONS DATA FROM WET SCRUBBING SYSTEMS
MWC
UNIT
RUN
INLET
OUTLET
REMOVAL
NAME
NUMBER
MERCURY
MERCURY
EFFICIENCY
(ug/dscm)
(ug/dscm)
(%)
LYON-NORD,
1
1
168
<49
>71
FRANCE
2
289
<50
>83
3
578
<49
>91
AVERAGE
345
<49
>82
2
4
177
49
72
5
177
76
57
6
140
60
57
AVERAGE
165
62
62
LYON-SUD.
1
3
457
72
84
FRANCE
4
568
<49
>91
AVERAGE
513
<61
>88
2
1
438
69
84
2
373
<49
>87
AVERAGE
406
<59
>86
BASEL,
1
1
252
16
94
SWITZERLAND
2
168
17
90
4
401
17
96
7
513
20
96
8
187
19
90
2
1
186
<13
>93
2
224
<14
>94
3
261
32
88
4
224
<13
>94
5
168
<13
>92
6
168
21
88
7
140
<13
>91
20
363
13
96
21
270
33
88
23
75
<13
>82
24
196
<13
>93
*A11 concentrations are at 7 percent 02.
257
-------
Intentionally Blank Page
258
-------
The work described in this paper was not funded by the U.S. Environmental
Protection Agency. The contents do not necessarily reflect the views of the
Agency and no official endorsement should be inferred.
CONTROLLING MERCURY EMISSIONS FROM RDF FACILITIES
by: Gary G. Pierce
Director, Environmental Permitting and Services
ABB Resource Recovery Systems
Windsor, CT 06095
ABSTRACT
Concern over the potential effects of mercury on food chains, public
health, and sport fisheries is resulting in close examination of sources of
mercury in the environment, in particular municipal waste combustors
(MWC's). Mercury emissions data from MWC's, mercury emission test methods
and the potential health effects of mercury are being scrutinized. The
U.S. EPA is studying potential mercury emission control technologies and
developing specific mercury emission standards for MWC's.
The potential air emissions of mercury from the processing of
municipal solid waste (MSW) at refuse derived fuel (RDF) facilities are
effectively minimized. The combination of processing MSW to remove
noncombustibles in the production of RDF, and the use of spray
dryer/electrostatic precipitator (SD/ESP) or spray dryer/fabric filter
(SD/FF) technology to control emissions from RDF-fired boilers, results in
extremely high reductions of potential mercury emissions.
Tests conducted at the Greater Detroit Resource Recovery Facility
(GDRRF) and at the Mid-Connecticut Resource Recovery Facility (MID-CT) in
Hartford, CT, each employing the ABB waste processing technology, indicate
that a significant portion of the mercury-bearing components are removed
from MSW through magnetic separation and subsequent processing steps prior
to combustion of the resultant RDF. High mercury removal efficiencies were
achieved at the RDF facilities in Biddeford, Maine (> 99 percent) and at
MID-CT (98.3 percent), both of which are equipped with SD/FF. Outlet flue
gas mercury concentrations recorded at the RDF facility in Honolulu,
Hawaii, and other RDF facilities equipped with SD/ESP, are among the lowest
values reported.
Preceding page blank
-------
Statistical analysis of the data from the U.S. EPA/Environment Canada
(U.S.EPA/EC) test program conducted at MID-CT indicates a good correlation
between mercury removal by the flue gas cleaning system and decreasing flue
gas temperature (spray dryer outlet) and increasing carbon content of the
fabric filter ash. Increased sorbent-to-acid-gas ratio (stoichiometric
ratio) tended to result in increased mercury emissions. Questionable
validity of the percent carbon in flyash data from the test program
dictates that additional research into the mechanisms of mercury removal is
necessary.
MERCURY EMISSIONS FROM MWC's
Mercury is emitted to the atmosphere from natural (non-anthropogenic)
sources and from human (anthropogenic) sources. Non-anthropogenic sources
include oceans, volcanoes, forest fires and terrestrial mineral deposits.
Fossil fuel combustion, the manufacturing and application of paints and
pesticides, the manufacturing and disposal of batteries and fluorescent
lamps, and municipal solid waste incineration are human activities which
generate mercury emissions. Mercury is also used in the manufacturing of
chlorine and caustic as well as in non-household consumer batteries used
for medical, industrial, and military purposes.
Based on published estimates, mercury emissions from MWC's in the
United States represent a small percentage of the annual global atmospheric
mercury burden. The Electric Power Research Institute (EPRI) estimates
total annual global emissions of mercury to be between 5 and 6 million
kilogram^, about 30 to 55 percent of which is derived from anthropogenic
sources. EPRI's estimate that fossil fuel combustion by utilities in the
United States amounts to approximately 75,000 kilograms of mercury annually
agrees with an estimate published by the New Jersey Chapter of the Clean
Water Action/Clean Water Fund. The latter source also contains an
estimate that 34,000 kilograms of mercury were emitted from MWC's in the
United States in 1989. If these estimates are correct, mercury emissions
from MWC's in the United States accounted for from 0.6 to 0.7 percent of
the global total in 1989.
Mercury emissions from MWC's are directly related to the presence of
mercury-bearing components in the MSW stream during combustion. According
to a recent study by Franklin Associates, approximately 88 percent of the
mercury contained in MSW comes from household batteries, primarily alkaline
and mercuric zinc "button" batteries. This study also estimates that
electric lighting components contribute 5 percent, thermometers 2 percent,
thermostats 2 percent, pigments 2 percent, and other components, which
include dental uses, special paper coating, mercury light switches, and
film pack batteries, approximately 1 percent of the total amount of
mercury. It is estimated that the mercury content of ^SW is on the order
of approximately 0.003 to 0.004 pounds per ton of MSW. Efforts are
underway to reduce this concentration by the reformulation of consumer
products.
260
-------
METHODS OF MINIMIZING MERCURY EMISSIONS FROM MWC's
There are four potential means of minimizing mercury air emissions
from MWC's. These methods are identified as follows:
1. The removal of mercury from consumer products which ultimately
are disposed of and contained in the incoming MSW stream.
2. The removal of mercury-bearing components from the MSW stream
prior to reaching the MWC (i.e. source separation).
3. The removal of mercury-bearing components from the MSW stream at
the MWC.
4. The use of post-combustion air pollution control technology at
MWC's to remove mercury compounds from the flue gas.
Brief discussions of the first two methods are presented below,
followed by a discussion of the effectiveness of RDF technology in
controlling mercury emission by the latter two means.
REMOVAL OF MERCURY FROM CONSUMER PRODUCTS
Substantial progress is being made to reduce the amount of mercury
discarded into MSW. The National Electrical Manufacturer's Association has
announced its intent by 1993 to substantially Reduce the mercury content of
alkaline batteries to 0.025 percent by weight. Such batteries are
currently available on the market in some areas. Mercury zinc "button"
batteries, which each contain approximately 50 percent mercury by weight,
are being replaced to some degree by silver oxide and zinc air batteries
which contain less mercury. In August 1990, the U.S. EPA banned the use of
mercury from all new interior latex house paints. Mercury discarded from
dental uses, paper coating, and film pack batteries is also projected to
decline rapidly or be eliminated. According to the Franklin Associates
study, the amount of mercury discarded in MSW will decrease by
approximately 65 percent between the years 1990 and 1995.
REMOVAL OF MERCURY BY SOURCE SEPARATION
Efforts to remove mercury-bearing components from the MSW stream prior
to reaching MWC's have almost exclusively consisted of household battery
recycling programs. Such programs associated with the Hennepin County, MN;
New Hampshire/Vermont Solid Waste Project; and the GDRRF have met with
limited success. While large quantities of batteries have been collected,
some of them have a limited mercury content, and most of them are not being
recycled. Rather, the majority are be^ng stored, or disposed of in
sanitary or hazardous waste landfills. The U.S. EPA did not adopt a
proposed requirement to remove mercury-bearing household batteries from the
incoming MSW stream to an MWC in adopting the New Source Performance
261
-------
Standards for MWC's on January 11, 1991. The Franklin Associates study
assumed that only 5 percent of the mercury in batteries in 1995 and 20^
percent in the year 2,000 will be recovered through source separation.
CONTROL OF MERCURY EMISSIONS AT RDF FACILITIES
Evidence indicates that RDF technology is effective in minimizing
potential air emissions of mercury which are generated as a result of
mercury contained in the incoming MSW stream. Such facilities differ from
more common mass burning MWC's by processing MSW to remove noncombustibles
before burning, and by injecting the RDF into a boiler so that
approximately 50 percent of the fuel burns in suspension instead of all the
MSW burning on a grate. Research and developmental testing at the GDRRF
and the MID-CT Facility indicates that front-end processing of MSW to
produce RDF results in the removal of a significant percentage of the
mercury in the incoming MSW stream. Mercury emissions test data
demonstrates that the mercury content of the flue gas which is generated by
the combustion of RDF can be reduced by greater than 96 percent using SD/FF
technology. Low stack gas mercury concentrations have also been recorded
at RDF facilities equipped with DS/ESP technology.
REMOVAL OF MERCURY FROM MSW
ABB Resource Recovery Systems (ABB-RRS) is the leader in the design,
construction and operation of RDF facilities. Currently, ABB-RRS has three
operating RDF facilities employing essentially identical front-end
processing technology and boiler design features, but each with different
air pollution control technology. These facilities are the MID-CT in
Hartford, CT with DS/FF technology; the GDRRF in Detroit, MI with lime
addition and ESP technology; and the Honolulu Resource Recovery Venture
Facility in Honolulu, HI with DS/ESP technology.
The flow of waste through the waste processing system at an ABB
process facility is shown on Figure 1. The design includes complete
residue conveying and materials load-out systems. The process equipment is
arranged to minimize changes in flow direction and material transfer
points.
Incoming trucks are weighed and then directed to the MSW receiving
area where the waste is discharged on the receiving building floor. The
waste is spread and compacted in the storage area with large sized
front-end loaders and inspected for the presence of non-processible items
which are separated for removal from the facility.
Processible waste is loaded onto a feed conveyor by front-end loaders.
The solid waste is metered by the feed conveyor onto an inclined steel pan
conveyor for leveling. The inclined conveyor carries the waste onto a
horizontal, steel infeed conveyor which feeds the primary shredder. Prior
to the shredder, a picking station is located alongside each horizontal
feed conveyor to allow the operator to inspect the waste being carried to
262
-------
the primary shredder and to stop the conveyor and remove non-processible
items. Non-processible items are picked using a hydraulic grapple
controlled from the picking station control booth.
The primary shredder is a flail mill which performs coarse shredding
of the waste stream. Closed bags are opened, ferrous metal is exposed, and
glass containers are broken. Each primary shredder is enclosed in a
reinforced concrete room and equipped with a dedicated dust control system
consisting of an exhaust fan and fabric filter.
Upon completion of primary shredding, the waste is conveyed to the
ferrous separation system which consists of a primary electromagnetic drum
and a secondary drum type transfer magnet.
After the ferrous separation point, the waste is conveyed to the
primary separation unit which is a totally enclosed trommel rotary screen.
Two discharge streams are conveyed from the trommel. One of the streams is
a sized fraction consisting of small combustible products, aluminum cans,
and medium sized non-combustibles. This stream is transported to the
secondary separation system. The second stream is an oversized fraction
consisting primarily of paper and plastic which is conveyed to the
secondary shredder for size reduction.
The secondary separation unit consists of a totally enclosed,
two-staged trommel rotary screen which further separates the sized fraction
received from the primary separation units. The discharge undersized
streams consist of:
1. Process residue consisting of fine sand, glass, dirt, etc., which
is conveyed directly to the residue load-out area where further
recovery of combustibles may be provided; and
2. A sized fraction which is conveyed to the RDF collection
conveyor.
The oversized fraction from the secondary separation unit is conveyed to
the RDF storage area.
The oversized fraction from the primary separation unit is conveyed to
the secondary shredder. The secondary shredder is a high horsepower,
horizontal shaft machine with hammers and grates arranged and sized to
produce the desired RDF particle size. Hammer and grate shape and size are
changeable to control top size and optimum particle size for boiler
performance and to accommodate seasonal variations in waste composition.
RDF is conveyed to the RDF storage building where a front-end loader
distributes and stockpiles the RDF.
On a weight basis, the following approximate materials distributions
exist at an RDF facility during normal operations:
263
-------
1. Ferrous - 5 percent
2. Primary Trommel Overs (RDF stream) - 60 percent
3. Primary Trommel Unders (Residue stream) - 6 percent
4. Secondary Trommel Overs (RDF stream) - 23 percent
5. Secondary Trommel Unders (Residue stream) - 6 percent
Thus, waste processing converts approximately 83 percent of the processible
MSW stream to RDF, with the remainder removed as ferrous material (5
percent) and residue (12 percent).
Data from studies conducted at both the GDRRF and the MID-CT
facilities indicate that waste processing to produce RDF results in the
removal of mercury-bearing components and, hence, mercury from processible
MSW prior to combustion. This, in turn, results in a reduction in
potential mercury air emissions generated by combustion, as compared to
levels which would have occurred in the absence of processing.
MAGNETIC SEPARATION OF BATTERIES
ABB-RRS has conducted testing at the GDRRF to assess the fate of
mercury-bearing household batteries as a result of processing the incoming
MSW stream. Data was obtained during six days of field testing at the
facility during the week of February 5, 1990.
A test plan was carried out which involved placing color-coded
batteries into the processible MSW stream prior to processing. Annual
national sales information was used to determine the number and types of
batteries with which to "spike" a known weight of processible MSW. Having
spiked a 14.63-ton sample of waste with 221 batteries prior to processing,
each of the three resultant process streams (ferrous, residue and RDF)
could be collected and sorted to determine the number of batteries.
Color-coded batteries, along with any additional batteries originally in
the waste material, were recovered and their size and type recorded.
The results of the test program indicated that 31 of the 221
color-coded batteries (14 percent) were magnetically separated from the MSW
sample during processing. These batteries were almost exclusively
steel-jacketed alkaline batteries. Based on the reported mercury content
of all of the color-coded batteries, the mercury content of the
magnetically separated batteries was calculated to equal approximately 17.6
percent of the total amount of mercury contained in all of the batteries.
In a perceived effort to minimize mercury emissions, the facility was
mandated in November 1989 to operate in a bypass configuration whereby all
of the material which falls through the primary trommel is diverted to
landfill, rather than directed to the secondary trommels for normal
processing. Because of this operating requirement, useful data on battery
removal other than by the drum magnet system during normal operation of the
facility could not be obtained during the test program.
264
-------
PARTITIONING OF ELEMENTS BY PROCESSING MSW
The MID-CT Resource Recovery Facility, located in Hartford,
Connecticut, was recently tested to study the partitioning of elements in
MSW and the advancement of methods for processing refuse. The study was
conducted for ABB-RRS by the University of^Wisconsin-Madison with funding
supplied by ABB Resource Recovery Systems. The research was conducted
from the summer of 1988 through the end of 1989.
A portion of the research involved an examination of twenty-seven (27
elements) in MSW and in the end products (i.e. ferrous, residue, and RDF)
of the MSW processing system at the facility. Composite samples of the
streams for a five-day production period were reduced following a protocol
developed by the research team and then analyzed.
The results from the MID-CT portion of the test are presented in Table
I. The distribution analyses indicate that approximately 53 percent of the
mercury in the processible MSW stream was diverted from the RDF which was
combusted at the facility. The majority of this amount was removed via the
drum magnet system.
Because of the nature of the sampling methods and variability of MSW
and the process streams, the results of this study should be viewed in
relative rather than precise quantitative terms. Nonetheless, the research
supports the findings at the GDRRF in that a significant portion of the
mercury contained in the incoming MSW stream to an RDF facility is
magnetically removed prior to combustion of the RDF product.
POST COMBUSTION CONTROL OF MERCURY EMISSIONS
Emissions data from MWC's indicate that mercury levels in the flue gas
from RDF MWC's can be efficiently controlled using DS/FF or DS/ESP
post-combustion control technology. The U.S. EPA recently published a
mercury emissions databage which covered 40 MWC units at 28 different
plants in North America. Outlet flue gas mercury concentrations, type of
combustor, type of air pollution control technology, control device
operating temperature, and inlet dioxin/furan concentration were reported.
Where available, control device inlet flue gas mercury concentrations were
reported, and mercury removal efficiencies were calculated.
Mercury emissions data from ten MWC's (13 units) equipped with SD/FF
technology were contained in the U.S. EPA's database. The highest mercury
removal efficiencies were detected at the only two listed RDF facilities.
Mercury removal efficiencies at the Biddeford, ME facility were greater
than 99 percent and averaged 98.3 percent at MID-CT. The mercury emissions
data from the 13 performance tests of the U.S. EPA/EC test program at
MID-CT, for which the mercury removal efficiency averaged 98.3 percent, are
presented in Table II. The average inlet flue gas mercury concentration of
659 ug/dscm at 7 percent 0^ is approximately equal to the average of the
entire database. The outlet flue gas mercury concentrations at these two
265
-------
RDF facilities were among the lowest values reported, and much lower than
the average outlet concentrations for similarly equipped mass burn MWC's.
Outlet mercury emissions data were reported for six MWC facilities (12
units) equipped with DS/ESP technology. Three of the six MWC's were mass
burn facilities, and the other three were RDF facilities. Similar control
device operating temperatures were reported for all six facilities. Outlet
flue gas mercury concentrations at the mass burn MWC's ranged from
approximately 380 to 950 ug/dscm at 7 percent 0^. Mercury emission data
from the three RDF MWC's (6 units) is presented in Table III. Outlet flue
gas mercury concentrations for these RDF facilities ranged from 5 to 105
ug/dscm at 7 percent 0^, well below the range reported for the mass burn
MWC's. No inlet flue gas mercury concentration data were available for any
of the six facilities; hence, no mercury removal efficiencies could be
calculated.
OPERATIONAL FACTORS RELATED TO MERCURY EMISSIONS CONTROL
The primary purpose of the U.S. EPA/EC test program conducted at
MID-CT was to aid both agencies in the development of regulations for MWC
facilities. In addition to establishing an emissions baseline for a new
RDF-fired facility, the goals of the test program included evaluation of
design and operating parameters, and the establishment of design and
operating criteria for combustion and flue gas cleaning systems. MID-CT
was the third MWC selected for evaluation under EC's National Incinerator
Testing and Evaluation Program (NITEP), which had previously examined the
two-stage combustion facility in Parksdale, Prince Edward Island, Canada
and the moving grate mass-burn facility in Quebec City, Canada. The U.S.
EPA had previously analyzed data from the mass burn MWC's in Marion County,
OR and Millbury, MA, as well as the RDF MWC in Biddeford, ME.
A two-phased approach was used in the NITEP testing program at MID-CT.
Twenty-eight characterization tests were conducted in January 1989 to
provide information on an RDF MWC under a wide range of operating
conditions. These data were then used in the process of selecting process
scenarios to be examined during the performance testing phase. A total of
13 valid performance test runs were conducted during the period of February
13 to March 1, 1989. During some of the performance tests, the combustion
and air pollution control systems at the facility were deliberately
operated under "poor" conditions in order to assess the effect of various
parameters upon emission levels.
Streams and locations sampled during testing included the RDF feed,
pond water, lime slurry feed, grate siftings, economizer ash, dry and wet
bottom ash, fabric filter ash, air preheater inlet, spray dryer inlet and
fabric filter outlet. Concurrent with the sampling, a data acquisition
system was used to record process data and continuous emission monitoring
data from the spray dryer inlet, spray dryer outlet and fabric filter
outlet locations. The large volume of data collected was statistically
analyzed using single linear regression and multiple linear regression to
266
-------
develop coefficients of correlation and determination between emission and
process parameters. These data and determinations were used in the
development of the New Source Performance Standards (NSPS) and emission
guidelines for new and existing RDF MWC's, respectively, which were adopted
by the U.S. EPA on January 11, 1991.
In accordance with the Clean Air Act Amendments of 1990, the U.S. EPA
must adopt specific mercury emission standards for new and existing MWC's
by mid-November of 1991. The U.S. EPA is currently studying mercury
emissions data and potential mercury emission control technologies, such as
the injection of sodium sulfide or activated carbon into the combustion
flue gas. None of these technologies are currently being used at an MWC in
the United States and they have only been used at a limited number of MWC's
in Europe and Canada. Also under study are the factors which affect
mercury removal and explain the apparently enhanced mercury removal at RDF
MWC's as compared to mass burn MWC's.
Although stating that their exact effects are unclear, it is asserted
by the U.S. EPA that factors such as combustion efficiency and the type and
operation of air pollution control equipment affect mercury removal. It is
purported that good particulate matter control, low temperatures in the
control system, and the level of carbon in the flyash are associated with
enhanced mercury removal efficiency. It is reasoned that a combination of
low operating temperature of the particulate matter control device and
higher levels of carbon in the flyash enhance mercury adsorption onto the
particulate matter which is then removed from the flue gas by the
particulate matter control device. This theory would support the higher
mercury control efficiencies noted at RDF MWC's, as compared to mass burn
MWC's, in that suspension firing of RDF results in slightly greater amounts
of flyash containing higher levels of carbon than at mass burn MWC's.
However, the concentration of flyash in the uncontrolled flue gas stream at
RDF MWC's is significantly greater than at mass burn MWC's because of the
reduced excess air levels at which RDF MWC's are typically operated. Most
of the unburned carbon remaining at mass burn MWC's is contained in the
bottom ash stream and is unavailable for functioning in an adsorptive
capacity.
Data collected during the NITEP test program at MID-CT partially
supports the U.S. EPA's theory of mercury removal from MWC flue gas. Data
relevant to this discussion is presented in Table IV. The flue gas carbon
content data presented in Column 4 of Table IV for each performance test
was calculated by multiplying the tested values of percent carbon in flyash
by the measured rates of flyash produced during each test run.
Measurements of percent carbon in the ash were based on loss-on-ignition
measurements and ranged from 4.26 to 10.45 percent. It has been noted by
Radian Corporation that much of loss-on-ignition is suspected to be water
of hydration in the collected slurry solids and, therefore, using these ash
samples to estimate percent carbon in the ash may cause an overestimate.
267
-------
The carbon content of the ash collected in the boiler economizer, prior to
the introduction of lime slurry in the spray dryer, ranged from 0.96 to
4.39 percent.
Statistical analysis of the NITEP data from MID-CT using single
regression analysis indicated that the only individual operating variables
which showed a good correlation with mercury emissions removal were
decreasing operating temperatures of the air pollution control system, both
DS outlet (R = 0.550) and FF outlet (R = 0.525) and the fabric filter ash
rate (R - 0.687). The correlation between mercury removal and DS outlet
temperature is pictured in Figure 2.
Mercury removal did not correlate with percent carbon in flyash. As
previously mentioned, the data on percent carbon in flyash is suspect
because of potential interferences associated with the test method. It is
interesting to note that the calculated flue gas carbon content values
presented in Column 4 of Table IV a^e significantly higher than the rates
of carbon injection (6 to 105 mg/Nm ) used at mass burn MWC's in Europe to
demonstrate the effectiveness of carbon injection as a post-combustion
control technology. This fact may help explain the consistently high
mercury removal efficiencies achieved at MID-CT.
Multiple regression^analysis of the NITEP data from MID-CT showed a
very good correlation (R = 0.89) between mercury removal and decreasing
flue gas temperature (spray dryer outlet), increasing percent carbon in
flyash, and decreasing stoichiometric ratio. Stoichiometric ratio is
believed to affect mercury emissions in that calcium contained in excess
amounts of lime may react with mercuric chloride (HgC^) and form calcium
chloride (CaC^) , thus liberating mercury vapor. Although the results of
the multiple regression analysis tend to support the U. S. EPA's hypothesis
on the mechanism of mercury removal from the flue gas from MWC's, the
questionable validity of the percent carbon in flyash data does not permit
a firm conclusion to be reached. Additional research is clearly necessary
to gain a more thorough understanding of the operating variables which
affect the control of mercury at MWC's.
CONCLUSION
The control of mercury emissions from MWC's is an issue of
considerable debate. Although mercury emissions from MWC's represent a
small fraction of the global annual mercury emissions into the environment,
a great deal of focus is being placed on minimizing mercury emissions from
MWC's by reformulating consumer products which contain mercury and source
separation of household batteries. The U. S. EPA is evaluating mercury
emissions data and potential mercury emission control technologies in view
of a mandate to adopt mercury emission standards for MWC's by mid-November
1991.
268
-------
Data indicate that potential air emissions of mercury at RDF MWC's are
effectively minimized. Effective control of mercury is being accomplished
by a combination of processing MSW to remove mercury-bearing
noncombustibles and efficiently removing mercury from the MWC flue gas.
Although the precise mechanisms for mercury removal from MWC's is not
clear, it appears that good particulate removal, reduced control system
operating temperature, carbon level in flyash, and stoichiometric ratio of
scrubber reagent are factors which affect mercury removal. Additional
research is necessary to better understand the control of mercury emissions
from MWC's.
REFERENCES
1. Porcella, D., Electric Power Research Institute, "Mercury in the
Environment," EPRI Journal, April/May 1990.
2. New Jersey Chapter of Clean Water Action/Clean Water Fund Research and
Technical Center, "Mercury Rising: Government Ignores the Threat of
Mercury from Municipal Waste Incinerators," September 1990.
3. A. T. Kearney, Inc., and Franklin Associates, Inc., "Characterization
of Products Containing Mercury in Municipal Solid Waste in the United
States, 1970 to 2000," EPA Contract No. 60-W9-0040, January 1991.
4. Amos, C. K., "Getting Ready for the Mercury Challenge at Municipal
Waste Incinerators," Solid Waste and Power, April 1991.
5. Ham, R. K., and Hammer, V. A., University of Wisconsin-Madison,
"Elemental Composition and Partitioning of Metals at the Madison and
Hartford Resource Recovery Plants," July 1990.
6. White, D. M., and Nebel, K. L., Radian Corporation, "Summary of
Information Related to Mercury Emission Rates and Control Technologies
Applied to Municipal Waste Combustors," prepared for U. S. EPA,
September 1990.
269
-------
FIGURE 1
Municipal Solid Waste Processing - Single Line
INSPECTION/
REMOVAL
MSW
RECEIVING
0
jOX
FERROUS
METAL
MAGNETIC
SEPARATION
PRIMARY
TROMMEL
SECONDARY
SHREDDER~
rO
V
PRIMARY
SHREDDER
RESIDUE
J SECONDARY
TROMMEL
-------
FIGURE 2
Mercury Removal Efficiency
vs Temperature At SDA Outlet
>.
o
c
0)
o
e
lu
15
>
o
E
©
cc
Z3
o
0)
350
325
300
275
250
O)
a>
Q
-------
TABLE I MERCURY DISTRIBUTION AT MID-CT FACILITY
MSW 2.4 LB/1,000 TON MSW
FERROUS METAL 49%
PROCESS RESIDUE 4%
RDF 47%
TABLE II MERCURY EMISSIONS DATA FROM EPA/EC
TEST PROGRAM AT MID-CONNECTICUT FACILITY (2/89)
TEST RUN
INLET MERCyRY
(ug/dscm)
OUTLET MERCyRY
(ug/dscm)
REMOVAL
EFFICIENCY (%)
PT-02
PT-03
PT-04
PT-05
PT-06
PT-07
PT-08
PT-09
PT-10
PT-11
PT-12
PT-13
PT-14
740
579
636
661
597
597
659
667
745
683
498
557
945
6.9
22.7
14.7
7.1
12.2
7.7
4.5
15.1
8.8
18.9
3.5
11.7
14.2
99.1
96.1
97.7
98.9
98.0
98.7
99.3
97.7
98.8
97.2
99.3
97.9
98.5
AVERAGE
659
11.4
98.3
CORRECTED TO 7% 0„
272
-------
TABLE III MERCURY EMISSIONS DATA
FROM RDF MWC's WITH DS/ESP
NUMBER OUTLET MERCURY
FACILITY OF TESTS (ug/dscm)
HONOLULU, UNIT 13 5
HONOLULU, UNIT 2 3 7
SEMASS, UNIT 1 3 59
SEMASS, UNIT 2 3 105
WEST PALM BEACH, 3 56
UNIT 1
WEST PALM BEACH, 3 23
UNIT 2
CORRECTED TO 7% 02
273
-------
TABLE IV OPERATING DATA FROM EPA/EC
TEST PROGRAM AT MID-CONNECTICUT FACILITY
FLUE GAS
FABRIC FILTER CARBON
Hg REMOVAL OUTLET CONTENT STOICHIOMETRIC
TEST RUN (%) TEMPERATURE (°F) (mg/Nm ) RATIO
PT-02
PT-03
PT-04
PT-05
PT-06
PT-07
PT-08
PT-09
PT-10
PT-11
PT-12
PT-13
PT-14
99.1
96.1
97.7
98.9
98.0
98.7
99.3
97.7
98.8
97.2
99.3
97.9
98.5
225
282
287
220
253
223
245
284
223
285
247
233
247
NC
NC
735
194
810
378
143
803
321
640
148
NC
782
0.81
2.47
1.81
0.18
2.50
1.44
0.65
0.67
0.66
2.64
0.74
1.13
0.81
274
-------
The work described in this paper was not funded by the U.S. Environmental
Protection Agency. The contents do not necessarily reflect the views of the
Agency and no official endorsement should be inferred.
MERCURY EMISSION CONTROL
SODIUM SULPHIDE DOSING AT THE HOGDALEN PLANT IN STOCKHOLM
By: Christer Andersson, Hogdalen Plant
Bengt Weimer, Hogdalen Plant
ABSTRACT
Sodium sulphide dosing is employed at the Hogdalen Plant in Stockholm
with the aim of minimizing mercury emissions. Dosing has been in progress
since 1986, and has been employed continuously since the beginning of 1988.
The following typical problems were initially encountered
- Working environment
- Various handling problems
- Sludge precipitation
- Deposition at the dosing point
During 1988 - 1989, sodium sulphide was used for all three lines.
However, due to the low emission values on the two older lines (P1/P2,
VKW/Deutsche Babcock), even without sodium sulphide, dosing of sodium
sulphide is now employed only on line 3 (Martin/Werkle-Werk). The mercury
emission values are usually below 30 ^im/mm' of dry gas at 10% CO, and at a
collecting efficiency of approximately 80%. The annual cost of sodium
sulphide dosing is approximately SEK 4.60 per tonne of refuse burned
(including capital cost).
BACKGROUND
The first plant for domestic refuse incineration in Sweden and
Stockholm was built at the beginning of the century in Lovsta, north-west
of central Stockholm. The plant was a pure incineration plant. New furnaces
were successively built as the quantities of refuse increased. By the early
1960s, the quantities of refuse had increased so much that a new plant had
to be built to meet the need for incineration capacity. For transport
reasons, the new plant was to be sited south of central Stockholm.
275
-------
Construction of the Hogdalen Plant began during the latter part of the
1960s and was ready for commissioning in the summer of 1970. The plant
consisted of two boilers (VKW/Deutsche Babcock), each with a capacity of 11
tonnes of refuse per hour or 125 000 tonnes annually. The boilers generated
superheated steam (at 36 bar and 360°C) which was supplied to a 24 MW
turbo-generator. A two-stage electrostatic precipitator, with cyclone
batteries downstream of it, was installed for flue gas treatment. The
buildings and turbine were designed to enable two additional boilers to be
installed.
The plant was taken over by Stockholm Energi AB in the mid-1970s, and
was converted for simultaneous generation of electric power and heat for
district heating. Stockholm Energi AB is Sweden's third largest energy
utility, and owns hydro-electric power stations, fossil-fired power plants,
heat pump plants and shares in a nuclear power plant. The company is wholly
owned by Stockholm City and supplies Stockholm with electric power, heat
for district heating and town gas. The company has around 2500 employees
and the turnover in 1989 was SEK 4 billion. In 1985, a third refuse-fired
steam boiler (Martin/Werle-Werk) with a capacity of 15 tonnes of refuse per
hour and generating steam at 36 bar and 360°C was built in Hogdalen. As a
result, the capacity of the plant increased to 250 000 tonnes annually. A
system employing dry injection of sodium hydroxide and a fabric barrier
filter (Flakt Industri AB) downstream of it were installed for flue gas
treatment as shown in Fig. 1.
-------
MERCURY EMISSION CONTROL
In conjunction with the first emission measurements in the spring of
1986, it was found that the mercury emissions exceeded the guaranteed
values (50 pg/m1 of dry gas at s.t.p. at 10% CO,). The supplier, Flakt
Industri AB, carried out tests on employing almost 10 different additives.
Two of these - pulverized activated charcoal and sodium sulphide in a water
solution - were selected for further full-scale trials. Activated charcoal
was added to the flue gases upstream of the cooling tower, and sodium
sulphide solution was dosed into the cooling water in the cooling tower.
The results of the trials (Table 1) and commercial considerations led to
the installation of a system for the preparation and dosing and sodium
sulphide solution at the Hogdalen Plant.
Additives Average outlet concentration of total
Type Rate, mercury in mg/m' at s.t.p.
1/h t = 170 °C t = 150 °C
Na,S
solution 10 90 370
20 45 10
40 15 5
Pulverized
activated 5 120 55
charcoal
15 10* 5*
Table 1. Results of testing various additives.
The originally low or insignificant collection of mercury was
considered to be due to the low content of unburned carbon in the fly ash
from the boiler.
STAGE 1. SYSTEM DESCRIPTION:
Sodium sulphide is supplied to the plant in the form of flakes
(approx. 40% bound crystal water) in 25 kg plastic bags. The bags are cut
open with a knife and the sodium sulphide flakes are conveyed by a
pneumatic system (vacuum) to the storage silo.
277
-------
©
k\sJr™~
WATER
TO
SPRAYNOZZLES
©
©
©
1 SODIUM SULPHIDE BAG
2 STORAGE SILO
3 SCREW DOSING UNIT
4 PREPARATION TANK
5 MIXER
6 DOSING TANK
7 DOSING PUMP
Fig. 2. Flow diagram showing the Na,S plant
From the storage silo, the flakes are metered in batches by a screw
dosing unit to the mixing tank, in which the flakes are dissolved in water
with the assistance of a mixer to produce a solution with a concentration
of approx. 6%. After mixing to ensure that the flakes are dissolved, the
solution is discharged to a dosing tank. From the dosing tank, the solution
is delivered by a steel centrifugal pump to the injection nozzles in the
cooling tower. The pump capacity is controlled by a mechanical variator.
The injection nozzles are of two-media type, in which the solution is
atomized by means of compressed air. The nozzles are made of stainless
steel. The solution is sprayed into the upper part of the evaporative flue-
gas cooler. The temperature at the dosing point is 190 - 225°C, depending
on the load and the degree of boiler fouling. Dosing is normally carried
out at the rate of 20 litres of solution per hour.
DESCRIPTION OF PROBLEMS:
Hydrogen sulphide is emitted when the sodium sulphide bags are cut
open, which is perceived as a working environment problem. To eliminate
this, an agreement was concluded with the chemicals supplier that the
sodium sulphide would be delivered in 500 kg safe-bin containers. From the
container, the flakes are conveyed to the storage silo in an enclosed
pneumatic system. A vacuum pump (central vacuum-cleaning system) raises a
vacuum in the system and also in the storage silo, and the conveying air is
drawn in at the container. This system also gave rise to problems, since
278
-------
the pressure reduction caused the hygroscopic sulphide to absorb some water
from the conveying air, and the flakes were then "cemented together" into
large lumps. The screw dosing unit also gave rise to problems caused by
clogging by humid sulphide flakes. Due to the above problems, the operating
time of the system was very short. When a new flue gas treatment system was
purchased from Flakt Industri AB for the two older boilers (Fig. 3) of
basically the same design as the existing system, it was decided that an
entirely new sodium sulphide system which is common to all three lines
would be installed.
Fig. 3. Flow diagram for the flue-gas treatment system on P1/P2
STAGE 2. SYSTEM DESCRIPTION:
The preparation and dosing system is accommodated in a separate part
of the building, with its own comfort ventilation system (Fig. 4).
evaporative cooler
4. Water injection
5. Bag filter
6. Recycled dust silo
7. Dust silo
8. Dust humidifier
9. Lime silo
10. Economizer
11. Flue-gas fan
12. Silencer
13. Stack
1. Boiler
2. Economizer
3. Reactor with
279
-------
©
CD
Iffr^
j 00
WATER
TO SPRAWOZZLES
P1 P2 PJ
CO
©
©SODIUM SULPHIDE CONTAINER
©MIXING- AND STORAGETANK
©MIXER
©PROCESS VENTILATION
©CONNECTING PIPING
©INTERMEDIATE STORAGE TANK
©OOSINGPUMP (ONE PER LINE)
©DOSINGPUMP (RESERVE)
© FLOW METER
©
0 © EMBANKMENT
Fig. 4. Flow diagram for stage 2 of the Na,S plant.
The plant is controlled from a control desk located outside the
building, with a window provided for supervision. The full container is
placed on top of the mixing and storage tank by means of the overhead
travelling crane. The slide damper at the bottom of the container is opened
by means of a pneumatic actuator, and the container contents drop into the
tank. At the same time, a water valve is opened, and a valve is used to
separate the mixing tank from the intermediate storage tank, in order to
prevent undissolved sodium sulphide from being drawn into the pumps. The
mixing tank is equipped with a continuous mixer in the form of a propeller.
The mixing tank is provided with separate process ventilation in order to
eliminate the problem of smell. When the liquid level in the mixing tank is
normal, the water supply is shut off, and after 30 minutes of continued
mixing, the communicating line between the mixing and intermediate storage
tanks is opened. The pumps - one per boiler plus one on stand-by - are of
steel centrifugal type, with a mechanical variator for flow control. The
entire system, including tanks, pipes and valves, is made of stainless
steel. The sodium sulphide solution which is dosed to lines 1 and 2
(VKW/Deutsche Babcock, 1970) is mixed with additional water and is sprayed
into the combined conditioning towers/reactors by means of a nozzle lance
equipped with two spray nozzles. The inlet flue gas temperature is around
140°C, and when the water is sprayed in, the temperature drops to around
125°C. Since 1988, the evaporative cooler of line 3 (Martin/Werle-Werk,
1985) has been replaced by a heat exchanger (recovery boiler) to enable all
energy to be utilized and to allow the flue gases to be treated at as low a
temperature as possible. The sodium sulphide solution is sprayed in by two
nozzle lances equipped with two-media nozzles and located in a horizontal
280
-------
flue-gas duct between the heat exchanger and the reactors. The temperature
at this point is around 140°C. Compressed air is used for atomization. The
nozzles are made of Hastalloy C, and the nozzle lance is made of
conventional stainless steel.
DESCRIPTION OF PROBLEMS:
The first major problem that occurred on the new plant was that a
relatively large amount of sludge occurred when the sodium sulphide was
dissolved. The sludge originated from impurities in the sodium sulphide
flakes and also from metal ions that were precipitated out of the water.
This sludge clogged the pumps and nozzles. Although change-over to a better
quality of sodium sulphide reduced the problem, it did not solve it
entirely. To reduce further the risk of sludge formation, the water was
replaced by partially demineralized water. The flow rate from the dosing
pumps was difficult to adjust and was very sensitive to pressure changes.
To ensure more uniform dosing, the pumps were replaced by piston pumps. To
reduce the pressure variations from these pumps, the connections to the
pumps were changed to rubber hoses. In conjunction with the delivery tests
on the flue gas treatment system for lines 1 and 2 (VKW, Deutsche Babcock),
measurements were carried out with and without the addition of sodium
sulphide. In all cases, the mercury emissions proved to be very low (Table
2). For particulars of the measurement method, see Appendix 1.
Date Boiler Raw gas Clean gas
no:
Hg total,ug/m3 of
dry gas at s.t.p.
890420
2
200
1
890424
1
250
1
890511
2
190
2
890516
2
290
3
890519
2
180
2
Sulphide Filter Boiler
dosing temp. load
tonne of
1/h *C steam/h
20 125 31
20 125 31
20 142 31
0 142 30
0 125 31
Table 2. Mercury emission on P1/P2
In view of these results, and similar results from other plants in Sweden,
the dosing of sodium sulphide into these boilers was abandoned. The high
collection of mercury is due to the relatively high content of unburned
carbon in the flue gases. On boiler 3, a further problem of deposition on
nozzles and surrounding flue gas duct persisted. In order to solve these
problems, the nozzle lances were modified so that they were supplied with
preheated flushing air in order to prevent the condensation of flue gas on
the nozzle lances, with resulting dust deposition which, in turn, disturbed
the flow around the nozzle and led to further deposition in the flue gas
duct. In addition, the concentration of the sodium sulphide solution was
increased to around 10%. As a result, less water need be evaporated and, by
using a higher compressed air pressure, quick evaporation of the liquid
droplets is achieved.
281
-------
RESULTS
The data from all mercury measurements carried out on P3 is shown in
Table 3.
SUMMARY OF Hg MEASUREMENTS ON BOILER 3
AT THE HOGDALEN PLANT
DATE
RAW GAS CLEAN GAS
Hg total, ug/mJ of
dry gas at s.t.p.
SULPHIDE FILTER BOILER
DOSING TEMP. LOAD
Boiler load,
1/h °C tonne of steam/h
860211
-
86
-
173
52
860528
162
150
-
158
52
860825
-
58
20
151
54
860827
308
33
20
173
55
860829
414
51
20
174
54
860901
-
46
15
151
53
861216
-
25
20
168
55
870226
277
3
20
123
55
870309
111
94
-
168
55
870428
185
48
20
168
55
870915
327
27
20
169
55
871221
-
124
-
169
55
880412
238
122
-
169
55
880518
330
20
20
163
55
880824
108
16
20
155
55
880825
-
7
20
155
55
880825
-
20
-
155
55
881004
91
12
30
144
49
891018
141
127
20
132
39
891122
172
4
40
138
53
900206
-
8
20
138
55
900207
-
7
20
128
35
900208
-
5
40
127
35
900628
-
23
40
130
50
Table 3. Mercury emissions on P3.
High emission values were found in all cases of high flue-gas
temperature (above 155°C) and without sulphide addition. Reading 880825
without sulphide dosing was taken immediately after a measurement with
sulphide dosing, and the results are therefore uncertain. The results of
reading 891018 deviate substantially from other comparable readings. While
these readings were being taken, the boiler load was increased from minimum
load (approx. 30 t/h of steam) to full load (approx. 55 t/h of steam),
which resulted in a temperature increase. This may possibly be responsible
for the high emission value. For particulars of the measurement method, see
Appendix 1. In addition to the equipment at the Hogdalen Plant, other
equipment for the preparation and dosing of sodium sulphide solution has
282
-------
been installed In Sweden in Koping, Avesta and Umea. All plants are
equipped with dry flue-gas treatment systems (in Avesta, followed by a
condensation stage).
ECONOMICS
The cost of the existing plant in the Hogdalen Plant, including the
building, is estimated to be SEK 2 million. Apart from the problems
mentioned earlier, the maintenance requirements are very low. The time that
the operations personnel have to devote to the plant is confined to the
mixing of new solution and the cleaning of nozzles and nozzle lances twice
a week. The costs are summarized in Table 4. The total cost is SEK 4.60 per
tonne of fuel incinerated. This figure is based on the entire capital cost
being allocated to one line (P3). If sodium sulphide were dosed to the
other two lines, the cost would drop to SEK 3.20 per tonne of refuse
incinerated.
Capa- Mainten- Opera- Na,S Hater Total Gross refuse, Cost,
city ance tion tonnes SEK/tonne
300 50 50 150 2 552 120 000 4.60
Table 4. Summary of costs (SEK x 1000, unless otherwise specified).
283
-------
APPENDIX 1
2.5 Mercury
The mercury in the flue gases is mainly in gaseous form. During
sampling, part of the mercury is bound to the dust. The determination of
total mercury content involves adding the dust-bound mercury to the gaseous
mercury.
The gaseous mercury in the flue gases is determined in a sampled flow
from dust sampling. The gas sample is taken from the dust-free gas at the
outlet from the filter holder. The gas sample flow rate is 2 - 3 1/min. The
gas sample passes through three absorption flasks connected in series. The
first flask contains 10% by weight of sodium carbonate (soda) in distilled
water. The water-soluble mercury is absorbed here. The other two flasks
contain sulphuric-acid acidified sodium permanganate solution (6 g KMn04 in
10% by volume of HjS04) . Hater-insoluble mercury is absorbed in these two
flasks.
All test flasks are treated at around 450°C and the purity of the
absorption solutions is checked before they leave our analysis laboratory.
Tests on metallic mercury in the sampling equipment used have revealed
that Hg is not absorbed to any significant extent in the soda solution. On
the other hand, more than 98% of the Hg is normally absorbed in the first
permanganate flask. The second absorption flask serves only as a safety and
monitoring unit.
The absorption flasks are sealed off by means of ground glass stoppers
immediately after sampling, and are transported to the analysis laboratory.
Mercury bound to the dust is collected on a quartz-fibre flat filter.
Before sampling, the filters are treated at 300°C, conditioned in a
desiccator and weighed. After sampling, the filters are conditioned only in
the desiccator and are weighed, so that the mercury retained will not be
liberated.
The mercury bound during dust sampling is determined by the
composition of the sample of dust and filter material. The entire sample of
dust and filter material from the filter container is crushed to a
homogeneous powder sample before analysis. Carefully weighed analysis
samples are treated in a teflon bomb at 160°C for one hour with 50 ml of H20
and 5 ml of HNO,. The quantity of mercury in the solutions from the
decomposition of the solid materials and the absorption solutions from the
gaseous phase sampling are determined by flameless atom absorption. The
mercury is driven out of the solution with nitrogen after reduction with
boron hydride.
284
-------
The mercury in the flue gas is reported in three fractions:
Hgs is the mercury bound to the dust during sampling.
Hg is the water-soluble mercury which has been absorbed in the
eiQ
soda solution.
o
Hg is the mercury which has been absorbed in the permanganate
solution during sampling.
+ +
Hg is the calculated sum of H6 and Hg . Hg is a measure of the
reactive quantity of mercury in the' a?lue gas.
o +
Hgtot is the sum of Hg and Hg in the flue gas. This sum should
represent a measure of how much mercury leaves with the flue
gases from combustion, even though it is distributed
differently due to reactions in the plant and during sampling.
285
-------
Intentionally Blank Page
286
-------
The work described in this paper was not funded by the U.S. Environmental
Protection Agency. The contents do not necessarily reiiect the views o 1 the
Agency and no official endorsement should be inferred.
CONTROL OF MERCURY AND DIOXIN EMISSIONS
FROM UNITED STATES AND EUROPEAN
MUNICIPAL SOLID WASTE INCINERATORS
BY SPRAY DRYER ABSORPTION SYSTEMS
By
B. Brown
JOY TECHNOLOGIES INC.
Joy Environmental Equipment Company
Monrovia, California
and
K.S. Felsvang
A/S Niro Atomizer
Columbia, Maryland
ABSTRACT
Incineration of Municipal Solids Waste (MSW) as a method of
reducing disposal volume requirements and recovery of energy has
been practiced extensively for many years in Europe and more
recently in North America. Concerns of potential air pollution
from this incineration have resulted in the promulgation of
emission standards for a wide range of pollutants and the
subsequent application of Spray Dryer Absorption (SDA) flue gas
cleaning systems to control incinerator emissions. In Europe,
where emission standards were adapted several years ago, SDA
systems have been installed using both fabric filters or existing
electrostatic precipitators as dust collectors. Emission standards
have required stringent control of acid gases (HC1, HF, S02),
particulate matter, trace metals (in particular mercury) and
dioxins. SDA systems in operation have demonstrated the ability to
achieve the required levels of control when either an electrostatic
precipitator or fabric filter has been used as dust collector.
The paper will focus on the mercury and dioxin emissions, which
have been achieved at three European and two United States SDA
systems. The operating experience has shown that the SDA systems
equipped with a single rotary atomizer per absorber can achieve
high removal efficiencies of mercury and dioxins by proper control
of spray dryer outlet temperature and by using a patented dry
additive injection system.
287
-------
ABSTRACT
Incineration of Municipal Solids Waste (MSW) as a method of
reducing disposal volume requirements and recovery of energy has
been practiced extensively for many years in Europe and more
recently in North America. Concerns of potential air pollution
from this incineration have resulted in the promulgation of
emission standards for a wide range of pollutants and the
subsequent application of Spray Dryer Absorption (SDA) flue gas
cleaning systems to control incinerator emissions. In Europe,
where emission standards were adapted several years ago, SDA
systems have been installed using both fabric filters or existing
electrostatic precipitators as dust collectors. Emission standards
have required stringent control of acid gases (HCl, HF, S02) ,
particulate matter, trace metals (in particular mercury) and
dioxins. SDA systems in operation have demonstrated the ability to
achieve the required levels of control when either an electrostatic
precipitator or fabric filter has been used as dust collector.
The paper will focus on the mercury and dioxin emissions, which
have been achieved at three European and two United States SDA
systems. The operating experience has shown that the SDA systems
equipped with a single rotary atomizer per absorber can achieve
high removal efficiencies of mercury and dioxins by proper control
of spray dryer outlet temperature and by using a patented dry
additive injection system.
288
-------
INTRODUCTION
Incineration of municipal solid waste (MSW) has been utilized for
many years in Europe and more recently in North America to
substantially reduce the volume of waste to be disposed of while at
the same time producing energy in the form of steam or electricity.
As the application of incinerators has increased, concerns have
been raised over potential air pollution impacts associated from
their use. This has led to the promulgation of emissions limits
and control requirements for a wide range of pollutants, first in
the European countries, and more recently in the states and
provinces in North America. Spray Dryer Absorption (SDA) flue gas
cleaning systems have been successfully applied to control
incinerator air pollutants.
Joy Environmental Equipment Company and A/S Niro Atomizer have been
jointly developing SDA gas cleaning systems since 1977 and have
successfully applied this technology for control of acid gases,
particulates and trace element emissions from coal-fired boilers,
mass-burn and RDF-fired MSW incinerators and hazardous waste
incinerators. Today more than 24 MSW incinerator trains are in
operation employing this technology with an additional 22 trains
under construction. The majority of the operating units have
successfully completed performance tests and operate reliably while
maintaining the incinerators in compliance with local air pollution
regulations. This success can be attributed to Niro's vast
knowledge of spray drying and Joy's extensive experience in
application of particulate control technologies to a wide range of
processes.
The systems applied in Europe and North America, while utilizing
the same basic technology, incorporate many different design
features for each particular application. The majority of the
systems installed in Europe were retrofitted to existing
incinerators where increasingly more stringent air pollution
regulations require acid gas and trace metal emission controls on
both new and existing plants. These systems (where possible)
utilize existing electrostatic precipitators either as the main
dust collector downstream of the spray dryer or as a precollector
upstream of the spray dryer. A few of the newer installations
employ a fabric filter as the dust collector.
In North America the opposite is true. The majority of SDA
applications are on new incinerators. Whereas the first four
Joy/Niro systems to start-up in the United States have
electrostatic precipitators as dust collectors, the remaining
trains under construction are equipped with fabric filters.
This paper presents a description of the SDA process and addresses
key design parameters for successful applications. Design
parameters and emission requirements from three European and two
North American installations are presented. Results of mercury and
dioxin testing on the European and North American MSW incinerators
289
-------
are being reported. Effects on the removal efficiency of mercury
and dioxins of a dry additive injection system are documented.
SYSTEM DESCRIPTION
The Zurich, Amager and Kassel SDA systems are all retrofit
applications while the Semass and Palm Beach are complete new
incinerator installations. All plants incorporate design features
common to Joy/Niro acid gas cleaning systems (1,2,3). These
features include:
• Use of a single spray dryer per incinerator train.
Use of a single rotary atomizer per spray dryer to produce a
cloud of fine slurry droplets.
• Use of a single gas disperser to control the shape of the
droplet cloud and achieve mixing between the gas and the
slurry.
• Inclusion of a two-point product discharge design to ensure an
open gas passage.
• Sufficient spray dryer gas residence time to ensure adequate
product drying.
• Use of lime reagent to achieve acid gas removal.
The Zurich, Amager, Palm Beach and Semass plants are designed as
single-pass systems, whereas the Kassel plant has the capability to
operate in either the single-pass or partial-product recycle mode.
The Zurich and Amager systems have retrofitted an additive system
for mercury and dioxin control. Figure 1 shows a simplified flow
sheet of the basic single-pass system.
The SDA flue gas cleaning system consists of a reagent preparation
system, a spray dryer absorber, a dust collector and an ash
transport system. Typically there is one spray dryer and dust
collector per incinerator train and common reagent preparation and
ash transport systems per multiple trains.
Quick lime is delivered by truck and conveyed pneumatically to the
storage silo. From the storage silo the lime is sent to a paste or
detention slaker, where a 20-30% lime slurry is prepared. The lime
slurry is either stored in a lime milk tank or transferred directly
to the feed tank. The final dilution of the lime slurry is
accomplished either in a small dilution head tank located above the
spray dryer absorber or in the rotary atomizer utilizing a unique
dual-liquid distributor. By using the latter method, lime slurry
and water are mixed directly in the atomizer wheel. The dilution
in the European plants is controlled by a signal from the HCl-
analyzer located in the stack, while the American plants use an S02
stack monitor for control.
290
-------
•MAT
OOTID
.TAXI
rcu
Figure 1
Simplified Process Flow Diagram
From the head tank the lime slurry flows by gravity to the
atomizer. Control of the total liquid flow to the atomizer is
based on a spray dryer outlet temperature control. The atomized
lime slurry enters the spray dryer where it mixes with the hot
incoming gas, simultaneously reacts with acid gases present and
prior to exiting the spray dryer absorber. A single rotary
atomizer per absorber module makes the control of the process and
the uniform mixing of the slurry droplets and flue gas extremely
efficient. Operating experience has shown that SDA systems with a
single rotary atomizer per module are able to control spray dryer
absorber outlet temperatures to lower levels than systems with
multiple atomizing devices, such as dual-fluid nozzles. The
extremely good contact in the spray dryer absorber utilizing a
rotary atomizer is of utmost importance for achieving high mercury
and dioxin removal efficiencies.
A portion of the dry product is removed from the bottom of the
spray dryer while the majority of the product is carried over to
the dust collector. Additional acid gas absorption takes place in
the dust collector and dust is removed from the flue gas. The dry
product from the spray dryer absorber and the dust collector is
conveyed either mechanically or pneumatically to a waste disposal
silo.
For additional mercury and dioxin control, the SDA system can be
retrofitted with an add-on control system. This consists of a
patented, dry additive system (4) for injection of the additive
upstream of the spray dryer absorber. The previously mentioned
extremely efficient mixing of flue gas, additive and slurry
droplets inside the spray dryer absorber provides optimum
conditions for contact between the additive and the dioxins and
mercury present in the flue gas.
291
-------
As mentioned earlier in the system description, the unit at Kassel
recycles a portion of the dry product back to the feed system to
improve reagent consumption. The Amager system utilizes hydrated
lime as reagent; however, they are presently retrofitting a slaking
system utilizing powdered quicklime. The Kassel SDA is retrofitted
downstream of existing electrostatic precipitators, the Zurich
plant is utilizing the existing electrostatic precipitator (EP) for
dust control, whereas the electrostatic precipitator at the Amager
system was scrapped and replaced by a SDA and pulse-jet fabric
filter (PJF). The Semass and Palm Beach systems utilize Joy/BSH
rigid-frame precipitators downstream of the spray dryers.
Figure 2 presents design conditions for the various acid gas
pollutants and the required removal efficiencies. The European
plants are designed to achieve a set emission rate on a one-half
hour averaging period for each pollutant, while the Semass plant is
based on a three-hour averaging period. The Kassel SDA system has
precollector and is utilizing dry product recycle which increases
the chloride concentration in the feed slurry. Therefore, the
spray dryer outlet temperature is controlled to a minimum of
275° F to ensure proper drying.
In addition to the acid gas and dust removal requirements, some
regulatory agencies have placed limits on trace metal emissions
from MSW incinerators. The approach in Europe has been to separate
heavy metals in three major classes and set emission levels for
each of these classes. In some cases, specific metal emission
levels have also been set. The emission limits include both dust-
bound and vapor-phase emissions. Furthermore, European regulatory
agencies limit the emissions of dioxins from MSW incineration (6).
Figure 3 presents the trace metal emission requirements of the
European installations.
The permits for Semass and Palm Beach did not have specific stack
emission requirements for metals and dioxins, but that does not
mean these issues were not addressed. Semass has a 608 ng/Ntr
stack emission limitation as well as an ambient air quality
standard listed in its permit.
TEST RESULTS
Mercury
Zurich
After start-up, the SDA system in Zurich, Switzerland, was
operated at various outlet temperatures to verify the impact of
outlet temperature on efficiency of acid gas and mercury removal.
These tests confirmed earlier work carried out at the pilot plant
level; i.e., the lower the SDA outlet temperature, the higher the
removal efficiency for acid gases and mercury. The system was
successfully operated at SDA outlet temperatures as low as
230° F.
292
-------
Semass W. Palm Zurich Kassel Amager
Rochester, Beach, Josephstr. City of City of
Project Mass. Florida Switzerland Germany Denmark
Type Waste
Size, TPD
RDF
2x900
RDF
2x1000
MSW
400
MSW
2x300
MSW
300
Inlet Flue Gas
Conditions
Flue Gas Flow
(Nominal), acfm
Temperature, °F
Fly Ash, gr/scf
HCI
Average,
Maximum,
S02
Average,
Maximum,
HF
Maximum,
ppm
ppm
ppm
ppm
ppm
212,000 197,000 103,000
350
10.0
400
1,000
110
330
350 430-570
7.2
710
1,290
240
430
40
1.0
500
700
200
15
2x67,000 1x82,000
460 356-446
0.02
500
1150
200
500
10
3.5
750
1150
140
Figure 2
Design Conditions
293
-------
Project
Semass
Rochester,
Mass.
W. Palm
Beach,
Florida
Zurich
Josephstr.
Switzerland
Kassel
City of
Germany
Amager
City of
Denmark
Type Waste
Size, TPD
RDF
2x900
RDF
2x1000
MSW
400
MSW
2x300
MSW
300
Outlet Flue Gas
Conditions
Reference
Conditions
12% C02
dry
12% C02
dry
11% o2
dry
11% o2
dry
11% o2
dry
Temperature, °F
265
255
285
275
284
HCi, ppm
% removal
50
90
90
20
87-96
12
97-99
60
92-95
SO p ppm
% removal
100
65
70
75
50-75
30
79-94
105
25
HF, ppm
% removal
—
93
5
5
—
Particulate, gr/scf
0.03
.015
0.02
0.013
0.013
Dust Collector
EP
EP
EP
Two-
Field
Pulsejet
Baghouse
Pulsejet
Baghouse
Process Type
Single
Pass
Single
Pass
Single
Pass
Recycle
Single
Pass
Figure 2
Design Conditions
(Continued)
294
-------
Zurich
Amager
Kassel
mg/Nm3, 11% 02 dry
Class I <0.1 <0.1 <0.1
Class II <0.1
Class III <5 - <5
Class I Mercury, Cadmium, Thallium
Class II Arsenic, Cobalt, Nickel
Selenium, Tellurium
Class III Antimony, Lead, Chromium
Copper, Manganese, Vanadium
Cyanides and Fluorides
Figure 3
Trace Metal Emission Requirements
295
-------
Although a significant increase in mercury removal was achieved
by lowering the outlet temperature of the SDA, this was no^
sufficient to meet the stringent requirements of 0.1 mg/Nm
mercury outlet emission. In May 1986, a dry additive system
developed and patented by Niro Atomizer (4) was retrofitted
upstream the spray dryer absorber. The preferred additive is
active carbon.
Figure 4 shows detailed results of the tests that were done with
and without activated carbon at various SDA outlet temperatures.
All mercury measurements were made simultaneously at the SDA
inlet and the EP outlet. Both multi-point isokinetic traverses
and single-point isokinetic samples were taken simultaneously
over a four-hour period to determine if it was necessary to test
with full duct traverses «3 The three single-point3samples showed
an average of 144 ^.g/Nm compared to 145 sfg/Nm for the full
traverse. Therefore, further tests were run on the carbon
additive system using only single-point isokinetic sampling.
Figure 5 shows an averaged summary of the detailed results in
Figure 4. The table shows the effect of lowering the SDA outlet
temperature and also the pronounced effect of carbon additive.
Note that even with fluctuating inlet mercury concentrations, the
lower outlet temperature results in a mild improvement in
efficiency, the use of carbon additive, roughly doubles the
efficiency from the mid-40% to the high 80% range. To see the
effect in more detail, the tests in Figure 4 at 239° F and
230° F SDA outlet temperatures show wide variations in inlet
concentration. Without carbon addition, t^he outlet
concentrations vapr widely between 117 and 670 jrg/Nm , while th^
outlet concentrations fall in a narrow range between 29-68^g/Nm
when additive is injected.
Figure 6 shows tests in which the influence of additive injection
rate on mercury removal was investigated. It is confirmed that
low outlet mercury emissions are achieved regardless of inlet
mercury level. Based on these tests, it appears the additive
injection rate does not have a major impact on the outlet
emission concentration or removal efficiency. We feel this is
due in part to the fact the downstream particulate collector is
a precipitator. Therefore it is concluded that the low and
relatively uniform outlet Hg emission is due to the efficient
control of the mixing of flue gas, additive and liquid in the SDA
due to the single rotary atomizer concept. The absorber module
can be compared with a fully back-mixed reactor with almost ideal
mixing.
Amaaer
The Amager MSW incinerator is situated in the middle of
Copenhagen. The retrofit SDA/BH system was brought into
operation mid 1988. In August 1989, a test program was carried
out to characterize the mercury emissions. Figure 7 shows the
detailed mercury test results with the SDA operating at 284° and
296
-------
Hg concentration fjg/Nm3, dry
SDA
Outlet
Temp. Additive EP
°F mg/Nm3 SDA Outlet % Removal
537
390
27
343
237
31
680
417
39
284
0
558
414
26
406
335
17
1072
769
28
539
39
93
248
30
589
31
95
495
232
53
643
207
68
234
117
64
0
949
670
29
736
476
36
401
250
38
239
352
44
88
30
353
44
88
281
29
90
0
249
124
50
224
132
41
346
212
39
230
30
486
68
86
650
45
93
131
44
66
417
40
90
269
51
81
Figure 4
Zurich - Detailed Mercury Test Results
With Variable SDA Outlet Temperature
297
-------
Hg concentration (jg/Nm3, dry
SDA
Outlet
Temp. Additive SDA EP
°F mg/Nm3 Inlet Outlet % Removal
284 0 600 427 29
239 0 576 325 44
239 30 329 39 89
230 0 273 156 43
230 30 390 49 87
Figure 5
Zurich - Summary of Hg Test Results
Averages of Detailed Results (Figure 4)
298
-------
SDA
Outlet
Temp.
Additive
mg/Nm3
Hg concentration fig/Nm , dry
SDA
Inlet
EP
Outlet
% Removal
501
439
93
47
81
82
248
Average
470
70
85
15
229
176
239
33
11
22
86
94
SI
Average
215
22
90
30
539
589
154
m
39
31
20
23
93
94
87
SQ
Average
349
28
92
Figure 6
Zurich - Hg Results
Influence of Additive Injection Rate
299
-------
Hg concentration iig/Nm3, 10% 02
SDA
Outlet
Temp. Additive SDA PJF
°F mg/Nm3 Inlet Outlet % Removal
284
171
193
184
170
139
130
164
72
62
56
15
29
61
64
59
318
191
49
34
85
82
17
180
209
283
26
29
30
86
86
89
58
1276
268
65
19
95
93
260
19
70
354
165
137
159
169
167
185
27
40
25
45
20
5
6
92
76
82
72
88
97
97
Figure 7
Amager - Detailed Mercury Test Results
3n0
-------
260* F outlet temperature with a variation in the quantity of
additive injected. In Figure 8 these results are averaged and
summarized for clarity. Again with no additive injection, it is
clear that reducing the SDA outlet temperature is effective in
increasing mercury removal. With additive injection, mercury
emissions can be controlled to values much lower than the
required 100/(g/Nm .
Kassel
The SDA system at Kassel in West Germany differs from Zurich and
Amager, because this system has incorporated a partial dry solids
recycle system. Furthermore, the incinerator fly ash is
precollected upstream of the SDA system. These factors will
increase the chloride concentration in the feed slurry of the SDA
system. Potentially a recycle system has a negative effect on
mercury removal, as previously captured mercury will evaporate
from the recycle solids after re-injection into the SDA system.
This phenomenon has been confirmed by test work at the Leverkusen
incinerator plant, where only 5-10% mercury removal was achieved.
The Leverkusen system consists of a SDA and a downstream EP and
is designed with recycle.
In Kassel the required outlet emission of 100^(g/Nm3 of mercury
can be achieved with dry additive injection upstream of the spray
dryer absorber. Figure 9 shows x^ssults from recent measurements
at the plant. By adding 20 mg/Nm additive or more, the emission
can be controlled to a safe value under 100 ^g/Nm . Note that
due to spray drying considerations with high chloride
concentrations in the feed slurry, the SDA outlet temperature has
to be kept at 279° F.
Semass
The Semass unit has been described in detail previously (12) and
underwent emissions testing in April 1989 and September 1990.
This plant differs slightly from the European installations
because it used a prepared fuel as opposed to the mass burn
principal. Unfortunately, SDA inlet concentrations were not
measured so the efficiency of the SDA/EP system could not be
calculated. The unit operates without carbon additive and
achieves emission levels consistent with the European
installations just described. The mercury emissions vary between
48-140^g/Nm and are detailed in Figure 10.
Palm Beach
The RDF plant at Palm Beach was started in late 1988 and is very
similar to the configuration at Semass. Here again, SDA inlet
mercury concentrations were not measured so the removal
efficiency is not known. However, mercury emissions data taken
at the stack are quite good. Unit 1 has an average emission of
50,#g/Nm and Unit 2 was tested at 20x^g/Nm . Detailed results
are available in Figure 11.
301
-------
SDA
Outlet
Hg concentration pg/Nm , 10% 0,
Temp.
Additive
SDA
PJF
°F
mg/Nm3
Inlet
Outlet
% Removal
0
171
97
43
6
255
42
84
284
17
224
28
87
58
633
37
94
260
0
19
70
154
169
176
37
20
5
76
88
97
Figure 8
Amager - Summary of Hg Test Results
302
-------
SDA
Outlet
Temp.
Additive
mg/Nm3
Hg concentration fjg/Nm , dry
SDA
Inlet
PJF
Outlet
% Removal
279
0
9
20
47
64
898
336
324
179
297
582
175
57
19
52
35
48
82
89
82
Figure 9
Kassel - Summary of Hg Test Results
303
-------
SDA
Outlet
Temp.
Additive
mg/Nm3
Hg concentration
fjg/Nm3, dry, 12% CO3
April
1989
EP
Outlet
Sept
1990
EP
Outlet
Unit 1
270
57
270
42
270
Average
45
48
Unit 2
270
96
162
270
68
119
270
Average
54
73
140
140
Figure 10
Semass - Hg Emission Test
304
-------
SDA
Outlet
Temp.
Additive
mg/Nm3
Hg concentration yg/Nm , dry
SDA
Inlet
EP
Outlet
Unit 1
261
261
261
Not
Measured
Not
Measured
Not
Measured
Average
29
52
&
50
Unit 2
261
261
261
Not
Measured
Not
Measured
Not
Measured
Average
23
15
22
20
Figure 11
Palm Beach - Hg Emission
305
-------
The mercury emissions data from the five plants are summarized in
Figure 12. Shown on the figure are SDA inlet concentrations,
when measured, and the effect that treatment in the SDA system
provides. The X axis shows the effect of carbon additive. Note
that in all cases save one, the mercury emission level can be
kept below 100>^g/Nm . The one exception is the unit at Kassel,
which has a pulsejet baghouse recycle system.
Dioxins and Furans
There has been worldwide concern over dioxin and furan emissions
from municipal solid-waste incinerators. For the last 20 years
studies have been conducted to define the toxic effects of these
compounds. Therefore, we have attempted to summarize the emissions
levels and the efficiency of spray dryer absorption systems of
these waste incinerators. The measurements presented include three
European installations, as well as the Semass unit. Data was not
available for the West Palm Beach installation. The emissions
levels are expressed as totals of tetra through octa chlorinated
dibenzo-p-dioxins and chlorinated dibenzofurans, as well as toxic
equivalents appropriate to the country and time of test. EPA
currently requires a limit of 30 ng/Nm dry at 7% 02 to meet new
source standards, while other authorities are interested in toxic
equivalents. Although not in the scope of this paper, please refer
to Figure 13 for a comparison of various toxicity equivalencies.
This information plus a thorough description of the various toxic
equivalent methods is contained in Reference (5).
Zurich
Dioxin and furan measurements at Zurich were performed by dk-
Teknik of Denmark. Before performing tests at Zurich, they had
been involved in test programs on Danish incinerators on behalf
of the Danish Environmental Protection Agency (6, 7, and 8).
The tests were performed in accordance with procedures outlined
by the Nordic Method. Simultaneous samples were collected
isokinetically at the inlet to the spray dryer absorber and the
outlet of the precipitator. The filter temperature was kept at
a temperature of 248° F. The dioxin and furan content reported
include that found in the condensate, the XAD2 column, the quartz
wool filter and the methylenechloride rinsing liquid.
Sample preparation and isomer specific analyses were made by the
Department of Organic Chemistry, Umeaa University, Sweden. The
samples were analyzed for the 12 most toxic dioxins commonly
known as the "dirty dozen". From this analysis an equivalent
amount of 2,3,7,8 TCDD was computed according to the Eadon
Method.
The plant was operated steady-state for an extended time for each
test condition before measurements were carried out. The testing
took place over four days. During sampling the incinerator was
operated at steady load and with a low CO content of 20-30 ppm
306
-------
Hg Concentration
(mG/Nui')
600
LEGEND
500
Plant
Tene.°F
Filter
A
A
Zurich
230-248
EP
•
Amager
284
RJF
X
Amager
260
RJF
400
0
Kassei
279
RJF Recycle
D
Semass
270
EP
0
A
Palm Beach
261
EP
300
200
100
SDA
INLET
0 10 20
' • Additive (mg/Nm3)
STACK
30
40
50
60
70
80
Figure 12
Control of Mercury Emissions by Additive Injection
307
-------
11
EADON CANADA CALIFORNIA EPA-TEQs/87 l-TEOs/89
Data (or Above Figure (Concentrations in ng/dscm3@ 7% 02)
TEF SCHEME
SOURCE
SPECIES
DATA
EADON
CANADA
CALIFORNIA
EPA-TEF/87
l-TEF/89
2378-TCDD
0.30
0.30
0.30
0.30
0.30
0.30
TCDDf- (OTHER)
2.7
0
0.27
0
0.027
0
12378-P«CDD
0.79
0.79
0.39
0.79
0.39
0.39
PeCDO# (OTHER)
2.2
0
0.01 1
0
0.01 1
0
123478-HiCDD
0.16
0.0047
0.016
0.0047
0.0063
0.016
123678-HxCDD
0.39
0.01 2
0.039
0.01 2
0.015
0.039
123789-HxCDD
0.059
0.0018
0.0059
0.001 8
0.0023
0.0059
HxCDD# (OTHER)
3.6
0
0.0036
0
0.0014
0
1234678-HpCDD
0
0
0
0
0
0
HpCDD* (OTHER)
5.9
0
0.00059
0
0.000059
0
OCDO
9.5
0
0.00095
0
0
0.00095
TOTAL CDD«
1.1
1.0
1 . t
0.8
0.8
2378-TCDF
2.3
0.76
0.23
2.3
0.23
0.23
TCDF« (OTHER)
30
0
0.03
0
0.030
0
12378-PtCDF
4.2
1.4
0.42
4.2
0.42
0.2 1
23478-PtCDF
2.5
0.82
0.25
2.5
0.25
1 .2
PeCDF# (OTHER)
1 2
0
0.012
0
0.012
0
123478-HxCDF
1.6
0.016
0.079
0.047
0.016
0.16
123678-HxCDF
0
0
0
0
0
0
234678-HxCDF
0.46
0.0046
0.023
0.014
0.0046
0.046
123789-HxCDF
0.0095
0.000095
0.00048
0.00029
0.00009S
0.00095
HxCDF* (OTHER)
1 7
0
0.0086
0
0.001 7
0
1234678-HpCDF
0
0
0
0
0
0
1234789-HpCDF
0
0
0
0
0
0
HpCDF* (OTHER)
1 1
0
0.01 1
0
0.01 1
0
OCDF
0.41
0
0.000041
0
0
0.00041
TOTAL CDF*
3.0
1 . 1
9.1
1.0
1 .8
TOTAL TEQs
4
2
1 0
EPA-TEQs/87s 2
l-TEQs/89r 3
15%
1 0%
Contrlbujed by
Contributed by
2,3,7,8 • TCDD
2,3,7,8 - TCDD
Reference: Adapted from NATO/CCMS, 1988a.
Figure 13
Toxicity Equivalents in Emissions from a
Municipal Waste Incinerator
308
-------
(dry), which indicated that the combustion conditions were good
and stable.
The results of the dioxin testing are shown in Figure 14 and are
reported as total PCDD and PCDF, as well as toxic equivalents
according to the Eadon Method. The spray dryer was operated at
inlet temperatures of 392-428* F and outlet temperatures of
248-284" F. The results show a significant reduction of
approximately 75% when operating without carbon additive. This
reduction is effected by the intimate contact of the dry solids
and the treated gas in the spray dryer. We suspect the operation
of the precipitator does not contribute much to the adsorption of
dioxins. It should be noted that an additive injection rate of
18 mg/Nm increases the removal efficiency to 90% at 284° F and
to 98.5% at 248* F with 59 mg/Nm additive injection rate.
Amacrer
The Amager system is similar to Zurich except the dust collector
is a pulsejet fabric filter instead of a precipitator.
The sampling and measurement technique was the same as used for
the Zurich testing. Results of the test work are shown in
Figure 15. The test results are expressed as total PCDD and PCDF
and as Nordic toxic equivalents, which in most cases are very
similar to Eadon equivalents. The results confirm that a spray
dryer absorber with fabric filter operating at low spray dryer
absorber outlet temperature is highly efficient and could be
considered best available control technology for dioxins.
Comparing the outlet emissions without additive results in a
reduction of dioxins by approximately 50% when lowering the
outlet temperature from 284 to 260° F. Injection of additive at
260° F further reduces the dioxin emissions corresponding to
almost complete removal on a Nordic toxic equivalent basis.
It is interesting to note that one test was done under simulated
start-up conditions, where dioxin emissions from the incinerator
were much higher than normal due to non-ideal burning conditions.
It is significant in all cases, including the simulated start-up,
very high removal efficiencies and Nordic toxic equivalents of
less than 0.1 ng/Nm were achieved with carbon additive (10).
Kassel
Recent proposed German regulation requires a set limit of dioxin
emission of less than 0.1 ng/Nm toxic equivalents according to
the TE/NATO CCMS methods of August 1988 (5) .
Figure 16 shows results of dioxin removal tests at the Kassel
incinerator. With no additive injection, a dioxin removal
efficiency of only 60-65% is achieved. The relatively low
removal efficiency can be attributed to the absence of fly ash
due to the fly ash precollector and to the recycling of the dry
reaction product through the feed system. Earlier pilot plant
309
-------
Total PCDD + PCDF Toxic Equivalents, Eadon
ng/Nm3, dry ng/Nm , dry
SDA
Outlet
Temp. Additive SDA EP % SDA EP %
°F mg/Nm3 Inlet Outlet Removal Inlet Outlet Removal
0 306 77 74.8 7.7 1.9 75.3
284
18 223 33 85.2 7.5 0.79 89.5
248
0
59
277
455
69
5.0
75.1
98.9
6.9
6.0
1.8 73.9
0.09 98.5
Figure 14
Zurich - Dioxin Removal Test Results
310
-------
Total PCDD + PCDF
ng/Nm3, dry
Toxic Equivalents, Nordic
ng/Nm , dry
SDA
Outlet
Temp.
°F
Additive
mg/Nm3
SDA
Inlet
PJF
Outlet
%
Removal
SDA
Inlet
PJF
Outlet
%
Removal
0
132
2.1
98.4
2.8
0.076
99.7
6
283
1.2
99.6
4.8
0.0075
99.8
284
17
276
2.4
99.2
8.3
0.045
99.5
58
201
1.1
99.5
4.0
0.035
99.1
58
2170*
3.2
99.9
50.0
0.050
99.9
0
254
1.3
99.5
7.7
0.0047
99.9
261
19
154
0.4
99.8
5.0
N.D.
100
70
154
0.7
99.6
4.5
0.002
100
* Start-up Condition, Simulated
N.D. = Non Detectable
Figure 15
Amager - Dioxin Removal Test Results
311
-------
SDA
Outlet
Temp.
°F
275
Total PCDD + PCDF
ng/Nm3, dry
Toxic Equivalents
TE/NATO/CCMS
ng/Nm3, dry
Additive
mg/Nm3
SDA
Inlet
PJF
Outlet
%
Removal
SDA
Inlet
PJF
Outlet
%
Removal
0
380
151
60
9.58
3.46
64
19
134
12
91
3.21
0.19
94
19
238
8
97
5.11
0.15
97
47
298
9
97
5.53
0.13
98
105
359
7
98
5.94
0.07
99
Figure 16
Kassel - Dioxin Removal Test Results
312
-------
results at an incinerator in Denmark (8) showed that by operating
a spray absorption system with bag filter in recirculation mode
without precollection of fly ash, one could achieve approximately
90% dioxin removal at the same temperature level as in Kassel.
It is therefore believed that the presence of fly ash plays a
major role in dioxin removal in the SDA system. Figure 16 shows
that injection of additive can compensate for the lack of fly ash
and secure high removal efficiencies of dioxins and furans. The
results also show the 0.1 ng/Nm toxic equivalent limit can be
achieved.
Semass
The Semass plant has been tested three times in the last two
years for a variety of pollutants, including dioxin and furan.
The results of stack testing are shown on Figures 17 and 18.
Note that results are given as a summation of total dioxin and
furan or as toxic equivalents (EPA-TEF/87) . As in the case of
the mercury tests, the SDA inlet was not measured, so there is no
indication of SDA/EP dioxin/furan removal efficiency. However,
when compared to the Zurich installation (SDA/EP) with no
additive, the Semass emission levels are quite similar with the
exception of one test. The reason for the unusually high
emission level of this one test is unknown at this time. Based
on the test results at Zurich, it is safe to assume the use of
carbon additive could improve the emission levels at Semass.
Comparing the dioxin and furan removal test results from Zurich,
Amager, Kassel and Semass, it can be concluded that injection of
additive can secure a very low dioxin emission regardless of
operating mode (single pass vs. recirculation), type of filter
(fabric filter vs. electrostatic precipitator), spray dryer
outlet temperature and upset conditions.
CONCLUSIONS
The paper has described the mercury and dioxin removal efficiencies
achieved at five full-scale operating SDA systems in Europe and the
United States. The test results have clearly demonstrated the
ability of the SDA system with a single rotary atomizer per
absorber ^odule to achieve outlet emission levels of mercury below
0.1 mg/Nn^ and outlet dioxin and furan emissions of less than
0.1 ng/Nm toxic equivalents when using a dry additive injection
system.
313
-------
Total PCDD + PCDF
ng/Nm3, dry
SDA
Outlet EP EP EP
Temp. Additive Outlet Outlet Outlet
Oi
F mg/Nm3 April '89 Jan. '90 Sept. '90
Unit 1
Average 7.6 17.3
Unit 2
270 0 12.7 12.2
270 0 5.2 ~ 12.4
270 0 698.1 - 10.7
Average 238.7 11.8
Figure 17
Semass - Dioxin and Furan Test Results
314
-------
Toxic Equivalent EPA/TEF (2,3,7,8 TCDD)
Test Averages (ng/Nm3 @ 12% CO£
Test Date
Unit 1
Unit 2
Figure 18
Semass - Dioxin and Furan Test Averages
April 1989 January 1990 September 1990
2.469
2.320 - 1.0826
315
-------
REFERENCES
(1) Holler, Jens Thousig: Cleaning of waste incinerator flue
gases by Niro Atomizer spray dryer absorption system. Danish
Pollution Control Technology Symposium, Taiwan, October 1986.
(2) Donnelly, J.R., et.al.: SDA systems for MSW incineration -
European operating results. APCA Annual Meeting, New York,
New York, June 1987.
(3) Donnelly, J.R. and Felsvang, K.B.: Joy/Niro SDA-FGC systems -
North American and European operating experience.
International Conference on Municipal Waste Combustion, April
11-14, 1989, Hollywood, Florida.
(4) Moller, Jens Thousig, et.al.: Process for removal of mercury
vapor and/or vapor of noxious organic compounds and/or
nitrogen oxides from flue gas from an incinerator plant. U.S.
Patent 4.889.698, issued December 26, 1989.
(5) Bellin, Judith 8. and Barnes, Donald 6.: Interim Procedures
for Estimating Risks Associated with Exposures to Mixtures of
Chlorinated Dibenzo-p-dioxins and Dibenzofurans and 1989
Update. U.S. EPA, EPA/625/3-89/016, March 1989.
(6) Schmitz, Hans Joachia: Durchbruch bei Dioxin-Emissionen aus
der Mullverbrennung. WLB Wasser, Luft und Boden 6/1990.
(7) ViMcelsoe, J., and J. Nielsen, D.: Study of the gas sampling
procedure for the Danish incinerator dioxin program. Dioxin
88, Umea, Sweden.
(8) Nielsen, Kirsten Kragh, et.al.: Reduction of chlorinated
dioxins and furanes in the flue gas from incinerators with
spray absorbers and electrostatic precipitators. Chemosphere,
Vol. 19, p. 367-372, 1989.
(9) Nielsen, Kirsten Kragh, et.al.: Reduction of dioxins and
furanes by spray dryer absorption from incinerator flue gas.
Dioxin-85, 5th International Symposium on Chlorinated Dioxins
and Related Compounds, Bayreuth, FRG, Sept. 16-19, 1985.
(10) Jansson, B. and Bergvall, 6: Recommended methodology for
measurements of PCDD and PCDF in the Nordic countries. Waste
Management & Research 5, 251 (1987).
(11) Nielsen, Kirsten Kragh: Meeting the 0.1 ng toxic equivalent
limit using spray dryer absorber and fabric filter for flue
gas cleaning on incinerators. (unpublished) Dioxin 90,
Bayreuth, FRG, September 1990.
(12) Donnelly, J.R., et.al.: Semass Waste-to-Energy Plant. IGCI
Forum, 1988.
316
-------
(13) RDF-fired plant eases South Florida solid waste problem.
Power Magazine. October 1990.
317
-------
Intentionally Blank Page
318
-------
SESSION 8G
NOVEL/EMERGING FLUE GAS CLEANING TECHNOLOGY II
Co-Chairmen:
Charles B. Sedman
AEERL
U.S. EPA
Research Triangle Park, NC
R. Michael Hartman
ABB Resource Recovery Systems
Windsor, CT
Preceding page blank 319
-------
Intentionally Blank Page
320
-------
The work described in this paper was not funded by the U.S. Environmental
Protection Agency. The contents do not necessarily rellect the views of the
Agency and no official endorsement should be inferred.
RETROFIT ACID GAS EMISSIONS CONTROLS
FOR
MUNICIPAL WASTE INCINERATION;
AN APPLICATION OF DRY SORBENT INJECTION
by
Jan T. Zmuda
Peter V. Smith
R-C Environmental Services and Technologies
P.O. Box 1500
Somerville, New Jersey 08876
(908)-685-4915
ABSTRACT
The EPA has developed revised emission standards for existing municipal solid
waste incinerators. Compliance with these proposed standards will have a significant
impact on operating incinerator facilities. Retrofit of acid gas emissions controls will be
required for many of these units.
Dry sorbent injection has been successfully demonstrated on coal fired boiler
applications as a means of reducing sulfur dioxide emissions. Recently, a Dense Phase
Dry Sorbent Injection system was applied to an existing municipal waste incinerator to
provide acid gas emission controls. The results obtained from the successful
demonstration of the sorbent injection system on an existing municipal incinerator are
presented. Removal efficiencies of compounds such as HC1, S02, S03, and others by the
use of sorbent injection are shown.
An application case study will outline exhaust conditions, expected pollution
reductions, and capital and operating costs for various types of sorbents for a typical
municipal incinerator which will require acid gas emission controls.
Preceding page blank
-------
INTRODUCTION
On November 15, 1990 the Clean Air Act Amendments of 1990 were signed into
law by President Bush. The enactment of these Amendments has ushered in a new era
of regulatory activity with the goal of improving the quality of the environment. The
Amendments require the U.S. Environmental Protection Agency to develop regulations
to control the emissions of a wide range of chemicals and compounds. These regulations
will affect a broad range of industries, many of which have previously not been subject to
stringent air emission controls. The EPA is also provided with enforcement powers to
ensure that the regulations are followed by the affected industries.
One of the new provisions contained in the Amendments applies to Municipal
Waste Combustion Facilities (MWC). These new provisions required that the EPA
develop new emissions standards for these facilities. In fact, the first set of regulations
which were developed by the EPA apply to the control of emissions from municipal
waste combustion.
Two sets of regulations have been developed to control emissions from municipal
waste combustion. These are (1) those that apply to new facilities, ie. revised New
Source Performance Standards (NSPS) and (2) those which apply to existing sources, the
"Emissions Guidelines." The regulations which apply to existing facilities are the focus of
this paper.
Compliance with these proposed standards will have a significant impact on many
operating incinerator facilities. Retrofit of acid gas emissions controls will be required
for many of these units.
REGULATIONS
The new emissions guidelines proposed for existing facilities are the first issued
under the new Clean Air Act Amendments. As presently developed, these "Guidelines"
are only partially complete. The new emissions guidelines apply to all MWC units which
have a capacity of greater than 250 tons per day (TPD), per combustor. Within two
years, the EPA will issue additional regulations for combustors of 250 TPD, or less,
capacity. Furthermore, within one year, the EPA is required to develop additional,
specific numerical limits for mercury, cadmium, and lead emissions.
The Clean Air Act Amendments require that the emissions limits developed for
municipal waste combustion reflect the "maximum degree of reduction in emissions of air
pollutants ... that the Administrator, taking into account the cost of achieving such
emission reduction, and any non-air quality health and environmental impacts and energy
requirements, determines is achievable for new or existing units in each category."
The Amendments also state that the "emissions standards for existing units in a
category ... shall not be less stringent than the average emission limitations achieved by
the best performing 12 % of existing units in the category ...". This is referred to as the
322
-------
Maximum Achievable Control Technology or MACT.
The new emissions levels specified in the guidelines are further based on the
premise that the "best demonstrated technological basis" (BDT) for reducing emissions
differs based on the aggregate size of each facility. (The BDT is considered to be that
technology which achieves the greatest reduction in emissions, considering cost.)
Consequently, the emissions guidelines are categorized by the size of the total facility
(within which individual units of capacities greater than 250 ton/day are contained). For
facilities in which the total aggregate capacity is less than or equal to 1,100 tons per day
one set of regulations apply. Another set of regulations apply to plants with a total
aggregate capacity greater than 1,100 tons/day.
Figure 1 summarizes the new emissions guidelines for municipal waste combustors
which are greater than 250 ton per day per unit at facilities with an aggregate capacity of
greater than 1,100 tons per day. This table shows that for facilities within this range, the
required emission controls include 90 % reduction of HC1 or 25 ppmdv, 70 % reduction
of S02 or 30 ppmdv, and control of particulate emissions to 0.015 grains/dscf.
Figure 2 summarizes the emissions guidelines for municipal waste combustors
which are greater than 250 tons per day per unit with an aggregate capacity which is
equal to or less than 1,100 tons per day. For these facilities, less stringent requirements
apply. These include 50 % reduction of HC1, 50 % reduction of S02, and control of
particulate emissions to 0.03 grains/dscf.
Heavy metals for both size facilities are not individually regulated. It is believed
that control of particulate matter would provide sufficient reduction of heavy metals
emissions. Mercury, however, will be regulated in the future. Mercury requirements are
to be promulgated by November 15, 1991. At present it is not clear what form these
regulations will have.
The Guidelines also identify the best demonstrated technologies for reducing
emissions. These technologies are not presented as the required control equipment but
merely to serve as the technological and economic basis for the guidelines.
For facilities which have an aggregate capacity greater than 1,100 tons/day the
BDT includes:
• Good combustion practice for control of organics.
• Spray dryer followed by an electrostatic precipitator for control of acid gas
and metals.
For facilities with aggregate capacity less than or equal to 1,100 tons/day the BDT
includes:
• Good combustion practice for control of organics.
323
-------
• Dry sorbent injection followed by an existing electrostatic precipitator for
control of acid gas and metals.
Recently Waste Age magazine published a report on the status of municipal waste
combustion which included a survey of operational waste incinerator installations (both
waste-to-energy plants and incinerators). According to this survey, there are 128 waste-
to-energy facilities and 40 incinerator facilities in operation in the U.S. Of these,
approximately 100 units fit the criteria established by the current regulations and require
installation of emission controls. These facilities will have until 1996 to be in compliance
with the emissions guidelines.
The EPA has until 1992 to establish comparable emissions standards for smaller
combustors, those less than or equal to 250 tons/day per train.
An analysis of the operating waste incinerator installations which are rated at 250
tons/day or greater, at facilities which have a total operating capacity of less than 1,100
tons per day was made. Based on this survey, there are about 30 operating incinerator
"trains", at roughly 15 plants, which do not have acid gas controls. However, these units
all have existing particulate controls, primarily electrostatic precipitators.
These units would be likely candidates for retrofit dry sorbent injection
technology.
DRY SORBENT INJECTION TECHNOLOGY OVERVIEW
Dry Sorbent Injection, or DSI, is a means of acid gas control in which a totally
dry, highly pulverized reagent, or sorbent, is introduced into a flue gas stream containing
acid gases. Dry sorbent technology has been applied for acid gas control in two
fundamental forms. The first type utilizes the injection of a sorbent directly into the
combustor or furnace, in a high temperature regime. This type is sometimes referred to
as furnace sorbent injection, or FSI. The second type utilizes the injection of a dry
sorbent into the flue gas downstream of the combustion chamber. This process is a post-
combustion process and is sometimes referred to as duct sorbent injection. In this
approach sorbent can be injected in the economizer region of a boiler at relatively high
temperatures or in the flue gas duct upstream of the final particulate control device
where the temperatures are much lower. In both control approaches the reaction
products and any unreacted sorbent are captured in conventional particulate control
equipment such as an electrostatic precipitator or fabric filter. Disposal of the waste
products is accomplished in the same manner as existing fly ash disposal.
A brief discussion of these sorbent injection techniques will be provided below.
FURNACE SORBENT INJECTION
The technique of combustion zone injection for the control of acid gas emissions
is shown in Figure 3. In this approach a dry sorbent such as limestone or hydrated lime
324
-------
is injected into the combustor. Injection of the sorbent is accomplished in several
different methods. These include adding the reagent with the waste charge to the
furnace, pneumatically conveying the reagent with combustion air, or pneumatically
injecting the reagent through nozzles in the combustion zone just above the fuel bed.
Furnace sorbent injection has been utilized to control S02 emissions in utility and
industrial coal fired boiler applications since the mid 1980's. Several furnace sorbent
injection systems have been operating for over 5 years.
Limestone injected into the furnace at temperatures between 1,800 to 2,200 will
calcine to produce lime. The lime (either from the calcined limestone or as directly
injected) reacts with the acid gases.
The chemical reactions which occur include:
CaC03 + heat —> CaO + C02
or
Ca(OH)2 + heat —> CaO + H20
then:
CaO + 2HC1 —> CaCl2 + H20
CaO + 2HF —> CaF2 + H20
CaO + S02 + V202 —> CaS04
Within the furnace there are several temperature zones which yield optimal
removal efficiencies. The calcination of limestone to lime occurs in the 1,800 to 2,200 °F
temperature window. S02 removal is also greatest in this high temperature zone as well
as in a second zone of temperature between 900 to 1000 °F. HC1 and HF, however, are
more effectively removed at lower temperatures, downstream of the furnace. Figure 4
illustrates the removal efficiency of S02 as a function of temperature of the injection
zone. The different zones of optimal efficiencies are clearly shown.
One criteria which is used to measure the effectiveness of dry sorbent injection
system operation is the calcium to sulfur molar ratio. This ratio reflects the quantity of
sorbent required to achieve acid gas removal. This ratio, also referred to as the
stoichiometric ratio, is the moles of calcium injected divided by the moles of entering
acid gases. Figure 4 illustrates sulfur dioxide capture at a stoichiometric ratio of 2.0.
Furnace sorbent injection of limestone has been demonstrated commercially on
coal fired boiler applications for many years. This approach is considered to be
commercially proven.
Furnace sorbent injection has several advantages, particularly as a retrofit control
technology. These include:
• provides extended contact time for lime and acid gas reactions to occur
325
-------
from the furnace to the particulate control device.
• potential to reduce dioxin/furan formation by removing chlorine.
• provide some measure of corrosion prevention within furnace and heat
recovery sections.
• relatively easy to retrofit.
• low capital cost.
• high reliability.
• commercially demonstrated technology.
The potential disadvantages of furnace sorbent injection include:
• possible fouling and erosion of convective heat transfer surfaces.
• low utilization of sorbent.
• limited removal capability.
• significant increase in particulate loading which may require modifications
to existing electrostatic precipitators and ash handling equipment.
POST-COMBUSTION INJECTION
Post furnace sorbent injection involves the introduction of a dry sorbent material
into the process after the combustion zone. Figure 5 shows a schematic of the various
locations which can be utilized for post-furnace combustion controls. These include
injection into the economizer region of the boiler or in the flue gas duct downstream of
the boiler.
A number of different sorbents can also be utilized for this dry sorbent injection
approach depending upon the location of the injection point. Each sorbent can provide
a different degree of acid gas emission reduction.
Hydrated lime can be injected into the economizer zone at a temperature of
approximately 800 to 1,000 °F. Hydrated lime can also be injected in the flue gas duct at
a temperature of 350 °F.
The chemical reactions which occur include:
Ca(OH)2 + 2HC1 > CaCl2 + H20
Ca(OH)2 + 2HF > CaF2 + H20
Ca(OH)2 + S02 + Vi02 > CaS04
In addition to hydrated lime, several sodium based sorbents have been utilized for
post-combustion injection. These include sodium bicarbonate and sodium
sesquicarbonate, or trona. These materials would be injected downstream of the
economizer, in the flue gas duct at a temperature of 350 °F.
The sodium reagents will not be discussed in this paper. It should be noted that
these reagents have demonstrated higher removal efficiencies at lower "equivalent"
326
-------
injection rates. However, the costs of these sorbents is substantially higher than those of
the calcium based reagents, and in general, disposal of calcium based waste products is
less difficult than disposal of sodium compounds.
In addition to reagent type and injection location, other factors which affect the
performance of these dry sorbent injection systems include quantity of injected sorbent
(stoichiometric ratio), contact time, flue gas moisture content, type of particulate control
device, and acid gas concentration.
In some cases, the retrofit of post-combustion sorbent injection systems have
included additional design features, some of which address the variables previously
mentioned. One such modification is the use of flue gas humidification. Humidifying
the flue gas enhances the collection efficiency of acid gases and conditions the flue gas to
improve the capture of particulates in a downstream ESP. Downstream humidification is
applicable for both calcium and sodium based sorbents.
Post-combustion sorbent injection also has a number of advantages as a retrofit
control technology. These include:
• easy to retrofit.
• simple design and operation.
• low capital cost.
• commercially demonstrated technology.
• higher removal efficiencies and utilizations (compared to FSI).
Potential disadvantages include:
• utilization of sorbent is low.
• limited removal capability.
• significant increase in particulate load to existing particulate controls.
A dry sorbent injection system is a very simple system to retrofit. The equipment
required includes a sorbent storage silo(s), feeder(s), compressed air system, pneumatic
conveying piping, and miscellaneous valves and fittings. Other modifications may include
upgrades to electrostatic precipitator ash handling equipment as well as possible
improvements which may be required to the electrostatic precipitator to insure that
existing particulate emissions standards are met, even with the increased dust loadings.
DENSE-PHASE DRY SORBENT INJECTION
One application of the dry sorbent injection system utilizes a dense-phase sorbent
transport and injection system. The term "Dense-Phase" refers to the method of
pneumatic transport of the reagent to the injection zones. Typical pneumatic conveying
is accomplished using dilute phase conveying, ie. there is a large amount of transport air
required to convey a specific mass rate of solids. In dense-phase transport, extremely
low volumes of air are used to transport and inject the same mass rate of solids. Since
327
-------
the volume of transport air is low, there are no impacts on the combustion process
(temperatures, volumes, etc.).
The dense phase transport also provides greater transport distances, in smaller
diameter pipe, as compared to typical pneumatic transport. This has two benefits. The
first is that lower operating power consumption is required. The second benefit is that
installation is simplified since the reagent storage equipment can be located at more
remote locations where space limitations are less critical. Figure 6 compares dilute
phase and dense phase pneumatic transport characteristics.
This process has successfully been applied to coal fired boiler applications with
sorbent injection directly into the combustion zone of the boiler.
The dense phase dry injection system was retrofitted on a full scale municipal
waste-to-energy facility for a program to demonstrate the feasibility of this technology for
control of acid gases. The demonstration site has several operating combustor units,
each greater than 250 tons/day. The total aggregate size of all of the operating
combustor units would place the site in the "greater than 1,100 ton per day" category.
The demonstration dense phase dry sorbent injection system is designed to be a
transportable system which can be moved to various sites as required. The mobile unit
consists of a flatbed truck which contains a storage silo and silo fill system, dense-phase
transporter, compressor with dryer and receiver, and a control panel. This system can be
quickly and easily set up. Delivery of sorbent is accomplished with a 20 ton capacity
tanker truck. This truck provides the sorbent to fill the silo as necessary using a remote
transport system.
A sketch of the dense phase mobile demonstration unit is shown in Figure 7.
DENSE PHASE SORBENT INJECTION PERFORMANCE
A demonstration program of the dry sorbent injection system on a municipal
waste combustor began in August 1990. The injection of sorbent was accomplished using
a dense phase injection into multiple locations in the furnace. Hydrated lime was used
as the sorbent. Figure 8 contains a summary of the chemical analysis of the hydrated
lime used during the demonstration program.
The demonstration system was located approximately 350 feet from the furnace.
Transport of the sorbent to the injection ports was via a 1 Vi" I.D. carbon steel pipe.
A preliminary evaluation of the performance of the incinerator facility was
conducted to establish baseline information of acid gas emissions and to establish
sorbent injection rates. Based on this evaluation, the dense phase sorbent injection
system was sized to provide from 500 to 2,500 lb/hr of sorbent. However, for much of
the demonstration period the unit was operated at about 1,000 lb/hr.
328
-------
The uncontrolled emissions levels from the incinerator were determined to be:
HC1 700 ppmdv corrected to 7 % 02
S02 200 ppmdv corrected to 7 % 02
In terms of mass rates, the uncontrolled emissions were determined to be 374 lbs/hr. for
HC1 and 175 lb/hr. for S02 for the baseline data.
Emissions data was obtained during various phases of the demonstration program.
It should be pointed out that uncontrolled emissions data could not be obtained with the
dense phase system in service. Consequently, the uncontrolled emissions data from the
baseline evaluation was used to establish the performance of the dry sorbent injection
system.
The results of the demonstration program showed that with an injection rate of
1,000 lbs/hr. of hydrated lime into the furnace the following emissions levels were
achieved:
HC1 350 ppmdv corrected to 7 % 02
S02 80 ppmdv corrected to 7 % 02
HC1 mass emissions were 180 lb/hr., which represents a reduction of 53 % from
uncontrolled levels. S02 mass emissions were 70 lb/hr., which represents a reduction of
60 % from the uncontrolled level. Again, the percentage reductions are based on
emissions data taken from a different period of testing.
The injection rate of 1,000 lb/hr. of hydrated lime represents a normalized
stoichiometric ratio of approximately 1.65 using the baseline uncontrolled acid gas
emissions levels.
HC1 emission reductions in excess of 75 % with S02 emission reductions in excess
of 80 % were recorded during different phases of the demonstration program. Hydrated
lime feed rates were not recorded during these tests. However, based on recorded
operating parametric data, it is believed that the hydrated lime feed rate was
approximately 1,200 lb/hr. This corresponds to a normalized stoichiometric ratio of
about 2.1.
Still lower outlet emissions were obtained at higher hydrated lime feed rates.
However, insufficient operating data was recorded to report these lower emissions now.
Particulate control for this incinerator is accomplished with an electrostatic
precipitator. Specific tests of precipitator performance (inlet-outlet) were not conducted.
However, an emissions monitor was used to track outlet opacity. No deterioration of
precipitator performance was exhibited during the demonstration program.
329
-------
No adverse effects were evident in the operation of the furnace or downstream
equipment as a result of dense phase dry sorbent injection. The existing ash handling
equipment was adequate to remove the additional dry waste products resulting from the
sorbent injection system. This equipment showed no adverse effects. Overall combustor
system reliability was maintained.
By way of comparison, data presented by Beckman and Spahn showed 70 % HC1
and 80 % S02 reduction by use of a multiple-location dry sorbent injection system on a
400 ton/day facility. Injection of hydrated lime at this site is accomplished in three
locations: at the furnace entrance, between the superheater and convection section, and
in the ductwork between the convection zone and the economizer. The approach here is
to take advantage of the acid gas removal efficiencies provided by the different
temperature zones.
Insufficient data was recorded to allow us to present data on reductions of heavy
metals or mercury at this time.
APPLICATION CASE STUDY
A typical municipal combustor can be used to illustrate the costs and operating
requirements of a dense phase sorbent injection system. Figure 9 contains a summary of
the performance variables associated with this "typical" 250 ton per day incinerator and
the associated dry sorbent injection system. The system design is based on hydrated lime
injection into the furnace. The equipment required for a dense phase sorbent injection
system includes:
• reagent storage silo with bin activator and feeder
• pneumatic transport piping
• compressors for transport air
• injection ports
Figure 10 summarizes the sizes of this equipment for a 250 ton/day facility.
The economics of the dense phase system can be summarized as:
Our 250 ton per day plant, operating at 85 % capacity, would handle approximately
77,600 tons per year of waste. The annual incremental cost of the air pollution
equipment, represented as annual cost per ton of waste would be $ 4.85 per ton.
Item
Equipment amortized (20 yrs)
Annual Cost
Operation & Maintenance
Hydrated Lime @ $80/ton
Waste disposal @ $50/ton
$ 50,000
$ 65,000
$ 149,000
$ 112.000
$ 376,000
330
-------
It should be noted that while the numbers used are fairly conservative, actual
costs for a specific site could differ, particularly for facilities with multiple incinerator
units.
CONCLUSIONS
The application of dry sorbent injection has been demonstrated to be a viable
emissions control strategy for retrofit on existing municipal waste combustion units. Dry
sorbent technology is simple to retrofit and has substantially lower capital costs as
compared to installation of spray dryers. It is simple to operate. No liquids are
circulated, all materials including the sorbent and waste products are dry.
In summary:
• dry sorbent injection is an effective means to achieve compliance for acid
gases for municipal waste combustors which are larger than 250 TPD at
facilities whose aggregate capacity is less than or equal to 1,100 TPD
• dry sorbent injection is easy to retrofit.
• dry sorbent injection is simple to operate.
• when combined with flue gas cooling and an ESP control of particulate
matter, heavy metals, CDD/CDF, and acid gases can be achieved.
• dry sorbent injection is an economical, effective method to achieve
compliance as required by the established EPA Emissions Guidelines.
The mobile dense-phase demonstration unit can be readily transported and set-up
at an existing facility to demonstrate the suitability of this technology on a site-specific
basis. The demonstration unit, coupled with a detailed analytical and operational test
program is a cost effective means of demonstrating the effectiveness of dry sorbent
injection technology and optimizing the design strategy for a particular municipal waste
incinerator facility. The use of the demonstration unit will permit the identification of
the impact of this type of acid gas control technology on the existing electrostatic
precipitator and ash handling equipment.
331
-------
REFERENCES
1. Offen, G. R. et al, "Assessment of Dry Sorbent Emission Control Technologies,"
Journal of the Air and Waste Management Association. Volume 37, No. 8, August
1987.
2. Smith, P. V. and Landreth, R. R., "Economics and Operating Experience of a
Furnace Limestone Injection System," Presented at IGCI Forum 90, March 1,
1990, Baltimore, MD.
3. Landreth, R. R. and Smith P. V., "Retrofit of Sorbent Injection Technology on an
Older Coal Fired Boiler," Presented at the 82nd Annual Meeting of the Air and
Waste Management Association, June 25-30, 1989.
4. Kiser, Jonathan, "A Comprehensive Report On the Status of Municipal Waste
Combustion," Waste Age. November 1990.
5. Environmental Protection Agency, "Emission Guidelines: Municipal Waste
Combustors, 40 CFR Part 60, [AD-FRL 3694-5].
6. Beckman, Arthur and Spahn, David, "Dry Lime Injection for Acid Gas Control in
Municipal Waste Incinerators," Presented at the 82nd Annual Meeting of the Air
and Waste Management Association, June 25-30, 1989.
332
-------
Figure 1
Summary of Guidelines for Existing Facilities
Facility Aggregate Capacity: > 1,100 TPD
HC1
so2
Particulates
Opacity
Organic emissions1
Technology Basis:
Notes:
1. Organic emissions are measured as total dioxin/furans.
2. All emission levels are at 7% 02, dry basis.
90 % or 25 ppmv
70 % or 30 ppmv (24 hr.avg)
0.015 grains/dscf
10 % (6 minute average)
60 ng/dscm
Good combustion practice followed by a spray dryer and electrostatic
precipitator
Figure 2
Summary of Guidelines for Existing Facilities
Facility Aggregate Capacity: < 1,100 TPD
HC1 50 % or 25 ppmv
S02 50 % or 30 ppmv (24 hr.avg)
Particulates 0.03 grains/dscf
Opacity 10 % (6 minute average)
Organic emissions1 125 ng/dscm
Technology Basis: Good combustion practice followed by dry sorbent injection and electrostatic
precipitator
Notes:
1. Organic emissions are measured as total dioxin/furans.
2. All emission levels are at 7% 02, dry basis.
333
-------
Figure 3
Furnace Sorbent Injection
Slack
Figure 4
Temperature Effects On Removal Efficiency of Sulfur Dioxide
sat
Air
Heater
Upper
Furnace
Econ
Superiiea ter/Rehea ter
100
Zone 3
m
u
s
Zone 2
Zone 1
500
0
1000
2000
1S00
3000
2500
SORBEIfT INJECTION TEMPERATURE. °F
334
-------
Figure 5
Post Combustion Sorbent Injection
[ *sp ]
w
\su
wire calcium
Figure 6
Dense Phase vs. Dilute Phase Characteristics
Characteristics
Dense Phase
Dilute Phase
Solids/Air Ratio
30+ : 1
5:1
Design Pressure, psig
40-100
o
i
Operating Pressure, psig
50
5
Pipe Velocity, ft/min.
300 - 500
3000 - 4000
Line size, ID
1 - 3
6 - 12
Horsepower
60
250
Conveying Distance, ft.
800 - 1000
200 - 300
Air Use, scfm
250
2000
335
-------
Figure 7
Dense Phase Mobile Demonstration Unit
mi
-n
i: \
r
ii tt •
inss
Figure 8
Hydrated Lime Chemical Analysis
Component
% bv weight
Calcium Hydroxide, Ca(OH)2
90 -95
Calcium Carbonate, CaC03
3 - 5
Magnesium Carbonate, MgC03
1.5- 2.5
Silica, Si02
1.0- 1.4
Iron, Fe203
0.2- 0.4
Aluminum, A1203
0.3- 0.6
Moisture, H20
0 - 2
336
-------
Figure 9
Operating Conditions of a "Typical" Incinerator
Incinerator:
Capacity 250 tons/day (77,600 tons/year)
Capacity factor 85 % annual on-line operation
Flue gas 50,000 acfm @ 400 F
HC1 150 lb/hr.
S02 75 lb/hr.
Sorbent Injection System:
Hydrated lime 500 lb/hr. at 95 % purity
(1,860 tons/year)
Waste ash 600 lb/hr. additional waste
(2,230 tons/year)
Figure 10
Dense Phase Dry Sorbent Injection System
Major Equipment List
Description
30 tons capacity ( > 5 days)
2,400 cu.ft. capacity
2 @ 10 cu.ft. (one operating)
2 @ 100 cfm (one operating, one
with receiver and dryer.
1 V2" I.D. carbon steel pipe.
Equipment
Silo
Transporter
Compressor
Transport pipe
337
-------
Intentionally Blank Page
338
-------
The work described in this paper was not funded by the U.S. Environmental
Protection Agency. The contents do not necessarily reflect the views of the
Agency and no official endorsement should be inferred.
PILOT PLANT TESTS TO OBTAIN HC1 and Hg reduction in emissions
produced by a Municipal and Hospital Solid Waste Combustor
A. Magagni, G. Boschi
AMNIUP - Padua, Italy
INTRODUCTION
In July 1990 a Decree of Environment Ministry had fixed limits for the emissions of
industrial plants in general and for MSW incinerators in particular. Each plant that does
non comply with these limits has to present a treatment gas project to the Region
administration to reach these limits within two years the end of 1992.
Padua Incinerator, that treats MSW together with Hospital Waste (25% of total amount)
and expired medicines (2% of the total amount) complies the law for all parametrs except
for Dust, Mercury and Hydrochloric Acid.
PLANT GENERAL DESCRIPTION
Combustion Chamber
Hospital waste are not dumped in the receiving bunker as municipal waste but, packed in
a polietilene bag introduced in a cardboard box or in a polietilene drum, are directly put
into the feed shaft with a conveyer belt.
Primary and secondary air of combustion may be conveniently adjusted and their
distribution system is designed to create a good turbulence to realize an effective mixing
between comburent and combustible to maintain a more uniform heat distribution all over
the chamber.
Post-Combustion Chamber
The flue gas, coming from combustion chamber, passes in the post- combustion chamber
that, according to italian law, must guarantee the follow operating conditions:
- 02 free in wet fumes (downstream the chamber) : >6% in vol.
- average gases velocity in chamber entrance section : >10 m/s
Preceding page blank 339
-------
- contact time
- temperature
- combustion efficiency
>2 s.
>950°C
> 99.9%
This chamber is provided with a burner that is automatically started as temperature of
fumes goes under 950°C. In Padua incinerator temperature is normally around 1050-
1100°C without the necessity of using the burner.
Heat recovery
Thermoelectric group consists of a vertical superheater working mainly by radiation.
Vapour generator is a tube nest type, vertically disposed, operating by convection.
Produced vapour is sended to a turbine to produce electric energy.
Dry Scrubbers
Flue gases coming from air mixer enter in the two reactors designed for dry adsorption of
acid gases. Dry scrubbers are composed of two parallel units. Downstream each reactor
is installed a cyclone for partial recycle of reagent. The gas control is obtained by
injection of Calcium Hydroxide through nozzles that realize a chemi-adsorption of acid
gases on line, giving as products salts as CaCl2, CaF2, CaSC>3, CaS04.
Adsorption efficiency is depending on: specific surface of the dry adsorbent per cubic
meter of gas; concentration of polluting acid gases; contact time; homogeneity of mixing;
temperature of reaction; water content of gases.
In our case temperature of reaction is in the range 250-300 °C too high to obtain good
efficiencies of acid gases scrubbing.
Since 1991 we started to use Sodium Bicarbonate. In Table 1 and 2 are summarized the
removal efficiencies versus stochiometric ratio for Sodium Bicarbonate and Calcium
Hydroxide.
Tab. 1 - Calcium Hydroxide versus Sodium Bicarbonate HC1 removal efficiencies
Reactive
Stochiometric ratio
HC1 removal efficiency %
Sodium Bicarbonate
1
80
1.2
90
1.5
97
Calcium Hydroxide
6.7
90
340
-------
Tab. 2 - Calcium Hydroxide and Sodium Bicarbonate: chemical reactions for HC1
removal and fly ash production.
Calcium Hydroxide
Ca(OH)2 + 2 HC1 -»¦ CaCl2 + 2tH20
Stoichiometric calculations
74/73 = 1.014 kg di Ca(OH)2 for 1 kg of HC1
1 kg of HC1 - 1.52 kg of CaCl2
Fly ash produced for a 90% yield
1 kg of HC1 -> 1.52 kg of CaCl2 + 5.78 kg of Ca(OH)2 = 7.3 kg of fly ash
Sodium Bicarbonate
NaHC03 + HC1 -»¦ NaCl + tC02 + tH20
2 NaHC03 -»¦ Na2C03 + tC02 + tH20
Stoichiometric calculations
84/36.5 = 2.301 kg of NaHC03 for 1 kg of HC1
1 kg of HC1 - 1.6 kg of NaCl
Fly ash produced for a 97% yield
1 kg ofHCl - 1.6 kg of NaCl + 0.73 of Na2C03 = 2.33 kg of fly ash
ESP
Solid particles - dust from combustion chamber, salts produced by reaction between acid
gases and lime, and unreacted lime - are carried by the flue gas to an electrostatic
precipitator.
Fly ash, classified as a toxic waste by italian law for the high content in Lead and
Cadmium, is sended to a stabilization-solidification plant in Modena (Italy).
341
-------
Pilot Plant
The pollutants that do not comply with Emission limits fixed by Italian normative are:
particulate matter;
Hydrochloric acid;
Mercury.
Because of this, we realized a pilot plant consisting of a wet system, an ESP and a
Baghouse. These three devices are connected to the actual plant as shown in Fig. 1.
Actual PJmt
COMBUSTION
CHAMBER
POST-COMB
Fig. 1
Pilot Plant
BOILER
ill
BAG
HOUSE
DRV
SCRUBBER
ESP
II
WET
SYSTEM
ESP
iii
IIliAT
EXCIIANG.
STACK
ACTIVATED
CARBON
0 sampling points
342
-------
Pilot Plant Description
The plant has been realized to treat a gas flow of 4000 Nm3/h.
Dedusted gas (by the ESP or the Baghouse) enter in a QUENCH to cool down the gas
from 210 °C to 70 °C.
The Quench has an anti - acid heat shock resistance fettling.
The gas enters at the temperature of 70 °C in a wet scrubber divided into two stages (an
acid washer stage and a basic washer stage).
The tower, containing special plates, has been realized in plastic material reinforced with
glass fibre.
In the bottom stage water is used to obtain HC1 adsorpion. The solution is almost entirely
recycled so that the pH is very low. A specific pH control system will provides to the
automatic immission of fresh water to mantain HC1 concentration constant.
A small amount of solution (5-10% of total amount in cycle) containg HC1, heavy metals
his sended to existing water treatment plant.
Air, coming out from the bottom stage of the column, passes to the second washer stage
in which a basic solution of water is used to get rid residual HC1 and of S02.
The caustic soda solution leads to the formation of Na2S03 (sodium sulphite).
The basic solution is sended to an oxidation tower in which air is injected for conversion
of sulphite to sulphate.
The fue gas, after the exit of the column passes a dimister.
The saturated flue gas is than sended to an heat excharger to bring temperature to 70 °C so
that percentage relative humidity drops to 35-45% . Treatet gas are sended to the existing
stack.
Some test will be conducted using an activated carbon adsorber.
The results of some of the tests are summarized in Fig. 2, 3 and 4.
343
-------
00
Sodium Bicarbonate
stochiometric ratio = 1.5
: HC1 = 50 mg/Nm3
; Hg = 800 jig/Nm3
.. dust = 100 mg^Nm3
Wet
Scrubber
HC1 < 10 mg^Nm3
Hg = 50 jig/Nm3
dust < 10 mg/Nm3
HC1= 1200mg/Nm3
Hg = 1000 jig/Nm3
dust = 900 mg/Nm3
flue gas
(from boiler) l'"1'''' *¦'
ESP
stack
iMIllMiiSiiiiiiliiiiiiilll
iiiliili: i
fly ash = 110 kg/h
Dry Scrubber
liquid waste
(Chlorine < 1.200 m^yl
Mercuiy » 4 mg^l)
Fig. 2
-------
w flueg^s
(from boiler)
HC1 = 1200mg/Nm3
Hg = 10CX) jig/Nm3
dust = 900 mg
-------
flue ©is
(from boiler)
Sodium Bicarbonate
stochiometric ratio = 1.5
HC1= 1200 mg/Nm3
Hg= 1000 ng/Nm3
dust = 900 mg/Nm3
Dry Scrubber
HC1< 10mg/Nm3
Hg = 800 jig/Nm3
dust < 10 mg/Nm3
Wet
Scrubber
Baghouse
fly ash = llOkg/h
HC1< 10mg/Nm3
Hg = 50 fig/Nm3
dust < 10 mg/Nm3
stack
Fig. 4
f
liquid waste
(Chlorine < 1.200 mg4
Mercury » 4 mg/1)
-------
Conclusions
The best configuration for Padua Plant are the one shown in Fig. 2 and 3.
Advantaged and disadvantages are summarized in the following Tables.
Boiler -+ ESP -+ Wet Scrubber -+ Stack
Advantages
Disadvantages
- a low production of fly ashes;
- no need for a chemical reactive for HC1
removal;
- high concentrations of chlorine in waste
waters (Italian law fixes a limit of 1200
mg/1)
Boiler -+ Dry Scrubber -+ ESP -+ Wet Scrubber -~ Stack
Advantages
Disadvantages
- low concentrations of chlorine in waste
waters (< 1200 mg/1)
- use of a chemical reactive
- high production of fly ash
Probably the best configuration for Padua Plant is intermediate betwen the two shown
above. That is, the quantity of Sodium Bicarbonate could be reduced to a quantity equal
to 0.7 the stochiometric value so that the Chlorine present in liquid waste could be kept
lower than 1200 mg/1 (limit fixed by Italian regulations) and the quantity of fly ashes
produced could be keeped to acceptable values.
347
-------
Intentionally Blank Page
348
-------
SESSION 9A: SAMPLING AND ANALYSIS I
Co-Chairmen:
James D. Kilgroe
AEERL
U.S. EPA
Research Triangle Park, NC
Stellan Markland
Institute of Environmental Chemistry
University of Umea
Umea, Sweden
Preceding page blank
349
-------
Intentionally Blank Page
350
-------
The work described in this paper was not funded by the U.S. Environmental
Protection Agency. The contents do not necessarily reflect the views of the
Agency and no official endorsement should be inferred.
MERCURY EMISSION MONITORING
ON MUNICIPAL WASTE COMBUSTION
by
Hartmut Braun
Kernforschungszentrum Karlsruhe GmbH
Karlsruhe, Federal Republic of Germany
and
Andreas Gerig
Seefelder Messtechnik GmbH & Co.
Seefeld, Federal Republic of Germany
ABSTRACT
In waste incineration, mercury is the only heavy metal to be released
as a gas, mostly as mercury(II) chloride, because of its high volatility.
Continuous emission monitoring is possible only when mercury occurs in its
elemental form. Various possibilities of converting Hg(II) into Hg(0) have
been studied and tested on a laboratory scale and in the TAMARA refuse
incineration pilot facility. Continuous mercury emission measurement appears
to be possible, provided mercury is converted in the flue gas condensate
precipitated. The measuring results obtained on two municipal solid waste
and on one sewage treatment sludge incineration plants show that the mercury
monitor is a highly sensitive and selective continuously working instrument
for mercury emission monitoring.
Preceding page blank
351
-------
INTRODUCTION
In waste incineration, mercury is the only heavy metal to be released
as a gas because of its high volatility. This makes it the real problem
element in waste incineration. Elemental mercury, which constitutes the bulk
occurring in waste, is oxidized quantitatively to mercury(II) chloride by
the hydrogen chloride gas present; in the flue gas only this compound is to
be found [1,2]. This fact is important in designing suitable offgas purifi-
cation techniques. If the offgas is cleaned in scrubbers, it is possible
that the plant will emit not only the Hg(II) component, which is not precip-
itated quantitatively, but also elemental mercury reconstituted as a result
of reduction in the scrubbing water.
The mercury emission limits defined in the German Clean Air Regulations
(TA Luft) cannot always be observed reliably, as the authors found in their
measurements. However, today's sporadic sampling technique for mercury is
unable to monitor the functioning capability of the flue gas purification
equipment.
In addition, the amount of mercury on the input side and, consequently,
also contained in the offgas is subject to major variations. Discontinuous
measurements of the kind occasionally carried out by the monitoring agencies
provide information only about the emission status while the sample is being
taken. No satisfactory monitoring of the real emission situation is possible
in this way. There is need for developing a continuous method of measurement
allowing mercury emission and pollution to be monitored, independent of the
type of mercury compound.
DISCONTINUOUS MERCURY MEASUREMENT TECHNIQUES
Continuous monitoring of the current emission and pollution limits with
respect to mercury is not possible at the present time. Monitoring for these
limits is done in the discontinuous mode by these two techniques [3]:
(1) Absorption in Solutions
Irrespective of the type of compound involved, mercury can be absorbed
quantitatively from the offgas in strongly acid oxidizing scrubbing
solutions. For this purpose, the sample gas flow is passed through two
series-connected impingers filled with a peroxodisulfate solution
352
-------
acidified with nitric acid, or a permanganate solution acidified with
sulfuric acid. After the mercury content in the absorption solution has
been analyzed, the corresponding mercury emission is calculated in the
light of the gas volume.
Adsorption onto Solids
Various solid adsorbers are available for the detection of elemental
and bound mercury. Iodized activated carbon is particularly suitable
for this purpose. The gas stream sampled is passed through an adsorber
tube filled with iodized activated carbon. The adsorber is decomposed
chemically, and the mercury emission is calculated from the amount of
mercury adsorbed, with the gas volume being known. In addition, specia-
tion of the mercury is possible if various adsorber tubes, those for
mercury chloride and those for elemental mercury, are combined (see
Fig. 1). Selective HgCl2 separation can be achieved in a particularly
effective way by a strongly basic anion exchange resin, such as Dowex
1X8. Dowex covered with chloride ions in an appropriate pretreatment
step can be used to remove more than 98 % of HgC^ from a gas stream at
sorption temperatures of 120 #C. Hg(0) is practically unresorbable.
Combining the two solid adsorbers, Dowex and iodized activated carbon,
makes for an efficient speciation of the forms of mercury existing in
the gas phase [3].
Filter
Condenser
Pump
Gasmefer
Flowmeter
DOWEX 1X8
Iodized charcoal
Figure 1. Measuring device for mercury speciation
353
-------
The discontinuous measuring techniques referred to above can be carried
out only at considerable expense. No analytical information can be obtained
at the sampling point, as the samples can only be analyzed in a laboratory.
And yet, there is need for these techniques, especially when balances must
be established at various points in an offgas purification plant, or if it
is a matter of speciation. In those cases, valuable information is obtained
about the functions of various plant components in offgas purification.
CONTINUOUS METHODS OF MERCURY MEASUREMENT
Elemental mercury measurements representing the state of the art are
carried out by means of resonance absorption at X = 253.7 nm. This technique
is used, e.g., in work place monitoring. However, no mercury compounds can
be detected in this way.
In emission measurements conducted in incineration plants, i.e., in
those cases in which mercury exists in a bound form, no direct determination
is possible. Instead, the mercury compounds would have to be converted
continuously into the elemental form if resonance absorption was to be used.
Various techniques of converting bound mercury into elemental mercury
are known from measurements of mercury concentrations in the ambient air
[e.g. 4,5]:
(1) One technique, which is also customarily employed in laboratory analy-
ses, is the addition of a reducing agent to a sample containing mer-
cury, which reliably causes the reduction of Hg(II). This method lends
itself well to monitoring liquid streams and is used for controlling
scrubber circuits and the liquid effluent of refuse incineration
plants.
(2) Another method makes use of the reducing action of activated carbon for
converting Hg(II) into Hg(0) at temperatures around 330 #C. In the
presence of hydrogen chloride, which occurs in gaseous effluents espe-
cially of refuse incineration, lime may be added to stop any reverse
reaction. The carbon converter can also be used to transform reliably
into the elemental form the low mercury contents of the ambient air.
354
-------
(3) Another method operates at high temperatures of around 800 *C, which
cause a direct conversion of Hg(II) into Hg(0). A reverse reaction with
hydrogen chloride must be excluded. Thermal decomposition can be em-
ployed without any restrictions only if the existence of hydrogen chlo-
ride can be largely excluded. As a rule, this is true in environmental
impact monitoring.
So far only the first method [6-9] has been used to measure mercury
emissions downstream of power plants and refuse incineration plants. All
these variants have some fundamental shortcomings:
Method 1 suffers from the drawback of continuously consuming chemicals,
which makes it expensive; the equipment needs permanent maintenance and
supervision.
Method 2 requires regular exchanges of the adsorber pipe as soon as the
adsorber mix has been spent and, as a consequence, can be used meaning-
fully only in so-called clean gas at hydrogen chloride concentrations
<100 mg/m^.
Because of a possible recombination with hydrogen chloride, method 3
works safely only in the absence of HC1. In addition, there are techni-
cal problems because of the high temperatures.
In our development of a mercury analyzer, a different possibility is
utilized to convert the non-detectable molecular form into the detectable
elemental form, namely the reaction of mercury bearing flue gas with its own
condensate, which is obtained by cooling the sample gas stream to the re-
quired level. Naturally, this condensate contains reducing constituents,
especially dissolved SO2. The contact of mercury bearing offgases with the
condensate causes Hg(II) to be directly reduced to Hg(0) and to be expelled
with the gas stream. It can then be measured by means of resonance absorp-
tion.
355
-------
EXPERIMENTAL INVESTIGATIONS OF MERCURY REDUCTION
VARIANT 1: Reduction with Stannous Chloride
Gas containing mercury must first be absorbed in a solution before
being reduced to its elemental state by the addition of stannous chloride or
borohydride solutions or sulfurous acid. The sample gas stream is passed
through an impinger with an acid solution, which provokes the adsorption of
Hg(II) in a first step; in a second step, Hg(0) is released after reduc-
tion. Within the framework of our development of a continuously measuring
analyzer for mercury bearing solutions, fundamental studies were conducted
in the laboratory. Experience has also been accumulated in practical appli-
cation in refuse incineration plants [10].
VARIANT 2: Reduction with Carbon
Gases containing mercury(II) are completely reduced to their elemental
form when passed through tempered carbon, e.g. in the form of activated
carbon, and through carbon bearing solids, respectively, such as a lime-
carbon mixture. The sample gas stream is drawn through such an adsorber tube
and, after conversion, the mercury is detected by means of resonance absorp-
tion. The hydrogen chloride concentrations normally encountered in the
gaseous effluents of refuse incineration plants affect the conversion reac-
tion. A suitable adsorber must be used to separate the HC1. For this pur-
pose, e.g., carbon may be mixed with an appropriate amount of lime and this
mixture may be used simultaneously as an HC1 adsorber and Hg converter .
VARIANT 3: Thermal Decomposition
The mercury(II) halides existing in the gaseous effluent or ambient air
are decomposed completely, forming elemental mercury, when passed through a
sufficiently heated temperature zone. Thermal decomposition can be felt
already at temperatures around 600 *C. Complete decomposition occurs at
reaction temperatures of 800 *C at gas retention times of only 0.4 s. The
presence of hydrogen chloride counteracts the decomposition reaction. The
mere presence of 100 mg of HCl/m3 more or less shifts the balance of forma-
tion back to the side of Hg(II). If HC1 is assumed to be present, this
component must be removed on all accounts. This can be achieved, e.g., by
means of a tempered HC1 adsorber, such as lime or silica gel, upstream.
However, the adsorber efficiency needs to be monitored permanently.
356
-------
Direct thermal decomposition in the heated absorption cell of a spec-
tral photometer, while the unit is being used for measurement, is possible
if sufficient retention times can be maintained and hydrogen chlorides can
be safely excluded. However, this is difficult to achieve technically.
VARIANT 4: Reaction with the Condensate
This procedure is characterized by the conversion of Hg(II), brought
about solely by the reducing components existing in the flue gas. These are,
above all, sulfurous acid, but also certain substances contained in the dust
[11]. Consequently, the reduction of Hg(II) can succeed only if the element
is brought into close contact with a condensate of the flue gas. For this
purpose, the sample gas stream is cooled down slightly, thus causing conden-
sate to form. The sample gas stream is passed through the condensate, as in
an impinger. The elemental mercury formed is directly expelled again with
the sample gas stream, entering the absorption cell of the spectral pho-
tometer. The reactor is equipped with an overflow to discharge the spent
condensate. This reactor design ensures that there is always fresh conden-
sate to maintain the reducing effect. This makes for a strictly continuous
mode of operation.
The reducing component(s) in the flue gas condensate, and their mecha-
nisms of action, will not be discussed in this paper. Suffice it to say that
this reducing property exists only in condensate in statu nascendi. In mod-
ern, smoothly functioning flue gas purification systems, the reducing effect
of the clean gas may no longer be sufficient. In that case, adding a small
amount of a reducing agent, such as a stannous chloride solution, is enough
to make the conversion complete.
CONTINUOUS MERCURY MEASUREMENT IN TECHNICAL PLANTS
ANALYTICAL COMPARISON BETWEEN CONTINUOUS AND DISCONTINUOUS MEASUREMENTS
In a test setup in the TAMARA refuse incineration pilot plant of the
Karlsruhe Nuclear Research Center, with a throughput of 200 kg of solid
municipal waste/h, the mercury monitor has been tested. The sample gas
stream was extracted through a heated quartz glass probe ana passed into the
cell of a commercial mercury analyzer (both after dust removal by means of
357
-------
quartz filter and, alternatively, directly) through the reactor in which the
condensate collects. For this purpose, the monitor developed for measure-
ments of maximum permissible work place concentrations by Seefelder Mess-
technik GmbH, with a measuring range between 0 and 1200 /jg/m3, was used. The
signal generated by resonance absorption was recorded and stored in a com-
puter. Figure 2 shows the test setup for continuous mercury measurement.
heating tube
stack
Hg-Monitor
DP
recorder
reactor
Figure 2. Test setup for continuous measurement
In parallel with this step, another partial gas stream was extracted
for discontinuous sampling and passed through solid adsorbers ensuring
precise speciation of the mercury as Hg(II) and Hg(0). The four-hour aver-
ages obtained in this way were to be used to monitor the concentrations
obtained on line.
Table 1 contains a summary of the results both of the discontinuous
measurements and of the data measured on-line as averaged by the computer.
These results were obtained in various TAMARA measurement campaigns, in
measurements in a sewage treatment sludge incineration plant, and in meas-
urements conducted in a technical scale incineration plant for solid munici-
pal waste. The good agreement between mercury concentrations in the flue gas
obtained continuously and those determined discontinuously is evident. This
proves that the conversion of Hg(II) to Hg(0) is complete during the contin-
uous sampling process. Figure 3 is a graphic plot of the data obtained.
358
-------
TABLE 1. COMPARISON OF MEASURED DATA (in /ig/m3)
(discontinuous, Hg(11)/Hg(0) with Dowex/iodized
activated carbon; continuous with Hg analyzer;
T=TAMARA, S=sewage sludge, M=municipal waste)
Test Date Time Hg discont. Hg cont.
No. (II) (0) total mean
T49
19.09.89
11-15
8
1
9
8
i
15
20.09.89
11-15
8
<1
9
10
t
16
21.09.89
7-11
7
1
8
9
±
3
22.09.89
9-13
8
1
9
9
t
2
T59
08.11.89
11-14
21
8
29
28
±
29
09.11.89
13-16
8
3
11
9
t
11
14.11.89
13-16
26
4
30
24
t
1
T10
27.03.90
7-9
8
7
15
18
t
4
28.03.90
10-13
18
60
78
92
t
161
29.03.90
11-14
14
28
42
49
t
55
T30
20.06.90
10-13
36
9
45
46
±
13
21.06.90
10-13
16
6
22
38
t
10
SOI
19.02.90
9-12
26
152
178
170
20.02.90
9-13
26
139
165
170
21.02.90
9-13
41
152
193
230
01.03.90
9-13
57
145
202
180
HOI
15.10.90
10-14
20
6
26
22
17.10.90
10-14
17
10
27
40
19.10.90
9-13
68
57
125
160
30.10.90
10-14
170
19
189
190
31.10.90
10-14
74
71
145
135
250
200
150
<2 100
O)
O)
150
200
250
100
0
50
Hg ug/m3 Adsorption tubes
E TAMARA ¦ Sewage sludge A Municipal waste
Figure 3. Analytical comparison of continuous versus
discontinuous measurements
359
-------
DISCUSSION OF THE CONTINUOUS MEASUREMENT SERIES
TAMARA Pilot Plant
Some typical series of clean gas measurements performed on the stack of
the TAMARA plant will be shown below. The plots are the curves of on-line
measurements of the flue gas scrubbing water in the first scrubber stage (Hg
in mg/1, scale 10 times enlarged) and the curves of the flue gas measurement
(Hg in ng/rt?) . The time resolution is 5 minutes; the measured points repre-
sent averages over the past 5 minutes.
Figure 4 very impressively shows that mercury is not distributed homo-
geneously in the refuse, but is detected as a discrete event. This had been
known from earlier continuous measurements in the flue gas scrubbing water
Mercury Emission
TAMARA
400
350
300
o>
250
E
O)
X
X
200
o
¦»—
m
b
150
O)
n>
X
100
50
0
29.03.90
V
|
A
1 An
[ 1
Hp
u
Ul
V
Time of day
Figure 4. On-line measurement curves showing mercury
concentrations in the scrubber and at the stack
360
-------
Mercury Emission
TAMARA
600
500
400
300
200
100
27.03.90
28.03.90
1
1^:
tat
K
O)
E
O)
X
x
o
00
E
O)
a.
O)
X
Time of day
Figure 5. On-line measurement curves showing mercury
concentrations in the scrubber and at the stack
of a refuse incineration plant [10]. However, the measurements performed in
the flue gas show the mercury peaks even more clearly. Sometimes peak levels
exceeding 500 /xg/m3 were measured. Deflections of the scrubber signal fol-
low, much broadened and with long decay times.
The example of two days of measurements is to show the doubtful nature
of discontinuous random sample measurements. On March 27, 1990 (Fig. 5,
measuring time 7-9 a.m.), a very low emission level (discontinuous) of
15 /xg/m3 was obtained. It is obvious that a minimum emission level was sam-
pled on that occasion. On the next day, March 28, 1990 (Fig. 5, measuring
time 10 a.m. - 1 p.m.), the value measured of 78 /xg/m3 turned out to be much
higher. Continuous measurement in this case reveals peaks existing during
the measuring period. So, discontinuous mercury measurements do not lend
themselves to emission monitoring, nor to producing information about the
effectiveness of flue gas purification equipment.
361
-------
Mercury Emission
TAMARA
CD
E
CD
x
x
o
co
E
o>
CD
x
G00
500
100
300
200
100
20.06.&
1
\ JL A
n Lv j
wb
T
(S
(S
Time of day
Figure 6. On-line measurement curves showing mercury
concentrations in the scrubber and at the stack
The law prescribes mean values obtained over one half hour and two
hours for assessing the emission situation. The effects of the peaks will be
shown below, again by the example of one day of measurement. One June 20,
1990, the sampling period considered (Fig. 6, measuring time 9-11 a.m.)
includes a five-minute mean of 550 ng/rn^. The computer produces these half-
hour mean values:
9:00- 9:30
9:30-10:00
10:00-10:30
10:30-11:00
60 ± 97
176 ± 187
52 ± 10
42 ± 2
This leads to a two-hour mean value of 77 + 112 /jg/m^. The occurrence o^
high peak emissions for very short periods of time thus does not necessarily
have to be a source of concern.
362
-------
Sewage Treatment Sludge Incineration Plant
The findings made in emission measurements at an incineration plant of
sewage treatment sludge are quite different (Fig. 7; the time resolution is
1 minute). Little had been known so far about mercury in connection with
such plants. Our measurements indicate that mercury is present practically
in a homogeneous distribution. The pronounced fluctuations, with peak emis-
sion levels, customarily found in the incineration plants of solid municipal
waste do not exist in this case. The concentration level in the flue gas of
incineration plants for sewage treatment sludge is equal to that found in
incineration plants for solid municipal waste. Consequently, the flue gas
purification systems must meet the same requirements. In contrast to solid
municipal waste incineration plants, elemental mercury is detected already
as a constituent of the unfiltered gas. This is due in particular to a
different HC1/S02 ratio in the flue gas.
Mercury Emission
Sewage Sludge Incineration Plant
3001
28.02.90
200-
100-
1030
1&00
Time of day
19:00
13:00
Figure 7. On-line measurement curves showing mercury
concentrations at the stack
363
-------
Solid Municipal Waste Incineration Plant
Our measurements performed in a technical scale plant proved for good
that mercury emissions are detected completely by the mercury monitor. In
the course of the measurement campaigns, the flue gas purification system
was operated in two modes so as to allow the optimum setting for mercury
removal in the scrubber to be found (see Fig. 3). Figure 8 shows the record
of one day. The results indicate that the legal limits can be observed reli-
ably provided that an instrument for continuous emission monitoring is
available.
Mercury Emission
Municipal Waste Incineration Plant
1501
30.10.90
100-
eo
ji
^5)
o>
X
50-
154)0
18:00
21:00
04)0
Time of day
Figure 8. On-line measurement curves showing mercury
concentrations at the stack
364
-------
CONCLUSION
The measured results obtained so far allow the statement to be made
that continuous measurement of mercury emissions in refuse incineration
plants is possible. The conversion of mercury chloride into elemental mer-
cury required for this purpose seems to be fully possible by passing the
substance through flue gas condensate. The measuring results obtained on two
municipal solid waste and on one sewage treatment sludge incineration plants
show that the mercury monitor is a highly sensitive and selective continu-
ously working instrument for mercury emission monitoring. A commercially
viable piece of equipment is being developed jointly with the licensee.
The work described in this paper was not funded by the U.S. Environmental Protection Agency and therefore
the contents do not necessarily reflect the views of the Agency and no official endorsement should be
inferred.
REFERENCES
[1] H. Braun, M. Metzger, H. Vogg:
Zur Problematik der Quecksilber-Abscheidung aus Rauchgasen
von Mullverbrennungsanlagen.
Mull und Abfall 18, 62-71 and 89-95 (1986).
[2] H. Braun, M. Metzger, H. Vogg:
Die Verbesserung der Abscheidung von Quecksilber aus Rauchgasen
der Abfallverbrennung.
in: K.J. Thome-Kozmiensky: Mullverbrennung und Umwelt 2, EV-Verlag
fur Energie- und Umwelttechnik GmbH, Berlin (1987), pp. 532-553.
[3] H. Braun, M. Metzger:
In-situ mercury speciation in flue gas by liquid and solid
sorption systems.
Chemosphere 16, 821-832 (1987).
and in: MeBtechnik bei der Mullverbrennung, Schriftenreihe
VDI-Kommission Reinhaltung der Luft Vol. 8 (1988), pp. 145-171.
[4] H.B. Cooper, G.D. Rawlings, R.S. Foote:
Measurement of mercury vapor concentrations in urban atmospheres.
ISA Transactions 13, 296-302 (1974).
365
-------
[5] F.P. Scaringelli, J.C. Puzak, B.I. Bennett, R.L. Denny:
Determination of total mercury in air by charcoal adsorption
and ultraviolet spectrophotometry.
Analyt. Chem. 46, 278-283 (1974).
[6] E. Ljungstrom, B. Hall:
Speciation of mercury in flue gas.
Environ. Sci. Technol. 24, 108-111 (1990).
and in: MeBtechnik bei der Miillverbrennung, Schriftenreihe
VDI-Kommission Reinhaltung der Luft Vol. 8 (1988), pp. 173-182.
[7] J.D. Herbell, P. Luxenberg, D. Ramke:
Betriebsergebnisse der 3. Stufe der Rauchgas-Reinigungsanlage
der SAV Ebenhausen.
GVC-FachausschuBsitzung Abfallbehandlung, Bad Soden, Nov. 10-11, 1988.
[8] T. Murakawa:
Simultaneous Hg and N0X removal in wet scrubber.
in: K.J. Thome-Kozmiensky: Miillverbrennung und Umwelt 3, EV-Verlag
fiir Energie- und Umwelttechnik GmbH, Berlin (1989), pp. 479-487.
[9] Y. Ogaki, Y. Fujisawa, T. Miyachi, J. Yoshiy:
Mercury removal from flue gas in municipal solid waste incineration
plants.
in: K.J. Thome-Kozmiensky: Miillverbrennung und Umwelt 3, EV-Verlag
fiir Energie- und Umwelttechnik GmbH, Berlin (1989), pp. 489-497.
[10] H. Braun, R. Seitner:
Mercury analyzer for continuous surveillance of the waste water
of refuse incineration plants.
Proc. ISWA '88 5th Int. Solid Wastes Conference, Copenhagen
(1988), Vol. 2, p. 319
[11] P. Dransfeld, H. Braun:
Abscheidung von Quecksilber.
GVC-Vortrags- und Diskussionstagung Entsorgung von Sonderabfallen
durch Verbrennung, Baden-Baden, Dec. 4-6, 1989.
366
-------
The work described in this paper was not funded by the U.S. Environmental
Protection Agency. The contents do not necessarily reflect the views of the
Agency and no official endorsement should be inferred.
A NEW METHOD FOR SAMPLING HALOGENATED DIOXINS
AND RELATED COMPOUNDS IN FLUE GASES.
by: Marklund, S., Fangmark, I. and Rappe, C.
Institute of Environmental Chemistry.
University of Umea, S-901 87, Sweden.
ABSTRACT
A new sampling method has been constructed that cools the
flue gas directly in the stack to avoid chemical formation
and/or degradation reactions. The sampling method is optimized
for adsorption of organic micropollutants in gases, particulates
and in aerosols. The sampler is simple and easy to handle. The
cooled probe, the impinger system, and the aerosol trap are as-
sembled on a trolley running on a rail. This means that it is
very easy to use a traversing procedure on a cross-sectual area
of the duct. The probe can be equipped with a buil^-in pitot.
The sample velocity can be adjusted to equal the gas velocity in
the duct during the whole sampling period and at every sampling
point. Another advantage is the ability to measure hot gases.
With a titanium probe and a quartz glass line flue gases with
temperatures up to 1100'C can be sampled.
INTRODUCTION
Polychlorinated dibenzo-p-dioxins (PCDDs) and polychlori-
nated dibenzofurans (PCDFs) as well as other chlorinated pollu-
tants in stack emissions from municipal solid waste (MSW) incin-
erators have become a matter of scientific and public concern
during the last years. The maximum emission of PCDDs and PCDFs
from MSW incinerators in Sweden has been legislated to 0.1 ng/m^
(norm. d.g. 10% CO2) based on International Toxicity Equivalents
(I-TEQ). High requirements must be met during the flue gas sam-
pling procedure and also by the analytical techniques used to
measure each toxic PCDD/F isomer with high specificity and a
367
-------
sensitivity, to a level below 10 pg/rn-^. Currently most countries
use their sampling procedure for measuring PCDDs and PCDFs in
flue gases from combustion sources. Data from different coun-
tries are therefore often compared without any consideration to
different sampling, analytical, and calculation methods.
The partitioning of PCDDs and PCDFs between the vapor and
the particle phase is very dynamical in nature. It depends on
the partial pressure of the compounds, the temperature of the
flue gas, the available surface area, and the nature, of the fly
ash particles. Thus the partitioning ratio will shift toward
more particle bound PCDDs and PCDFs in the cooling regions
within the incinerator and smoke stack. The ratio is also chang-
ing when the temperature drops in the sampling train. A conden-
sation aerosol is formed in the condenser when the equilibrium
vapor concentration (EVC) at the temperature of the condenser is
exceeded. This condensation aerosol is of physical dimention
that most easily passes through sampling equipment (around 0.5
pm aerodynamic diameter). An efficient stack sampling method for
PCDDs and PCDFs must collect both the vapor phase and the par-
ticle bound components, as well as the condensation aerosol,
that could be formed in the sampling train. Failure to collect
the particle bound PCDDs and PCDFs will result in an underesti-
mation of the PCDD/F emission. Futhermore failure to retain pre-
sampling spikes will have a negative effect on the method accu-
racy due to difficulties in correction of these losses.
Practical "in-the-field" experiments (1,2' have demonstrated the
following weaknesses with the existing flue gas sampling meth-
ods :
1. High temperature at the filter house (as in the EPA sampling
train 120-160"C). This can cause formation and/or degrada-
tion reactions of the native and labelled PCDDs and PCDFs
isomers on the filter yielding erroneous results.
2. Very few sampling methods are optimized for collection of
aerosols (3) .
3. Sampling trains in wich most of the PCDDs and PCDFs are
found in the rinsing fraction (US-EPA method 23) had to be
treated with extreme care, otherwise losses of sample during
the rinsing procedure would occur.
4. Most sampling methods were unnecessarily complicated and
difficult to use in practice.
368
-------
METHODS
A new sampling apparatus with which most of the above prob-
lems are remedied has been constructed (3), and is shown in Fig-
ure 1.
TO PUMP
PUFP-ADSORBENT
225a AEOROSOL FILTER
m ¦¦¦ ¦!¦¦¦" I WIWII ...II
PUFP-ADSORBENT
WATER COOLED
GLASS PROBE
CONDENSATE
IMPINGER
COLLECTION
FLASK
STACK
Figure 1. Schematic drawing of the modified sampling train.
The advantages of this samplin strategy include:
1. Rapidly cooling of the flue gas to 5'C in the front end of
the probe. This eliminates formation/degradation reactions
in the sampling train.
2. The probe can be equipped with a pitot that improves the
ability for isokinetic sampling.
3. The sample train is assembled on a movable trolley which
thereby makes it possible to sample from a multitude of
points within the stack.
369
-------
4. Optimization for aerosol adsorption. The second impinger has
a nozzle and a water soluble solvent, ethylene glycol, which
increases the adsorption of small particles. The combination
adsorbent - aerosol filter - adsorbent also improves the
collection of small volatile aerosols.
5. Easy in handling. The compartments are composed of four
glass components: the glass probe, two impingers, and a
glass housing for two polyurethane foam plugs with an
aerosol filter between them.
TABLE 1 PROPOSED SPIKING PROTOCOL FOR SAMPLING OF PCDDS AND PCDFS
IN STACKS, WITH EMISSION RESULTS COMPENSATED FOR INCOM-
PLETE SAMPLING RECOVERY.
Sample spike
Internal standard
MS group 1
Clean up spike
MS group 1
Syringe spike
Recovery standard
MS group 1
13
13
13C12-2,3,7,8-TCDD
C12_l,2,3,7,8-PeCDD
C12-l,2,3,6,7,8-HxCDD
13,
13
37C1 2,3,7,8-TCDD
C12~l,2,3,7,8,9-HxCDD
C12-2,3,7,8-TCDF
13c12_2,3,4,7,8-PeCDF
13c12_1'2'3'6'7'8_HxCDF
13
C12~l,2,3,7,8-PeCDF
MS group 2
MS group 2
MS group 2
13
13
c12-1'2'3-4'6,7,8-HpCDD
13c12-2.3,4,6,7,8-HxCDF 13C12-1,2,3,7,8,9-HxCDF
C12-1,2,3,4,6,7,8-HpCDF
13
C12-l,2,3,4,7,8,9-HpCDF
MS group 3
MS group 3
MS group 3
13,
13,
-12
-OCDD
Br,CI-DD or -DF
-12
-OCDF
CONCLUSIONS
Validation studies between
recommended sampling methods in
al(l,2) show comparable results
chlorinated components. But the
this sampling method and other
Europe presented by Marklund et
both for PCDDs, PCDFs, and other
studies took place before the
370
-------
sampling method was modified with the aerosol filter, so losses
of small particles may have occurred in all sampling trains.
One important conclusion from the two validation studies of
sampling techniques is identification of the need for an appro-
priate spiking protocol. If, as in the Scandinavian countries,
the regulation prescribes emission data, compensated for sam-
pling recoveries(4), as many labelled PCDD and PCDF congeners as
possible, should be included in the sampling spike. Table 1
shows the spiking protocol used at the University of Umea.
Another conclusion is that the sampling spike should be ap-
plied in the first cold region in the sampling train. Experi-
ence from the validation studies clearly demonstrates, that
spikes added on the glasswall inside the cooled probe can vary
largely and are not related to the losses of the sampling spike
applied in the cold part of the sampling train, nor to the
losses of native isomers. Explanations may be attributed to
losses in the evaporation of the spike mixture solvent, irre-
versible adsorption onto the glass walls, or backflushing when
the probe is introduced into the stack.
REFERENCES
1. Marklund, S., Rappe, C., Soderstrom, G., Ljung, K.,
Aittola, J.-P., Vesterinen, R., Hoffren, H., Benestad, C.,
Jebens A., and Oehme, M. "Internordic" Method Calibration
for Sampling and Analysis of Dioxins and other Chlorinated
Organic Compounds. In: Proceedings of the International
Conference on Municipal Waste Combustion Volume 1. Minister
of Supply and Services., Canada. 1989 p.14.
2. Marklund, S., Soderstrom, G., Ljung, K., Rappe, C.,
Kraft M. and Hagenmaier H.-P. Parallel Sampling Using Vari-
ous Sampling Techniques at a Swedish MSW Incinerator.
Accepted for publ. in Waste Management & Research 1991.
3. Fangmark, I., Wikstrom, L.-E., Marklund, S. and Rappe, C.
Studies on Sampling Methods for Particle Bound PCDD and PCDF
in Stack Emission. Chemosohere 20: 1339, 1990.
4. Jansson,B. and Bergwall, G. Recommended Methodology for Mea-
surement of PCDD and PCDF in the Nordic Countries. Waste
Management & Research 5: 251, 1987.
The work described in this paper was not funded by the U.S.
Environmental Protection Agency and therefore the contents do
not necessarily reflect the views of the Agency and no official
endorsement should be inferred.
371
-------
Intentionally Blank Page
372
-------
SESSION 10A: SAMPLING AND ANALYSIS II
Co-Chairman:
Stellan Markland
Institute of Environmental Chemistry
University of Umea
Umea, Sweden
James D. Kilgroe
AEERL
U.S. EPA
Research Triangle Park, NC
Preceding page blank
373
-------
Intentionally Blank Page
374
-------
The work described in this paper was not funded by the U.S. Environmental
Protection Agency. The contents do not necessarily reflect the views of the
Agency and no official endorsement should be inferred.
Assessment of an On-Line CI-Mass Spectrometer as a
Continuous Emission Monitor for Sewage Sludge Incinerators
K. R. Campbell, D. J. Hallett and R. J. Resch
ELI Eco Technologies Inc.
14 3, Dennis Street, Rockwood, Ontario, CANADA NOB 2K0
and
J. Villinger and V. Federer
V&F Analyse-Und Messtechnik GMBH
A-6060 Absam, Maderpergestrasse 18, Austria
ABSTRACT
ELI Eco Technologies Inc. tested two sewage sludge
incinerators using regulatory methods and a V&F CIMS-500 chemical
ionization mass spectrometer. Correlations between dioxins and
dibenzofurans from the regulatory MM5 trains and the continuous
readings from the CIMS-500 for chlorobenzenes and chlorophenols
were noted. As well, correlations between chlorinated organics and
other volatile organics were obvious under poor combustion
conditions.
ELI Eco Technologies Inc. recently completed an extensive
survey of organic chemical emissions including VOCs,
chlorobenzenes, chlorophenols, chlorinated dioxins and
dibenzofurans from two sewage sludge incinerators. The program was
funded by the Municipality of Metro Toronto, Environment Ontario,
and Environment Canada. Contaminants were measured by regulatory
methods (ASME Modified Method 5) and simultaneously with the
continuous mass spectrometer. The purpose of the study was to
provide regulatory testing and at the same time evaluate the
usefulness of the CIMS-500 mass spectrometer in assessing
emissions.
Continuous monitoring of emission concentrations at two sewage
slide incinerators with the V&F CIMS-500 mass spectrometer
correlated well with data obtained using regulatory methods. The
most important difference found between the regulatory method (ASME
Modified Method 5) and measurement with the CIMS-500 was that the
results were available immediately using the continuous mass
spectrometer system, and the higher emissions from one day to the
next were noted at the time, when some corrective action could have
been taken.
Preceding page blank
375
-------
The regulatory method requires triplicate sampling for day-
long periods, lengthy delays for lab analysis, and yields only one
number per test for each contaminant. With current regulatory
testing, it is difficult to establish correlations between
incinerator operating conditions and stack emissions, let alone
attempt to correct them. Continuous emission monitoring with the
CIMS-500 would make it possible to optimize combustion by
measurement of various hydrocarbons, reduce dioxin and dibenzofuran
emissions by monitoring chlorobenzene and chlorophenol emissions,
and track emissions year-round.
Three days of testing at a sewage sludge incinerator in
January 1990 showed chlorinated dioxin and dibenzofuran emissions
that can respectively be called low, high and medium, relative to
each other. The CIMS-500 data for both chlorobenzene and
chlorophenol concentrations correlated well with that trend.
Although the mass spectrometer system cannot measure the extremely
low concentrations of dioxin and dibenzofuran emissions, it can
measure the part per billion levels of chlorophenols that occur in
conjunction with dioxins and dibenzofurans, thus providing an
indirect measurement of dioxin and dibenzofuran emissions. Figure
1 shows correlations of dioxin and furan concentrations by ASME MM5
versus chlorophenol and chlorobenzene concentrations by CIMS-500.
A slightly better correlation results using the chlorobenzene data.
Although continuous data is available for the chlorobenzene and
chlorophenol concentrations, the correlations are limited by the
three data points available from the ASME MM5 trains for dioxins
and dibenzofurans.
Other correlations were noted between chlorophenol and
chlorobenzene concentrations and those for benzene, toluene,
xylene, and phenol. These correlations reflect the poor combustion
conditions that occurred on the test day with the highest dioxin
emissions. The continuous analysis of organic emission with an
installed mass spectrometer system would therefore allow the
incinerator operator to optimize combustion using benzene, toluene,
and xylene readouts, and further optimize to achieve the lowest
dioxin and dibenzofuran emissions by measurement of chlorobenzenes
and chlorophenols. As well, a continuous record of emission data
would be produced all year, rather than three days per year.
Figure 2 shows the continuous monitoring of benzene, toluene
and xylene, with dichlorophenol (DCP), trichlorophenol (TCP) and
tetrachlorophenol (TeCP) on the first graph and with chlorobenzene
(CB) , dichlorobenzene (DCB) and trichlorobenzene (TCB) on the
second graph. Note that the benzene, toluene and xylene
concentrations are shown 10 times reduced. It is quite obvious
from this graph that there are correlations between the compounds
that were measured. Note that the feed to the incinerator was off
at 12:10 and at a reduced rate at 13:30. Bringing the system back
on line probably contributed to the high results obtained. Breaks
in the continuous monitoring correspond to instrument calibration
periods.
376
-------
In summary, continuous emission monitoring using the CIMS-500
mass spectrometer has been successful in comparison with ASME MM5
sample train measurement and should be useful in optimizing
combustion conditions, minimizing emissions of chlorinated dioxins
and dibenzofurans, and providing continuous emission data for
incinerators.
377
-------
DIOXINS & FURANS vs DICHLOROPHENOL
IASVC MMS TRAIN v. CAG-€QO MONITOR]
2JS -
Total CDF
2£ -
2.4 -
T4CDF
18 -
U> -
Total CSD
14 -
U -
T-4CDD
PSCDF
aa -
as -
02 -
atM
ass
aoE
a
Dicfaloropbvtwl CaKcstndM (ppmj
DIOXINS & FURANS vs DICHLOROBENZENE
(ASVC MM5 TRAIN vm CIMS-500 MONITOR)
23 -
Total CHF
2£ -
2.4 -
22 -
2 -
18 -
IS -
14 -
12 -
T4CDD
1 -
06 -
0.4 -
02 -
004
01
006
0
PidiloriitMi n tii mi Cuufntrtloo (ppm)
37«
-------
SEWAGE SLUDGE INCINERATOR CIMS-500 DATA
09 -
OJ -
CL6 -
OS -t
"POL/TO
03 -
000
BOO
Jaooafy 22. B90
SEWAGE SLUDGE INCINERATOR CIMS-500 DATA
BTX ft CUnrnlwnwM CuateiUntioai
LB -r
17 -
LS -
IS -
L4 -
CB
13 -
L2 -
U -
as -
as -
07 -
as -
as -
0L4 -
03 -
02 -
01 -
TCB
BOO
1500
0)00
1200
1740
1400
January 22. B90
379
-------
Intentionally Blank Page
380
-------
SAMPLING AND ANALYSIS OF AIR TOXICS FROM MUNICIPAL WASTE COMBUSTORS
by: Justice Manning, P.E.
USEPA/ORD/CERI
Cincinnati, Ohio 45268
John Chehaske
Pacific Environmental Services, Inc.
Herndon, Virginia 22070
Paul Fraley, P.E., Dave Wetmore, and Eric Hoi Tins
Pacific Environmental Services, Inc.
Mason, Ohio 45040
ABSTRACT
The new federal rules addressing emissions from large municipal
incinerators will significantly increase the scope and complexity of periodic
emissions testing at these facilities. New and existing facilities with a
capacity of more than 250 tons per day will be required to test for metals
emissions, organics, and acid gases. Specific testing for mercury and other
metals may be addressed in a forthcoming proposed rule for municipal waste
combustors. This paper provides a brief introduction to the principal test
methods which are currently available to meet the expanded emission
measurement requirements. Test methods presented here include: EPA Method 5
(particulates), EPA Proposed Method 0012 (multiple metals), EPA Proposed
Method 23 (PCDD/PCDF), EPA Proposed Method 26 (hydrochloric acid), and a brief
summary of wet chemistry methods for oxides of sulfur and nitrogen.
This paper has been reviewed in accordance with the U.S. Environmental
Protection Agency's peer and administrative review policies and approved for
presentation and publication.
INTRODUCTION
Traditional emission sampling (or stack testing) methods have long been an
established fact of life for Municipal Waste Combustors (MWC). Stack testing
has usually been the best means of verifying compliance with a variety of
federal, state, and local regulations which stipulate maximum atmospheric
emission levels from waste incineration facilities. Stack testing is also an
important tool in assessing unit operating performance, control system
performance, and efficiency and accuracy of continuous emission monitors.
New and forthcoming federal regulations will add substantially to the
scope, frequency, and complexity of required testing. However, most of the
emission tests which are "new" to municipal waste combustion are established
test methods which have been previously developed and applied to the
measurement of emissions from other sources. In some cases, these methods
have been specifically modified for application to the testing of municipal
Preceding page blank
-------
waste combustors. This presentation is intended to provide a brief
introduction to the principal test methods currently available to meet the
expanding emission measurement regulations applicable to municipal waste
combustors.
Emissions from MWCs include a complex mixture of pollutants that can affect
public health. The three general subclasses of pollutants within MWC
emissions consist of metals, organics, and acid gases. In total, MWC
emissions contain 100 or more pollutant components. Although it is
theoretically possible to measure all of the components, such a task would be
extremely burdensome, expensive, and quite impractical. EPA has determined
that it is unnecessary to measure MWC emissions as an entity; it makes far
more sense to develop standards for certain components of MWC emissions. The
current list of regulated emissions from MWCs includes: particulates (PM),
MWC metals, dioxin/furans, sulphur dioxide (SO,), and hydrogen chloride (HC1).
For new MWCs, emissions of oxides of nitrogen (NO ) are also regulated. Of
these substances, continuous emission monitors (CEMs) are recommended (and in
some cases may be required) for S02 and N0X. Forthcoming proposed regulations
(MWCII) will address MWC metals and stipulate emissions testing methods, for
mercury, cadmium, and lead. Control of MWC organics is demonstrated through
the measurement of dioxin/furans and control of acid gases is demonstrated
through the measurement of HC1 and S02. Standardized emission sampling
methodologies exist for all of these pollutants.
A summary of available sampling procedures (including instrumental or CEM
methods) are presented in Table 1. The measurement methods are primarily
those in general use. Most are EPA methods from the Federal Register\2 or
"Test Methods for Evaluation of Solid Waste," SW-846.3 It is important to
note that, in practice, all sampling methods are in various stages of
development and subject to processes of evaluation and refinement. In
addition, state or local agencies may require specific procedures or
modifications to established methods. Before undertaking or contracting for
any type of emission testing, the current status of EPA methods, as well as
local regulatory requirements, should be confirmed.
All EPA methods are published in the Federal Register (40 CFR Part 60, 61).
Copies of individual methods are also available from the Emissions Measurement
Branch, USEPA, Mail Drop 19, RTP, NC 27711. The contact person for Method
0012 (multiple metals) is Tom Ward, (919) 541-3788. For Method 23
(PCDD/PCDF), contact Gary McAlister at (919) 541-1062. For Method 26
(Hydrochloric Acid), contact Foston Curtis at (919) 541-1063.
TYPES OF TESTING
Stack testing can be divided into two categories: continuous testing and
periodic testing. Generally, continuous testing is effective only for
constituent gases and stack gas opacity, although some particulate matter
samplers are available for specific processes and needs. Continuous emission
monitoring systems (CEMS) are used as an operations tool to monitor or
332
-------
TABLE 1. STACK SAMPLING METHODS
Pollutant
Pri nciple
Procedure
Part i culate
Isokinetic collection of a 1 hr sample on glass fiber
filter at 120 i 14° C. Train includes: heated
probe, optional cyclones, heated filter, impingers,
flow control and gas volume metering system.
Visual determination of opacity.
Instrumental measurement of opacity (optical
densi ty).
EPA Method 5, 17
EPA Method 9
Multiple Metals
(cadmium, lead)
Collection on glass fiber filter, nitric
acid/hydrogen peroxide and potassium permanganate
impi ngers.
EPA Proposed Method
0012
Mercury
Collection in iodine monochloride or acidic
permanganate impingers in M5-type train.
EPA Method 101A
PCDD/PCDF
Collection by sorption on XAD2 resin with MM5 train,
analysis by GC/MS
EPA Proposed Method 23
Hydrochloric Acid
Collection in acidified water impinger solution in
M5-type train or midget impingers.
EPA Proposed Method 26
SU-846 Methods
0050 or 0051
Sulfur Oxides
Instrumental using UV, NDIR, or fluorescence
Collection in isopropanol (SO^) and hydrogen peroxide
(SO.,) impingers of M5-type train.
EPA Method 6C
EPA Method 6,8
N i trogen Oxi des
Instrumental using chemiI luminescent analyzer.
Collection in evacuated flask containing sulfuric
acid and hydrogen peroxide.
Collection in impinger solutions of potassium
permanganate.
EPA Method 7E
EPA Method 7,7A,7B
EPA Method 7C,7D
maintain combustion air, ensure flue gas path integrity, and control
stoichiometric ratios in control systems. As analytical instruments, CEMS
provide a means of continuously determining the compliance status of the
source. Under the new MWC standards, CEMS are to be utilized for the
measurement of S02 and N0x emissions as well as to monitor ongoing adherence
to operating standards required of certain MWCs.
Periodic testing is conducted for regulatory compliance, unit performance
testing, calibration and validation of CEMS accuracy, and research and
development (R&D) studies. Periodic tests include both wet chemistry methods
and instrumental techniques. Wet chemistry methods are essentially sample
capture techniques which require off-site analysis and measurement of the
pollutants of concern. Instrumental tests, if they are available or feasible,
provide on-site analysis for target pollutants. For the pollutants discussed
in this paper, most of the readily available methods are wet chemistry
techn i ques.
383
-------
METHOD 5: MEASUREMENT OF PARTICULATES
The basic method for. measurement of particulate matter in stack gas,
whether for compliance or performance testing, is that specified in EPA Method
5 and the associated Methods 1-4.1 Moreover, Method 5 constitutes the basic
equipment and sampling methodology on which many other sampling methods are
based. Many of these tests can be described as a variation of Method 5 and,
in some cases (such as multiple metals) the sampling objectives can be
combined into a single, modified Method 5 which measures both particulates and
the additional pollutants.
SAMPLING EQUIPMENT AND PREPARATION
The Method 5 sampling train is shown schematically in Figure 1. The
essential elements are: a borosilicate or quartz-lined, temperature-
controlled probe equipped with a Pitot tube and thermocouple (for measuring
stack gas flow rate); a sharp-edged button hook nozzle; a glass or quartz-
fiber filter, supported in a glass or Teflon filter holder, inside an oven,
and immediately at the outlet of the probe; an impinger train or condenser to
remove water; and a pump and metering system.
Hardware suitable for use in EPA Method 5 is commercially available from a
number of suppliers, and no reason exists to consider a nonstandard train
configuration for compliance/performance measurement of particulate matter.
Quartz glass probe liners can be used if the temperature does not exceed
900°C(1650°F). Water-cooled probes with quartz or borosilicate glass liners
should be used if the stack temperature is expected to exceed 900°C; at
temperatures below 480°C, borosilicate glass can be used without cooling. The
metal probe option allowed in Method 5 is not recommended for MWC testing, due
to the potential for corrosion.4
EPA Method 5 operationally defines particulate matter as any material that
is collected on the filter when stack gas is withdrawn isokinetically through
a temperature controlled, glass lined probe and high efficiency, glass-fiber
filter. If the only purpose for which the sample is to be used is
determination of particulate matter emissions, regular or quartz glass may be
chosen as the fiber filter medium. Note, however, that if chemical analysis
procedures, e.g., metals, are to be applied to the collected particulate
material after its mass has been gravimetrically determined, the filter medium
selected must be shown to be free of analytical interferences and/or
background contamination. Component cleaning and system preparation is
described in detail in Method 5.
SAMPLING/SAMFLE VOLUMES
Procedures for selecting sampling locations and for operation of the train
are specified in detail in Method 5 and associated Methods 1-4. The procedure
indicates a sampling rate of 0.5 to 1.0 cubic feet per minute (cfm) or 14 to
28 liters per minute (1pm) for a minimum of 1 hour total sampling time. The
384
-------
1. Sampling nozzle
2. Sampling probe sheath
3. Heated sample probe liner
4. Cyclone assembly (optional, for heavy particulate load situations)
5. Out of stack filter assembly
6. Heated filter compartment
7. Impinger case
8. First impinger filled with H^O (100 ml)
9. Greenburg-Smith (or modified Greenburg-Smith) impinger filled with
HpO (100 ml)
10. Tnird impinger - dry
11. Fourth impinger - filled with H20 absorption media (200-300 gm)
12. Impinger exit gas thermometer
13. Check valve to prevent back pressure
14. Umbilical cord - vacuum line
15. Pressure gage
16. Coarse adjustnent valve
17. Leak free pump
18. By-pass valve
19. Dry gas meter with inlet and outlet dry gas meter thermometer
20. Orifice meter with manometer
21. Type S pitot tube with manometer
22. Stack temperature sensor
Figure 1. EPA Method 5 Particulate Sampling Train
385
-------
minimum total sample size is, thus, 30 cubic feet (ft3) or 0.85 cubic meters
(m3). Sampling is performed isokinetically.
Depending on the particulate matter emission standard with which a given
facility must comply, a sampling period longer than the 1 hour indicated in
the Method 5 protocol may be necessary. For example, at an emission rate of
0.03 grains per dry standard cubic foot (gr/dscf), the limit specified in some
regulations, a larger sample of stack gas may be required to ensure that the
precision of the compliance/performance determination is adequate. The
necessary stack gas sample size (and sampling time) for a given test must be
calculated by dividing the minimum weighable particulate mass (taking into
account the filter tare weight) by the particulate mass emission rate
specified in the permit. A rule-of-thumb is that the particulate weight
should be at least 1% of the filter tare weight.
ANALYSIS
Analytical procedures for particulate analysis are described in detail in
Method 5. The exposed filter is desiccated to dryness and weighed. The
acetone probe/nozzle wash is transferred to a tared beaker, evaporated to
dryness and weighed. The net weight gains of the filter and beaker are summed
and divided by the sample volume to obtain a concentration. The emission rate
is the product of the concentration and the stack gas flow rate. The standard
expression of concentration is grains per dry standard cubic foot (gr/dscf) or
milligrams per normal cubic meter (mg/Nm3). Emission rates are reported as
pounds per hour (lb/hr), kilograms per hour (kg/hr), or pounds per pound of
throughput (lb/lb).
Some agency regulations may require analysis of the impinger contents for
condensible particulates. Quantification of condensible vapors passing
through the filter, or demonstration of the integrity of the filter during
sampling may be possible with this analysis. Either an extractive procedure,
to separate solids from crystallized compounds, or an evaporative routine may
be required.
QUALITY ASSURANCE AND QUALITY CONTROL PROCEDURES
The comprehensive QA/QC procedures for Method 5 can be found in the USEPA
"Quality Assurance Handbook for Air Pollution Measurement Systems: Volume
III."6 Procedures are specified in Reference Method 5 that must be used to
calibrate and leak check the sampling equipment. The analytical balance used
for the gravimetric determinations also must be calibrated regularly.
In applying Method 5, a QC blank sample, consisting of an aliquot of the
acetone solvent used for sample recovery must be analyzed. The method does
not require a blank filter QC sample. However, generation of a field blank
filter is standard practice. To ensure comparability with the sample filters,
this blank may be mounted in a filter holder and held at 120° + 14°C for a
time equal to the total sampling run time. Although not stipulated in the
Federal Register, standard practice is to perform Method 5 determinations in
triplicate (three separate stack sampling runs).
386
-------
The equations to be used to calculate the percent isokineticity and the
particulate matter emission rate are presented in Method 5. Isokineticity
must be in the range of 90% to 110%.1 Some agencies require that this
objective be met for each sampling point, or for a specified fraction
(percent) of the sampling points, while others require that only the average
isokineticity over each run meet the objective.
Data quality objectives for precision of replicate determinations are not
included in Method 5. Usual practice is to compute the mean of the three
separate determinations and compare this mean value with the emission
limitations. Some agencies may require that each of the three measurements
fall within the emission limitations.
LIMITATIONS OF THE TECHNIQUE
The greatest limitation of Method 5 is the batch nature of the sampling.
Continuous sampling is impossible, and even frequent sampling is economically
and logistically prohibitive. In the case of performance and R&D testing, the
time delay between sampling and analysis can create a problem, as a 24-48 hour
minimum time lag occurs for results.
Some tests may require sampling upstream of control devices; consequently,
additional considerations and method modifications may be necessary. If, for
example, a Method 5 traverse is impossible, a traverse for velocity only may
be made, and isokinetic sampling may be performed at a single representative
point away from the duct walls. High temperatures and corrosivity can also be
a problem. Water cooled probes or exotic metal probe liners, i.e. inconnel®,
may be necessary. These options must be approved by the appropriate
regulatory agency for compliance testing.
ALTERNATIVE TECHNIQUES
EPA Method 171 may be used as an alternative to Method 5 in some
situations. The basic difference in the two methods is that the Method 17
filter is contained in a stainless steel holder, which is attached to the
front of the probe. This places the filter in the stack and defines the
particulates at stack conditions, i.e. temperature and pressure. Operating
procedures are exactly the same as for Method 5.
No other alternative particulate evaluation methods will generate the
precise information determined by Methods 5/17. However, some devices exist
which enable the continuous monitoring of relative emissions. The most common
of these is the opacity monitor. This instrument has a transmitter mounted on
one side of the duct, and a photodetector or reflector on the opposite side.
A light beam is projected between the two, and the dim nished intensity of the
beam reaching the photodetector determines the percent opacity of the stack
gas. Applicable agency regulations may require an opacity monitor. A means
of visually determining opacity is presented in EPA Method 9.1
387
-------
METHOD 0012: MULTIPLE METALS
DESCRIPTION OF METHOD
The most comprehensive means of sampling for metals is the USEPA proposed
Method 0012, "Methodology for the Determination of Metals Emissions in Exhaust
Gases from Hazardous Waste Incineration and Similar Combustion Processes."12
This method is applicable for simultaneously determining emissions of: total
chromium (Cr), cadmium (Cd), arsenic (As), nickel (Ni), manganese (Mn),
beryllium (Be), copper (Cu), zinc (Zn), lead (Pb), selenium (Se), phosphorus
(P), thallium (Tl), silver (Ag), antimony (Sb), barium (Ba), and mercury (Hg).
The basic sample system is a Method 5 particulate train operated in the
same manner as for particulate determinations. The stack sample is extracted
isokinetically through a glass lined probe, glass-fiber filter, absorbing
solutions of dilute nitric acid in hydrogen peroxide (HNOj/H^) in two
impingers, and acidic potassium permanganate (KMnOJ in one (or two)
impingers. Particulate emissions can be determined by recovering the filter
and front half of the sample train for gravimetric analysis. Metals are
determined by laboratory analyses of the particulate catch and the impinger
solutions, using appropriate detectors and analytical equipment.
SAMPLING EQUIPMENT AND PREPARATION
The sampling equipment is basically the same as for Method 5. The nozzle,
probe, and filter holder are of borosilicate or quartz glass construction, and
the filter is of quartz-or glass-fiber, supported in the holder by a Teflon
frit. The condensing system consists of four to six impingers. When high
moisture levels are expected, the first impinger will be empty and serve only
as a knock-out for H?0. The first two absorbing impingers contain a solution
of 5% nitric acid (HN03) and 10% hydrogen peroxide (H202) in water. The
following impingers (one or two) contain an absorbing solution of 4% potassium
permanganate (KMnOJ in 10% sulfuric acid (H2S0J. The KMn04 solution must not
contact the glassware to be analyzed for Mn ana the two absorbing solutions
must not be mixed. The last impinger contains a drying agent to aid in
determining moisture levels in the stack and to protect the sample pump and
meter. A standard metals sampling train is shown in Figure 2.
Metals train glassware must be thoroughly washed and soaked in nitric acid
before use. Detailed cleaning procedures are given in proposed Method 0012.
SAMPLING/SAMPLE VOLUMES
The sample train should be operated exactly as a Method 5 particulate
train. Great care must be taken to ensure no mixing of the impinger
solutions during the leak checks. The use of a glass nozzle also merits
caution in handling the train, especially when changing ports. Sampling times
should be sufficiently long to capture minimum amounts of the target metals.
Generally, sampling for two (2) hours at a rate of 14 to 28 liters per minute
is recommended. The metals and their in-stack detection limits are shown in
388
-------
Alt glata aampla axpoaad aurfact lo hara.
(Excapt whan Taflon Altar aupporl la uiad.).
LO
00
Tharmoooupla chtd(
•J* Vatv*
Thamiocoup)
Glaaa prob# Rnac
Probt
RavaraaTyp#
Pltol Tuba
i
Empty (Optional Knocfcoul)
b%hno3/io%h2o2
4XKmiO 4/IO%H2S04
Bypaat
Valva
ThtrmooouplM
Figure 2.
Standard Metals Sampling Train
-------
Table 2. These should be used when calculating sample volumes, to ensure that
the analytical limit is less than the emission limit.
The sample recovery procedures must be performed with care. Five
individual samples are recovered:
Fi1ter;
Acetone probe wash;
Front half rinse - front of the filter holder, the nozzle, probe, and
any connecting glassware are rinsed with exactly 100 ml of 0.1N
nitric acid;
Impingers 1 and 2 (and 3 if 1 is a knockout) and back half - recover
and measure the contents of the impingers and rinse the back half of
the filter holder, connecting glassware, and impingers with exactly
100 ml of 0.1N nitric acid.
TABLE 2. IN-STACK METHOD DETECTION LIMITS (ug/m3)
FOR TRAIN FRACTIONS USING ICAP AND AAS
Metal
Front Half/Fraction 1
Probe and Filter
Back Half/Fraction 2
Impingers 1-3
Fraction 3
Impingers 4-5
Total Train
Antimony
7.7 (0.7)*
3.8 (0.4)*
11.5 (1.1)*
Arseni c
12.7 (0.3)*
6.4 (0.1)*
19.1 (0.4)*
Bari um
0.5
0.3
0.8
Beryl Ii um
0.07 (0.05)*
0.04 (0.03)*
0.11 (0.08)*
Cadmium
1.0 (0.02)*
0.5 (0.01)*
1.5 (0.03)*
Chromi um
1.7 (0.2)*
0.8 (0.1)*
2.5 (0.3)*
Copper
1 .4
0.7
2.1
Lead
10.1 (0.2)*
5.0 (0.1)*
15.1 (0.3)*
Manganese
0.5 (0.2)*
0.2 (0.1)*
0.7 (0.3)*
Mercury
0.05**
0.03**
0.03**
0.11**
Nickel
3.6
1.8
5.4
Phosphorus
18
9
27
Selenium
18 (0.5)*
9 (0.3)*
27 (0.8)*
SiIver
1.7
0.9
2.6
Thai Iium
9.6 (0.2)*
4.8 (0.1)*
14.4 (0.3)*
1 i nc
0.5
0.3
0.8
( )* Detection limit when analyzed by GFAAS.
"Detection limit when analyzed by CVAAS.
390
1.
2.
3.
4.
-------
5. Impingers 3 and 4 (or 4 and 5 if 1 is a knockout) - recover,
measure the contents of, and rinse the impingers and
connecting glassware with exactly 50 ml of 8N HC1.
Detailed recovery procedures are given in the EPA proposed Method 0012.7
All sample containers must be pre-cleaned, using the glassware cleaning
procedure in the Method. Sample volumes should be marked and containers
capped and sealed with Teflon tape for transport to the laboratory.
ANALYSIS
The filter and acetone probe wash may be gravimetrically analyzed to
determine particulate emissions, as given in Method 5. Anything which
contacts the filter (forceps, balance tray, etc.) must be free of possible
metallic contamination.
The filter and probe wash residue are acid digested, using Parr Bomb or
microwave digestion techniques. The HN03/Hp02 impinger solution, the KMn04
impinger solution, the digested filter, ana the front half rinse are all
analyzed for Hg by cold vapor atomic adsorption spectroscopy (CVAAS). Except
for the KMn(L solution, the other samples are analyzed for Cr, Cd, Ni, Mn, Br,
Cu, Zn, Pb, Se, P, T1, Ag, Sb, Ba, and As by inductively coupled argon plasma
emission spectroscopy (ICAP) or atomic adsorption spectroscopy (AAS).
Graphite furnace atomic adsorption spectroscopy (GFAAS) may be used for Sb,
As, Cd, Pb, Se and T1 if greater sensitivity is reguired. Detailed analytical
procedures are presented in the draft Method 0012.
QUALITY ASSURANCE AND QUALITY CONTROL PROCEDURES
QA for the sampling train is the same as that for particulate sampling and
is presented in the Method 5 procedures given in the Federal Register.1 As
with Method 5, triplicate measurement runs are standard practice. Method 5
calculations for isokirieticity and emission rates are also used. During
sampling, the KMn04 impingers must be monitored visually, as a change in color
of the absorbing solution indicates exhaustion of the reagent.
Field blanks of the absorbing solutions, the recovery reagents and an
unused filter are prepared. Measured volumes of blank solutions are
required to perform blank analytical corrections on the samples. The filter
blank is not required to be mounted in a holder or heated.
The ICAP, AAS, and GFAAS instruments must be calibrated according to
manufacturers specifications. Mercury standards are prepared using procedures
outlined in the "Standard Methods for Water and Wastewater," 15th Edition,
Method 303F.2 The other standards are prepared using procedures in the draft
Method 0012.7 All samples are analyzed in duplicate, and matrix spikes are
performed on one sample with each analytical procedure.
391
-------
LIMITATIONS OF THE TECHNIQUE
Continuous analysis for metals is not possible with this method. The
possibility of incomplete reaction in the impingers, as well as cross-
contamination and introduction of other artifacts, is minimal if procedures
are followed.
ALTERNATIVE TECHNIQUES
No single comprehensive alternative exists for the Draft Method. However,
several Federal Register Methods exist, which can be used to measure emissions
of individual metals or multiple metal combinations. All are configured from
a Method 5 sample train, some without filters, but each with a unique
combination of absorbing solutions. The Federal Register Methods and their
respective target metals are: Method 12 - Lead1; Method 101 and 101A -
Mercury2; Method 104 - Beryllium2; and Method 108 - Arsenic2. In the case of
MWCs, the most important of these concerns the measurement of Mercury. While
Method 101A is usually acceptable for testing incinerators, Method 101 is
generally not recommended. If a separate test for Mercury is required, the
Multiple Metals sample train can be greatly simplified by eliminating the
KMnO^ impingers.
METHOD 23: MEASUREMENT OF PCDD/PCDF
METHOD DESCRIPTION
USEPA Proposed Method 23, "Determination of PCDDs and PCDFs From Stationary
Sources," is recommended as the most appropriate method for measuring
tetrachloro- through octachloro-polychlorinated dibenzofurans (PCDFs).
Example target PCDDs and PCDFs are listed in Table 3.4
Stack emissions are sampled isokinetically with a Method 5 train which has
been modified by the addition of a condenser (to cool the gases) and a sorbent
cartridge (to capture the gaseous organic compounds). Particulates are
collected on an inert heated filter, and vaporous emissions are collected on
the cooled sorbent material. Following sampling, the individual sample
components (probe wash, filter and cartridge) are recovered and transferred to
the laboratory for analysis. The samples are cleaned to remove potential
interferences and analyzed for PCDDs and PCDFs by gas chromatography/high
resolution mass spectrometry (GC/HRMS).
TABLE 3. PCDDS AND PCDFS ANALYZED FOR IN MWC STACK GASES4
DIOXINS
2,3,7,8 Tetrachlorodibenzo-p-dioxin (2,3,7,8 TCDD)
Total Tetrachlorinated dibenzo-p-dioxins (TCDD)
1,2,3,7,8 Pentachlorodibenzo-p-dioxin (1,2,3,7,8 PCDD)
Total Pentachlorinated dibenzo-p-dioxins (PCDD)
1,2,3,4,7,8 Hexachlorodibenzo-p-dioxin (1,2,3,4,7,8 HxCDD)
392
-------
TABLE 3. (cont.)
DIOXINS (cont.)
1.2.3.6.7.8 Hexachlorodibenzo-p-dioxin (1,2,3,6,7,8 HxCDD)
1.2.3.7.8.9 Hexachlorodibenzo-p-dioxin (1,2,3,7,8,9 HxCDD)
Total Hexachlorinated dibenzo-p-dioxins (HxCDD)
Total Heptachlorinated dibenzo-p-dioxins (HpCDD)
Total Octachlorinated dibenzo-p-dioxins (OCDD)
FURANS
2,3,7,8 Tetrachlorodibenzofurans (2,3,7,8 TCDF)
Total Tetrachlorinated dibenzofurans (TCDF)
1,2,3,7,8 Pentachl.orodibenzofuran (1,2,3,7,8 PCDF)
2,3,4,7,8 Pentachlorodibenzofuran (2,3,4,7,8 PCDF)
Total Pentachlorinated dibenzofurans (PCDF)
1,2,3,4,7,8 Hexachlorodibenzofuran (1,2,3,4,7,8 HxCDF)
1.2.3.6.7.8 Hexachlorodibenzofuran (1,2,3,6,7,8 HxCDF)
1.2.3.7.8.9 Hexachlorodibenzofuran (1,2,3,7,8,9 HxCDF)
Total Hexachlorinated dibenzofurans (HxCDF)
Total Heptachlorinated dibenzofurans (HpCDF)
Total Octachlorinated dibenzofurans (OCDF)
Proposed Method 23 has evolved primarily from Method 0010 of SW-846.3
Changes may have been made since publication of this paper. The entire method
is continually being evaluated and revised, and changes have been made since
the publication of this paper. The Emission Measurement Branch (MD-19),
Technical Support Division, USEPA, Research Triangle Park,North Carolina
27711, should be contacted for current developments before commencing dioxin
testing on MWCs.
SAMPLING EQUIPMENT AND PREPARATION
The MM5 sampling train is shown schematically in Figure 3. The components
are essentially the same as for Method 5, except with the addition of a
condenser and sorbent cartridge1. The sorbent cartridge is filled with
Amberlite XAD-2 resin, which has been thoroughly cleaned. Sorbent cartridge
cleaning and preparation procedures are described in Method 0010.3 The
condenser and sorbent cartridge may be constructed as a single unit (Figure 4)
or as separate components (Figure 5). The preferred practice is to construct
them as separate units to facilitate sample recovery.
The filter must be free of organic binders and cleaned with toluene. The
nozzle is constructed of nickel, nickel plated stainless steel, or glass.
Glass is recommended, in the presence of HC1, due to HCl's corrosive effect on
metal nozzles. All components must be meticulously cleaned with solvent (e.g.
acetone, methylene chloride), prior to testing, and sealed from contamination
with Teflon tape or hexane cleaned aluminum foil. Detailed glassware cleaning
procedures are described in Section 3A of "Manual of Analytical Methods for
the Analysis of Pesticides in Human and Environmental Samples" (see SW846).3
393
-------
u>
Thermometer
Heated Area/*^ Filter Holder
Sorbent Trap
Silica Gel
l-^t-
Temperature | |Stack Wall
Sensor N.
Probe
Reverse-Typpr
Pltot Tube
Thermometer
Check Valve
Pi tot Manometer
Recirculation Pump-»
Thermometers
Orifice
Vacuum Line
n
lmplngers Bath
-Pass Valve
Main Valve
Gas Meter A1r Tight Pump
Figure 3. Modified EPA Method 5 Sampling Train
-------
Flow Direction
Retaining Spring
8mm Glass Cooling Coil
Adsorbent
Glass Wool Plug
Glass Fritted Disc
28/12 Ball Joint
Fritted Stainless Steel Disc
Glass Water
Jacket
15 rim Solv-Seal Joint
(or 28/12 Socket Joint)
Figure 4. Combined Condenser - Adsorbent Cartridge Unit
-------
Coildaillo*
Sotbenl Trap
Flue Got Flow
-------
To avoid contamination, an enclosed clean area at the job site should be
dedicated to assembly of the sample trains and recovery of the samples. During
preparation, assembly, and disassembly of the sampling train, all component
openings where contamination can occur must be sealed with Teflon, hexane
rinsed aluminum foil, or ground glass caps. No sealant greases shall be used
on the sampling train.
SAMPLING/SAMPLING VOLUMES
Sampling procedures are exactly the same as those for Method 5 particulate
sampling.1 The stack should be traversed and sampled isokinetically. Since
the M23 train is very cumbersome and fragile, the sampling location must be
carefully selected and "prepared. Leak checks are performed in the same manner
as for Method 5. During sampling, the XAD-2 resin should not exceed 20°C
(68°F) to insure efficient capture of the PCDDs and PCDFs. The cartridge
should be wrapped in foil to shield the resin from light.
A minimum sample volume of 3 dscm (105.9 dscf) is required for the
determination of the destruction and removal efficiency (DRE) of PCDDs and
PCDFs from incineration systems.4 Additional sample volumes may be necessary
if, for example, unusually low concentrations are expected. Presented in
Method 0010 are the calculations to determine specific sample volumes, based
on refuse charging rates and expected DREs.3
Sample recovery is extremely critical. The entire train, or the individual
components, should be sealed with hexane cleaned foil, ground glass caps, or
Teflon tape before transporting to the recovery area. The individual samples
to be recovered consist of the following:
a. Sample train rinse - the solvent rinses of the probe, nozzle, cyclone
(if used), connecting lines between probe and filter, the front half
of the filter holder, the back half of the filter holder, and the
connecting lines from the filter holder to the resin cartridge
(including condenser, if separate). Each component is rinsed three
times with acetone followed by three rinses of methylene chloride.
b. Particulate catch - filter and cyclone catch (if used).
c. Resin cartridge.
d. Quality assurance rinse - all components mentioned in part a. (see
above) are rinsed into one container with toluene and analyzed
separately.
Detailed recovery procedures are presented in EPA Proposed Method 23.1
Basically, all components, except the adsorbent cartridge, are recovered to
precleaned amber glass bottles for transport to the lab. Liquid levels are
noted, bottles are capped and sealed and chain of custody data sheets are
prepared to accompany the samples. The resin cartridge is tightly sealed with
Teflon tape or ground glass caps, wrapped in foil and stored on ice until
397
-------
ready for analysis. Samples are carefully packed for shipment, as leakage or
breakage may void the test.
ANALYSIS
In the laboratory, all samples are spiked with a solution of method
internal standard (MIS) compounds prior to extraction. These MIS compounds
contain standards to analytically quantitate the PCDD and PCDF concentrations.
A typical set of calibration compounds is presented in Table 4.4 Samples are
prepared for analysis using a Soxhlet extraction apparatus. A detailed
procedure for sample preparation and extraction is given in EPA Proposed
Method 23.1
The primary analytical method for PCDDs and PCDFs, as specified in Proposed
Method 23, is GC/MS, using fused-silica capillary columns. Samples are first
analyzed, using a column coated with DB-5, to determine the concentration of
each isomer of PCDDs and PCDFs. If tetra-chlorinated dibenzofurans (TCDFs)
are detected, a separate analysis is performed with an SP2331 coated column to
measure the 2,3,7,8-TCDF isomer.
The GC/MS system must provide sufficient response and chromatographic
separation to achieve the minimum detection limits for the subject compounds.
PCDD and PCDF levels are quantified using response factors generated by the
standard compounds spiked prior to sample extractions. Proposed Method 231
contains detailed procedures for calibration and calculations.
TABLE 4. LIST OF ANALYTES, METHOD INTERNAL STANDARDS, SURROGATES, AND
RECOVERY INTERNAL STANDARDS FOR DIOXIN/FURAN ANALYSIS
Analyte
Compounds In
Calibration Standard
Method
Internal Standard3
Recovery
Internal Standard
Tetra-CDD
Tetra-CDF
2,3,7,8-TCDD
2,3,7,8-TCDF
iC«,-2,3,7,8-TCDD
C^-2,3,7,8-TCDF
37,
1
'XI,-2,3,7,8-TCDD or
CI-1,2,3,4, -TCDD
Penta-CDD
Penta-CDF
Penta-CDF
1.2.3.7.8-PeCDD
1.2.3.8.9-PeCDF
2,3,4,7,8-PeCDF
13C
13^12
13^12
tA2.
1.2.3.7.8-PeCDD
1.2.3.8.9-PeCDF
1,2,3,7,8-PeCDD
Hexa-CDD
Hexa-CDD
Hexa-CDD
Hexa-CDF
Hexa-CDF
Hexa-CDF
Hexa-CDF
1,2,3,4,7,8-HxCDD
1.2.3.6.7.8-HxCDD
1.2.3.7.8.9-HxCDD
1,2,3,4,7,8-HxCDF
1,2,3,6,7,8-HxCDF
2.3.4.6.7.8-HxCDF
1.2.3.4.8.9-HxCDF
13C
13^12
13^12
13^2
13^12
13^12
13^12
—Ui
•1,2,3,6,7,8-HxCDD
1,2,3,6,7,8-HxCDD
1,2,3,6,7,8-HxCDD
1,2,3,6,7,8-HxCDD
•1,2,3,6,7,8-HxCDD
1,2,3,6,7,8-HxCDD
1,2,3,6,7,8-HxCDD
13
C«2"1,2,3,6,7,8-
HxCDD
Hepta-CDD
Hepta-CDF
Hepta-CDF
1,2,3,4,6,7,8-HpCDD
1.2.3.4.6.7.8-HpCDF
1.2.3.4.7.8.9-HpCDF
13C -1
l3c12-
13c12-
U2-J.
,2,3,4,6,7,8-HpCDD
,2,3,4,6,7,8-HpCDD
,2,3,4,6,7,8-HpCDD
Octa-CDD
Octa-CDF
OCDD
OCDF
j3°1 -OCDD
C1,-OCDD
Added to sample prior to extraction
3Added to sample at time of injection into GC/MS
398
-------
QUALITY ASSURANCE AND QUALITY CONTROL TECHNIQUES
Sampling train calibrations and leak checks are the same as those given for
Method 5. At least one blank sample train is required in the procedure. The
blank train will be assembled in the recovery area and will accompany the
working train to the sampling site. The probe and filter holder will be
heated and maintained for the duration of the actual test, but no gas will be
passed through the system. Leak checks will be performed concurrently with
the actual sampling system, and recovery will be identical to that of the
other MMS trains. In addition, field blanks of sample containers, filters,
resin cartridges, and rinse solvents should be reserved for analysis, should
the blank train become contaminated or be unacceptable.
Analytical blanks should include samples of all solvents and materials used
in the extraction and preparation process. A quantity of XAD-2 resin from
each lot should be reserved for analysis. Blind quality control standards in
the expected range of the facility samples should be prepared and analyzed
with the samples. A bias in the analysis is determined by comparing GC/MS
results of the MIS surrogates to the known spiked concentrations.8
LIMITATIONS OF THE TECHNIQUE
The batch nature of this type of sampling limits actual emission
determinations to the testing periods. Refuse feed rates and unit operating
conditions during testing must be monitored closely and should be typical of
actual process rates.9 Continuous process monitoring and periodic chemical
analysis of the refuse can then be employed to demonstrate long term
conformity to the operating conditions which were in effect during testing.
Implementation of good combustion practices and operator
training/certification programs can help reduce organic emissions (including
emissions of dioxins/furans and their precursors) by promoting more thorough
combustion of these pollutants.10 Continuous monitoring of carbon monoxide
concentrations, maximum combustor load limits, and. flue gas temperatures at
the inlet to air pollution control devices, provides a means of documenting
adherence to the desired operating conditions.
Most of the problems common to stack testing methodologies are magnified by
the demands of testing for dioxins and furans. During sampling, the
possibility of incomplete sorption exists if the XAD 2 resin is allowed to
exceed 20°C. Another major problem is contamination. Total sample catches
often weigh only a few nanograms, and such small sample quantities demand a
much more rigorous approach than routine stack sampling.
ALTERNATIVE TECHNIQUES
No widely accepted alternatives to MM5 exist for the quantification of
PCDDs and PCDFs in flue gas.
399
-------
METHOD 26: MEASUREMENT OF HYDROCHLORIC ACID
METHOD DESCRIPTION
The collection of hydrogen chloride (HC1) in stack gas emission samples
from municipal waste combustors is described in the EPA Proposed Method 0026.
This method represents a simplified alternative to the traditional M5
configuration which can be applied when the pollutants of concern are in a
gaseous form. Smaller (midget) impingers are used and isokinetic sampling is
not required. This method has been recently reviewed and modified and the
USEPA Emissions Measurement Branch should be consulted before proceeding with
testing for HC1.
SAMPLING EQUIPMENT AND PREPARATION
A schematic of the sampling train used in this method is shown in Figure 6.
Stack gases are drawn through a borosilicate glass probe with a Teflon filter
installed at the probe outlet. A borosilicate, three-way glass stopcock is
connected directly to the outlet of the probe and inlet of the first impinger.
The probe and stopcock are heated to prevent any condensation up to the inlet
of the first impinger. Stack gases are then drawn through a sequence of 30 ml
midget impingers. In the first two impingers, a 0.1N sulfuric acid (H2S0J
solution collects the HC1 sample. This is followed by one impinger containing
a 0.1N Sodium hydroxide (NaOH) solution, which protects the sampling system
against corrosion by Cl2. Leak checks, pre-test calculations, and sampling
parameters are all described in the proposed Method.
SAMPLING/SAMPLE VOLUMES
In the method, sampling is recommended at a rate of 2 liters/min. for a
period of one hour. Shorter sampling times may introduce a significant
negative bias in the HC1 concentration. Sample recovery for HC1 consists of
combining the H2S04 impinger solutions and rinses in a leak-free glass or
polyethylene bottle, for transport or storage (refrigerated samples can be
stored for up to four weeks before analysis). Bottles should be marked at
liquid level and sealed to prevent contamination and leakage.
ANALYSIS
Analysis of samples is by ion chromatograph (IC) for CI" ions. The volume
of each sample is determined or adjusted with distilled, deionized water
(e.g., to 100 ml). An aliquot of each sample is then analyzed and a
concentration in ug/ml is determined. Total ug/sample can then be
ascertained, and the stack concentration can be calculated. Calculations are
presented in the Method.
QUALITY ASSURANCE AND QUALITY CONTROL PROCEDURES
A set of two audit samples must be concurrently analyzed with the
compliance samples to evaluate the technique of the analyst and the standards
400
-------
Stack Wnll
Purnn
3-Way Glnsfi Stopcock
SS Sheath
Mno Wosl Implnnor or Dryinn Tubn
lloatocJ Area
Filter Holdor
Gas
Flow
Healed Glass Uncr
Silica Gel
Empty
0.1 N NaOH
Flnln Motor
Vacuum Gniifjo
l_~— Pump
& Sampling (Figuro 1G)
Q3 VcriUnji (Rn»ro 1B)
Dry GaB Motor
Surno Tank
Purging (Fi(]ure 1A)
Figure 6. EPA Proposed Method 26 - HC1 Sampling Train
-------
preparation. Audit samples are available through written requests to EPA
regional offices.
As is conventional with other sampling methods, tests should include
triplicate measurement runs. The IC instrument must be calibrated according
to manufacturers' specifications. Blank samples of the H?S0, and rinse water
must be generated during field testing to be analyzed with tne stack samples.
LIMITATIONS OF TECHNIQUE
As with other stack sampling techniques, this approach provides a "snap-
shot" or short-term profile of emissions. However, triplicate replication of
sampling runs is intended to address this concern. Proposed requirements for
continuous monitoring of selected combustion parameters will further serve to
demonstrate long term compliance with optimum combustion practices in effect
during the test periods.
This method is designed to minimize possible analytical interference of the
HC1 sample with C12. However, it is possible that other volatile compounds in
the stack gases may produce chlorine ions upon dissolution during sampling
(after passing through the filter as a gas).
ALTERNATIVE TECHNIQUES
Proposed Methods 0050 and 0051 are available when more exacting
measurements of HC1 and Cl2 are required. Method 0051 also utilizes midget
impingers and is very similar to proposed Method 26. The additions of a
second impinger containing NaOH permits measurement of both HC1 and Cl2.
Proposed Method 0050 utilizes conventional impingers in a Method 5 type
sampling train. This method collects the sample isokinetically and is,
therefore, particularly suited for sampling at sources (such as those
controlled by wet scrubbers) emitting acid particulate matter (e.g., HC1
dissolved in water droplets). This method is also designed for collection of
both HC1 and Cl2.
Instrumental analyzers are available for HC1 determinations and their
practicality as a CEM has been demonstrated.11 It is probable that they will
accepted in some states for MWC compliance testing.
SAMPLING METHODS FOR OXIDES OF SULFUR AND NITROGEN
Instrumental Methods are available for S02 (EPA Method 6C) and for N0X (EPA
Method 7E). These methods can be incorporated into CEMS which are now
generally required for monitoring of S02 and N0X emissions. However,
chemistry methods may still be required from time to time for calibration or
relative accuracy tests of the CEMS.
402
-------
SULFUR OXIDES METHOD DESCRIPTION
Sampling methods are designed to address two sulfur oxides of primary
concern: sulfur dioxide (SOJ and sulfur trioxide (S03). S02 is, by far, the
more prevalent of the two. Generally, most methods screen out S03 so as not
to bias SO? determinations. The most commonly used method for S02 collection
and analysis is EPA Method 6.
SAMPLING EQUIPMENT AND PREPARATION
The basic sampling train consists of a borosilicate glass or stainless
steel probe with a glass wool plug, a series of midget impingers, and a pump
and dry gas meter to control and measure sample flow rates. The
first impinger contains isopropyl alcohol (I PA) and has a glass wool plug in
the neck. The second and third impingers contain a solution of 3% hydrogen
peroxide (H202). The fourth impinger is empty and is followed by a fifth
impinger containing a drying agent or a drying tube. The resultant sampling
train is shown on Figure 7.
SAMPLING/SAMPLING VOLUMES
A sample of stack gas is extracted at a constant rate, nominally 1 liter
per minute for 20 minutes at a single representative point in the duct. The
S03 is absorbed in the IPA, with the glass wool plug preventing carryover of
any sulfuric acid mist. The S02 is oxidized to H2S04 in the H202 impingers.
Sample volumes are measured with the dry gas meter. Post test leak checks are
mandatory, and care must be taken not to allow backflushing of the impingers,
as this will void the test. The IPA is discarded, and the H202 impingers are
recovered and rinsed into a single borosilicate glass bottle. Liquid level is
marked, and the bottle is sealed for transport to the laboratory.
ANALYSIS
Analysis for S02 is colorimetric titration. In the lab, the sample is diluted
to 100 ml, and a color change indicator (e.g., thorin) is added. Barium
perchlorate (Ba(C10J2) is added via burette until a color change occurs. The
S02 concentration is then calculated from the volume of titrant and volume of
stack gas metered. Alternatively, the analysis may be performed by ion
chromatograph (IC). S02 is most often reported as a concentration, ppm or
mg/dscm, but may be converted to a rate, lb/hr. The complete sampling,
analytical, and calculational procedures are presented in Method 6.1
QUALITY ASSURANCE AND QUALITY CONTROL PROCEDURES
Method 6 tests are typically run in triplicate. Duplicate titrations,
which must agree within + 1% or + 0.2 ml, are performed on each sample. Audit
samples are available from the EPA and must be analyzed concurrently with the
stack samples.12 The titrant must be normalized against a sulfuric acid
solution which has been standardized against a primary standard. Calibration
procedures for the metering system are given in the Method.
403
-------
PROBE iENO PACKED
WITH QUARTZ OH
PYREX WOOL]
¦Jf STACK WALL
1^ *
THERMOMETER
MIDGET IMPINGES
fcc
GLASS WOOL
NEEDLE VALVE
SIUCA GEL
DRYING TUBE
ICE BATH
THERMOMETER
PUMP
SURGE TANK
VACSUK VACUUM
CAUSE
SAUCE
CRITICAL
ORIFICE
IHGETAMK
(Optional using critical orifice to control sample rate)
Figure 7. EPA Method 6 S02 Sampling Train
404
-------
METHOD DESCRIPTION FOR OXIDES OF NITROGEN
EPA Method 7, 7A, 7B, 7C, and 7D are all applicable for determining N0X
emissions in incinerator stack gases. Methods 7, 7A, and 7B all involve
extracting a gas sample with an evacuated flask containing a dilute sulfuric
acid hydrogen peroxide absorbing solution.
The differences in the three methods are only in the preparation and
laboratory analysis of the samples. Colorimetric titration is used for
analysis in Method 7, while ion chromatograph (IC) and ultraviolet
spectrophotometry (UVS) are used in Methods 7A and 7B respectively. Of these,
Method 7A is the most widely employed.
The main disadvantage to Method 7, 7A or 7B sampling is the extremely short
duration (typically less than 30 seconds) of individual tests. Several sets,
of at least three (3) tests each, must be run to ensure determination of
average emissions.
Methods 7C and 7D employ an impinger train to absorb NO compounds in an
alkaline, potassium permanganate solution. The two methods differ only in
analysis, which is colorimetric titration for Method 7C and IC for Method 7D.
The gas sample is extracted through a glass lined probe and
a series of restricted orifice impingers containing the KMn04 absorbing
solution. A pump and dry gas meter are used to maintain and measure flow.
QUALITY ASSURANCE AND QUALITY CONTROL PROCEDURES
QA/QC procedures include calibration of sampling equipment prior to
testing, generating blank samples, and the use of audit samples, which are
available from the EPA.12 IC and UVS calibration techniques are given in the
appropriate Methods. Usual practice requires triplicate tests and replicate
laboratory analyses.
405
-------
REFERENCES
1. Code of Federal Regulations, Title 40, Part 60. Appendix A.
2. Code of Federal Regulations, Title 40, Part 61. Appendix B.
3. USEPA, Office of Solid Waste. Test Methods for Evaluation of Sol id Waste -
Physical/Chemical Methods. SW-846. Washington, D.C. 1986.
4. Haile, Clarence L. and Harris, Judith C. Guidelines for Stack Testing of
Municipal Waste Combustion Facilities. EPA-600/8-88-085. Research
Triangle Park, NC. June 1988.
5. New Jersey Administrative Code. 7:26-10.7(d).
6. USEPA. Quality Assurance Handbook for Air Pollution Measurement Systems:
Volume III Stationary Source Specific Methods. EPA/600/4-77-027b.
Feburary 1984.
7. U.S. EPA. Methodology for the Determination of Metals Emissions in Exhaust
Gases from Hazardous Waste Incineration and Similar Combustion Processes.
EPA Draft. Washington, DC. August 29, 1989.
8. Margeson, J.H.; Knoll, J.E.; Midgett, M.R.; Wagoner, D.E.; Rice, J.;
Homolya, J.B. "An Evaluation of the Semi-VOST Method for Determining
Emissions from Hazardous Waste Incinerators." JAPCA 37 (1987), p. 1067.
9. Visalli, J.R. "A Comparison of Dioxin, Furan, and Combustion Gas Data from
Test Programs at Three MSW Incinerators." JAPCA 37 (1987), p. 1451.
10. USEPA. "Standards of Performance for New Stationary Sources: Municipal
Waste Combustors (Proposed)." 54 FR 52251. December 20, 1989, p. 52263.
11. Entropy Environmentalists, Inc. "Emission Test Report-Municipal Waste
Combustion-Continuous Emission Monitoring Program." EMB Report 88-MIN-07C.
Research Triangle Park, NC. January 1989.
12. USEPA. EPA/AREAL/QAD/RMEB, MD - 77B. Research Triangle Park, NC.
406
-------
Development of Source Testing. Analytical and Mutagenicity Bioassav Procedures
for Evaluating Emissions from Municipal and Hospital Waste Combustors
by: R.R. Watts,-'- P.M. Lemieux,^ R.A. Grote,^ R.W. Lowans,^ R.W. Williams,^
L.R. Brooks, ^ S.H. Warren,^ D.M. DeMarini,! and J. Lewtas^
-'-Health Effects Research Laboratory and ^Air and Energy Engineering
Research Laboratory, U.S. Environmental Protection Agency; ^Environmental
Health Research and Testing, Inc., Research Triangle Park, NC 27709
ABSTRACT
Incineration is currently being utilized for disposal of about 10 percent
of the solid waste generated in this country, and this percentage will likely
increase as land disposal declines. Siting of new incinerators, however, is
often controversial because of concerns related to the possibility of adverse
health effects and environmental contamination from long-term exposure to
stack emissions. Specific concerns relate to the adequacies of a) stack
emission testing protocols, b) existing regulations, and c) compliance
monitoring and enforcement of regulations. These U.S. EPA Laboratories are
cooperatively conducting research aimed at developing new testing equipment
and procedures that will allow a more comprehensive assessment of the complex
mixture of organics that is present in stack emissions. These efforts are
directed specifically toward development of source testing equipment and
procedures, analytical procedures, and bioassay procedures. The objectives of
this study were to field test two types of high-volume source dilution
samplers, collect stack samples for use in developing analytical and
mutagenicity bioassay procedures, and determine mutagenicity of organics
associated with emission particles from two municipal waste incinerators.
Data are presented for particle concentrations and emission rates, extractable
organic concentrations and emission rates, and Salmonella (Ames) mutagenic
potency and emission rates. The mutagenic emission rates and emission factors
are compared to other incinerators and combustion sources.
This document has been reviewed in accordance with the U.S. Environmental
Protection Agency's peer and administrative review policies and approved for
presentation and publication.
407
-------
INTRODUCTION
Solid waste management is increasingly becoming a public issue in many
urban areas as city managers and engineers seek new ways of handling the
nearly 250 million tons of trash that is generated in the U.S. each year.
Approximately 160 million tons is produced by individual households and
neighborhood businesses. Landfilling accounts for about 80 percent of solid
waste disposal, and incineration and recycling each account for about 10
percent (1). Since 1979, however, 3500 landfills have been closed (2).
Furthermore, the U.S. Environmental Protection Agency projects that more than
30 percent of the existing landfills will close within 5 years. This will
result in an overall yearly capacity loss of 56 million tons. At current
construction rates, additional landfill space will be available for only 20
million tons, resulting in a significant shortfall of disposal capacity (1).
Alternative solutions to these problems will include concentrating more effort
into source reduction and recycling and also increased utilization of
incineration through expansion of existing municipal waste combustion (MWC)
units and construction of new ones. This incineration option is further
reinforced in many areas because of problems that have been encountered with
soil and ground water contamination resulting from runoff and seepage from
waste landfills.
The siting of new incinerators, however, is one of the most controversial
environmental issues today. The adequacies of existing regulations, stack
testing protocols, and compliance monitoring associated with incinerators are
being challenged, especially in areas being considered for siting of new
incinerator facilities. Of immediate concern to residents, of course, are
stack emissions and their effect on air quality in nearby neighborhoods and
areas downwind from the incinerator. The accompanying problems of disposal of
bottom ash and precipitated fly ash in landfills are also a primary concern in
many areas. The possibility of adverse health effects and environmental
contamination that could result from long-term exposure to stack emissions has
become an important issue. Research is needed to directly assess the
potential health effects of incinerator emissions through biological studies
of the complex mixtures that are emitted.
These U.S. EPA engineering and health Laboratories are cooperatively
conducting research aimed at developing additional sampling and assay
procedures for assessing possible health and environmental risks from MWC.
This research includes developing MWC stack sampling equipment and analytical
techniques that can be used to analyze the complex mixture of organics that is
present in emissions. Stack emission samples are being analyzed by the Ames
Salmonella typhimurium bacterial mutagenicity bioassay (3,4). These
procedures represent simple, sensitive, rapid, and reliable screening tools
for mutagenic substances. This approach, therefore, diverges from assessments
which combine emission studies that determine individual compounds or a class
of compounds (5-8) with toxicological data for those direct emissions.
Analyses for specific organics (e.g., the chlorinated dibenzodioxins and
dibenzofurans) will continue to be important in assessing potentially
hazardous emissions. However, approaches that analyze incinerator emissions
as a whole complex mixture by using assays (e.g., mutagenicity bioassay) that
408
-------
are not compound specific but are targeted to specific health effects are
needed. Such assays are especially relevant for combustion emissions, which
include products of incomplete combustion and pyrolysis products, because
these emissions contain thousands of compounds. The potential hazards
associated with such a complex mixture might be underestimated if only a few
species were included in a risk assessment determination. Mutagenicity
bioassay was also chosen for this study because chemicals and mixtures
recognized as human carcinogens are generally mutagenic in short-term tests
(9), and combustion emissions from other sources have been shown to be both
mutagenic in this assay and carcinogenic in rodents. It has also been
reported that chemicals that are rodent carcinogens across several species and
organ sites (trans-species carcinogens) are generally also mutagenic in
Salmonella tvuhimurium (10). Other biological studies were planned, e.g.,
rodent cancer studies, and will be reported elsewhere.
A study using the Ames Salmonella/microsomal mutagenicity bioassay to
analyze MWC stack emissions was reported by Kamiya and Ose in 1987 (11).
These investigators used a small sampling apparatus to collect fine particles
(<7 /zm) and gaseous organics in 10-15 m^ of stack gas emissions. They sampled
a continuously operating modern incinerator with a complete combustion system
and a discontinuous batch-type incinerator, where combustion was incomplete.
Both incinerators were located in Aichi Prefecture, Japan. The authors found
significant mutagenicity in the stack gas emissions from the latter MWC unit
and, in fact, concluded "the emission gases from batch-type incinerators are
mainly responsible for atmospheric pollution." They also identified and
quantified 13 polynuclear aromatic hydrocarbon (PAH) compounds in the same
emission gases and found a high correlation between mutagenic activity and PAH
concentrations.
Ahlborg and Victorin (12) analyzed emission samples from four MWC sites
using mutagenicity bioassay. Results were reported in revertants per
megajoule of fuel and were compared to other Swedish combustion sources.
These authors also concluded that MWC can cause relatively high emission of
organic mutagens and potentially carcinogenic compounds if well controlled
combustion conditions are not used.
More recent studies of incineration have also used Ames mutagenicity
bioassay to characterize emissions. Driver, et al. (13) found that the
mutagenic potencies of the stack fly ash from a medical pathological waste
incinerator and from an adjacent industrial boiler were similar when both
combustors were operated under "normal" conditions. This finding was
unexpected considering the vastly different fuels for the medical waste
incinerator and the industrial boiler burning No. 6 residual fuel oil. This
assay was also able to determine significant increases in mutagenic emissions
resulting from "upset" burn corditions caused by auxiliary burner failure.
Linak et al. (14) and DeMarini et al. (15) determined mutagenicity emission
factors for incineration experiments with the pesticide Dinoseb in a fuel-
oil/xylene solvent. Results with either air staging or air staging and
reburning were similar to those measured for the burning of fuel oil for
residential heating.
409
-------
The primary goal of the work reported here was to begin development of
procedures to assess mutagenicity of emissions from municipal or hazardous
waste incinerators. Specific objectives were to a) field test two types of
high-volume source-dilution samplers, b) collect stack samples from three
incinerator units for use in developing analytical and mutagenicity bioassay
techniques required for analyses, c) determine the mutagenicity of organics
associated with particle emissions, d) determine the stack emission rate for
organic mutagens, and e) identify specific compounds/mutagens that are present
in emission samples. This paper presents data on the mutagenic activity of
stack gas emission samples and emission rates from a municipal waste
incinerator, a municipal waste/hospital medical pathological waste incinerator
and an incinerator dedicated to hospital medical pathological waste combustion
(HMPWC). These data are compared to results from earlier MWC research and to
other combustion sources (11,12,13,16). Details of the field sampling portion
of this study will be described elsewhere (17).
EXPERIMENTAL
SAMPLE COLLECTION
Two prototype source dilution samplers were field tested at three sites in
three USA states. These samplers, the 10 cfm (0.28 m^/min) Source Dilution
Sampler (SDS)(18) and the 100 cfm (2.83 m^/min) Baghouse/Dilution Tunnel
sampler (19) , were designed to collect the large gram or even kilogram
quantities of sample required for in vitro and in vivo toxicological studies
including mutagenicity bioassay, animal carcinogenicity studies; and for
identification of principal organic components responsible for
mutagenic/carcinogenic activity. Both samplers bring outdoor air into a
mixing chamber to dilute emission gases in a 10:1 ratio; i.e., 100:10 for the
SDS sampler and 1000:100 for the baghouse sampler. This dilution process
simulates the flue gas quenching that occurs upon emission from the stack to
the atmosphere and was incorporated into the design to give organic vapors a
sufficient sampler residence time to condense on fine particles prior to being
collected on a particle filter or trapped in the baghouse. Both units also
have the possibility of collecting semi-volatile organics on XAD-2 or other
cartridge filters placed downstream from the particle filters, however, XAD-2
collections were not used in the present study.
In the SDS sampler, outdoor air was heated to approx. 75°F (24°C) and was
HEPA (high efficiency particulate air filter) and charcoal filtered prior to
entering the dilution chamber. Diluted particles were collected on round
(approximately 26 in. [66 cm] diameter) Teflon-impregnated glass-fiber (TIGF)
filters. A modified SASS cyclone used at the sampler inlet removed particles
>2.5 /zm prior to dilution and collection.
The baghouse/dilution tunnel sampler was also designed to filter heated
outdoor air through HEPA and charcoal filters prior to mixing with the stack
gases in a mixing chamber. Mixed gases passed into a baghouse unit containing
a felt filter cartridge. This filter fabric consisted of a Gore-Tex^ membrane
backed with 100 percent Nomex^ fiber. Periodically, inlet gas flow was
interrupted long enough to use a reverse pulse of high pressure nitrogen to
410
-------
dislodge particles from the filter into a glass collection bulb positioned at
the bottom of the unit.
Source samples were collected by the SDS sampler in two field studies
during September and November 1988 from a MWC unit (site A) and during
December of 1988 from a MWC/HMPWC at site B. The baghouse sampler was field
tested for the first time during the second sampling study at site A and then
tested again during the sampling study at site B. The baghouse heater for the
dilution air was not installed prior to sampling at site A. A third sampling
study, which utilized only the baghouse sampler, was conducted during August
1990 at the HMPWC unit located at site C. The baghouse sampler at this site
C, with the dilution air heater installed, was primarily used to gain further
experience with operation of this modified sampler and to attempt to collect
sufficient sample size for mutagenicity and carcinogenicity determinations.
An experimental sampler was also installed upstream from the baghouse at site
C to draw a small portion of the sample/dilution air mixture through a 142 mm
TIGF particle filter. This small sampler flow was approximately 300 percent
of isokinetic. Sampling was not conducted at isokinetic rates at any of the
three sites.
The incinerator at site A had two refuse-fired boilers, each with a
capacity of 100 tons/day. Each combustion unit had a reciprocating stoker, an
economizer, and an electrostatic precipitator (ESP). Combustion emissions,
were vented into a common stack. The sampling probe was inserted into the
emission gases from one boiler just prior to their entrance to the base of the
stack. Stack gas temperature measured after the ESP unit was 425°F (218°C).
The incinerator at site B consisted of two 50 ton/day Consumat starved-air
combustors with a common ESP and stack. The unit burns primarily municipal
waste and about 3-5 tons/day of hospital wastes. Stack gas temperature
measured just after the ESP unit was 483°F (251°C). Sampling was conducted in
the stack just downstream from the ESP outlet.
The HMPWC unit at site C consisted of a single 6.8 ton/day Consumat
starved-air combustion system. No air pollution control devices were used on
this unit. Stack gases leave the secondary chamber and are vented to the
atmosphere through the stack. The incinerator is operated for an 8 h period
and burns approximately 500-800 waste boxes/day with each weighing an average
of 17.5 lb (7.9 kg). Sampling was performed with the baghouse sampler probe
placed in the transfer duct between the secondary combustion chamber and the
stack, where the gas temperature was approximately 1800°F (982°C). Stack
emissions were sampled at a lower rate than used at sites A and B due to the
requirement of maintaining a maximum dilution tunnel inlet temperature of
450°F (232°C) so as to avoid decomposition of Teflon parts in the sampler.
Site C emissions were also diluted with outside air to a greater extent than
at the other sites due to the lower sampling rate. In addition to the
baghouse samples, experimental particle samplers were fitted to the dilution
tunnel before and after the baghouse. These samplers pulled dilution tunnel
gases through a 142 mm TIGF filter at rates of approx. 2.6 ft^/min (4.4 m^/h).
The post-baghouse TIGF particle sampler pump failed, however, soon after
start-up and did not collect a sample.
411
-------
EXTRACTION AND FRACTIONATION
The SDS filters were cut into small pieces, placed in a glass or Teflon
container along with 150 mL of dichloromethane (DCM) and sonicated (Branson
117 V 50/60 Hz Sonicator) for 10 min. at 77°F (25°C). This process was
repeated two additional times. Combined extracts were filtered in an all-
glass apparatus through a 0.45 pm Teflon filter. Extracts were concentrated
by rotary vacuum evaporation at 95°F (35°C) and transferred to a volumetric
flask where volumes were adjusted to 10 mL. Aliquots were removed for
gravimetric determination of extractable organic mass (EOM) and for solvent
exchange to dimethyl sulfoxide (DMSO) prior to bioassay (20). Gravimetric
determinations of EOM for each sample extract were performed using two or
three 0.5 mL aliquots from the known volume of extract. Aliquots were placed
in tared aluminum weigh pans and solvent was evaporated in a hood. Sample
pans were then equilibrated overnight in a desiccator prior to final weighing.
Baghouse particles were Soxhlet extracted for 24 h with 850 mL of DCM.
Solvent cycle time was 13.5 min. Extracts were filtered, concentrated, volume
adjusted and aliquoted for gravimetric and bioassay determinations as above.
Aliquots representing specific amounts of EOM were solvent exchanged to
DMSO and bioassayed whole or were fractionated on a nonaqueous ion-exchange
column prior to DMSO solvent exchange and bioassay. The fractionation scheme
utilized a quaternary ammonium styrene ion-exchange resin, Bio-Rad AGMP-1, to
class fractionate DCM extracts of incineration samples. This solid phase
extraction (SPE) procedure is described in more detail elsewhere (21) and is a
modification of an earlier procedure (22) . The modified procedure utilized 1
mL of washed and activated resin in a 1 cm i.d. column. Sample aliquots of
<500 n1 of DCM extracts were placed on the column and four 16 mL eluates were
separately collected. The sequential elution solvents were: 1) DCM, 2)
methanol, 3) methanol saturated with CO2 and 4) 10 percent trifluoroacetic
acid in methanol. Gravimetric analyses on aliquots of the four fractions
showed that fraction 1 contained approximately 70 percent of the organic mass
placed on the column. Fraction 1, utilized for determinations of mutagenic
potency and emission rates, was concentrated and prepared for bioassay as
previously described for whole sample extracts.
MUTAGENICITY TESTING
The Ames Salmonella tvphimurium histidine reversion assay (3) with strain
TA98 was used for mutagenicity bioassay. Samples having sufficient EOM to
meet the minimum detectable limits of the assay were tested at a minimum of
five doses using triplicate plates with and without Aroclor-induced rat liver
S9 metabolic activation (+S9 and -S9) at each dose. Duplicate plates were
used for samples with limited quantities. The minimum amount of sample for
testing was approximately 0.5 mg. Sample extraction/elution solvents were
exchanged to DMSO to make bioassay stock solution concentrations of 1 mg/mL.
Spontaneous counts for TA98 were 25-50 colonies per plate after a 72-h
incubation. Mutagenicities were confirmed by streaking revertant colonies
onto minimal medium supplemented with biotin, but not histidine. A set of
positive controls was incorporated in each experiment. These controls
included 2-aminoanthracene (0.5 /ig/plate) and 2-nitrofluorene (3.0 /ig/plate)
412
-------
with and without S9, respectively. A negative bioassay control utilized in
each experiment consisted of a DMSO blank. Controls, consisting of blank
filters and/or laboratory blanks, were extracted and analyzed in parallel with
actual samples.
The revertant colony versus microgram of EOM data were plotted and
analyzed by linear regression. The slope values (revertants//ig) from these
dose-response determinations were used to calculate revertants/mg of
particles, revertants//ig of EOM, revertants/m^ of emission gas, and stack
emission rates in revertants/min.
RESULTS AND DISCUSSION
The mass burn municipal waste incinerator at site A was sampled in
September 1988 with the SDS sampler and again in November of the same year
with both the SDS and baghouse samplers. The SDS sampler collected five
filter samples containing particle masses of 1.3 to 3.8 g during the first
sampling study and one filter containing 3.4 g of particles during the second
sampling trip. Collection periods were from approximately 1 to 4.5 h. Table
1 lists the volumes collected during each period and also shows particle
concentrations of 35 to 102 mg/m^ of stack gas. The percent EOM values ranged
from 0.26 to 1.72 percent with a mean of 0.77 percent. The EOM concentrations
in stack gases were 0.16 to 1.61 mg/m^. Particle and EOM emission rates were
calculated for the six SDS samples collected during the two sampling periods
at site A. Average emission rates were 359 mg/min for EOM and 41.4 g/min for
particles. These rates were calculated with a stack gas emission rate of 644
wet std. m-^/min.
The weather conditions and consequently the nature of the municipal waste
being burned were quite different for the two sampling studies at site A.
Rainy weather during the second sampling study resulted in wet trash being fed
to the two MWC units and, consequently, emission gases contained a higher
percentage of water. This burn condition plus intake of water saturated
dilution air (10:1 ratio of air to sample) led to condensation of very
corrosive gases that damaged the stainless steel transfer lines and sampler
pumps. Condensation problems were also encountered inside the baghouse
sampler which caused collection of a black liquid/particulate mixture in the
sample jar. These liquid baghouse samples were deemed unuseable for
mutagenicity analysis.
Three of the baghouse liquid condensate samples were subjected to ion
speciation analyses. Atomic absorption (AA) was used for cations and ion
chromatography (IC) for anions. Table 2 results of these analyses show a 3
percent chlorine content which would indicate corrosion problems resulted from
formation of hydrochloric acid. Fluorine was present in concentrations of 88
- 207 ppm. Nitric and sulfuric acids were also indicated by the IC results of
200 and 450 ppm concentrations respectively for these anions. Chromium and
nickel concentrations were each in excess of 1000 ppm. The presence of
significant quantities of fluorine and chlorine indicates a high likelihood of
finding halogenated organic species in stack emissions.
413
-------
The SDS samples collected during the same second study at site A were also
different in character from those collected during the September sampling
study. Only one SDS sample (identified as 3/2 in Table 1) was suitable for
analysis. The DCM extract of this filter was corrosive and mass
determinations on aliquots of the extract resulted in aluminum weigh pans
being corroded and mass determinations voided because of sample leakage.
Polypropylene weigh pans, therefore, had to be utilized for these EOM
analyses. A baghouse particle sample, which was obtained only after
dismantling the filter housing assembly and physically rapping to dislodge
particles, consisted of 102 g of wet filter cake. DCM extracts of this sample
were also found to be corrosive. Mutagenicity assay of samples collected from
both types of samplers failed because extracts proved to be too cytotoxic.
A sampling study at the site B MWC/HMPWC was initiated before analyses
were completed for site A samples. Results for three SDS filters are reported
in Table 1. Particle concentrations in stack emissions averaged 55 mg/m^.
This value is a close comparison with the average particle concentration of 64
mg/m^ found at site A. The percent EOM and EOM concentrations (mg/m^) in the
stack emissions were also similar to those from site A. The smaller particle
and EOM emission rates are partly due to the lower stack emission rate of 393
std m-^/min. These Table 1 values from sites A and B, however, are all
remarkably similar. Corrosion problems were again encountered while sampling
with the SDS sampler, and extracts of the SDS filters from site B were also
found to be cytotoxic. The baghouse sampler at site B failed to collect a
sufficient amount of particles for bioassay analysis.
A non-aqueous solid phase extraction (SPE) scheme (21) , which was
developed to separate acidic fractions from neutral organics, was applied to
selected extracts from sites A and B. These highly acidic fractions are often
too cytotoxic to measure any mutagenic activity. The first SPE column
fraction, the DCM neutral fraction, was found to contain approximately 70
percent of the organic mass placed on the column. Moreover, this fraction did
not exhibit cytotoxicity in the Ames bioassay.
Table 1 also shows results for the HMPWC unit at site C. Four samples
represent baghouse particles and one sample is a composite of three 142 mm
TIGF filters collected by the experimental sampler with the intake positioned
in the mixing chamber upstream from the baghouse sampler. The particle
concentrations calculated for the baghouse samples at this site are
approximately half of the amounts from sites A and B; however, the 59.7 mg/m^
particle concentration determined for the composited TIGF particle filters is
very similar to the 64 mg/m^ average for site A and the 55 mg/m^ average for
site B. The 0.15 average percent EOM for the baghouse samples are also about
half of the average value for site B (0.33) and one-fifth of the site A value
(0.77). The TIGF percent EOM of 1.04 is again similar to that observed from
site A. Additional testing of the baghouse sampler is being planned in order
to make further comparisons between the particles collected by the baghouse
and those collected on a particle filter; e.g, the TIGF filter media used in
the SDS sampler.
414
-------
Table 3 shows mutagenic measurements for emissions expressed as revertant
colonies (rev)/mg of particles and rev//ig of EOM, mutagenicity concentrations
(rev/m^) and emission rates (rev/min) for samples taken from incinerators at
sites A and C. Results are shown for whole samples (unfractionated) and DCM
neutral fractions (fraction 1) from SPE column separations. Three SDS filters
(2,3, and 4) from site A were composited prior to bioassay to provide
sufficient sample size for further chemical characterization studies on high
performance liquid chromatography (HPLC) subfractions of the first elution
(neutral fraction) from the SPE column. This bioassay directed fractionation
and chemical characterization work is a continuing effort to identify specific
compounds and/or compound classes that are responsible for observed
mutagenicity. The small (142 mm) TIGF filters from site C were also combined
to provide sufficient sample size for bioassay. The mutagenic potency data
for the site A filters show divergent values for the whole sample extracts.
Filter 1 gave a +S9 value of 27 rev/ug compared to 0.95 rev/ug for the
composite sample (filters 2,3, and 4). The fractionated composite sample,
however, shows a potent first (neutral) fraction from the SPE column,
indicating the possibility that the 0.95 value for the whole sample was
depressed due to the presence of acidic cytotoxic components that inhibit the
expression of mutagenicity. This type of bioassay requires that the
Salmonella cells remain viable during incubation. A fractionation step,
therefore, may be necessary for some or all incinerator samples to remove such
interferences prior to the mutagenicity assay. Site B whole samples were also
cytotoxic, and analyses of SPE fractions failed to demonstrate mutagenicity.
The emission factors in Table 4 were developed to make comparisons with
previous MWC studies and other combustion sources. The mutagenicity per hour
values from site A closely match those reported for an incinerator in Aichi
Prefecture, Japan (11). Better comparisons, however, can be made for emission
results reported as revertants per kg of fuel. A study of four MWC units in
Sweden reported approximately 10,000 to 100,000 revertants per megajoule (MJ)
of fuel (12). Conversion of these values to rev/kg of fuel basis was made
using the factor of 11.614 MJ/kg of municipal solid waste (MSW). These values
ranged from 1-12 x 10^ rev/kg, which compare with the 1.6-4.4 x 10^ rev/kg
values from the MWC at site A and 0.7 x 10^ rev/kg from the HMPWC at site C.
Data for industrial and utility boilers and power plants using oil, coal, and
wood fuel (16) show emission factors ranging from 0.03-0.2 x 10^ rev/kg of
fuel. The similarity of these emission factors indicates that the mutagenic
potency of emissions may not be greatly affected by the fuel source. The
combustion condition or "completeness of combustion" combined with the
effectiveness of pollution control equipment operating at the incinerator may
be much more relevant factors to consider for controlling the mutagenicity of
incinerator emissions.
Table 4 emission factors reported for gasoline and diesel vehicles range
from 1-40 x 10^ rev/kg of fuel (16). A comparison of the 0.5 x 10^ rev/hour
rate for a gasoline catalyst car (24) with 1000 x 10^ rev/hour from a MWC unit
(11) shows that MWC emission to be equivalent to approximately 2000 gasoline
catalyst cars. However, it should be noted that comparisons between emission
sources may be misleading. Automobile exhausts contain highly mutagenic, but
not necessarily strongly carcinogenic species, and MWC emissions may contain
415
-------
strongly carcinogenic species, i.e., chlorinated dibenzodioxins, that are not
mutagenic (12).
The data reported here should be considered as pilot study information
obtained during initial field testing of the SDS and baghouse/dilution tunnel
samplers. Isokinetic sampling was not used at any of the incineration sites.
Problems were encountered during each sampling event (e.g., condensation
inside sampler housings and corrosion damage) which are currently being
addressed. The analytical operations of sample extraction, gravimetric
determinations, and bioassay also encountered problems related to the
corrosive nature of extracts and to cytotoxic species that often prevented or
otherwise affected mutagenic potency analyses on the whole or unfractionated
extracts. A fractionation scheme had to be developed before bioassay could be
successfully performed.
Engineering modifications have been made on the sampling equipment, and
further field testing is needed in order to continue the development of
incinerator emission samplers. The quality assurance aspects of field
sampling, sample handling, and analytical procedures all need additional
study. The SPE fractionation method used and perhaps other similar methods
need to be further examined for their suitability in removing cytotoxic
species and separating organics into class fractions. Additional bioassay-
directed fractionation procedures (e.g., HPLC coupled with microsuspension
bioassay (25)) are being used to subfractionate the neutral fraction for
further examination by mass spectroscopy in order to identify and quantify the
principal organic mutagens.
SUMMARY
Two types of incinerator stack samplers were field tested at three sites.
Emission samples were collected from municipal waste and hospital medical
pathological waste incinerators. Organics associated with emission particles
were bioassayed for mutagenicity using the Ames plate incorporation assay.
Stack emissions, which were characterized by particle concentration and
percent extractable organic mass, were similar although there were large
differences in the nature of the fuel materials consumed at the various sites.
Emissions, characterized by mutagenicity emission factors, also were similar
for these same sites and were similar to factors previously reported for a
hospital medical pathological incinerator and for industrial and utility
boilers burning coal, wood, and oil. The mutagenicity of incinerator
emissions, therefore, may not be greatly affected by the fuel source. Burn
conditions and pollution control devices are likely to be more important
considerations for ensuring safest possible emissions.
Mutagenicity concentrations reported here for a MWC unit are also similar
to those reported by other researchers. The mutagenicity of sample fractions
from a MWC sample extract indicates the presence of potent organic mutagens
associated with particle emissions. However, additional research is needed to
identify such species and to determine their emission rates. Additional
studies are also needed to further improve: sampling equipment, sampling and
416
-------
sample handling procedures, analytical and sample preparation methods, and
bioassay procedures.
417
-------
REFERENCES
1. Wingerter, E.J. Are landfills and incinerators part of the answer?
Three viewpoints. EPA Journal 15:22, 1989.
2. Steinhart, P. Down in the dumps, Audubon 88:102, 1986.
3. Maron, D.M., Ames, B.N. Revised methods for the salmonella mutagenicity
test. Mutat. Res. 113:173, 1983.
4. Kado, N.Y., Langley, D., Eisenstadt, E. A simple modification of the
Salmonella liquid incubation assay. Increased sensitivity for detection
of mutagens in human urine. Mutat. Res. 121:25, 1983.
5. Ballschmiter, K., Zoller, W., Scholz, C., Nottrodt, A. Occurrence and
absence of polychlorodibenzofurans and polychlorodibenzodioxins in fly
ash from municipal incinerators. Chemosphere 12:585, 1983.
6. Wakimoto, T., Tatsukawa, R. Polychlorinated dibenzo-p-dioxins and
dibenzofurans in fly ash and cinders collected from several municipal
incinerators in Japan. Environ. Health Perspect. 59:159, 1985.
7. Davis, I.W., Harrison, R.M., Perry, R., Ratnayaka, D., Wellings, R.A.
Municipal incinerator as source of polynuclear aromatic hydrocarbons in
environment. Environ. Sci. Technol. 10:451, 1976.
8. Ozvacic, V. , Wong, G., Tosine, H., Clement, R.E., Osborne, J. Emissions
of chlorinated organics from two municipal incinerators in Ontario. JAPCA
35:849, 1985.
9. Waters, M.D., Garrett, N.E., Covone-de Serres, C.M., Howard, B.E., Stack,
H.F. Genetic toxicology of some known or suspected human carcinogens.
In: F.J. de Serres (ed.), Chemical Mutagens: Principles and Methods for
their Detection. Vol. 8, New York, Plenum Press, 1983. pp. 261-341.
10. Ashby, J., Tennant, R.W., Chemical structure, salmonella mutagenicity
and extent of carcinogenicity as indicators of genotoxic carcinogenesis
among 222 chemicals tested in rodents by the U.S. NCI/NTP. Mutat. Res.
204:17-115, 1988.
11. Kamiya, A., Ose, Y. Mutagenic activity and PAH analysis in municipal
incinerators. Sci. Total Environ. 61:37, 1987.
12. Ahlborg, U.G., Victorin, K., Impact on health of chlorinated dioxins and
other trace organic emissions. Waste Management & Research 5:203 (1987).
13. Driver, J.H., Rogers, H.W., Claxton, L.D., Mutagenicity of combustion
emissions from a biomedical waste incinerator. Waste Management 10:177,
1990.
418
-------
14. Linak, W.P., McSorley, J.A., Hall, R.E., Srivastava, R.K., Ryan, J.V.,
Mulholland, J.S., Nishioka, M.G., Lewtas, J., DeMarini, D.M. Application
of staged combustion and reburning to the co-firing of nitrogen-
containing wastes. Hazard. Waste Hazard. Materials, in press, 1991.
15. DeMarini, D.M., Houk, V.S., Lewtas, J., Williams, R.W., Nishioka, M.G.,
Srivastava, R.K., Ryan, J.V., McSorley, J.A., Hall, R.E., Linak, W.P.
Measurement of mutagenic emissions from the incineration of the pesticide
dinoseb during application of combustion modifications. Environ. Sci.
Technol. in press, 1991.
16. Lewtas, J. Genotoxicity of complex mixtures: strategies for the
identification and comparative assessment of airborne mutagens and
carcinogens from combustion sources. Fundamental and Applied Toxicology
10: 571, 1988.
17. Lemieux, P.M. Private communication. U.S. EPA, AEERL, MD-65, Research
Triangle Park, NC 27711.
18. Steele, W.J., Williamson, A.D., McCain, J.D. "Construction and operation
of a 10 CFM sampling system with a 10:1 dilution ratio for measuring
condensable emissions," EPA-600/8-88-069 (PB88-198551), April 1988.
U.S. EPA, AEERL, Research Triangle Park, NC 27711.
19. Lemieux, P.M., McSorley, J.A., Linak, W.P. A prototype baghouse/dilution
tunnel system for particulate sampling of hazardous and municipal waste
incinerators. In: Remedial Action Treatment and Disposal of Hazardous
Waste, Proceedings of the Fifteenth Annual Research Symposium, U.S.
Environmental Protection Agency, Risk Reduction Engineering Laboratory
Cincinnati, OH 45268, EPA/600/9-90/006, 1990.
20. Williams, R., Pasley, T., Warren, S., Zweidinger, R., Stead, A.,
Chappell, J., Watts, R., Claxton, L. Selection of a suitable extraction
method for mutagenic activity from wood smoke impacted air particles.
Int. J.Environ. Anal Chem. 34:137, 1988.
21. Williams, R., Brooks, L., Taylor, M., Thompson, D., Bell, D., DeMarini,
D., Watts, R. Fractionation of complex mixtures using an ion-exchange
methodology. To be presented at the 1991 EPA/A&WMA International
Symposium, Measurement of Toxic and Related Air Pollutants, A&WMA,
Pittsburgh, PA (1991).
22. Bell, D.A., Karam, H., Kamens, R.M. Nonaqueous ion-exchange separation
technique for use in bioassay-directed fractionation of complex mixtures:
application to wood smoke particle extracts. Environ. Sci. Technol. 24:
1261, 1990.
23. Lofroth, G., Georgios, L., Rudling, L. Mutagenicity assay of emission
extracts from wood stoves: comparison with other emission parameters.
Sci. Total Environ. 58: 199, 1986.
419
-------
24. Lofroth, G. Salmonella/microsome mutagenicity assays of exhaust from
diesel and gasoline powered vehicles. Environ. Int. 5: 255, 1981.
25. DeMarini, D.M., Williams, R.W., Brooks, L.R., Taylor, M.S. Use of
cyanopropyl-bonded silica for bioassay-directed fractionation of organic
extracts from incinerator emissions, (in preparation).
420
-------
TABLE 1. PARTICLE AND EXTRACTABLE ORGANICS IN INCINERATOR EMISSIONS.
Site A, MWC Site B, MUC/HMPUC Site C, HMPUC
(SDS filters ) (SDS filters ) (Baghouse samples )TIGF
Filter No. 1 2 3 4 5 3/2 7 21 23 1 2 3,4 5 1,3,4
3
std m
collected 40.7
3
part. mg/m
stack gas 93.4
% EOM 1.72
3
EOM mg/m
stack gas 1.61
a
EOM mg/min
from stack 1037
part, g/min 60.1
from stack
a
Stack gas emission rates (wet basis): site A = 644 std. cubic meters/min (flue gas temperature
of 425 deg. F); site B = 393 std. cubic meters/min (flue gas temperature of 483 deg. F); and
site C = 120.6 std. cubic meters/min (flue gas temperature of 1400 deg. F)
57.3 80.3 12.8 63.8 43.1 68.8 140.9 141.6 72.5 268.1 214.2 359.4 3.57
34.9 41.1 101.6 36.1 78.9 78.5 52.5 33.2 33.1 16.2 62.8 8 59.7
0.62 0.76 0.82 0.43 0.26 0.31 0.23 0.46 0.25 0.14 0.04 0.15 1.04
0.22 0.31 0.83 0.16 0.21 0.24 0.12 0.15 0.08 0.02 0.03 0.01 0.62
142 200 535 103 135 94 47 59 9.8 2.7 3.1 1.4 75
22.5 26.5 65.4 23.2 50.8 30.9 20.6 13.0 4.0 2.0 7.6 1.0 7.2
421
-------
Table 2. CATIONIC AND ANIONIC SPECIATION RESULTS FOR BAGHOUSE LIQUID
CONDENSATES COLLECTED FROM INCINERATOR A.
AA Results (p-onO
Sample Cd Cr Pb Zn Cu Ni
1 6 500-1000 -100 >50 25-50 -1000
2 9 >1000 >100 >50 25-50 >1000
3 5 >1000 50-100 >50 50-100 >1000
IC Results (dpih)
F Cl(%) N03 S04
1
88
3
—
268
2
108
3
210
456
3
207
3.4
203
418
422
-------
Table 3. MUTAGENIC POTENCY, CONCENTRATIONS, AND STACK EMISSION RATES.
Revertant colonies per
a
Fract. +/-S9 mg part, ug EOM cubic m min
Sample
Site No.
1
1
2,3 & 4
2,3 & 4
whole
whole
whole
whole
2,3 & 4 neutral
2,3 & 4 neutral
BH 1 whole
BH 1 whole
BH 2 whole
BH 2 whole
BH 3&4 whole
BH 3&4 whole
BH 5 whole
BH 5 whole
TIGF 1,3&4 whole
TIGF 1,3&4 whole
458 26.6 42800 276 x 10
339 19.7 31700 204 x 10
6.9 0.95
1.41
1.37
304 1.96 x 10
25.4 3.48 1115 7.18 x 10
5
353 48.3 15479 100 x 10
5
351 48 15383 99 x 10
5
2.95 1.2 98 0.12 x 10
5
6.11 2.49 202 0.24 x 10
1.03
1
0.44 1.07
0.94 2.3
2.69 1.82
1.77 1.2
7.78 0.75
18.77 1.81
23 0.03 x 10
22 0.03 x 10
27 0.03 x 10
59 0.07 x 10
21 0.03 x 10
14 0.02 x 10
465 0.56 x 10
1121 1.35 x 10
Stack gas emission rates (std. cubic meters/min): site A = 644;
site C = 120.6
423
-------
Table 4. TYPICAL MUTAGENIC EMISSION FACTORS FROM VARIOUS COMBUSTION SOURCES.
Revertants per
a b
Source hour kg fuel MJ
MUC; site A 1656 x 10 4.4 x 10 3.8 x 10
6 5 4
MUC; site A neutral fraction 600 x 10 1.6 x 10 1.4 x 10
6 5
HMPUC; site C 3.4 x 10 0.7 x 10 580
6
MUC (11) 1008 x 10
c 5 4
MUC (12) 1-12 x 10 1-10 x 10
Industrial and utility boilers
and power plants (16)
5
oil 0.03 x 10 70
5
coal 0.06 x 10 230
5
wood 0.20 x 10 1000
Automobiles and trucks (16)
5
diesel vehicles 40 x 10
5
diesel trucks/buses 40 x 10
5
gasoline-noncatalyst 10 x 10
5
gasoline-catalyst 1 x 10
6
wood stove (23) 6 x 10
6
gasoline car (24) 0.5 x 10
6
diesel car (24) 6 x 10
a
waste burn rates (kg/min): site A = 63; site C = 8.4
b
megajoule: literature value or calculated for MUC using 11.614 MJ/kg of
municipal solid waste (MSU)
c
calculated revertants/kg MSU using factor of 11.614 MJ/kg MSU
424
-------
SESSION IOC". FLUE GAS CLEANING SYSTEM PERFORMANCE
Co-Chairmen:
R. Michael Hartman
ABB Resource Recovery Systems
Windsor, CT
Charles B. Sedman
AEERL
U.S. EPA
Research Triangle Park, NC
425
-------
Intentionally Blank Page
426
-------
The work described in this paper was not funded by the U.S. Environmental
Protection Agency. The contents do not necessarily rellect the views of the
Agency and no official endorsement should be inferred.
A REVIEW
OF ACTIVATED CARBON TECHNOLOGIES
FOR REDUCING MSW INCINERATOR EMISSIONS
by
Mar*jor*ie J . Clarice
Environmental Consultant
1795 Riverside Drive, #5F
New York, NY 10034
ABSTRACT
Though activated carbon is, by no means, a newcomer to the pollution
control field, having been used as a water purifier and more recently
demonstrated as a flue gas cleaner on power plants, it is now attracting
considerable attention in Europe as a means to reduce further the quantity of
toxic organic and metal emissions from new and existing municipal waste
combustors. Since activated carbon is a potentially important future
emissions control technology for MWCs in the US, particularly for removal of
mercury and dioxin, this paper will discuss the impetus which has motivated
the experimentation with various activated carbon technologies which is now
taking place, will describe how some of the activated carbon systems (e.g.,
post-emissions control fixed carbon bed and injection of carbon with scrubber
reagent) being tested now function and where they fit in existing pollution
control trains, and will present available performance data and emissions
reductions actually achieved for each system.
Preceding page blank
427
-------
INTRODUCTION
Activated carbon technologies are, and have been for decades, commonly
used as a method for filtering a wide variety of pollutants from both gaseous
streams and effluent streams. In one database alone, EPA lists 171 oils and
hazardous materials, including pesticides, oils, fuels, organics (including
hexachlorobenzene and three chlorophenols), metallic compounds, radioactive
compounds, and various halides (chlorides, sulfates, fluorides), for which
activated carbon is listed as an appropriate adsorbent and control technology.
(Ref. 27)
THE BASIS FOR ACTIVATED CARBON TECHNOLOGY
Activated carbon, or activated coke, functions as a very effective
adsorbent due to the fact that its particles or fibers contain extraordinarily
large internal pore surface area which can bond adsorptively to a very broad
range of substances. The internal surface area of the active coke/carbon used
on large-scale applications today are between 300 to 800 m2 per gram of
activated carbon. (Ref. 22) Grains having an internal surface area up to
approximately 300 mz/gram are referred to as activated cokes; above this
value, and as high as 1500 mVg, the material is called activated carbon.
(Ref. 3) For purposes of reference, the 800 square meters of pore space in
one gram of activated carbon is equivalent to a tenth of a city block.
In addition to having large internal pore surface area on which to bond
impurities in gases, some activated carbons also possess catalytic properties.
For example, this effect has been used to achieve reduction of oxides of
nitrogen by adding ammonia or ammonium hydroxide. (Ref. 22)
Once the activated carbon has been in place for a period of time it
becomes saturated with the impurities it has been adsorbing. At this point,
it becomes necessary to regenerate the carbon.
HISTORY AND USES OF ACTIVATED CARBON TECHNOLOGIES
The first use of activated carbon for extraction of harmful substances
from flue gases and sewage occurred in 1909. Activated coke filters were
first used in respirators to purify air in World War I. (Ref. 22) Since then
much effort has been expended to develop more efficient and specialized
activated carbon products for distinctive tasks. Such diverse tasks have
included the purification of water for drinking water (Ref. 6), treatment of
industrial effluents, (Ref. 9) and the treatment of landfill leachate (Ref.
8). Other research has involved the study of process kinetics and the
development of models for predicting the removal efficiency of these
materials. (See Refs. 4, 18, 21, and 29) Investigations have been pursued to
determine the optimal configurations and methods of using activated carbons,
(see Refs. 14, 15, 16), and optimal conditions under which these materials
would produce the greatest removal efficiency, (see Ref. 12). Finally, still
further research has concentrated on methods of regenerating the adsorbent
(Refs. 20, 23, 24)
428
-------
In the last decade activated coke has been demonstrated as an air
pollution control device on various pilot and full-scale, large applications,
most commonly for power stations. The first large-scale activated carbon bed
using lignite coke was used on a power plant, located behind ten reactors,
each of which has a throughput of 250,000 Nm3/hour, at Garath, near
Duesseldorf. (Ref. 22) This pilot plant, with 65,000 m3/hr capacity was used
for S0B adsorption and NOx reduction. (Ref. 3)
UPSURGE IN INTEREST IN ACTIVATED CARBON FOR MSW INCINERATORS --
REGULATIONS IN EUROPE
Why all the current interest in activated carbon to reduce emissions
from incinerators? While it is true that advances over the last ten years in
combustion efficiency, furnace design, operating techniques, and add-on
emission control devices (e.g., multi-field ESP's, baghouses, and scrubbers)
have improved the state-of-the-art as much as two orders of magnitude insofar
as emissions are concerned, it is also true that the incineration industry is
facing an increasing challenge from solid waste solutions that are more energy
and natural resources-efficient and less environmentally deleterious on a
process lifecycle basis, and which are now officially recognized by EPA and
many states as superior: reduction, reuse, recycling, and organics
composting.
In Europe these realities, combined with a heightened public awareness
and concern about environmental issues in general, have contributed to the
promulgation of increasingly more stringent standards, regulations, and goals
for incinerator emissions. In Sweden and the Netherlands, increased standards
were the direct result of findings of unacceptably high levels of dioxins,
traced to MSW incinerator emissions, in the food chain. Thus, commencing with
Sweden in 1986, the Netherlands, Austria, Denmark, Germany, and Sweden have
adopted as either a standard or a goal a 0.1 ng/Nm3 dioxin toxic equivalent
level.
More stringent standards for heavy metals have also been issued.
Previous standards for mercury were as high as 130 ug/Nm3 for Austria,
Switzerland, and Germany. However, at present the emission limit is 50 ug/Nm3
in Austria for mercury and for cadmium/thallium. (Ref. 11) The maximum
emission limits during test sampling for new and existing incinerators set in
the Netherlands in 1989 also included a level of 50 ug/m3 each for cadmium and
for mercury. The Dutch government also set more stringent standards for NOx
(70 mg/m3 or 52 ppm corrected to 7% 02), S0Z (40 mg/m3 or 21 ppm corrected),
HC1 (10 mg/m3 or 9.4 ppm) and particulates (5 mg/m3 or 0.003 gr/dscf). (Ref.
iO)
In order for new and existing plants to meet these emission levels, the
recent state-of-the-art practiced in the U.S. (the dry scrubber/baghouse) is
not considered sufficiently reliable. Thus, various activated carbon and
other emission control technologies are being examined for purposes of
retrofitting older incinerators and complementing the conventional emissions
control devices on new plants so as to achieve further reductions in mercury,
dioxins, NOx, and other emissions of concern.
429
-------
The most recent baseline regulations issued in Germany (the 17th
regulation pertinent to the Bundes Emissionsschutz Gesetz — their Clean Air
Act), requires that new incinerators of municipal solid waste and similar
materials must, under normal operation, meet a 0.1 ng/Nm3 dioxin toxic
equivalent level and that they meet a level of 0.05 mg/Nm3 for Cadmium plus
Thallium, and 0.05 mg/Nm3 for Mercury (all at 11% 02 dry). The regulation for
another group of metals (Sb, As, Cr, Pb, Co, Cu, Ni, Mn, V, and Sn) is 0.'5
mg/Nm3, particulate, hydrocarbons, and HCl emissions are each limited to 10
mg/Nm3, the S02 limit is 50 mg/Nm3 and the HF limit is 1 mg/Nm3.
New plants (those with permits issued after December 1, 1990) will have
to meet the regulation immediately. Older existing incinerators (those with
permits issued under TA Luft 1974) will have to comply with these new
regulations by March 1, 1994; those newer existing incinerators (with permits
issued under TA Luft 1986 or those not issued permits prior to December 1,
1990) will have to comply by March 1, 1996. According to Bernd Johnke, an
engineer with the German environmental research agency (UBA), the regulations
were developed based on the information that sufficient technology exists now
for new plants to meet the standards, and it is fully expected that all plants
will be able to comply with the regulation.
A case in point is a newly constructed large-scale incinerator in Bonn,
which was designed with an ESP and wet scrubber. Since the facility did not
receive its permit before December 1, 1990, it is now being retrofit in order
to achieve the mercury and dioxin limits. The new system positioned after the
existing pollution control train involves a fabric filter (100°C inlet) into
which activated carbon and lime will be injected. (Ref. 13)
By way of comparison, the New Source Performance Standards (NSPS)
recently issued by USEPA are considerably higher for all pollutants of
concern. For example, the final standards require only large plants to meet a
total dioxin/furan limit of 30 ng/dscm. Though these units are not directly
comparable to toxic equivalents, it is safe to say that this new U.S. limit is
at least one and possibly two or more orders of magnitude higher than those
currently in effect for all plants in several European countries based on data
elsewhere in this paper.
The NSPS for specific heavy metals have not yet been established, but
draft standards for mercury, cadmium, and lead are expected soon, since the
new Clean Air Act requires that limits for these metals be promulgated by
November 15, 1991. Some discussion has centered around setting the U.S.
standard to equal the previous European standard of roughly 130 ug/dscm, a
value over two and one-half times the current European standard. The NSPS for
NOx is If0 ppmv (7% 02) 24-hour block average. This level is over three times
the Dutch limit.
430
-------
A DESCRIPTION OF THE ACTIVATED CARBON TECHNOLOGIES
INJECTION OF ACTIVATED CARBON WITH LIME
This technology is essentially a hybrid of two emissions control
mechanisms: alkaline absorption and activated carbon adsorption. After
injection of Sorbalit and hydrated lime, dust collection onto the activated
carbon begins in the scrubber and continues in the flue gases downstream. Of
similar significance insofar as adsorption of pollutants is concerned, is the
further adsorption which takes place on the cake which adheres to the fabric
filter. Unfortunately, ESP's do not have the latter capability. After the
carbon is injected, some is collected, some remains in circulation, allowing
for fresh carbon to replace saturated carbon.
Well suited to retrofit into systems already possessing an acid gas
scrubber, injection of activated carbon has been demonstrated on a number of
pilot and existing full-scale plants. According to Flakt removal efficiency
for dioxins and for mercury corresponds to that obtained with granular static
bed filters of activated coke. In addition, Flakt contends that injection of
activated carbon uses less activated carbon than the static carbon bed system.
FINAL STAGE ACTIVATED CARBON BED
Much of the experimentation with "back-end" activated carbon beds for
MSW Incinerators has been in Germany and Austria. There are mainly two types
of active coke/carbon on the market today having similar adsorptive capacity,
but different catalytic activity. The formation coke, marketed by Mining
Research, and based on mineral coal, has a higher catalytic activity and is
also ten to twenty times as expensive as its chief competitor, the hearth-type
furnace coke sold by Rheinbraun, which is based on lignite. This activated
carbon bed system is positioned after all other add-on emissions control
devices in a pollution control train.
Flue gas cleaning is accomplished by drawing the flue gas through the
bed, whereupon the pollutants of concern accumulate on the abundant surfaces
of the pore spaces. From investigations by the firm Hugo Petersen and the
Technical University of Berlin it was found that mercury, other heavy metals,
and large molecular hydrocarbons, such as dioxin and furan, are adsorbed in
the first layers of the reactor. Sulfur dioxide is displaced by heavy metals
in the coke pores and then redeposited in the following layers. HCl and HF
are, in turn, displaced by S02 from the pores and are adsorbed in the
following, unfilled layers of the bed. All pollutants are fully adsorbed when
the filter is appropriately large. (Ref. 3)
One present potential drawback of upscaling the technology is the
possibility of fires, particularly in a bed sized for a full-scale, large
incinerator. Activated carbons are naturally flammable because of their high
carbon content, and may tend to self-ignition at temperatures from 80 to 400°C
when oxygen is present. (Ref. 3) According to Justus Engelfried of the EPEA
Umwelt Institut in Hamburg, a large incinerator processing 300,000 TPY would
require an activated carbon bed of 1000 square meters 40 meters high.
However, the ignition point of the furnace coke is at least 390°C and of the
431
-------
formation coke is roughly 350°C (Ref. 19), so in back-end installations, where
the operating temperature of the bed is usually less than 140°C and sometimes
as low as 110°C, the risk would seem to be small under normal operating
conditions. In addition activated carbon beds can be monitored for
temperature and the difference in CO before and after the reactor. However,
in one case a fire in an activated carbon bed was reported after major upset
conditions lasting 24 hours caused back-end temperatures to rise. (It was
easily extinguished.) This experience underscores the need for well-trained,
attentive operators. (Ref. 19)
REGENERATION/DISPOSAL OF ACTIVATED CARBON
Several alternatives are available for treating and/or disposing of
saturated activated carbon. The total amount of charged or used-up coke
represents about 1% in relation to the total amount of solid waste processed.
One method of disposal involves depositing the saturated carbon in an
incinerator. This is deemed economically advantageous and would result in a
destruction of organics collected by the carbon. It is important, however,
that sufficient emissions reduction potential exists in the incinerator's
pollution control train to absorb the additional load represented by the
disposed activated carbon. A method for regeneration involves separate
thermal treatment in a moving bed desorber, where the carbon is brought into
contact with a circulating gas whose temperature lies in the range of 350°C to
500°C. (Ref. 3) This process desorbs the mercury and catalytically reduces
the dioxins, with subsequent condensation at 20°C at which point the mercury
precipitates. (Ref. 22) A third method involves mechanical cleaning or
desorption to regenerate the coke and retain its properties. (Ref. 19) A
fourth alternative is to use the saturated carbon as an additive for
converting flue dust into glass, for example, using the ABB process DEGLOR.
(Ref. 2) Because the furnace coke is so inexpensive, it is usually
incinerated, however, the formation coke is usually regenerated in a
mechanical screening process. (Ref. 19)
A DESCRIPTION OF PILOT AND FULL-SCALE INVESTIGATIONS TO-DATE
INJECTION OF ACTIVATED CARBON
NIRO Atomizer
One of the early innovators in applying activated carbon technology to
MSW incineration was NIRO Atomizer. At the Zurich Josefstrasse plant, a full-
scale plant with a large conventional ESP, later retrofitted with a NIRO spray
dry scrubber, activated carbon injection testing began in 1987. Four sets of
samples were collected at the inlet and outlet of the SDA system and analyzed
for total dioxin content. Table 1 indicates the inlet, outlet, and removal
efficiencies for the tests which measured the effectiveness of lowering outlet
temperature from 285°F to 245°F and the injection of various concentrations of
activated carbon. In addition, both toxic equivalents and total dioxins were
measured. At low SDA outlet temperatures it was found that the use of small
amounts of activated carbon substantially lowered dioxin emissions to as
little as 0.08 ng/Nm3 Toxic equivalents and improved system collection
efficiency to as much as 99.6%. (Ref. 5)
432
-------
Table 1. Zurich Pilot Plant Dioxin Emissions and Removal
SDA Outlet Temp. °F
248
120
0
248
120
0.75
285
140
0
285
140
0.25
Additive Rate gr/acf
Dioxin Emissions, ng/Nm3
2,3,7,8 - TCDD Eadon Equiv.
Total Dioxin + Furan
1.3
60
0.08
5.0
1.5
66
0.84
35.2
Dioxin Removal %
2,3,7,8 - TCDD Eadon Equiv.
Total Dioxin + Furan
91.3
92.6
99.5
99.6
92.4
92.3
90.3
90.3
On the basis of these experiments, the 400 TPD Zurich incinerator was
retrofit with this technology to increase the level of control for mercury and
dioxin and was tested for both of these pollutants. The system for injection
of dry activated carbon is situated upstream of the spray dryer absorber. The
carbon then mixes intimately with the flue gas and lime slurry droplets inside
the SDA providing optimum conditions for contact between the activated carbon
and the dioxins and mercury present in the flue gas.
After start-up the retrofit was tested to verify pilot plant figures.
Though it was determined that the lower the SDA outlet temperature the lower
the dioxin and mercury emissions (the system was successfully operated at
temperatures as low as 110°C), the removal achieved without using activated
carbon was not sufficient to meet the then applicable 0.1 mg/Nm3 mercury
outlet limit (note the current limit is 0.05 mg/Nm3). Table 2 shows the
summary of results of 25 mercury tests at Zurich with and without using
activated carbon. (All mercury measurements were made simultaneously at the
SDA inlet and in the stack, and are assumed to be 11% 02.) These data show
that there is a significant effect of using activated carbon. Also, when
using activated carbon injection, a low mercury emission was achieved
regardless of fluctuations in the inlet mercury level. For example, at 30
mg/Nm3 carbon injection level and at an outlet temperature of 110°C, the inlet
mercury level fluctuated from .131 to .650 mg/Nm3 and the outlet concentration
varied between .040 and .068 mg/Nm3 (approximately the current regulation
level) with an average removal rate of 83%. (Ref. 7)
The dioxin emissions at the Zurich retrofit were measured by the Danish
EPA-recognized firm, dk-Teknik, Denmark, sampled according to the Nordic
protocol, and analyzed by Umea University. Steady-state conditions, at a
constant, low CO level (20-30 ppm dry) were maintained for an extended time
before measurements were taken. The results showed that without the activated
carbon the quenching of flue gas temperature (from roughly 210°C to 130°C) in
the scrubber resulted in a 75% removal efficiency for dioxin. But by adding
activated carbon, removal efficiency is increased to 90% at 140°C using 18
mg/Nm3 activated carbon, and to 98.5% removal at 120°C and 59 mg/Nm3 activated
carbon. Table 3 summarizes these results. (Ref. 7)
-------
Table 2. Mercury test results at Zurich Josefstrasse
Hg concentration ug/Nm3 dry
SDA
outlet Additive SDA ESP % removal
temp.°C mg/Nm3 Inlet outlet
140 0 537 390 27
343 237 31
680 417 39
558 414 26
406 335 17
1072 769 28
Average 599 427 28
120 30 539 39 93
589 31 95
Average 564 35 94
115 0 495 232 53
643 207 68
234 117 50
949 670 29
736 476 35
401 250 38
Average 576 325 44
115 30 352 44 88
353 44 88
281 29 90
Average 329 39 89
110 0 249 124 50
224 132 41
346 212 39
Average 273 156 43
110 30 486 68 86
650 45 93
131 44 66
417 40 90
269 51 81
Average 391 50 87
434
-------
Table 3. Dioxin Test Results at Zurich Josefstrasse
Total PCDD -r PCDF Eadon Tox. Equivalents
SDA Active
outlet carbon
temp. mg/Nm3 SDA ESP % SDA ESP %
°C inlet outlet rem. inlet outlet rem
140 0 306 77 74.8 7.7 1.9 75.3
18 223 33 85.2 7.5 0.79 89.5
120
0
59
277
455
69
5.0
75.1
98.9
6.9
6.0
1.8
0.09
73.9
98.5
In 1988 NIRO also retrofit its activated carbon injection technology on
the 300 TPD Volund rotary system incinerator at Amager in Copenhagen.
Downstream of the carbon injection is a SDA and pulse-jet fabric filter. As
with the Zurich incinerator, tests showed that with no activated carbon
injected, the reduction of SDA outlet temperature had a clear effect on
mercury removal. However, the summary of the 19 tests of this full-scale
system undertaken in 1989, Table 4, shows that with sufficient additive
injection (as much as 70 mg/Nm3) coupled with low outlet temperature (as low
as 127°C), mercury emissions were reduced by as much as 97% and were
controlled to a level one-tenth of the new 0.05 mg/Nm3 German limit. (Ref. 7)
Table 4.
Summary of Mercury Test Results - Amager, Denmark
Mercury concentrations ug/Nm3, 10% C0Z
SDA Activated
outlet Carbon
temp °C mg/Nm3
SDA inlet
FF outlet
% removal
140 0
171
97
43
6
255
42
84
17
224
28
87
58
633
37
94
120
0
19
70
154
169
176
37
20
5
76
88
97
The dioxin sampling and analysis for Amager were the same as for Zurich.
The dioxin removal figures (Table 5) show that outlet emissions without
activated carbon are reduced about 50% by lowering the outlet temperature from
140°C to 127°C. Injection of activated carbon increased dioxin removal
efficiency further to virtually 100%. It is important to note that in one
test, where start-up conditions were simulated, the inlet dioxin value
increased ten-fold to 50 ng/Nm3, but the outlet value was 0.05 ng/Nm3 Toxic
equivalent, half the new German regulation. (Ref. 7)
435
-------
Table 5.
Dioxin Test Results - Amager
SDA Activated
outlet Carbon
temp. mg/Nm3
°C
Total PCDD + PCDF
ng/Nm3, dry
SDA FF %
inlet outlet rem.
Nordic Toxic Equiv.
ng/Nm3, dry
SDA FF %
inlet outlet rem.
0
132
2.1
98.4
2.8
0.076
99.7
6
283
1.2
99.6
4.8
0.0075
99.8
17
276
2.4
99.2
8.3
0.045
99.5
58
201
1.1
99.5
4.0
0.035
99.1
58
2170"
3.2
99.9
50.0
0.050
99.9
127 0 254 1.3 99.5 7.7 0.0047 99.9
19 154 0.4 99.8 5.0 N.D. 100
70 154 0.7 99.6 4.5 0.002 100
"" Simulated Start-up condition
The Kassel, Germany activated carbon retrofit installation, achieved
similar results to Zurich and Amager with up to 97% removal and as little as
.005 mg/Nm3 outlet value for mercury at 127°C and 70 mg/Nm3 activated carbon
added (see Table 6).
Table 6. Summary of Mercury Test Results - Kassel, Germany
Mercury Concentrations ug/Nm3, dry
Activated
Carbon SDA inlet FF outlet % removal
mg/Nm3
137 0
898
582
35
9
336
175
48
20
324
57
82
47
179
19
89
64
297
52
82
The dioxin results were also similar with up to 99% removal for dioxin/furan
and as little as 0.07 ng/Nm3 emission for TCDD Toxic equivalents (see Table
7). (Ref. 7)
It was concluded from these studies of Zurich, Amager, and Kassel that
the injection of activated carbon upstream of a spray dryer absorber results
in a very low dioxin emission regardless of operating mode (single pass as at
Zurich and Amager vs. recirculation as at Kassel), type of dust collector (ESP
as at Zurich and fabric filter as at Amager and Kassel), spray dryer outlet
temperature, and upset conditions. Insofar as mercury removal is concerned, a
combination of low flue gas temperature and the presence of activated carbon
in the scrubber reagent corresponded to an increase in removal efficiency and
SDA
outlet
temp.°C
436
-------
a decrease in emissions, in many cases below the current European emission
standard for mercury of 50 ug/Nm3.
Table 7. Dioxin Test Results - Kassel, Germany
Total PCDD + PCDF Toxic Equiv.
SDA Activated ng/Nma ng/Nm3
outlet Carbon
temp. mg/Nm3 SDA FF % SDA FF %
°C inlet outlet rem. inlet outlet rem.
0
380
151
60
9.58
3.46
64
19
134
12
91
3.21
0.19
94
19
238
8
97
5.11
0.15
97
47
298
9
97
5.53
0.13
98
105
359
7
98
5.94
0.07
99
Research-Cottrell/Teller
Another of the activated carbon injection systems, the Research
Cottrell-Teller system, (Ref. 25) has achieved significant reductions of
mercury from a medical waste incinerator in Skovde, Sweden. This incinerator
processes waste at a rate of roughly 25 tons per day, the combustion
temperature is approximately 1000°C achieving a 99.99% reduction in
combustibles; and the temperature of the flue gas is reduced by thermal
recovery to 140-150°C. The all-dry emission control system has two basic
components: (1) a Dry Venturi, into which are blown hydrated lime, activated
carbon reagent, and Tesisorb bagcake modifier, and (2) a pulse jet fabric
filter, where the cake is permitted to accumulate for 4 to 6 hours between
cleaning cycles. The bag filter with a 3-10 mm cake, in essence, serves as
the major reactor for acid gas, mercury, and dioxin recovery. The activated
carbon was considered more suitable for this application because the
anticipated flue gas temperature (135-150°C) exceeded the flyash adsorption
temperature.
The performance tests for the Skovde Hospital Incinerator were conducted
by Miljokonsulterna of Nykoping, Sweden, a testing group approved by the
Swedish environmental authorities. For mercury, inlet values ranged from 294
to 10,170 ug/Nm3, (as much as ten times as much as expected in the design
stage) and outlet values ranged from 5 to 25 ug/Nm3 (all values at 10% C02).
The mercury guarantee was 80 ug/Nm3. Table 8 shows the results of the four
tests broken down for inlet, outlet, and percent removal for mercury metal,
salts, vapor, and total.
With respect to dioxin, two test results in Table 9 showed the
considerable effect on removal efficiency of adding an activated
carbon/lime/Tesisorb reagent, with stack temperatures in the range of 136 to
138°C. Overall removal efficiency was 93% to 99%, and the addition of carbon
more than doubled removal efficiency achieved by the Tesisorb and lime alone.
437
-------
Table 8. Specific Mercury Emissions at Skovde Hospital Incinerator
Inlet
Outlet
ug/Nm3 (10%C0j,)
ug/Nm3
(10%COH
Hg Hg Hg
Hg
Hg
Hg
Metal Salts Vapor
Total
Metal
Salts
Vapor
Total
3.0 286 4.9
254
0.05
7.3
1.6
9.0
19.0 9993 160
10510
0.05
23
1.5
25
2.6 705 191
922
O
o
3.0
3.5
6.5
3.0 389 122
514
o
o
2.8
1.2
4.5
Hg Metal
98.3
99.7
99.6
99.7
Hg Salts
97.4
99.8
99.6
99.3
Hg Vapor
77.3
99.1
98.2
99.1
Total
97.0
99.8
99.3
99.1
Table 9. Dioxin Emissions/Removal at Skovde Hospital Incinerator
ng/Nm3 (10% C02) TCDD Equivalent
Test 2
Test 4
Test 5
Activated carbon kg/hr
0
2
0.2
Ca(0H)2 kg/hr
10
39
60
Tesisorb kg/hr
2.2
10.5
o
o
Inlet TE
5.7
9.2
21.3
Outlet TE
2.5
0.6
0.2
Removal
44%
CO
CO
CO
&
99.1%
The dioxin TCDD equivalents guarantee was 2 ng/Nm3. Chlorobenzenes were
reduced in two tests 75% and 94%.
Flakt
One of the first tests of the activated carbon/lime injection systems on
an old municipal solid waste incinerator in Geiselbullach, in Bavarian
southern Germany, began in January 1989. At this plant two lines process 144
TPD. Used in this system was a reagent consisting of 95-97% lime and 3-5%
activated carbon. Since this was one of Flakt's first tests, the incinerator
outlet temperature of 200 - 220°C was not lowered, involving a slight risk of
auto-ignition (Ref. 2). Three tests for dioxin removal efficiency were
conducted during the period the activated carbon system was being evaluated.
For all tests inlet values were below 5.0 ng/Nm3 Toxic Equivalents and all
outlet values met the 0.1 ng/Nm3 limit. (Ref. 13) Additionally, measurements
taken by Prof. Hutzinger through the TUV Bayern showed inlet values of 2.2 ng
TE/m3 and outlet values clearly below 0.1 ng TE/m3, demonstrating that
Sorbalit, even in high temperatures of 180 to 220°C, could be suitable for
removal of dioxin and furan from flue gas. (Ref. 17)
438
-------
With respect to mercury, in spite of the high temperatures, which
occasionally reached 240°C, the mercury emission which had normally been in
the vicinity of 0.2 to 0.25 mg/m3, clearly sunk to below 0.1 mg/m3. (Ref. 17)
Another plant which has had success in using the activated carbon
scrubber additive, Sorbalit, in 2% proportion to 98* lime, is the municipal
waste incinerator Berlin-Ruhleben. This plant has seven lines at 12 tons/hour
each, totalling over 2,000 TPD. Each line processes about 50,000 to 60,000
m3/hour of flue gas. In 1988 the plant was retrofitted with a Flakt spray dry
absorber and fabric filter to replace the Lurgi Electrofilter. The full-scale
test of the efficiency of Sorbalit involved over 40 tests for mercury and five
isomer-specific tests for dioxin/furan, using two boilers. (Ref. 28) ITU
conducted both the sampling and analysis for the German Environmental Agency
(UBA) using the VDI 3499 regulation.
The mercury inlet values (all at 11% 0a, dry) ranged from 165.7 ug/Nm3
to 546.1 ug/Nm3, averaging 339 ug/Nm3 for Boiler 2 and 307 ug/Nm3 for Boiler
3. The outlet values ranged from 31.3 ug/Nm3 to 111.7 ug/Nm3. averaging 75.5
ug/Nm3 for Boiler 2 and 63.7 ug/Nm3 for Boiler 3. Thus, the removal
efficiency for mercury in these tests were 78% for Boiler 2 and 79% for Boiler
3. (Ref. 28)
The five dioxin tests at Berlin-Ruhleben, summarized in Table 10,
demonstrated consistent achievement of the 0.1 ng/Nm3 limit.
Table 10.
Dioxin Emissions and Removal Rate for Berlin-Ruhleben
5/7/90
boiler 3
5/9/90
boiler 3
5/11/90
boiler 3
5/14/90
boiler 3
5/17/90
boiler 2
Inlet
Sum PCDD+PCDF 212.56 297.88 1042.41 442.74 214.28
Int. Toxic Eq. 4.82 6.93 23.77 10.01 4.67
Outlet
Sum PCDD+PCDF 2.29
Int. Toxic Eq. 0.034
2.73
0.044
5.15
0.082
4.54
0.062
1 .58
0.022
Removal Efficiency
Sum PCDD+PCDF 98.92% 99.08% 99.51% 99.98% 99.26%
Int. Toxic Eq. 99.29% 99.36% 99.65% 99.38% 99.53%
Another plant at which Sorbalit is used is the 15,000 TPY SVA Schoneiche
special waste incinerator with rotary kiln and secondary combustion chamber,
and Flakt spray dry absorber system with fabric filter. This was the first
full-scale application of Sorbalit to a special waste incinerator. Sorbalit
with 2% activated carbon as well as sodium sulfide has been continuously used
since 1989, after the plant began operations and initial tests had showed
"increased mercury concentrations and ... indefensible dioxin and furan
emissions". (Ref. 26) In order to reduce the risk of auto-ignition of the
carbon, the maximum bag filter temperature used is 160°C. (Ref. 2)
439
-------
An extensive program of measurements was undertaken during the months of
March, April, August and September 1989 and January and February of 1990.
Though acceptable results were obtained in the first two months, the
combustion of automobile shredder waste with a mercury content of up to 61.8
mg/kg caused unacceptably high mercury emissions for the August tests. In
addition, the four August tests showed dioxin/furan BGA toxic equivalent
emission levels of between 3 and 6 ng/Nm3. As a result, the lime was
exchanged for Sorbalit and sodium sulfide, and removal efficiency and
emissions improved considerably. By December 1989 a removal efficiency for
mercury of more than 90% had been achieved with inlet levels ranging from 175
to 1200 ug/Nm3 (averaging closer to 300 ug) and outlets ranging from less than
5 to 50 ug/Nm3, averaging about 25 ug/Nm3. By January 1990 dioxin/furan BGA
toxic equivalent emissions had been reduced by two orders of magnitude to
below 0.1 ng/Nm3, with one test at that level and three others below 0.03
ng/Nm3. Total dioxin was below the detection limit and total furan was below
0.025 ng/Nm3. (Ref. 26)
Other plants have been tested with activated carbon injection. These
include the Palm Beach and SEMASS RDF incinerators, reported by J0Y/NIR0 in
another paper at this conference, as well as installations by LURGI and
Research-Cottrell/Teller to be reported at the 1991 AWMA Meeting in Vancouver.
FINAL STAGE ACTIVATED CARBON BED
Hugo Petersen
To take advantage of the different properties and costs of these two
types of activated carbon, one firm, Hugo Petersen of Wiesbaden, Germany, has,
during the past seven years, developed and tested a two-stage active coke
filter for application on natural gas, fuel oil, and coal-fired plants,
municipal solid waste incinerators, hazardous waste incinerators, and sewage
sludge boilers. (Ref. 3) The first stage consists of the cheaper furnace coke
and serves to extract particulate, oxides of sulfur, sulfuric acid,
hydrochloric acids, hydrofluoric acids, dioxins and furans, PAH's, PCB's,
mercury, and other heavy metals. The second stage, usually consisting of the
more expensive, but more highly catalytic formation coke, serves to reduce
nitrous oxide via the injection of ammonia/ammonium hydroxides. The water
vapor content of the flue gas does not seem to affect the efficiency of
adsorption, since the adsorption of water vapor takes place at temperatures
below 80°C. (Ref. 22)
The first tests of the Hugo Petersen activated carbon beds on solid
waste incinerators were carried out on two plants in Germany: at Hamburg-
Stapelfeld using the catalytic formation coke, and at Duesseldorf-Flingern
using hearth-type furnace coke. Both tests indicated similar removal
efficiencies for a.'.l substances sampled. (Ref. 22)
The tests at Flingern were carried out using a hearth-type furnace coke
with basic fixed-bed filters; subsequently, a multi-channel filter with
furnace coke was used. According to Bernd Johnke, two coke bed depths were
tested: 75 cm and 150 cm. With an inlet TCDD Toxic Equivalent value of 3
ng/Nm3 after a semi-dry scrubber, both carbon beds achieved outlet values of
440
-------
between 0.02 and 0.06 ng/Nm3 Toxic Eq. (a removal efficiency of 98% to 99.3%).
Air volume intakes on the pilot bypass facility were 1600 m3/hour. (Ref. 22)
The tests at Stapelfeld were carried out on a fixed-bed reactor with a
volume intake of approximately 100m3/hour with the flue gas at roughly 140°C.
The test period was over three months with a total of 1,467 operating hours.
The following table shows the PCDD/PCDF removals achieved at Stapelfeld in
ng/Nm3 (Ref. 22 - Test l)(Ref. 3 - Test 2)
Table 11. Dioxin emissions/removal at Stapelfeld
Pollutant INLET OUTLET REMOVAL
EFFICIENCY
Test 1
Sum PCDDs 27.10 < 0.134 99.5%
Sum PCDFs 102.47 < 0.010 99.9%
TCDD Toxic Equivalents 2.633 <0.0061 99.8%
(International System)
Test 2
Sum PCDDs 51.4 0.091 99.8%
Sum PCDFs 216.4 0.007 100%
Reductions in PCB concentration was equally impressive with a range of
99.1 to 99.5% removal efficiency over five tests.
With respect to mercury, tests were carried out on Stapelfeld finding
inlet values ranging from 61 to 129 ug/Nm3, averaging 83 ug/Nm3, and outlet
values ranging from <.4 to 7.1 ug/Nm3, averaging 1.74 ug/Nm3. The removal
efficiency ranged from 90 to over 99%, averaging 97.4%. (Ref. 22)
Though the activated carbon bed has been shown to reduce significantly
the emissions of dioxin and mercury, it is also capable of reducing nitrous
oxide. At the Garath power plant, when ammonia/ammonium hydroxide was added
to the activated carbon, NO emissions were reduced by about 50% using hearth-
type furnace coke, achieving values of 150-220 mg/Nm3 (171-251 ppm corrected
to 7% Oz). Using the more highly catalytic formation coke, emissions were
reduced to 100 mg/Nm3 (114 ppm at 7% 02).
SGP-VA
The Austrian company, Simmering Graz Pauker, has been using back-end
activated carbon bed technology for reducing dioxin and furan emissions since
1987, and tests have been conducted on a pilot plant (processing 10,000 to
15,000 Nm3/hr) at the EBS special waste incinerator in Vienna and also at a
laboratory scale (200 Nm3/hr). The EBS incinerator was also equipped with an
efficient three-stage scrubber.
Test results at the Vienna pilot plant (Ref. 11), operated from January
to December 1990, given in original units, are summarized below:
441
-------
Table 12. Dioxin and Mercury emissions and removal at Vienna
Inlet
Outlet
Removal
Sum PCDD+PCDF ng/Nm3
TCDD International Toxic Eq.
291.645
4.2675
4.008
0.0530
98. Q%
98.8%
Mercury mg/Nm3
up to 0.5 0.01 - 0.03 94-98%
Other MSW incineration plants in planning and under construction which
employ the SGP-VA activated carbon bed (Ref. 11) include:
6 lines, 155,000 Nm3/hr each, with heat recovery
ESP operating at 170°C
2-stage wet scrubber,
first wet stage for HC1, HF, and heavy metals;
second alkaline stage for S02 and HF, operated at 60°C
Activated carbon adsorber - 5 x 15 x 16 meters, for dioxin, and
remaining HC1, HF, S02, and heavy metals, operated at 110°C
SCR-DeNOx operated at 180°C, followed by
A stack and water treatment plant.
Start-up for first two lines in January, 1992. (Ref. 1)
o Mannheim, Germany - 1 line, 155,000 Nm3/hr, with baghouse, 2-stage wet
scrubber, activated carbon adsorber, and SCR-DeNOx. Start-up in 1993.
Other planned installations include activated carbon adsorbers on a new
hospital waste incinerator in Dordrecht, Netherlands (start-up August, 1991),
a totally reconstructed hospital waste incinerator at the University of
Heidelberg (start-up March 1991), and at a special waste incinerator in
Rotterdam (start-up December 1991).
It is clear that use of activated carbon to reduce the emissions of
mercury and dioxin/furan as well as other pollutants from MSW and other
incinerators is well on its way to becoming a new state-of-the-art. Both the
injection and the fixed-bed technologies have provided an additional one to
two orders of magnitude of reduction in these emissions above what has already
been achieved by using more conventional emission control devices such as wet
or dry scrubbers and dust collectors (ESPs and fabric filters). In addition,
both technologies have achieved emissions levels considerably lower than the
new U.S. standard for dioxin using a variety of existing emission control
configurations.
Table 13 summarizes the lowest achieved emissions from the 12 activated
carbon-equipped plants/pilots/units for which data was available. The table
o
AVR Rotterdam
CONCLUSIONS
442
-------
Table 13. A PARTIAL LISTING OF MERCURY AND DIOXIN REMOVAL AT
INCINERATORS USING ACTIVATED CARBON TECHNOLOGIES
WITH REFERENCE NATIONAL STANDARDS
Plant/Country Year Mercury emissions Dioxin Emissions
ug/Nm3 ng/Nm3
EMISSION STANDARDS Total Toxic Equiv.
Netherlands 50 0.1
Austria 50 0.1
Denmark 0.1
Germany 50 0.1
Sweden 0.1
U.S. 30
ACTIVATED CARBON INJECTION INSTALLATIONS (lowest reported values)
Zurich, Switz pilot1 1987
Zurich, Switz retrofit2•14 1987
Amager, Denmark 3-ls 1988
Kassel, Germany °-s 1987
Skovde, Sweden 7
Geiselbullach, Germany ° 1989
Berlin-Ruhleben Unit 2 1990
Berlin-Ruhleben Unit 3 12 1990
Schoneiche, Germany 10 1989
5.0
0.08
31
5.0
0.09
5
0.4
<0.002
19
7.0
0.07
4.5
0.2
<100
<0.1
41.1
1.58
0.022
31.3
2.29
0.034
5
0.03
FIXED BED ACTIVATED CARBON INSTALLATIONS (lowest reported values)
Flingern, Germany
Stapelfeld, Germany 11 1987
Stapelfeld, Germany
Vienna-EBS 1990
0.02
<0.4 <0.144 <0.007
0.098
0.01 4.008 0.053
Notes: mercury test dioxin test
1.
Outlet
temp:
120°C
Injection
rate:
0.75 gr/acf
2.
Outlet
temp:
120°C
Injection
rate:
30
mg/Nm3
59 mg/Nm3
3.
Outlet
temp:
127°C
Injection
rate:
70
mg/Nm3
19 mg/Nm3
5.
Outlet
temp:
137°C
Injection
rate:
47
mg/Nm3
6.
Outlet
temp:
135°C
Injection
rate:
105 mg/Nm3
7.
Outlet
temp:
140°C
Injection
rate:
n. a.
8.
Outlet
temp:
136°C
Injection
rate:
0.2 kg/hr
9.
Outlet
temp:
220°C
10.
Outlet
temp:
160°C
11.
Outlet
temp:
140°C
12. All four dioxin tests on this unit were at or below .082 ng/Nm3 TE and
5.15 ng/Nm3 PCDD + PCDF;
11 mercury tests fell at or below 50 ug/Nm3.
13. All four mercury tests were at or below 25 ug/Nm3 with three below 10
ug/Nm3
14. Three other mercury tests at this condition met the 50 ug/Nm3 standard.
15. The two mercury tests at this condition were 5 and 6 ug/Nm3; all seven
dioxin tests using activated carbon were at 0.05 ng/Nm3 TE or less and at
3.2 ng/Nm3 or less for PCDD + PCDF.
443
-------
indicates the outlet flue gas temperature and the amount of activated carbon
required to achieve these values. Generally, outlet temperatures were in the
120 to 140°C range. All mercury emission values are below the current
European standard of 50 ug/Nm3 and all but one were below the 0.1 ng/Nm3
standard for TCDD Toxic Equivalents. Though these figures are the lowest
values reported in a series of tests, many other samples also met the current
limits as indicated in the notes for Table 13.
Is activated carbon technology the wave of the future for minimizing MSW
incinerator emissions? It certainly appears to be the case for Europe.
However, for the United States, since the regulatory authorities are generally
not requiring nearly as stringent emission limits for new or existing plants
as their counterparts Europe, there is less impetus for vendors to design
plants to achieve the lower emission rates that are being demonstrated over
the past few years in Europe. Therefore, it may be some time before those who
purchase incinerators in this country become aware of and create a widespread
demand for these new technologies.
The work described in this paper was not funded by the U.S. Environmental
Protection Agency and therefore the contents do not necessarily reflect the
views of the Agency and no official endorsement should be inferred.
ACKNOWLEDGEMENTS
I am appreciative of several people who went out of their way to provide
me with hard-to-find information: Barbara Zeschmer-Lahl, Bernd Franke,
Michael Wilken, Bernd Johnke, Bert Brown, E. Holzerbauer, Kurt Carlsson,
Walter Panknin, Aaron Teller, Mike Johnston, David Minott, David Hoecke, and
Justus Engelfried.
REFERENCES
1. AVR (Afvalverwerking) Rijnmond advertising literature for Rotterdam MSW
incinerator.
2. Carlsson, Kurt "Flakt Filsorption Process for the Collection of Dioxins
after Refuse Incineration Plants"
3. Cleve, U., "Application of Carbon Based Adsorbers for Washing of Flue
Gases", HUGO PETERSEN, Wiesbaden, Germany, presented at the 1989
Incineration Conference, Knoxville, TN., May, 1989.
4. Costa, Enrique et. al., "Kinetics of adsorption of phenol and P-
nitrophenol on activated carbon" Proceedings of the Second Engineering
Foundation Conference on the Fundamentals of Adsorption Santa Barbara,
CA May 4-9, 1986. Published by Engineering Foundation, New York, NY
1987 pp. 195-198.
444
-------
5. Donnelly, J.R., and Jens Moller, "JOY/NIRO MSW Incinerator FGC Systems
European Experience - An Update", Air Pollution Control Association 81st
Annual Meeting, Dallas, Texas, June 19-24, 1988.
6. Falcones, Ian R., et. al., "Using activated carbon to remove toxicity
from drinking water containing cyanobacterial blooms" Journal of the
American Water Works Association, v. 81 #2, Feb. 1989, pp. 102-5.
7. Felsvang, K. S. et. al., "Control of Mercury and Dioxin Emissions from
European MSW Incinerators by Spray Dryer Absorption Systems using Rotary
Atomizers", presented at AIChE's 1990 Summer National Meeting, San
Diego, CA August 19-22, 1990
8. Forgie, D.J.L. "Selection of the most appropriate leachate treatment
methods. Part 2: A Review of recirculation, irrigation, and potential
physical-chemical treatment methods" Water Pollution Research Journal
of Canada, v. 23 #2, 1988 pp. 329-40.
9. Gupta, G. S., et. al., "Treatment of effluents of carpet industry in
Bhadaki" Research and Industry v. 33 #2, June 1988 pp. 132-8.
10. Holmes, William personal communication. (Appendix I of letter DGM/A, #
0779512 - 7/19/89 from Mr. Compaan, Dutch Environmental Ministry)
11. Holzerbauer, Dr. E., SGP-VA, personal communication, February 12, 1991.
12. Ibrado, A. S. and D. W. Fuerstenau "Adsorption of the cyano complexes of
Ag (I), Cu (I), Hg (II), Cd (II), and Zn (II) on activated carbon",
Minerals and Metallurgical Processing v. 6 #1, Feb. 1989 pp. 23-8.
13. Johnke, Bernd. Dipl. Ing. with the Umwelt Bundes Amt, Berlin, Germany.
Personal communication, February, 1991.
14. Kaneko, K. "Dynamic Hg (II) adsorption characterization of iron-oxide-
dispersed activated carbon fibers", Carbon v. 26, #6, 1988, pp. 903-5.
15. Kaneko, K. et. al., "Chemisorption-assisted micropore filling of NO on
Cu, Ni, and Co oxide-dispersed activated carbon fibers" Applied Surface
Science (1985)
16. Kusakabe, Katsuki, et. al., "Rate of Reduction of nitric oxide with
ammonia on coke catalysts activated with sulfuric acid", Fuel v. 67, #5,
May 1988, pp. 714-8.
17. Nethe, Luiz-Peter, "Ein Weg zu weniger Quecksilber und Dioxin" Umwelt
Magazin, November, 1990, pp. 60-1.
18. Noll, Kenneth E., Fang, Kenneth et. al., "Adsorption Characteristics of
Activated Carbon, XAD4 resin and molecular sieves for the removal of
hazardous organic solvents" 1987 Proceedings of the Second Engineering
Foundation Conference on the Fundamentals of Adsorption Santa Barbara,
CA May 4-9, 1986. Published by Engineering Foundation, New York, NY
1987 pp. 441-50.
445
-------
19. Panknin, Walter. Steinmuller. personal communication, February 20,
1991.
20. Recasens, F., et. al., "Desorption Processes: Supercritical fluid
regeneration of activated carbon" AIChE Journal, v. 35 #6, June, 1989
pp. 951-8.
21. Roda, I. G., "Mass transfer in the adsorption of dissolved substances by
a fluidized bed of activated carbon" Soviet Journal of Water Chemistry
and Technology v. 10 #1, 1988. pp. 46-9.
22. Shamekhi, Reza and Walter K. Panknin, "Research into Activated Carbon
Technology on Harmful Organic Substances, Heavy Metals and NOx Control",
83rd Annual Meeting of the Air and Waste Management Association,
Pittsburgh, PA. June 24-29, 1990.
23. Tan, Chung-Sung, and Liou Din-Chung "Regeneration of activated carbon
loaded with toluene by supercritical carbon dioxide" Separation Science
and Technology v. 24 #1-2 Jan-Feb. 1989, pp. 111-27.
24. Tan, C. S. and Liou D. C., "Modeling of desorption at supercritical
conditions" AIChE Journal v. 35 #6 June, 1989. pp. 1029-31.
25. Teller, Aaron J. et. al., "Emission Control from Hospital Waste
Incineration" presented at the Hazardous Materials Control Research
Institute's Great Lake 90 Conference, Cleveland, OH., September 26-28,
1990.
26. Tienes, Anton. "Special Refuse Incineration Plant Schoneiche", Report
issued by BC Berlin-Consult GmbH, February 12, 1990.
27. USEPA Oil and Hazardous Materials Technical Assistance Data Systems
(0HMTADS Computer Database) 1987
28. Wilken, Michael and A. Beyer. "Praktische Konzepte zur Verminderung der
Bildung von Polychlorierten Dibenzodioxinen und Dibenzofuranen bei
kommunalen Mullverbrennungsanlagen" by ITU-Forschungsgesellschaft
Technischer Umweltschutz for Umweltbundesamt, Berlin, Germany. April,
1990.
29. Wilson, Owen J., "Radon Transport in an activated charcoal canister"
Nuclear Instruments and Methods in Physics Research, Section A" v. 275
#1 Feb. 1, 19R9. pp. 163-171.
446
-------
EMISSIONS CONTROL OF POLYCHLORINATED DIBENZO-P-DIOXINS AND
POLYCHLORINATED DIBENZOFURANS AT MUNICIPAL WASTE COMBUSTORS
by: Shiaw C. Tseng, Wojciech Jozewicz
Acurex Corporation
P. 0. Box 13109
Research Triangle Park, NC 27709
Charles B. Sedman
Gas Cleaning Technology Branch, MD-04
Air and Energy Engineering Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
ABSTRACT
This paper gives the results of an analysis of available
emission data of polychlorinated dibenzo-p-dioxins and
polychlorinated dibenzofurans (PCDD/PCDF) from municipal waste
combustors (MWCs) to evaluate the effectiveness of various air
pollution control devices on PCDD/PCDF removal. The effects of
flue gas temperature, recycling fabric filter ash, and process
additives such as ammonia and Tesisorb™ powder were also analyzed.
The analysis shows that MWCs equipped with a spray dryer
followed by fabric filters can achieve PCDD/PCDF removal
efficiencies (REs) of 97% and higher. A RE of 94% has been
achieved at a combustor equipped with a Thermal DeNO, system
followed by a spray dryer and fabric filters. MWCs equipped with
a duct sorbent injection system followed by fabric filters can
potentially achieve a RE of 99%. A combustor equipped with a spray
dryer followed by electrostatic precipitators (ESPs) has achieved
a RE of 64%. Neither a duct sorbent injection system followed by
ESPs nor a furnace sorbent injection system followed by ESPs could
effectively remove PCDD/PCDF. PCDD/PCDF were not effectively
removed from MWCs equipped with ESPs as the only devices to control
air pollution.
This paper has been reviewed by the Air and Energy Engineering
Research Laboratory, U.S. Environmental Protection Agency, 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 by the Agency.
For presentation at the Second International Conference on
Municipal Waste Combustion in Tampa, FL, April 15-19, 1991.
447
-------
INTRODUCTION
Due to the increasing costs and complexities of landfilling
operations, disposal of municipal solid waste (MSW) through
combustion has gained increasing popularity. However, toxic air
pollutants such as polychlorinated dibenzo-p-dioxins (PCDDs) and
polychlorinated dibenzofurans (PCDFs) have been detected at almost
every municipal waste combustor (MWC).'1"31 Emissions of these
pollutants into the atmosphere have caused serious concerns over
possible adverse impacts on the environment and human health.
Municipalities across the U.S. have encountered strong resistance
in the form of NOT-IN-MY-BACKYARD attitudes from residents living
in the neighborhood of the potential sites, despite the fact that
more MSW is generated and less landfill space is available for
disposal.
Currently, no system has been developed solely for the purpose
of controlling the emissions of PCDD/PCDF from MWCs. The present
practice to control their emissions is in conjunction with the
control of particulate matter (PM) and acid gases. To control PM
emissions, electrostatic precipitators (ESPs) and fabric filters
(FFs) are commonly used. At three facilities, Tesisorb* powder was
injected into the flue gas to condition the filter cake and to
reduce the pressure drop across the FF. To control the emissions
of acid gases (HC1 and S02) , lime-based sorbents are utilized in
three different technologies: furnace sorbent injection (FSI), duct
sorbent injection (DSI), and spray dryer (SD) absorption. To
control the emissions of nitrogen oxides (N0„) , Exxon's Thermal
DeNOx process is installed in three MWCs in California. Ammonia
(NH3) was injected into the upper section of the combustor and
reacted with NO, to form nitrogen (N2) and water.
This paper gives the results of an analysis of available
PCDD/PCDF emissions field test data from three types of MWCs
including mass burn (MB), modular (excess-air and starved-air), and
refuse-derived fuel (RDF) . The main purpose is to see how various
air pollution control devices (APCDs) and their operating
conditions affect the control of PCDD/PCDF emitted from MWCs. The
effect of temperature and the impact of adding Tesisorb powder or
NH3 on controlling PCDD/PCDF emissions will be examined. The
influence of recycling FF ash into the flue gas on PCDD/PCDF
emissions control will also be analyzed.
Tesisorb is a trade mark of R-C Environmental Services and
Technologies.
448
-------
EMISSION DATA SOURCES
The PCDD/PCDF emissions data collected for this work were
taken from reports published by the U.S. Environmental Protection
Agency'1"31 and Environment Canada.<4) Two phrases, "uncontrolled
emission" and "controlled emission" are constantly used throughout
the text. The uncontrolled emission is the amount of PCDD/PCDF
measured before the APCDs. The controlled emission is the amount
of PCDD/PCDF measured after the APCDs. Unless otherwise specified,
the emissions are presented in nanograms per dry, standard cubic
meter (ng/dscm) of flue gas and normalized to 7% 02.
UNCONTROLLED PCDD/PCDF EMISSIONS
The typical levels of uncontrolled PCDD/PCDF emissions from
MWCs of different types and capacities are summarized in Table l.(S)
The overall capacities for the large, mid-size, and small MWC
plants are greater than 600 tons per day (tpd), between 200 and 600
tpd, and less than 200 tpd, respectively.'61 For mass burn
refractory combustors, typical uncontrolled PCDD/PCDF emissions are
4,000 ng/dscm. At large mass burn waterwall combustors, typical
emissions are 500 ng/dscm. Mid-size mass burn waterwall and
modular excess air combustors have typical uncontrolled emissions
of 200 ng/dscm. The typical emissions from three types of
combustors, small mass burn waterwall, RDF, and mass burn rotary
waterwall, are 2,000 ng/dscm. Typical emissions at modular starved
air combustors are 400 ng/dscm.
RESULTS AND DISCUSSION
EFFECTIVENESS OF SD/FF
The effectiveness of the APCD consisting of a spray dryer
followed by fabric filters (SD/FF) on PCDD/PCDF removal can be seen
from the data in Table 2. Tests were conducted at the following
facilities, Biddeford (in Biddeford, ME), Mid-Connecticut (in
Hartford, CT), Quebec City (in Quebec, Canada), and Marion County
(in Marion, OR). Combustors at the first two facilities were fired
with RDF. The Quebec City data were generated from the pilot plant
tests conducted at its Unit No. 3 combustor which was a mass burn
unit with waterwall. The Marion County combustors are mass burn
units with waterwall. The results in Table 2 show that the
calculated REs are higher than 95%, regardless of the differences
in mechanical designs of these combustors. Also included in this
table are the data from the Penobscot facility (in Orrington, ME)
where the combustors burned RDF and were equipped with waterwall,
although the uncontrolled emissions were not available and the RE
could not be calculated.
EFFECTIVENESS OF THERMAL DENO, SYSTEM
Exxon's NH, based Thermal DeNO* process has been installed at
three California facilities: Commerce (in Commerce), Stanislaus
449
-------
County (in Crows Landing), and Long Beach (in Long Beach). These
MWCs are equipped with SD/FF to control acid gases and PM
emissions. The PCDD/PCDF emissions from these facilities are shown
in Table 3. At the Commerce MWC, the RE is 87.7% for the tests
conducted in 1987 when the combustor burned residential refuse.
The RE is 94.5% for the tests conducted in 1988 when the combustor
burned commercial and residential wastes. The RE is 99.6% for the
tests conducted in 1988 when the combustor burned commercial
refuse. The uncontrolled emissions were not reported from MWCs at
the last two locations and the REs cannot be calculated.
Comparison of the overall data in Table 3 with those in Table 2
indicates that the installation of the Thermal DeNO, system
followed by the SD/FF does not appear to affect the control of
PCDD/PCDF emissions.
EFFECTIVENESS OF DSI/FF
Several MWCs employed a duct sorbent injection system followed
by fabric filters (DSI/FF) to control air pollution. The PCDD/PCDF
emissions data from such MWCs are shown in Table 4. The
effectiveness can only be seen by the pilot plant tests conducted
at Quebec City No. 3 combustor, because no uncontrolled emissions
were reported from other MWCs. During the pilot plant tests, the
temperature of the flue gas entering the FF was changed from about
204 to 111°C. The results shown in Table 4 indicate that almost
all of the PCDD/PCDF were removed.
EFFECTIVENESS OF SD/ESP
The effectiveness of APCD consisting of a spray dryer followed
by electrostatic precipitators (SD/ESPs) can be seen from the data
obtained from the tests conducted at Unit No. 2 combustor at
Millbury, MA. The emission data are shown in Table 5. The RE is
64.3%. The performance of the SD/ESP at the Portland facility (in
Portland, ME) cannot be calculated because uncontrolled emissions
were not reported. Comparison of the data in Table 5 with those in
Table 2 indicates that the SD/ESP system is much less effective at
controlling PCDD/PCDF emissions than the SD/FF system.
EFFECTIVENESS OF DSI/ESP
The effectiveness of APCD consisting of a duct sorbent
injection system followed by electrostatic precipitators (DSI/ESPs)
can be seen from the test data obtained at Dayton Unit No. 3
combustor in Dayton, OH. The results listed in Table 6 show that
the controlled emissions are higher than the uncontrolled
emissions, indicating that such a system cannot remove PCDD/PCDF.
However, it has to be noted that the uncontrolled enissions are
very low as compared to all the data presented earlier.
Furthermore, comparison of the data in Table 6 with those in Table
3 indicates that DSI/ESP is much less effective than DSI/FF to
control PCDD/PCDF emissions.
450
-------
EFFECTIVENESS OF FSI/ESP
The effectiveness of the APCD consisting of a furnace sorbent
injection system followed by electrostatic precipitators (FSI/ESPs)
can also be seen from the results of tests conducted at the Dayton
Unit No. 3 combustor. The data are shown in Table 7. Three tests
were conducted at a flue gas temperature of about 201°C (measured
at the inlet of the ESPs) and another three tests were at 148"C.
The lower flue gas temperature was achieved by spraying water into
the duct. The test results show that at both temperatures more
PCDD/PCDF are generated after the flue gas passed through ESPs, an
indication that such an APCD cannot remove PCDD/PCDF. Also listed
in Table 7 are the data reported by the combustor located in
Alexandria, VA, where no uncontrolled emissions were reported and
the effectiveness of the APCD cannot be calculated.
EFFECTIVENESS OF ESP
The effectiveness of ESPs alone to control PCDD/PCDF emissions
can be seen from the data listed in Table 8. These data were
reported by the following facilities: Dayton Unit No. 3, Peekskill
(in Peekskill, NY), North Andover (in North Andover, MA), Oswego
County (in Fulton, NY), and Pinellas County (in St. Petersburg,
FL). With the exception of the tests conducted at the Peekskill
facility, all other test data show that more PCDD/PCDF have been
generated as the flue gas passed the ESPs, indicating that the ESPs
installed at these MWCs cannot remove PCDD/PCDF. The Peekskill
data indicate that the average RE is about 46.9%.
Although ESP is the most popular APCD installed in the MWCs,
most of the units reported only the controlled emissions. This is
especially true for the older units. The following facilities did
not report the uncontrolled emissions: Pigeon Point (in Pigeon
Point, DE) , Lawrence (in Lawrence, MA), Oneida County (in Rome,
NY), Pope/Douglas (in Alexandria, MN) , Quebec City Unit No. 3
(pilot plant tests), Red Wing (in Red Wing, MN), Tulsa (in Tulsa,
OK), Akron (in Akron, OH), Albany (in Albany, NY), Hampton (in
Hampton, VA), Chicago-NW (in Chicago, IL), Hamilton-Wentworth (in
Hamilton, Toronto, Canada), Niagara Falls (in Niagara Falls, NY),
Philadelphia-NW (in Philadelphia, PA), and Saugus (in Saugus, MA).
The effectiveness of ESPs cannot be evaluated from these MWCs.
EFFECT OF RECYCLING FF ASH
FF ash was recycled into the SD during the pilot plant tests
conducted at Quebec City No. 3 combustor. The effect on PCDD/PCDF
emissions can be seen from the emission test data shown in Table 9.
The flue gas temperatures are almost the same, at around 140°C.
With ash recycling, the two-run averaged uncontrolled emission is
2,157 ng/dscm and the RE is 99.9%. Without ash recycling, the two-
run averaged uncontrolled emission is 1,764 ng/dscm and the
controlled emissions are below the detection limit, giving a RE of
100%. These results indicate that practically no apparent effect
is observed when FF ash was recycled.
451
-------
EFFECT OF TESISORB POWDER
Tesisorb powder was injected in the flue gas stream at the
Marion County, Commerce, and Dutchess County (in Poughkeepsie, NY)
facilties. The original intention of using this material was to
condition the filter cakes and to reduce the pressure drops across
the filter bags so that the bags did not require as frequent
cleaning. All the MWCs are mass burn units. The corresponding
APCDs installed in the combustors are SD/FF, Thermal DeNO„/SD/FF,
and DSI/FF.
Listed in Table 10 are the test results from the above
facilities. These data indicate that, on the average, REs of 97.5
and 94% are achieved at the Marion County and the Commerce
facilities, respectively. No uncontrolled emissions were reported
from the Dutchess County facility, and the RE cannot be calculated.
Comparison of the data in Table 10 with those in Tables 2: and 3
indicates that the addition of Tesisorb powder into the flue gas
transport duct does not affect the removal of PCDD/PCDF.
EFFECT OF FLUE GAS TEMPERATURE
The effect of flue gas temperature on PCDD/PCDF removal can be
seen from the test data summarized in Table 11 which includes the
data from two MWCs, one from the pilot plant tests conducted at
Quebec City Unit No. 3 and the other from Dayton Unit No. 3. At
the former facility, the pilot plant was equipped with DSI/FF to
control acid gases and PM emissions. The pilot plant test results
indicate that the uncontrolled PCDD/PCDF emissions were lowered to
887 ng/dscm when the flue gas temperature at the inlet of the FF
was 111° C, the lowest temperature tested. The uncontrolled
emissions are 2 or 3 times higher at other test temperatures, 204,
141, and 121°C. The flue gas temperature was reduced by spraying
water into the mixing chamber between the combustor and the pilot
plant APCD.
The Dayton No. 3 combustor was equipped with FSI/ESP to
control acid gases and PM emissions. The uncontrolled PCDD/PCDF
emissions are 38.1 and 14.2 ng/dscm when the flue gas temperatures
at the inlet of the ESPs were at 201 and 148°C, respectively. The
flue gas temperature was reduced by spraying water into the flue
gas. The results indicate that, as the flue gas cools in the
transport duct, the uncontrolled PCDD/PCDF emissions decrease.
At this Dayton combustor, the PCDD/PCDF emissions were also
measured without injecting sorbent into the flue gas duct. In this
case, only ESPs were used to control air pollution. The data were
taken while the flue gas temperatures at the inlet of the ESPs were
299, 278, and 202°C. The corresponding uncontrolled emissions
decreased from 252, to 214, and to 32.8 ng/dscm.
The above results indicate that the uncontrolled PCDD/PCDF
emissions can be reduced if the flue gas temperature is lowered.
452
-------
CONCLUSIONS
Available emission data of PCDD/PCDF from MWCs were analyzed
to see how APCDs and their operating conditions affect the removal
of PCDD/PCDF.
The analysis shows that the PCDD/PCDF REs are 97% or higher
for the tests conducted at the Biddeford, Mid-Connecticut, and
Marion County facilities. These combustors are equipped with a
SD/FF. A RE of 94% has been achieved at the Commerce combustor
equipped with a Thermal DeNOx/SD/FF system. MWCs equipped with a
DSI/FF system can potentially achieve a RE of 99%, as suggested
from the pilot plant test data obtained at Quebec City No. 3
combustor. A RE of 64% has been reported at the Millbury No. 2
combustor equipped with a SD/ESP system. Data obtained from tests
conducted at Dayton No. 3 combustor, utilizing either a DSI/ESP or
a FSI/ESP system, indicate that neither of these two control
techniques can remove PCDD/PCDF. PCDD/PCDF were not removed, in
general, from MWCs equipped with ESPs alone to control air
pollution.
Recycling of FF ash to the SD, or addition of NH3 or Tesisorb
powder to the flue gas did not have an apparent effect on PCDD/PCDF
removal. An important parameter to control PCDD/PCDF emissions is
the flue gas temperature through the APCDs. Field test data
indicate that the uncontrolled emissions could be reduced if the
flue gas temperature entering the PM control devices (FF and ESP)
was lowered.
Many MWCs did not report or were not required to report the
uncontrolled emissions, and the RE could not be calculated. The
analysis is based on a rather limited amount of data, and the
conclusions stated above should not be deemed final. More data
and tests are needed to see whether the above conclusions can be
generalized to any MWCs in the U.S.
GLOSSARIES
air pollution control device
duct sorbent injection
duct sorbent injection followed by electrostatic
precipitator
duct sorbent injection followed by fabric filter
electrostatic precipitator
fabric filter
furnace sorbent injection
APCD
DSI
DSI/ESP
DSI/FF
ESP
FF
FSI
453
-------
FSI/ESP
furnace sorbent injection followed by
precipitator
electrostatic
MB
mass burn
MB/R
mass burn refractory
MB/RWW
mass burn rotary waterwall
MB/WW
mass burn waterwall
MEA
modular excess-air
MSA
modular starved-air
MSW
municipal solid waste
MWC
municipal waste combustor
PCDD
polychlorinated dibenzo-p-dioxin
PCDF
polychlorinated dibenzofuran
PM
particulate matter
RDF
refuse-derived fuel
RE
removal efficiency
SD
spray dryer
SD/ESP
spray dryer followed by electrostatic
precipitator
SD/FF
spray dryer follower by fabric filter
REFERENCES
(1) U.S. Environmental Protection Agency, "Municipal Waste
Combustion Study—Emission Data Base For Municipal Waste
Combustors," EPA/530-SW-87-021b, June 1987.
(2) U.S. Environmental Protection Agency, "Municipal Waste
Combustors—Background Information For Proposed Standards:
Control of N0X Emissions," Vol. 4, EPA-450/3-89-27d (NTIS
PB90-154873), August 1989.
(3) U.S. Environmental Protection Agency, "Municipal Waste
Combustion Assessment: Combustion Control at Existing
Facilities," EPA-600/8-89-058 (NTIS PB90-154931), August 1989.
(4) Environment Canada, "The National Incinerator Testing and
Evaluation Program: Air Pollution Control Technology," Summary
Report EPS 3/UP/2, September 1986.
^54
-------
(5) U.S. Environmental Protection Agency, "Municipal Waste
Combustors—Background Information for Proposed Standards:
Post-Combustion Technology Performance," Vol. 3.
EPA-450/3-89-27C (NTIS PB90-154865), September 1989. p. 1-5.
(6) Reference 3. p. 3-2.
Table 1. Typical Uncontrolled PCDD/PCDF Emissions'51
Combustor Type
Emission'*'
(ng/dscm at 7% 0,)
Mass burn refractory (MB/R)
4 ,000
Large mass burn waterwall (MB/WW)
500
Mid-size mass burn waterwall
Modular excess air (MEA)
200
Small mass burn waterwall
Refused-derived fuel (RDF)
Mass burn rotary waterwall (MB/RWW)
2,000
Modular starved air (MSA)
400
Note: "" All concentrations reflect the sum of the tetra-
through octa-chlorinated PCDD/PCDF homologues only.
455
-------
Table 2. PCDD/PCDF Emissions From MWCs Equipped With SD/FF.<3,i)
Location of NWCs (type/design) FF Inlet Uncontrolled Controlled PCDD/PCDF
Temperature Emission RE
(°C) (ng/dscm) (%)
Biddeford (RDF) 137
Mid-Connecticut (RDF) 133
Quebec City Unit No. 3 (MB/WW)
pilot plant tests
with FF ash recycled 140
pilot plant tests
without FF ash recycled 139
Marion County (MB/WW) 133
Penobscot (RDF/WW) 146
903 4.38 99.5
1,019 0.66 99.9
2,157 1.26 99.9
1,764 ND ND = not detected (below the detection limit)
Value estimated from other tests.
-------
Table 3. PCDD/PCDF Emissions From MWCs With Thermal DeNOx/SD/FF. <3>
Location of MWCs (type/design) FF Inlet Uncontrolled Controlled PCDD/PCDF
Temperature Emission RE
(°C) (ng/dscm) (%)
Commerce (MB/WW)
1987 tests, combustor burning
residential refuse
135
28.1
3.47
87.7
v_n
-^1
1988 tests, combustor burning
commercial and residential
refuse
1988 tests, combustor burning
commercial refuse
NR(a)
NR(a)
446
783
9.59
2.78
94 .5
99.6
Stanislaus County (MB/WW)
Unit No. 1.
Unit No. 2.
145
146
6.25
6.53
Long Beach (MB/WW)
152 NR = not reported.
Value estimated from other tests.
-------
Table 4. PCDD/PCDF Emissions F-rom MWCs Equipped With DSI/FF.<3'i)
Location of MWCs (type/design)
FF Inlet
Temperature
(°C)
Uncontrolled Controlled
Emis s ion
(ng/dscm)
PCDD/PCDF
RE
(%)
Quebec City No. 3 (MB/WW)
(pilot plant tests)
204
141
121
111
1, 597
2, 277
2, 361
887
7.35
0.49
ND
2.43
99.55
99. 98
100 .00
99.75
Claremont (MB/WW)
Unit No. 1
Unit No. 2
225 ND = not detected (below the detection limit).
Value estimated from other tests.
-------
Table 5. PCDD/PCDF Emissions From MWCs Equipped With SD/ESP. <3)
Location of MWCs (type/design) FF Inlet Uncontrolled Controlled PCDD/PCDF
Temperature Emission RE
(°C) (ng/dscm) (%)
Millbury Unit No. 2 (MB/WW) 123 170 59.2 64.3
Portland (MB/WW) 141 173
-fc-
ui
vo
Table 6. PCDD/PCDF Emissions From MWCs Equipped With DSI/ESP . <3>
Location of MWCs (type/design) FF Inlet Uncontrolled Controlled PCDD/PCDF
Temperature Emission RE
(°C) (ng/dscm) (%)
Dayton Unit No. 3(MB/R) 152 5.31 57.2 -6,360
-------
Table 7. PCDD/PCDF Emissions From MWCs Equipped With FSI/ESP.'3'
Location of MWCs (type/design) FF Inlet Uncontrolled Controlled PCDD/PCDF
Temperature Emission RE
(°C) (ng/dscm) (%)
Dayton Unit No. 3 (MB/R)
201
148
38.1
14 .2
1, 481
659
-4,182
-6,490
4>
ON
o
Alexandria (MB/WW)
183(a)
54 .86
Note:
-------
Table 8. PCDD/PCDF Emissions From MWCs Equipped Only With ESP.'3'
Location of MWCs (type/design) FF Inlet Uncontrolled Controlled PCDD/PCDF
Temperature Emission RE
(°C) (ng/dscm) (%)
Dayton Unit No. 3 (MB/R)
299
252
17,
100
-9,140
278
214
14,
500
-6,990
202
32 .8
866
-2,640
Peekskill (MB/WW)
Apr/1985 tests, combustor=normal
237(a)
NR )
107 .3
Nov/1985 tests, combustor=start-up
215
11,432
9,
570
15.4
Nov/1985 tests, start of campaign,
combustor=normal load
229
478
263
44 .5
Nov/1985 tests, end of campaign,
combustor=normal load
244
617
179
69.0
Nov/1985 tests, combustor=high load
234
438
126
71.7
Nov/1985 tests, combustor=low load
225
228
148
34 .1
North Andover (MB/WW)
313
245
362
-43.1
Oswego County (MSA)
combustor=normal, start of campaign
257
175
353
-118.0
combustor at mid-range temp., 954°C
251
195
301
-53 . 4
combustor = normal, end of campaign
255
359
412
-30.9
low combustor temp., 899°C
242
732
819
-12 .0
Pinellas County (MB/WW)
281
54
100
-95.0
Note: Value estimated from other tests,
(b) nr = not reported.
-------
Table 9. Effect Of Recycling FF Ash.C)
Location of MWCs (type/design) FF Inlet Uncontrolled Controlled PCDD/PCDF
Temperature Emission RE
(°C) (ng/dscm) (%)
Quebec City Unit No. 3(MB/WW)
with FF ash recycled 140 2,157 1.26 99.9
without FF ash recycled 139 1,764 ND 100.0
Note: ND = not detected (below the detection limiit)
-------
Table 10. PCDD/PCDF Emissions From MWCs With Tesisorb Powder Injection.<3>
Location of MWCs (type/design) FF Inlet Uncontrolled Controlled PCDD/PCDF
Temperature Emission RE
(°C) (ng/dscm) (%)
Marion County (MB/WW)
equipped with SD/FF
combustor = normal
combustor = shut down
combustor = start up
133 43.0
145 61.6
149 435
1.26 95.7
1.91 97.4
3.36 99.5
Commerce (MB/WW)
equipped with Thermal DeNOx/SD/FF
1987 tests, combustor burning 135 28.1 3.47 87.7
residential refuse
1988 tests, combustor burning NR 783 2.78 99.6
commercial refuse
Dutchess County (MB/RWW)
equipped with DSI/FF
Unit No. 1 192(b) 4 .83
Unit No. 2 197(b) 17.90
Note:
-------
Table 11. Effect Of Flue Gas Temperature. i3-H
Location of MWCs (type/design)
FF Inlet Uncontrolled Controlled PCDD/PCDF
Temperature Emission RE
(°C) (ng/dscm) (%)
->
ON
Quebec City Unit No. 3(MB/WW)
pilot plant equipped with DSI/FF
Dayton Unit No. 3 (MB/R)
with FSI/ESP
with ESP
204
141
121
111
201
148
299
278
202
1, 597
2,277
2, 361
887
38.1
14 .2
252
214
32.8
7 .35
0.49
ND (a)
2 .43
1, 481
659
17,100
14,500
866
99.55
99. 98
100
99.75
-4,182
-6, 490
-9,140
-6,990
-2,640
Note: ND = not detected (below the detection limiit)
-------
DESIGN AND OPERATION OF PULSE-JET FABRIC FILTERS
FOR INCINERATION AIR POLLUTION CONTROL
By: W. GREGG
MIKROPUL ENVIRONMENTAL SYSTEMS
ABSTRACT
Pulse-jet fabric filters have gained acceptance as a viable
control method for air emissions from incineration. A pulse jet
can capture fine particulate emissions with far less energy
expenditure than a wet scrubber. Their initial cost is also less
than electrostatic precipitators. With the addition of sorbent
chemicals injected upstream of the pulse-jet, these fabric filters
can also achieve compliance with regulations governing gaseous
emissions. Pulse-jets have been used on a wide variety of
incineration applications, including infectious hospital waste, low
level radioactive waste, various hazardous materials, and simple
volume reduction of bulk waste.
However, incineration is a vigorous application for pulse-jets
or any fabric filter. The aggressive and corrosive gas stream
created by incineration can easily attack the pulse-jet components.
Operation of incinerators and their pollution control devices,
including pulse-jets, is often less than what is considered
acceptable practice. Also, selection of pulse-jets and their
components for incinerator off-gas filtration is frequently done on
the basis of lowest initial cost. These combined factors have
resulted in reduced pulse-jet and pulse-jet component service life
and repeated incidents of non-compliance with air emission rules.
To insure adequate service life and to maintain emissions at
a compliance level, it is important that a pulse-jet fabric filter
be designed, equipped and operated with the unique conditions
imposed by incineration kept in mind. Selection of materials of
construction for this fabric filter and its components must be
based on worst case scenarios. Still, the cost of such materials
should be reasonable. Design and sizing of a pulse-jet must
recognize the wide variety of conditions a single incinerator both
"sees" and, in turn, creates. Changes in feed stock quantity and
type are two main reasons for these wide variety of conditions,
frequent down-time on the incinerator is another reason. Shut-down
and start-up of the incinerator lead to more problems with a pulse-
jet filter than on-line operation.
The paper will review the various options one has in pulse-jet
design, filter media selection, and choice of materials of
construction. Recommendations will be made on which of these are
most appropriate for various incinerator applications. New
465
-------
-2-
technology, which has only recently been utilized in incinerator
pulse-jet filters, will be highlighted. Guidelines for correct
operation of both the pulse-jet and the incinerator will also be
offered.
INTRODUCTION
Pulse-jet fabric filters (or baghouses) are being successfully
utilized as the control technology for incinerator emissions for a
broad range of applications. Municipal, industrial, hospital,
hazardous, and even radioactive waste incineration emission can be
effectively and efficiently collected using pulse-jet (and other)
fabric filters. Their inherent design flexibility, their excellent
particulate capture and their relatively low capital and energy
costs have recently resulted in the selection of pulse-jets over
the more traditional incinerator emission controls, i.e.
electrostatic precipitators and wet scrubbers. With the addition
of absorbents, such as lime or sodium bicarbonate, injected
upstream of the fabric filter, acid gases produced by incineration
can also be controlled. Various studies, conducted in both Europe
and North America, have shown that even organic and heavy metal
toxics, either formed or liberated by incineration, can be
effectively captured with pulse-jet filters.
There is no question that when compared with the pulse-jet
fabric filter, other technologies do have some advantages. Control
systems with electrostatic precipitators certainly consume less
electrical power than fabric filters and are more tolerant of gas
temperature excursions. Though not exempt from maintenance,
(including extensive replacement of internal parts), the
precipitator is also generally considered to require less cost for
this category than fabric filters. Regular replacement of filters
is required by pulse-jets and other fabric filters.
Wet scrubbers are more effective in controlling acid gases
than absorbent addition systems employing pulse-jet or other
baghouses. Scrubbers can have a substantial initial cost
advantage, too. They also enjoy an artificial advantage in the way
the liquid waste stream is viewed. Often this scrubber waste
blowdown can be discharged directly to a municipal sewer. Dry
waste, whether from a fabric filter or a precipitator, containing
the same type and amount of contaminants as the liquid stream, must
usually be handled as a special or even hazardous waste. This
might change, of course.
466
-------
-3-
Yet fabric filters, and particularly pulse-jet baghouses, are
being chosen as the control technology for many incinerator
installations. The primary reason for their acceptance is their
ability to capture particulate, even low and sub-micron particles,
efficiently and at a relatively low cost, both in capital and
operating expenses. Whether a particulate liberated during
incineration, a acid gas molecule or a heavy metal condensate,
pulse-jets (and other fabric filters) can capture these emissions.
However, to achieve this potential for high capture efficiency
and to maintain it, the selection, design, fabrication,
installation and operation of a pulse-jet fabric filter must all be
done carefully. Failure to do so can lead to excess emissions,
high operating expense and significantly hinder operation of the
whole incineration system. The purpose of this paper is to prevent
these from occurring. By offering the design and operational
guidelines listed below, it is the author's hope that future pulse
jet filter incinerator emission control systems will be a success.
DISCUSSION
The focus of this paper will be pulse-jet dust collectors. In
general, the guidelines given will also apply to other fabric
filters, i.e. shakers or reverse-gas dust collectors. Pulse-jets
are used more often than the other fabric filters for control of
incinerator emissions since they are more economical, particularly
for lower gas volumes.
Figure 1 is an artist's cross-sectional view of a pulse-jet
dust collector (MikroPul's design). Some license has been taken to
show all components. Different manufacturer's use differing
designs but the basic principles are common. "Dirty" gas is passed
through the collector's filter elements and the particulate
entrained in the gas is captured by the fibrous filter media. The
"cleaned" gas then passes through the collector. The filter
elements require support, e.g. a wire cage, to prevent collapse
from the pressure differential across these filters.
As the captured particulate matter increases on the surface of
these filter elements, the pressure differential increases. To
prevent it from becoming excessive, some of the particulate must be
removed from the filters. This is accomplished by compressed air
pulse cleaning. A certain percentage of the filter's have
compressed air rapidly injected into their hollow center. This
produces a pressurized volume inside of the filter element, at a
pressure greater than the pressure differential occurring
467
-------
-4-
previously. This reverse differential produces rapid flexing of
the filter fabric and a reverse flow of air through the fabric.
The combination of flexing and reverse flow dislodges the majority
of the accumulated particulate and this particulate or dust falls
into a collection hopper positioned beneath the filter elements.
Gravity is relied upon to remove the dust from this hopper, through
a rotary valve or other device that provides a gas seal, thus
minimizing gas leakage in the dust removal process.
The pulse cleaning is repeated at regular intervals or on a
demand basis, determined by the level or the rate of rise of the
pressure drop across the filters. In either case, all filters will
be cleaned repeatedly.
Pulse-jets are utilized in various configurations for
incinerator emission control. The simplest is a stand alone system
where the pulse-jet is essentially the only control device. This
configuration is typically used for simple volume reduction of
waste, usually by industrial users who are discarding packaging
materials. Plastics are not supposed to be incinerated in such a
system, though inevitably some are. The gas temperature must be
reduced prior to entering the pulse-jet filter. With these simple
systems this is usually accomplished by passing the incinerator
exhaust gas through a waste heat recovery boiler or similar heat
exchanger or by simply diluting the gas with ambient air.
A more refined approach and one capable of collecting acid
gases is injection of absorbent chemical, e.g. hydrated lime, into
the incinerator exhaust stream prior to the pulse jet. The use of
temperature controlling equipment, such as a heat recovery system,
and the addition of moisture to the gas stream, which reduces
temperature further, will allow the pulse-jet collector control
system to achieve removal efficiencies of over 90% for HC1 gas.
This has been demonstrated. SOz removal with such a system is less
efficient, normally in the 60-70% range, though better results have
been reported. See Figure 2.
The use of a spray dryer absorber prior to the pulse-jet
filter is a more rigorous approach than dry absorbent injection.
Such systems require slurry preparation systems in addition to the
spray dryer. They are significantly more exjoensive than dry
injection systems, but may be more effective in acid gas removal.
In large incinerator control systems, where absorbent chemical
costs can be significant, the better utilization of the absorbent
by such systems can justify the added expense. See Figure 3.
A hybrid system that is gaining popularity is the combined dry
filter/wet scrubber concept. Utilizing the fine particulate
468
-------
control provided by a pulse-jet filter and the excellent acid gas
removal efficiency of the wet scrubber, these systems provide
excellent capture efficiencies for dry particulate, heavy metals
and most organics. Since particulate loading to the scrubber is
minimal, low pressure drop packed or spray towers can be used,
keeping energy costs low. See Figure 4.
The key to all these systems is the pulse-jet filter. In
order for the control system to work effectively and, hence, for
the total incinerator system to work well, the dust collector must
be operating correctly with a minimum of downtime. To achieve
this, the pulse-jet must be designed and operated with the unique
demands of incineration in mind.
Incineration is a rigorous application for pulse-jet filters.
Wide swings in gas volume, temperature and moisture content,
frequent downtime, variation in particulate character, changes in
acid gas concentration and the potential of combustion of collected
dust all contribute to causes of pulse-jet downtime and excess
emissions.
What are the causes of such downtime and increased emissions
with pulse-jets on incinerator applications? The following is a
partial list of items that can lead to difficulties:
1. Fire
2. Filter failure
3. Filter blinding/High aP
4. Corrosion
5. Dust discharge problems.
Fire was a persistent cause of problems with incinerator
fabric filters and precipitators. Typically, such fires occurred
in simple systems where waste was burned to reduce volume. In a
pulse-jet fabric filter, fires will damage the filters, the wire
cages, and warp the tubesheet and other components. A severe fire
or repeated fires will require nearly complete replacement of the
dust collector.
When an operator of a pulse-jet or other fabric filter
experience a fire, his/her first response is to order "fireproof"
filters. Frankly, most fabrics that are used for filtering
incinerator exhaust do not support combustion. It is the collected
dust that catches fire not, at least initially, the filter bags.
To prevent fires, one must eliminate at least one of the three
items that contribute to combustion; fuel, ignition source and
oxygen. To stop fires, the same procedure must be followed. Since
469
-------
incinerators tend to operate with significant excess air,
controlling oxygen is difficult. Eliminating in-leakage of air has
some effect. Preventing fuel and an ignition source occurring in
the dust collector is the other approach. The best way to
accomplish this is to insure complete combustion of the waste in
the incinerator. Utilizing waste heat recovery boilers also helps
since large hot cinders can be stopped by impaction on the heat
exchanger elements, and the cooler gas will cool cinders. Long
duct runs, with multiple elbows, also promote cooling and, in some
cases, help complete combustion. Various industries, such as
primary metals and wood products have fire prevention experence.
They use mechanical collectors, place the induced draft fan
upstream of the dust collector and use spark and/or fire detection
and extinguishing systems. These extinguishing systems use C02,
dry chemicals or other agents. Fast acting isolation dampers are
often required with these systems. They do add substantially to
the cost of a fabric filter.
The simplest extinguishing system, the water deluge system
with temperature activated sprinkler, is not recommended for
incinerator pulse-jets. While acceptable for systems where the
collected dust is just char, an absorbent chemical injection system
should not use sprinkler systems. Water would turn a fabric filter
collecting hydrated lime into a large lime slaker.
Fortunately, fires in incinerator pulse-jet collectors are
less prevalent than they once were. Retention chambers for
organics destruction, waste heat recovery, more emphasis on
obtaining complete combustion, absorbent injection (perhaps the
absorbent acts as an inerting material, helping to remove fuel from
the fire equation) and the auxiliary equipment associated with
absorbent injection systems probably all have contributed to
reducing the fire potential.
One operating practice that will help reduce fire potential in
pulse-jet filters is to keep the dust collection hoppers free of
material. This helps remove the fuel factor from the fire
equation.
FILTER FAILURE
Filter failure is generally considered to be a breach in the
integrity of the filter, causing excessive emissions. Improper
design, fabrication, and installation of the filters and/or the
dust collector will also cause excessive emissions. However, this
discussion will focus on failures during service, rather than at
start-up.
/. ir\
-------
-7-
Filter failure is caused by a loss of strength of the filter
fabric, eventually leading to a hole or tear in the fabric. This
hole or tear allows excess emissions to pass through the filter.
The holes will enlarge over time from the flexing caused by pulse
cleaning and from abrasion by the passing dust.
Dust abrasion can wear holes through filters as will repeated
contact with other objects in the dust collector. These include
another filter, walls of the collector or the wires of the support
cage. These failures are not unique to incineration, though
corrosion of the wire cage will accelerate wear, particulary if a
wire weld breaks.
Loss of strength of the filter is caused by chemical and/or
over temperature degradation of the fibers. The agressive
chemicals present in incinerator exhaust will attack filter
fabrics. Operation of a dust collector at temperatures greater
than its filters' can withstand degrade the fiber's that make up
that fabric. In either case, the weakening of the fibers combined
with the flexing caused by filter cleaning will induce the eventual
failure.
To avoid failure, the filter fabric must be compatible with
the gas conditions and constituents. One has to select the right
filter or control the gas conditions and constituents or do both.
With incineration systems, usually both must be done.
There are only a few fibers or fabrics that are suitable for
incineration exhaust streams and each has some drawbacks. Chart 1
lists these fabrics and a general description of their
capabilities. In addition, some other constraints must be
recognized. Ryton will be attacked by nitric acid, bromine, and
even oxygen in quantities greater than 10-15% of the gas.
Fiberglass will be readily attacked by HF and in the woven form
must be operated at lower gas/filter area (air/cloth) ratios than
felted fabrics. It is prone to flexing failure. Other acidic gas
constituents will attack glass as well as the remaining fabrics,
with the exception of Teflon. However, such attack can be avoided
by absorbent addition and maintaining the gas temperature above the
acid gas dewpoint temperature. More on this later.
Homopolymer acrylic is typically used with lower temperatures,
and absorbent injection and humidification, since it will tolerate
higher moisture gas streams. Nomex is particularly susceptible to
chemical attack but has been used successfully with simple (no
plastics) waste volume reduction systems. P-84 is a newer fiber
and there is limited experience with this fabric with incinerator
applications. It is not as tolerant of the various agressive
471
-------
-8-
chemcials in incinerator gas when compared to Ryton or Teflon. Its
higher temperature limitation offers advantages when considering
temperature excursions of the gas stream. P-84 felt also is a
superior filter media compared to woven fiberglass, Ryton or
Teflon, and can achieve very restrictive outlet emission levels.
Goretex is an expanded PTFE membrane that is laminated to
various felt or woven fabrics, such as those listed here or even to
fabrics made from expanded PTFE. It is a proprietary product of
W.L. Gore and Associates. Since the patents on this product are
near expiration, similiar products are being introduced by other
suppliers. The PTFE membrane usually provides excellent filtration
(low emission levels), low pressure differential and extended
filter life. However, the fabric substrate must be compatible with
the gas conditions and constituents if this type of filter is used.
Cost of the filters made from these fabrics vary considerably
and this cost factor has tended to favor some of these fabrics.
Woven fiberglass has been a prominent choice because of its lower
initial cost and also because of its high upper temperature limit.
In some cases, it has been misapplied, and been replaced with some
of the other fabrics listed. However, rather than simply exclude
woven fiberglass, it is important to remember that all the fabrics
listed have failed prematurely or had excessive emission with some
incinerator or similar gas stream. One must be careful to
accurately characterize the constituents and conditions of the
incinerator exhaust and be sure that the filter selected is
compatible with that exhaust. One must also be wary of the
tendency to select a fabric with a heavy emphasis on its cost.
FILTER BLINDING/HIGH aP
Filter "blinding" is a problem that occurs with all fabric
filters. The nature of both the incinerator emissions and the
efforts to treat those emissions make incinerator dust collectors
more succeptible to this condition. "Blinding", for the purposes
of this discussion, will refer to any condition that overly
restricts gas flow through the filters. Due to the high pressure
drop caused by blinding and subsequent loss of gas flow (when this
excess aP affects fan performance), blinding is the leading cause
of filter replacement.
The main cause of blinding is the interaction of the
particulate in the gas with the filter fabric. Water, either as a
vapor or in droplet form, often affects this interaction. In some
cases, dust particles will agglomerate on the filter's surface to
472
-------
-9-
form a highly impermeable dust cake. Often this cake is difficult
to remove by filter cleaning. In other cases, particulate
penetrates deep within the filter's media before being captured.
This restricts the gas passages within the media.
A "mechanical" problem that will cause filter blinding is
insufficient or improper cleaning. In a pulse-jet filter, filter
cleaning may be improved by altering the pulse pressure, the pulse
duration and/or pulse frequency. Misalignment of the pulse
distributing pipes or improper operation of the pulse valves will
contribute to blinding and must be corrected.
If filter cleaning is not the problem (or the solution) than
other techniques must be used. Most of these are preventative
measures and have little effect on blinding that has already
occurred with operating filters. Sometimes shutting down the
fabric filter (stop gas flow) and pulse-cleaning the filters for an
extended period will result in partial recovery from the blinded
condition.
The use of dry chemical filter aids, applied to the filters
prior to exposure to incinerator exhaust, does help prevent
blinding. Continuous seeding with such filter aids can also
improve filter performance though this increases the equipment and
operating effort needed. Limestone is a common precoat material
but various silicates have superior performance. Neutralite, from
BHA Group, is one such filter aid. The chemicals used for sorbent
injection, such as hydrated lime, have some precoat effectiveness
but they are far from ideal.
Ammonia or sulfur trioxide gas injection upstream of fabric
filters collecting coal fly-ash have demonstrated an improvement in
filter dust cake, lowering a P. This technique, if proven effective
with incineration ash, would be economically justified for large
incinerator systems.
The design of the dust collector's inlet configuration has
been shown to affect collector performance. Standard pulse-jet
designs use a solid or perforated baffle placed in the path of the
collector's inlet to "diffuse" the gas and particulate streams.
These designs produce poor gas distribution in the collector,
creating high turbulence, dust re-entrainment, high filter aP and,
even abrasive wear of the filters. Inlet configurations such as
that shown in Figure 1 (MikroPul's patented Cascadairm inlet
diffuser), or similar concepts distribute the inlet gas stream
evenly throughout the collector. Gas turbulence and dust re-
entrainment are greatly reduced, resulting in lower filter aP.
473
-------
-10-
Various fabric treatments can reduce the tendency towards
blinding. Singeing of felt fabrics is a simple, cheap technique
that can reduce filter surface blinding. Silicone and Teflon
emulsions have been shown to improve resistance to blinding. HCE
and Tuflex II, from Mendardi-Criswell, are, respectively, such
fabric treatments. Constructing fabrics with finer fibers and with
lower permeability (porosity) improves the fabric's filtration
performance. The finer fibers increase the surface area fine
particles can adhere to and the lower porosity resists particle
penetration into the filter media.
The presence of high moisture will increase the tendency
towards blinding. Water droplets, from condensation or a poorly
operating spray dryer or spray cooler, will quickly cause blinding
of the filters. Maintaining liquid injection systems upstream of
the dust collector is important.
All these measures should be considered and most followed.
Yet, the best way to combat filter blinding is to recognize it can
occur. To minimize the effect of such blinding, one must size
(design) the dust collector properly. With pulse-jet collectors,
Air/Cloth ratios of 3-4:1 (ACFM/ft filter area) are preferred,
with the more demanding incinerator applications tending to the
lower figure.
CORROSION
Since incinerator exhaust contains corrosive chemical
constituents, the potential for corrosion of a fabric filter's
components is high. Though material selection is a standard
technique used to combat corrosion, the wide variety of gas
constituents and the possibility of high temperatures precludes the
use of many corrosion resistance approaches. For example, the use
of 300 series stainless steel is contraindicated due to its
susceptibility to chloride attack. While various high nickel
alloys have shown that they are capable of withstanding incinerator
exhaust, they are prohibitively expensive.
In general, pulse-jet fabric filters for incinerator
applications are fabricated from carbon steel. To resist
corrosive effects, it is important to establish operating criteria
that prevent the dust collector's operating temperature from being
at or below the acid dewpoint of the gas. The dust collector must
be fully insulated and in-leakage of cool ambient air must be
prevented. This means gasketed joints (flanges), doors and
hatchways, dust discharge devices, etc. should be as leak-free as
474
-------
-11-
possible. In addition, at start-up, the dust collector must be
preheated to correct temperature (above acid dewpoint). At shut-
down the collector's temperature must be maintained above dewpoint
while purging the system with hot, dry, acid-free gas until the
acid constituents are removed from the collector. The use of
hopper heaters (or heat-tracing) on the dust collector, preheating
and purge heating using the auxiliary fuel firing needed by the
incinerator, and, if necessary, adding a supplementary heating
system, will allow an incinerator operator to conform with these
guidelines. There practices have worked well with fabric filters
on coal-fired boilers. Unfortunately, many existing incinerator
dust collectors are not operated in this fashion. While the
collectors will be insulated and normally operated above acid
dewpoint, start-up and shut-down practices are not strictly adhered
to, and corrosive attack occurs. With absorbent injection systems,
the "dirty side" of the collector, the region upstream of
the filter surface, tend to survive. The absorbent acts to
neutralize the acid droplets formed. Also, this area tends to be
the hottest area of the collector.
The so-called clean air plenum is the area that suffers most
from corrosion, since absorbent is not present in this region and
it tends to be cooler. Eventually, all components will suffer,
including filters, resulting in excessive emissions and the need to
repair or replace components.
In some instances, various coatings have been applied to the
gas contact steel surfaces to add some additional corrosion
protection. To work, these coatings, primarily coal tar
derivatives, must be applied under proper ambient conditions to
oxide-free steel. This means dedicated abrasive cleaning. Also,
these coating systems have definite upper temperature limitations
and the gas must be kept below these limits. If temperature
excursions are expected, emergency cooling systems should be
included in the dust collector system. This is also a good idea
for protecting the filters. Coating the wire filter support cages
helps extend the life of the cages and the filters. Two cautions:
1) The localized heat concentration from dust collector heat
tracing may not be compatible with the coating. This must be
checked. 2) Such coatings are not substitutes for proper start-up,
operating and shut-down practices. Coatings will extend component
life when the dust collector is operated correctly.
DUST DISCHARGE
A problem that will occur with incinerator fabric filters is
475
-------
-12-
blockage or bridging of the collected particulates in the
collection hopper. Since the absorbent chemicals, e.g. hydrated
lime, will cake up in the presence of moisture, the dry absorbent
systems are most susceptible. Once a blockage occurs, it must be
cleared quickly, since the collector's operation is adversely
affected. Such blockage will lead to abrasion of the filters. To
prevent such blockages or at least reduce the frequency of them
occurring, the following guidelines are offered. First, the dust
discharge device or devices should be insulated, have a minimum of
air in-leakage and very important, run continuously. The size and
speed of these devices must be able to remove the collected dust
quicker than it enters the dust collector.
Since condensation will promote hopper blockages, it is
important to avoid condensation. This is not easy to do in
practice. Some of the dust will adhere to the collector's inside
surfaces. It tends to form an insulating layer and this results in
a thermal gradient being formed. Even with insulation and
operation of the collector above gas dewpoint, this thermal
gradient can permit localized condensation to occur. When this
occurs, this dust layer can grow, eventually causing a blockage.
Steep hopper walls, heat tracing and hopper vibrators will all help
prevent significant dust layers from accumulating. The coating
systems mentioned above may also help since their "slicker" surface
tends to reduce dust build up. Again, the compatibility of coating
systems, heat tracing and vibrators must be checked.
A technique that has worked in other dry material handling
systems is the addition of "flow aids". Silica gel is a common
flow aid that prevents hopper blockages. Blended with a dry
absorbent, it will probably improve the dry injection system
operation as well as the fabric filter's.
ADDITIONAL RECOMMENDATIONS
Following are some other practices and equipment suggestions
that will improve the operation of a pulse-jet fabric filter
venting an incinerator.
Since maintaining the filters in good operating condition is
critical, it is important to monitor that condition. Filter
pressure drop is the traditional monitoring method. Less used, but
recommended, are particulate emission detection devices.
Continuous emission monitors (CEM) are quite expensive but another
class of device, called broken bag detectors, are reasonable. They
will detect even small increases in particulate emissions, thus
giving an early warning of trouble. Investigating the cause of
such emissions can result in preventing a more difficult problem.
476
-------
-13-
Several firms make these detectors, including Auburn International,
Babbitt International and Anderson.
Investigating the cause of filter failure, rather than just
replacing filters, will result in increased filter life. The
suppliers of the dust collectors and the filters generally have
some capability to test filters but a third party opinion may also
be warranted. Several firms can perform filter evaluations. These
firms include ECC, ETS and Grubb Filtration Testing Services.
The various configurations of incinerator emission control
systems using pulse-jet filters have some operation difficulties
which occur more with a particular configuration. The spray dryer
absorber and dry injection with humidification systems are more
prone to filter wetting (and thus, blinding) and dust discharge
problems. Failure of the liquid sprays will lead to temperature
excursions so an adequate warning system, back-up cooling or gas
bypass systems are recommended.
The configuration with just the filter collector is, as noted
before, more prone to fire and also temperature swings. This
system, in some cases, may also be prone to higher particulate
emission levels so precoating and high efficiency bags are
warranted.
When a wet scrubber is used downstream of the pulse-jet
filter, it is usual to use a packed bed wet absorber. These are
prone to plugging if exposed to particulate. Thus it is extra
important to prevent emissions and detect them if they occur.
Broken bag detectors are certainly warranted with these systems.
CONCLUSION
Pulse-jet fabric filters have demonstrated that they have the
capability to successfully treat incinerator emissions. To insure
success of such a system, it must not only be well designed but
also operated and maintained properly. Monitoring the important
parameters of inlet and outlet gas temperature, filter ~ P and,
emissions will help prevent minor problems from becoming major
ones, as long as they are investigated quickly. Following the
guidelines discussed above will reduce the need for maintenance and
insure a system that will achieve compliance with emission
regulations.
477
-------
-14-
REFERENCES
The National Incinerator Testing and Evaluation Programs: Air
Pollution Control Technology; Summary Report, Environment Canada,
EPS 3/UP/2, September, 1986.
Condensed Manual and Handbook. Filter Media and Fabric Filter
Aspects; Lutz Bergmann, Filter Media Consulting, Inc. 1990.
Baghouse Applied To Incenerators; A. Cesta and H. Franco;
Proceedings, The User and Fabric Filtration Equipment IV
Conference, November, 1988.
Performance and Operation of Fabric Filtration Systems on Hazardous
Wastes Incinerators; P. Farber, W. Fischer, et al; Proceedings, The
User and Fabric Filtration Equipment IV Conference, November, 1988.
Experience with Air Pollution Control Equipment with Hazardous
Waste Incinerators; A. Buonicore, 90-33.2, Air and Waste Management
Association, Annual Meeting, June, 1990.
A Comparison of Air Pollution Control Equipment for Hospital Waste
Incinerators; D. Corbus; 90-27.4, Air and Waste Management Annual
Meeting, June, 1990.
Design Considerations for MSW Incinerator. APC Systems Retrofit;
J.R. Donnelly, 90-25.3, Air and Waste Management Annual Meeting,
June, 1990.
Improving Pulse-Jet Dust Collector Operation: The Effect of Gas
Stream Inlet Design: W. Gregg, Proceedings, Powder and Bulk Solids
Conference, June, 1990.
478
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before compter
1. REPORT NO.
EPA-600/R-92-209C
2.
3- £B 93-1241 Sb
4. TITLE AND SUBTITLE
Proceedings: 1991 International Conference on
5. REPORT DATE
November 1992
Municipal Waste Combustion, Volume 3.
1C, 2C, 3C, 4C, 6C, 7C, 8C, 9A, and 10A/C
Sessions
6. PERFORMING ORGANIZATION CODE
EPA/ORD
7. AUTHOR(S)
Miscellaneous
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
10. PROGRAM ELEMENT NO.
See Block 12
11. CONTRACT/GRANT NO.
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; 2/91-9/92
14. SPONSORING AGENCY CODE
EPA/600/13
1^SUP™TARYN°TESAEERL^r°iect officer is Theodore G. Brna, Mail Drop 62B,919/
541-2683. Vol. 1 contains Sessions P, 0,1A, 2A, 3A, 4A, 6A/B, 9C, and 10B. Vol. 2 con-
tains Sessions IB, 2B, 3B, 4B, 7A/B, 8A/B, and 9B.
16. abstract The ^ree"volumes document 82 presentations by authors from 15 countries
at the Second International Conference on Municipal Waste Combustion (MWC) in
Tampa, Florida, April 16-19, 1991. The Conference fostered the exchange of current
information on research concerning MWC, ash disposal and treatment, and flue gas
cleaning as well as unit operating experience, regulatory developments, and plant
siting considerations. Topics discussed included overviews on MWC from Canada,
Europe, and the U. S. ; MWC processes; dry/wet flue gas cleaning developments and
operating experience; ash characterization, treatment, utilization, and disposal;
chlorinated dioxin/furan control; environmental compliance; health risk; quality con-
trol/ assurance; municipal waste management; mercury emission control; sampling
and analysis; economic and social issues; and regulatory effects.
17.
KEY WORDS AND DOCUMENT ANALYSIS
a. DESCRIPTORS
b. IDENTIFIERS/OPEN ENDED TERMS
c. cosati Field/Gioup
Pollution Furans
Waste Disposal Health
Combustion Quality Assurance
Ashes Mercury (Metal)
Flue Gases Sampling
Halohydrocarbons Analyzing
Economics
Pollution Control
Stationary Sources
Municipal Waste
Dioxin
Health Risk
13 B
15E 06N
21B 13H. 14D
07B
14 B
07C
05C
18. DISTRIBUTION STATEMENT
19. SECURITY CLASS (This Report)
Unclassified
21. NO. OF PAGES
503
Release to Public
20. SECURITY CLASS (This page)
Unclassified
22. PRICE
EPA Form 2220-1 (9-73)
i
------- |