EPRI
Electric Power
Research Institute
Topics:
Participates
Electrostatic precipitators
Fabric filters
High-temperature filtration
Sulfur dioxide
Gaseous wastes
EPRI GS-7050
Volume 2
Project 1129-18
Proceedings
November 1990
                        Proceedings:  Eighth Particulate
                        Control  Symposium
                        Volume 2:
                        Baghouses and Particulate Control for New
                        Applications
                        Prepared by
                        Electric Power Research Institute
                        Palo Alto, California

-------
                                   REPORT     SUMMARY
                        SUBJECT   Fossil plant air quality control
                         TOPICS   Particulates
                                   Electrostatic precipitators
                                   Fabric filters
                                High-temperature filtration
                                Sulfur dioxide
                                Gaseous wastes
                      AUDIENCE   Environmental engineers and operators
                                   Proceedings: Eighth Particulate Control
                                   Symposium
                                   Volumes 1  and 2

                                   These proceedings describe the latest R&D efforts on improved
                                   participate control devices while treating traditional concerns of
                                   operational cost and compliance. Overall, participate control re-
                                   mains a key issue in  the cost and applicability of furnace sorbent
                                   injection, spray dryers, fluidized-bed combustion, municipal solid
                                   waste, and advanced power generation processes.
                   BACKGROUND
                      OBJECTIVE
                      APPROACH
                     KEY POINTS
Utilities confront increasingly stringent environmental regulations—including
proposed clean air legislation for SO2, NOX, and toxic air pollutants—as
well as growing pressures from competition. These conditions dictate a
need for upgrades to existing particulate controls, integration of particulate
devices into other pollutant-control systems, and construction of newer and
more-efficient control devices. The research community has responded to
these changes by producing a number of promising new technologies. Peri-
odic, comprehensive exchanges of information and ideas among devel-
opers, manufacturers,  and technology users stimulate and guide the
development process.

To provide a  forum for discussing the latest developments in particulate
control technology.

EPRI and EPA cosponsored the Eighth Particulate Control Symposium,
held  in San Diego, March 20-23,  1990, featuring more than  80 presenta-
tions.  Participants included approximately 350 representatives of utilities,
manufacturers, universities, architect/engineering firms,  and  research
organizations. Two parallel sessions emphasized fabric  filter and electro-
static precipitator (ESP) research. Several sessions addressed high-
temperature filtration as well  as the impact of new SO2 control processes
on baghouses and ESPs.

Symposium presentations highlighted several important issues in particulate
control technology. Participants noted that
• Underperforming ESPs continue to pose problems to  many utilities in
complying with particulate emissions regulations.
EPRI GS-7050S Vols. 1 and 2
                                   Electric Power Research Institute

-------
           • Fluidized-bed combustion and retrofit of in- and post-furnace sorbent
           injection processes can cause particulate collection problems,  such as
           excessive precipitator emissions and high baghouse pressure drops.
           • Relatively small pulse jet baghouses, which can be retrofitted into
           limited spaces, represent an attractive technology for utilities.
           • Utilities express growing interest in microcomputer-based ESP and
           baghouse models as well as expert systems.
           • Better understanding of dustcake characteristics and methods to
           modify the dustcake  can improve  ESP and baghouse operation.
           • Concern with control of toxic emissions, including volatile organic
           compounds and metals from fossil-fueled and waste-fired systems, con-
           tinues to mount among utilities.
           • Development of effective high-temperature particulate collectors re-
           mains a key problem for advanced power generation systems.
           Volume 1 contains papers on ESP technologies. Volume 2 focuses on
           fabric filter technologies and particulate controls for new applications
           (refuse-derived fuel, advanced SO2 control processes, and fluidized-bed
           combustion). EPRI and EPA sponsored other particulate symposia in
           1984 (EPRI  report CS-4404), 1986 (report CS-4918), and 1988  (report
           GS-6208).

PROJECT  RP1129-18
           EPRI Project Managers: Ramsay Chang;  Ralph F.  Altman
           Generation and Storage Division
           For further information on  EPRI research programs, call
           EPRI Technical Information Specialists (415) 855-2411.

-------
Proceedings: Eighth Particulate Control Symposium
  Volume 2: Baghouses and Particulate Control for New
                      Applications
                   GS-7050, Volume 2
                 Research  Project 1129-18

                Proceedings, November 1990

                    San Diego, California
                     March  20-23, 1990
                      Cosponsored by

             U.S. Environmental Protection Agency
             Office of Research and Development
                     401 M Street, SW
                   Washington, D.C. 20460

                     EPA Project Officer
                        G. Ramsey

        Air and Energy Engineering Research  Laboratory
          Research Triangle Park, North Carolina 27711

                           and

               Electric Power Research Institute
                    3412 Hillview Avenue
                  Palo Alto, California 94304

                   EPRI Project Managers
                        R. L. Chang
                        R. F. Altman

                 Air Quality Control Program
               Generation and Storage Division

-------
                                ORDERING INFORMATION
Requests for copies of this report should be directed to Research Reports Center
(RRC),  Box 50490, Palo Alto, CA 94303, (415) 965-4081. There is no charge for reports
requested by EPRI member  utilities and affiliates, U.S. utility associations, U.S. government
agencies (federal, state, and  local),  media, and foreign organizations with which EPRI has
an information exchange agreement. On request, RRC will  send  a catalog of  EPRI reports.
 Electric Power Research Institute and EPRI are registered service marks of Electric Power Research Institute, Inc

 Copyright <£> 1990 Electric Power Research Institute, Inc. All rights reserved


                                           NOTICE
 This report was prepared as an account of work sponsored in part by the Electric Power Research Institute, Inc.
 (EPRI)  Neither EPRI, members of EPRI, nor any person acting on their  behalf  (a) makes any warranty, express or
 implied, with respect to the use of any information, apparatus, method, or process disclosed in this report or that
 such use  may not infringe privately owned rights, or (b) assumes any liabilities with respect to the use of, or for
 damages resulting from the use of. any information, apparatus, method, or process disclosed in this report.

-------
                              ABSTRACT
The Eighth Symposium on the Transfer and Utilization of Particulate Control
Technology was held in San Diego, California, March 20 through 23, 1990.
This symposium, the fourth of its kind to be jointly sponsored by the
U.S. Environmental Protection Agency (EPA) and the Electric Power Research
Institute (EPRI), was designed to promote the transfer of results of particulate
control research and applications.

The symposium proceedings contain 80 papers presented by representatives
of utility companies, equipment  and process suppliers, university
representatives, research and development companies, EPA and other federal
and state agency representatives, and EPRI staff members.  Additionally,
individuals from fifteen countries presented information on worldwide
technological developments.  Electrostatic precipitators and fabric filters were
the major topics discussed during the symposium.  Other topics presented
included high temperature filtration, RDF incinerator emissions control,
system operation and maintenance,  and integrated  control processes.

Symposium cochairmen were Drs. Ramsay  Chang and Ralph Altman, Project
Managers in the Air Quality Control Program of EPRI's Generation and
Storage Division; and Geddes Ramsey, Project Officer in the Air Toxics
Control Branch of EPA's Air and Energy Engineering Research Laboratory.
Welcoming remarks were made for EPRI by Ramsay Chang and Dr. Ian
Torrens, Director of the Environmental Control Systems Department of
EPRI's Generation and Storage Division. The keynote address was given  by
Mr. George Green, Manager, Electric Operations Services, Public Service
Company of Colorado.

Plenary session addresses were given by Dr. Peter Davids, President, State
Agency for Air Pollution Control and Noise Abatement, North Rhineland-
Westphalia, Federal Republic of Germany; Mr. Ted Brna, Environmental
Engineer, U.S. Environmental Protection Agency; Mr.  Tom Bechtel, Director,
U.S. Department of Energy, Morgantown Energy Technology Center; and Mr.
Sabert Oglesby, President Emeritus, Southern Research Institute.

Papers from this conference are organized by session in two volumes as
follows:

Volume 1 contains papers presented in the sessions on: precipitator controls,
innovative pollution control technologies, precipitator modeling, fly ash/ESP
studies, ESP plate spacing, ESP rapping, ESP performance upgrading and  hot-
side precipitator studies. Except for papers on corona destruction of pollutants
in the innovative pollution control technology sessions, these papers are all
concerned with ESP technology.
                                   i i i

-------
Volume 2 contains papers presented in the sessions on:  low ratio baghouse
O&M experience, pulse-jet baghouse experience, particulate control for
AFBCs, particulate control for dry SO2 control processes, baghouse design and
performance, fundamental baghouse studies, high temperature filtration, and
control of emissions from RDF incinerators. Both fabric filters and ESPs are
discussed in the AFBC and dry SO2 control papers. The high temperature
filtration papers deal with ceramic barrier and granular bed filters. The rest of
the papers in Volume 2 are concerned with fabric filters on pulverized-coal-
fired boilers.
                                    IV

-------
                         TABLE OF CONTENTS
                               VOLUME 1

Precipitator controls, innovative pollution control technologies, precipitator
modeling, fly ash/ESP studies, ESP plate spacing, ESP rapping, ESP
performance upgrading and hot-side precipitator studies sessions.

Section                                                                   Page

Session 1A, New Controls for Precipitators -1

      Morris Tuck, Chairman

      Advanced Microprocessor Technology for Electrostatic                    1-1
      Precipitator High Voltage Control Systems
            E. H. Weaver and F.A. Gallo

      Case Study of a Hot-Side Precipitator Using Voltage Limit, Current        2-1
      Limit, Pulse Blocking, and Pulse Blocking with Background Power
            E. M. Drysdale, D. Wakefield, and J.Wester

      An Evaluation of the Energy Savings and Electrical Waveforms           3-1
      from the Intermittent Energization of Electrostatic
      Precipitators at Coal-Fired Stoker Utility Boilers
            P. Gelfand, J.A. Alden, D.J. McKay, and C.M. Richardson

Session 2A, New Controls for Precipitators  II

      Bill McKinney, Chairman

      Intermittent Energization Optimization on PSI Gibson Station            4-1
      Unit #1 Precipitator
            S. Szczecinki, J. Lantz, M. Neundorfer, and R. Pepmeier

      Full-Scale Demo of Intermittent Energization on a 500 MW               5-1
      Hot-Side Electrostatic Precipitator
            W.A. Harrison, R.P. Gehri, E.G. Landham, M.B. Tucker,
            and W. Piulle

      Experimental Evaluation of Improved Design of ESPs                    6-1
            B. Bellagamba, G. Dinelli, and E. Riboldi

-------
Section

      Delaying Sodium Depletion in Electrostatic Precipitators                  ^
      at Ghent Generating Station
            C.C. Robinson

Session 3A, Innovative Pollution Control Technologies

      Norm Plaks, Chairman

      DeNOX and DeSOX by PPCP and SPCP                                 8-1
            S. Masuda and J. Wang

      The Destruction of Volatile Organic Compounds by an Innovative        9-1
       Corona Technology
             G.H. Ramsey, N. Plaks, C.A. Vogel,W.H. Ponder, and L.E.Hamel

       Application of Corona-Induced Plasma Reactors to Decomposition        10-1
       of Volatile Organic Compounds
             T. Yamamoto, P.A. Lawless, K. Ramanathan, D.S. Ensor,
             G.H. Ramsey and N. Plaks

 Session 4A, Precipitator Model Studies

       Sydney Self, Chairman

       Requirements for a Precipitator Performance Expert System              11-1
              J.G. Musgrove and R.L. Jeffcoat

       An Integrated Electrostatic Precipitator Model for Microcomputers        12-1
              P.A. Lawless and R.F. Altman

        An Advanced Microcomputer Model for Electrostatic                   13-1
        Precipitators
              P.A. Lawless and N. Plaks

        Measurements Inside a Model Precipitator                              14-1
              A.L.H. Braam and W. Hiemstra

  Session 5A, Fly Ash/ESP Studies

        Scott Thomas, Chairman

        The Effects of Fireside Process Conditions on Electrostatic                15-1
        Precipitator Performance in the Electric Utility Industry
              H.J. Hall
                                    VI

-------
Section                                                                  Page
      Observations of Modeled and Laboratory Measured Resistivity           16-1
            R.E. Bickelhaupt

      Computer Model Developed to Predict ESP Performance                17-1
      Based on Coal Quality
            Wm. Borowy (No paper provided)

 Session 6A, Precipitator Plate Spacing Studies

      Joe Kaminski, Chairman

      Engineering Study of Wide Plate Spacing                              18-1
            K.S. Kumar and P.L. Feldman

      Increased Plate Spacing in Electrostatic Precipitators                     19-1
            K. Darby and D. Novogoratz

      Mechanism of Performance Enhancement in Wide                     20-1
      Plate Electrostatic Precipitators
            H. Elshimy and G.S.P. Castle

 Session 7A, ESP Rapping Studies

      Geddes  Ramsey, Chairman

      Characteristics of Rapping Acceleration of Precipitator Collecting         21-1
      Plates Before and After the Installation of Plate Straightening
      Devices
            J. Cummins

      Temperature Dependency of Magnetic Impact Rappers                  22-1
            M.W. Neundorfer, K.M. Artz, and M.A. McNabb

      Experimental Study of Ash Rapping of Collector Plates in a Lab-Scale     23-1
      Electrostatic Precipitator
            D.H. Choi, S.A. Self, M. Mitchner and R. Leach

 Session 8A, ESP Performance Upgrading Studies  1

      Wallis A. Harrison, Chairman

      Meeting Emission Levels Through Precipitator Upgrades                24-1
            S.F. Weinmann and K.R. Parker
                                   VI 1

-------
Section

       Operating Experience of the Rigid Frame Electrostatic Precipitators        25-1
       Installed at Metropolitan Edison Company's Portland Station
            P.G. Abbott, T.C. Schafebook and J.A. Brummer

       Modern Electrode Geometries and Voltage Waveforms                  26-1
       Minimize the Required SCAs
            K. Porle, S. Maartmann, M.O. Bergstrom and K. Bradburn

       ESP Design Concepts for Improving Performance and Reliability         27-1
            J.R. Meinders and R.E. Jonellis

Session 9A, ESP Performance Upgrading Studies  II

       Phillip  Lawless, Chairman

       Considerations in Rebuilding the Sibley Unit 1 Precipitator              28-1
            D.M. Greashaber and P.A. Miller

       Flue Gas Field Study, Model Study, and Post-Study Review to            29-1
       Improve the Performance of a Chevron ESP at Duke Power's
       Belews Creek Station
             S.L. Thomas and L.A. Zemke

       Experience with Dual Flue Gas Conditioning of Electrostatic             30-1
       Precipitators
             H.V. Krigmont and E.L. Coe, Jr.

 Session 10A, Hot-Side Precipitator Studies

       Richard Roberts, Chairman

       Modification and Conversion of the Nebraska  City Unit 1 Hot            31-1
       ESP to Cold-Side Operation
             A.W. Ferguson, R.C. Wicina, B.L. Duncan, K.A. Roth
             and R.M. Kotan

       Results of the Roy S. Nelson Unit 6 Hot-Side Precipitator                32-1
       Structural Evaluation
             C.R. Reeves, S.A. Johnson and R.L. Schneider

       Columbia Unit 2  Precipitator Hot to Cold Conversion                  ^ i
             M. Vakili and A.W. Ferguson
                                   VI 1 1

-------
Section                                                                  Page

 Session 11 A, Innovative Pollution Control Technologies

       Ralph Altman, Chairman

       Improved Carbon Particulate Control via Additive Injection            35-1
             D. Farrar, J. Reuther, W. Steiger, R. Schmitt, and
             R. van der Velde

       In-Line Particle Measurement Instrument for Power Generation        36-1
       System
             D.J. Holve (No paper provided)

-------
                                VOLUME 2

Low ratio baghouse O&M experience, pulse-jet baghouse experience,
particulate control for AFBCs, participate control for dry SO2 control
processes, baghouse design and performance studies, fundamental baghouse
studies, high temperature filtration, and control of emissions from RDF
incinerators  sessions.  Both fabric filters and ESPs are discussed in the AFBC
and dry SO2 control papers.

Section                                                                   Page

Keynote Address

      The Need for Continued Research and Development in                  K-l
      Particulate Control
            G. Green (No paper provided)

Session IB, Low Ratio Baghouse O&M Experience

      John Mycock, Chairman

      The Operation and Maintenance History at the City of Colorado           1-1
      Springs, Martin Drake #6 Reverse Gas Fabric Filter System
      ("Over a Decade of Excellence" 1978 to 1988)
            R.L. Miller and L.V. Hekkers

      1990 Update, Operating History and Current Status of Fabric Filters        2-1
      in the Utility Industry
            K.M. Gushing, R.L. Merritt and R.L. Chang

Session 2B, Pulse-Jet Baghouse Experience  I

      Richard Rhudy, Chairman

      Australian Experience with High Ratio Fabric Filters on Utility           3-1
      Boilers
            P.R. Heeley and C. Robertson

      A Ten-Year Review of Pulse-Jet Baghouse Operation and                4-1
      Maintenance at the H.R. Milner Generating Station
            B.R. Thicke

      Design and Performance Evaluation of a 350 MW  Utility                5-1
      Boiler Pulse-Jet Fabric Filter
            P.W. R. Funnell, P.R. Heeley, and S. Strangert

                                   xi

-------
Section                                                                   Page

       A Survey of the Performance of Pulse-Jet Baghouses for                  6-1
       Application to Coal-Fired Boilers, Worldwide
             V.H. Belba, T. Grubb, and R.L. Chang

 Session 3B, Pulse-Jet Baghouse Experience - II

       Brian  Thicke, Chairman

       Retrofit of Fabric Filters to Power Boilers                               7-1
             H.F. Johnson

       The EPRI Pilot Pulse-Jet Baghouse Facility at Plant Scholz                8-1
             K.J. Mills and R.F. Heaphy

       Pilot Demonstration of a Pulse-Jet Fabric Filter for Particulate            9-1
       Matter Control at a Coal-Fired Utility Boiler
             R.C. Carr and C.J. Bustard

 Plenary Session

       George Offen, Chairman

       Acid Rain Regulations in Germany and  their Effects                     Pl-1
             P. Davids

       Particulate Emissions Control and its Impacts on the Control of          P2-1
       Other Air Pollutant Emissions from Municipal Waste
       Combustors
             T.G. Brna

 Session 4B, Particulate Control for AFBCs

       Tom  Boyd, Chairman

       Baghouse Design Considerations Unique to Fluidized-Bed Boilers        10-1
             J.B. Landwehr, F.W. Campbell and J.G. Weis

       Fabric Filter Monitoring at the CUEA Nucla AFBC Demonstration        11-1
       Plant
             K.M. Gushing,  T.J. Heller, R.F. Altaian, T.J. Boyd,
             M.A. Friedman, and R.L. Chang
                                    XI 1

-------
Section                                                                  Page

      Electrostatic Precipitation of Particles Produced by Three Utility           12-1
      Fluidized-Bed Combustors
            E.G. Landham, Jr., M.G. Faulkner, R.P. Young, R.F. Altman
            and R.L. Chang

      Choice Between ESP and Baghouse for Pakistan's First Coal-Fired         13-1
      Power Plant
            G.M. Ilias (no paper provided)

Session 5B, Particulate Control for Dry SO2 Control Processes

      Michael Maxwell, Chairman

      Effects of  E-SOX Technology on ESP Performance                        14-1
            G.H. Marchant, J.P. Gooch, M.G. Faulkner and L.S. Hovis

      Identification of Low-Resistivity Reentrainment in ESPs                 15-1
      Operating in Dry Scrubbing Applications
            M.D. Durham, R.G. Rhudy, T.A. Burnett, J. DeGuzman,
            G.A. Hollinden, R.A. Barton and C.W. Dawson

      Electrostatic Precipitation of Particles Produced by Furnace Sorbent        16-1
      Injection at Edgewater
            R.F. Altman, E.G. Landham, E.B. Dismukes,
            M.G. Faulkner, R.P. Young and L.S. Hovis

      Proposed Demonstration of HYP AS on Duke Power's Marshall           17-1
      Station Unit 2: An Integrated Approach to Particulate Upgrades
      and SO2 Control
            K.W. Knudsen, R.C. Carr, and R.G.  Rhudy

Session 6B, Baghouse Design & Performance Studies -1

      Lou Hovis, Chairman

      Influence  of a Sock Between Supporting Cage and Bag on Filter           18-1
      Performance
            E. Schmidt and F. Loffler

      Accelerated Bag Wear Testing                                        19-1
            L. G. Felix, R.F. Heaphy, R.F. Altman, R.L. Chang
            and W.T. Grubb

      Collection of Reactive and Cohesive Fine Particles in a  Bag Filter         20-1
      with Pulse-Jet Cleaning
            E. Schmidt and F. Loffler
                                  XI 1 1

-------
Section
      DuPont's Engineering Fibers for Hot Gas Filtration Case                21-1
      Histories
            P.E.  Frankenburg
Plenary Session

      Walter Piulle, Chairman

      Advanced Power System Particulate Control Technology               P3-1
             T.F. Bechtel (No paper provided)

      Future Directions  in Particulate Control Technology                   P4-1
             S. Oglesby

 Session 7B, Baghouse Design & Performance Studies   II

       Bob Carr, Chairman

       Optimizing Baghouse Performance at Monticello Station with           22-1
       Ammonia Injection
             K. Duncan, R.  Watts, R.L. Merritt, P.V. Bush, W.V. Piulle
             and R.L. Chang

       Enhancing Baghouse Performance with Conditioning Agents:           23-1
       Basis, Developments and Economics
             S.J. Miller and  D.L. Laudal

       Baghouse Performance Advisor  A Knowledge Based Baghouse          24-1
       Operator Advisor
             J.P. Eckenrode, G.P. Greiner, E. Lewis, and R.L. Chang

       Efficiency of Fabric Filters and ESPs in Controlling Trace Metal           25-1
       Emissions from Coal-Burning Facilities
             R.C. Trueblood, C. Wedig, and R.J. Gendreau

  Session 8B, Fundamental Baghouse Studies

       Grady Nichols, Chairman

       The Structural Analysis of Dustcakes                                  26-1
              E. Schmidt and F. Loffler
                                    xiv

-------
Section                                                                   Page

      Effects of Additives and Conditioning Agents on the Filtration           27-1
      Properties of Fly Ash
            P.V. Bush and T.R. Snyder

      Particle Size Effects on High Temperature Dust Filtration from a         28-1
      Coal-fired Atmospheric Fluidized-Bed Combustor
            R.A. Dennis, L.D. Strickland, and  T.K. Chiang

      Generalization of Laboratory Dust Cake  Characteristics for Full-Scale      29-1
      Applications
            T.K. Chiang, R.A. Dennis, L.D. Strickland, and C.M. Zeh

Session 9B, High Temperature Filtration I

      Steve Drenker, Chairman

      High Temperature Filtration Using Ceramic Filters                     30-1
            L.R. White and S.M. Sanocki

      High Temperature Filter Media Evaluation                             31-1
            DJ. Helfritch and P.L. Feldman

      Pilot-Scale Performance/Durability Evaluation of 3M Company's         32-1
      High-Temperature Nextel Filter Bags
            G.F Weber and G.L. Schelkoph

      Particle Control in Advanced Coal-Based Power Generation Systems      33-1
            S.J. Bossart and C.V. Nakaishi

Session 10B, High Temperature Filtration  II

      Richard Dennis,  Chairman

      Performance of a Hot Gas Cleanup System on a Pressurized              34-1
      Fluidized-Bed Combustor
            J. Andries, J. Bernard, B. Scarlett and B.  Pitchumani

      Electrified Granular Filter for High Temperature Gas Filtration          35-1
            P.H. deHann, M.L.G. van Gasselt  and L.M. Rappoldt

      Nested  Fiber Filter for Particulate Control                              36-1
            R.D. Litt and H. N. Conkle
                                    xv

-------
Section                                                                  Page

Session 11B, Control of Emissions from RDF Incinerators

       Ted Brna, Chairman

       Particulate Emissions from Prepared Fuel (RDF) Municipal              37-1
       Waste Incinerators
            R. M. Hartman

       Condensible Emissions from Municipal Waste Incinerators              38-1
            A.S. Damle, D.S. Ensor and N. Plaks

       Treatment of Flue Gas and Resideus from Municipal and                39-1
       Industrial Waste Incinerators
            G.  Mayer-Schwinning
                                   xvi

-------
                Keynote Address
The Need for Continued Research and Development in
               Participate Control
                    G. Green
               (No paper provided)

-------
                      THE OPERATION & MAINTENANCE HISTORY
           OF THE CITY OF COLORADO SPRINGS, MARTIN DRAKE UNIT NO. 6
                       REVERSE GAS FABRIC FILTER SYSTEM

                         "Over A Decade Of Excellence"
                                 (1978  TO  1988)
                               Richard L. Miller
                       Particulate Technology Department
                            GE  Environmental  Systems
                             Lebanon, Pennsylvania
                               Leslie V. Hekkers
                            City  of Colorado  Springs
                            Department of Utilities
                        Environmental Services Division
                           Colorado Springs,  Colorado
                                    ABSTRACT

The City of Colorado Springs, Martin Drake Unit #6 reverse gas fabric filter
installation has successfully been in operation since September of 1978 and has
consistently demonstrated excellent overall performance and minimal required
maintenance while still utilizing the originally installed woven fiberglass
filter bags.  This paper briefly describes the baghouse start-up and operating
procedures along with the general operating and maintenance history of the
baghouse system which have contributed to the success of this utility industry
pioneer installation over the initial ten (10) year operating period.
                                       1-1

-------
                                  INTRODUCTION

The Martin Drake Generating Station is owned  and operated by the City of
Colorado Springs, Department of Public Utilities.  Unit No. 6 is a Babcock  and
Wilcock 85 MW, front fired, pulverized coal fired  pressurized furnace steam
generator rated at 685,000 Ibs. of steam per  hour  with  a  pressure of  1325 psi
at the superheater outlet and 950°F.  The  unit was originally equipped  with a
cold side electrostatic precipitator which  had been  retrofitted  with  a  sulfuric
acid gas conditioner in 1972.  Due to more  poor precipitator performance  and
more stringent environmental regulations Unit No.  6  was retrofitted with  a
reverse gas  fabric filter system.

The  Unit  No.  6 fabric  filter installation  was originally  part of a  research and
development  product  optimization  project  (1)  which was  awarded  to  the Buell
Emission  Control  Division of Envirotech Corporation  in  March, 1977.   Buell  was
subsequently acquired  by the General Electric Company  in  1981 and  is  now  known
as GE  Environmental  Systems.  The R&D program evaluated the design, various
filter materials  and operational  parameters of a full  scale fabric  filter
 system with  the  goal to  advance  the  state-of-the-art of fabric  filtration for
 combustion  sources.  This  paper  will address  the historical operation,
 performance  and  maintenance  aspects  of the Unit No.  6  fabric filter system  over
 the initial  ten  (10) year  time operating  period from start-up in September,
 1978 to September,  1988.   Examined  in detail  will  be the  condition  of the
 original  woven  glass filter  bags  which have successfully  been in operation
 throughout  this  extended time period with  minimal  replacements.

 Initial flue gas  operation  (2) of the Unit No. 6 fabric filter  system began on
 September 15, 1978 with  commercial  operation  two days  later on  September  17,
 1978.   The  unit  burns  a  Western,  low-sulfur,  bituminous coal from  the Routt
 County, Northwestern Colorado  area.  Typical  coal  and  ash analysis  is included
 in Table 1.

 Under the terms  of the contract,  the fabric filter system had to provide  an
 overall particulate collection efficiency  of  99.86%  which is equivalent to
 0.008 grains/ACF at an inlet dust loading  of  5.55  grains/ACF.   The  fabric
 filter system has met  all  contractual performance  requirements  with detail
 results of emission testing  conducted on  the  unit  presented  in  Table  3.
                                       1-2

-------
                              FABRIC  FILTER  DESIGN

The fabric filter system installed on Unit No. 6 consists of a single casing
with a total of twelve (12) compartments, each containing 198 filter bags per
compartment.  This equals a total of 2,376 bags for the installation.  Detailed
design parameters are presented  in Table 2.

Scope of Supply

The original scope of supply for Buell under this program was to supply the
fabric filter equipment; technical erection supervision; initial operation
advisory services; and conduct performance testing and research and development
program activities.  The City of Colorado Springs provided the foundations, ash
handling system,  induced draft fan,  hopper enclosures, piping and wiring, and
auxiliary equipment and erection.  The City also provided overall operation and
maintenance services for the fabric  filter system.

Bag Design

The original filter bags used are a  nominal  twelve (12) inches in diameter by
30'-6" in length.  The filtration material is a 9.5 oz./yd.^ woven fiberglass
with  a 10%  Teflon-B coating as supplied  by Menardi-Southern Company.  Each bag
contains seven  (7) variably spaced stainless steel anti-collapse rings and are
attached to the thimbles with a  stainless steel clamp.

General Design  Features

The design  of the baghouse incorporates  pneumatically operated poppet valves
for the inlets, outlets, re-inflation, and reverse gas valves.  One  (1) valve
for each function is provided per compartment.  The bypass consists of a double
louver damper located between the inlet  and outlet ducts.  A 3,000 CFM fan
provides pressurized air between the louvers to prevent dirty flue gas from
bypassing the baghouse due to damper leakage during normal operation.

Cleaning System Design

The baghouse cleaning system consists of two  (2) reverse gas cleaning fans with
one running during normal operation  and  the second fan as a standby.
                                       1-3

-------
Cleaning System Design Continued

The fans were sized for a maximum of 2.0 cfm/ft.2 of effective cloth area per
compartment.  This is equivalent to 36,000 ACFM flow rate at a 12.0 "W.C.
static pressure capacity.  The cleaning cycle utilized  is a batch cleaning
method which is initiated when the flange-to-flange pressure loss reaches a
preset level of 4.5 "W.C..  Each compartment is cleaned  for 4.5  minutes  for  a
total cleaning cycle of 55 minutes.  The compartments will remain in  the
filtration  mode until the baghouse pressure  loss  level  again reaches  this
preset pressure value or the  system can be programmed to remain  in  the cleaning
mode  continuously.

                               OPERATING HISTORY

 Initial  Operation

 The baghouse was  started  up without precoating  as the boiler startup  fuel was
 natural  gas.  The unit  was  started  (£)  up  in the  bypass  mode until  stable
 boiler operation  was  established with  the  flue  gas  temperature  well  above the
 moisture dew point at  an  inlet temperature  (4)  of 235°F. All compartments  were
 placed in the filtration  mode on natural gas until  the  entire baghouse was  at
 temperature and expanded  to  its  fullest structural  extent.

 At 3:30 PM on September 15,  1978 the  first  coal mill was placed on-line  and the
 first fly ash laden flue  gas  entered  the  baghouse.  The unit was operating  on
 100% coal by 5:25 PM with  an  opacity  of 1-1/2%  at a boiler  load of  20 MW.   On
 September 16 the baghouse  was placed  into  automatic operation at a  stabilized
 load of 70 MW.  Opacities  were  near zero with visible emissions at  levels  as
 low as 1%.  Uhen the baghouse was  placed  in  its first automatic cleaning cycle,
 opacity levels spiked as  high as 40-60%.   After approximately 2 weeks of
 operation, the filter bags developed  a more  stable  filter cake  and  opacity
 levels settled out to less than  2% during  cleaning  and  less  than 1% during
 normal operation.  The reverse gas  fan was  initially  set with only  a 10% open
 inlet damper to minimize  the  cleaning effectiveness.  At full  load  there was
 approximately five (5)  hours  between  cleaning  cycles with pressure  drops
 averaging 4.5 "W.C. across the bags.   By  October  3rd.,  the  compartment pressure
 drops were 4.5" before cleaning,  reducing  to a  level  of 1.5"  after  cleaning.
 Boiler loads were 85 MW with  4-1/2  hours  between  cleaning cycles.
                                        1-4

-------
In order to prevent structural damage to the baghouse during startup, the
baghouse compartments are placed into service from the center structurally
fixed point outwards to allow equal expansion of the steel as it is heated by
the incoming flue gas as shown in Figure 1.  Note: Compartments are placed in
service in a reverse sequence during boiler startup.  Compartments 5,6,11 & 12
first, then 4 & 10, followed by 3 & 9, then 2 & 8, finally 1 & 7.

Ten Year Overview

Bag Life.

During the first ten (10) months of operation there were only two (2) bag
failures reported which equates to only 0.084% of the total number of bags
installed.  At the four (4) year mark, 5% had failed plus an additional  5%
which were intentionally replaced for tests, bag evaluations, etc..  By March,
1984 with 5-1/2 years of operation, a total of 455 bags had been replaced, or
19.15% of the total quantity of original bags.  Of these, 198 test bags had
been replaced by Buell in early 1981.  These bags were Acid Flex coated bags as
supplied by Fabric Filters located in compartment #5.  These bags were replaced
due to excessive elongation of the filter bags beyond the point where they
could be properly tensioned.  This type of stretching occurred due to the fact
that the fabric was not pre-tensioned during manufacturing of the bags prior to
cutting/sewing, resulting in stretching as much as six (6) inches under
tension.  These bags were replaced with 10% Teflon-B bags as originally
supplied in the remainder of the baghouse.

At the end of the ten (10) year operating period, the total number of bags that
had been replaced due to failures was 552 or 23% of the total number of bags.
Bags failed within six feet of the bag thimble inlet and generally in groups of
two, three or four bags.  Rough handling during inspection, replacement and
compartment cleaning was the most probable cause.
 Also during this time period a large number of test bags had been installed in
compartments 3,4,9, and 10 by various bag suppliers.  These bags were not
included in the total number of failures.  Bag life by any measure has been
excellent.  Even after ten years of operation the original filter bags remain
in excellent shape retaining normal strength losses for this length of service.
                                       1-5

-------
Figure 2 shows strength loss vs time over the total life of the filter bags
since initial operation.  Recent bag tests (5) indicate that there is no
evidence of any chemical or thermal deterioration of the fabric.  The bags from
a flow standpoint are extremely good, exhibiting a generally porous cake
structure with low levels of moisture induced agglomerates.  Due to this
condition, the residual cake exhibits high discharge levels of  ash.
Baghouse Performance.

In January of  1979, extensive testing was performed to determine the  particle
size  distribution of hopper samples from all twelve (12) compartments and  also
from  bag  shakeout samples from baghouse compartments numbers 1,6,  and 12.  This
data  as presented in Figure 3 indicates that there is no significant  variation
of fly  ash particle size distribution between hopper samples, with an average
mass  mean diameter by  volume of 13.1 microns.  Of greater  interest is the  fact
that  the  ash  samples from the filter bags shows a definite  reduction  in
particle  size  distribution  in comparison to the hopper samples  with  an  average
mass  mean diameter by  volume of 7.2 .microns.  This indicates that  there  is a
 significant  amount of  classification of the ash occurring  in the hopper  area
 itself, with  the  larger particles  dropping out in the hoppers and  the finer ash
 portion passing  through the tubesheet and going to the bags.

 In May  of 1979,  formal  performance testing of the baghouse  was  conducted by an
 independent  test  company.   Tests were conducted to determine the overall
 efficiency  of the baghouse, inlet  and outlet dust loadings  and  in  addition,
 they were contracted to conduct baghouse inlet and outlet  particle size
 distribution tests.  These  tests indicated  (6) that an overall  efficiency  of
 99.93%  was  achieved  based on an inlet dust  loading of 1.84  Gr/ACF  and an outlet
 loading of  0.0013 Gr/ACF  (0.004 Ib/MM BTU).

 The  particle size distribution tests conducted at the baghouse  inlet  and stack
 outlets were measured  using cascade  impactors.  These tests demonstrated an
 average inlet size distribution of approximately 29 microns mean diameter  with
 an  outlet particle size distribution of 2.2 microns. A detail composite
 distribution curve showing  both inlet and outlet tests are  shown  in  Figure 4.
                                       1-6

-------
Pressure drops have been at moderate levels since startup.  Over the past four
(4) years, pressure drops have averaged from as low as 2.43 " W.C. at a load of
40 MW,  to as high as 5.19 " W.C. at 80 MW.  These figures are based upon normal
gross filtration operation.  Opacity has been less than five (5) percent for
the majority of time since startup with three (3) to four (4) percent opacity
on average when there are no bag problems.  These problems have generally been
holes in bags or bags slipping off the thimbles.
Baghouse Operating Procedures.

The baghouse startup operating procedures are incorporated into the boiler
startup procedures.  The boiler is fired with natural gas fuel  and is warmed up
to 400°F economizer gas outlet temperature before any flue gas  is admitted to
the compartments.  During this time, the baghouse bypass damper is open and the
compartment ash hopper inspection ports are open.  If there is  any leakage
through the compartment inlet and outlet valves,  the open inspection ports
allow ambient air to enter the compartments instead of flue gas out of the
compartments and minimizes the potential for condensation forming which could
lead to premature blinding of the bags.

The open baghouse bypass damper allows the boiler to be started up on clean
natural gas fuel and the compartments are put on line with clean hot flue gas
before firing coal.  When a 400°F economizer gas outlet temperaturehas been
reached, four compartments are put in service and the bypass damper is closed.
These four compartments are warmed up using a relatively high flue gas flow
rate which brings the compartment through the dew point quickly and sweeps the
moisture out of the compartments.  If there were more compartments in service
at this time, there would be a lower flow rate through each one and possible
more moisture condensing on the bags.

When the steam generator is on line, there is a higher flue gas flow and two
more compartments are put in service.  At 35 megawatts load, the boiler fuel is
switched to 100% coal and two more compartments are put in service.  This
continues as the flue gas flow increases until all the compartments are in
service.  Experience has shown this procedure to be effective at keeping the
moisture out of the compartments and helping extend the life of the bags.  The
shutdown procedure keeps all the compartments in service until  the forced draft
fans are shut down.  By this time, all the flue gas is purged and only low
moisture ambient air remains in the compartments.
                                       1-7

-------
                          BAGHOUSE  MAINTENANCE HISTORY

Bag Tensionlnq.

Bag tensioning is performed once per year to  an overall tension level of
approximately 75 pounds.  Some of  the original springs  have  also been replaced
with a spring of a larger diameter wire to eliminate permanent  set.   In
addition, bag clamps have been replaced as required due to breakage.

Compartment Weld Leaks.

During the first few years of baghouse operation  at Martin Drake,  the baghouse
had  numerous  compartment weld leaks that contributed to higher  than  expected
opacities and bag wear.   Some bag  failures were caused  by direct  impingement  of
fly  ash  against  the filter bags.   The cause of this ash abrasion was from
missing  and/or cracked  welds along the compartment walls and along seams  on the
tubesheet.  The  plant maintenance  department  has  resolved the majority  of the
 leaks  by seal  welding where required  in the compartments.

General  Repairs.

 Normal  maintenance  has  occurred  over  the last ten years of operation which
 included areas  such as  expansion joints, solenoid valves, replacement of
 various  damper cylinders,  recorder failures,  compressed air  dryer  maintenance,
 etc..
                             CURRENT SYSTEM UPGRADE

 In 1988 a major project (6)  was begun  to  upgrade  the reliability of the Unit #6
 baghouse.  This upgrade was  required due  to  the potential  environmental impact
 on Unit #6.  Deterioration of the bag  support frames as well  as the age of the
 bags themselves were causing bag problems.   Starting in late  1988,  a planned
 bag replacement program was  developed  to  enable the plant  to  start  replacement
 of the filter bags on a compartment by compartment  basis over an extended time
 period.  As of January 1,  1990, all but two  (2) compartments  have been replaced
 with new filter bags identical  to the  original  construction.
                                       1-8

-------
In spring of 1989, a modification to the bag support structure was designed by
The City of Colorado Springs to correct deflection in the bag support beams
which had occurred over the years.  Work began on this modification in August,
1989 along with installation of new replacement of bags for those which had not
been previously replaced.  The bag supports experienced extra deflection in the
center which allowed sagging due to the weight of the bags, filter cake and
tensioning loads over the last ten (10) years of operation.  The modification
consisted of adding support hangers from the support beam to the roof sections
on the end 33 bag racks.  The center 66 bag racks were modified with a new
support beam welded between the walls and on top of the original support beam.
The old beam was cut in the middle and drawn up level against the new beam and
welded.

                                    SUMMARY

The baghouse installation at the City of Colorado Springs, Martin Drake Unit #6
has been highly successful since its initial operation in September of 1978.
Performance testing has indicated emissions as low as 0.0013 GR/ACF (0.004
Ib/MM BTU) with an overall efficiency of 99.93%  and opacities of less than 4%
with average pressure drop levels of approximately 5.2" W.C. under full load
conditions.

The filter bags have lasted over ten (10) years of continuous operation with
only 23% of the original bags being replaced due to actual failures.  A total
bag replacement program has been initiated as part of an overall system upgrade
to prevent any potential environmental impact from the normal long term
deterioration of the Unit #6 baghouse system.  As of January 1, 1990, all but
two (2) compartments of original bags have been replaced.  Bag strengths in
these compartments are extremely good with no evidence of either chemical or
thermal deterioration and still exhibit a generally porous filter cake with low
levels of agglomerates.

Historically the baghouse system at the City of Colorado Springs has been
highly successful with minimal required maintenance and exceptional overall
performance due in part to the excellent care and scheduled routine maintenance
                 provided  by  the  plant  maintenance department.
                                       1-9

-------
                                ACKNOWLEDGMENTS

We wish to acknowledge the cooperation and effort that Charlie Boland,
Mechanical Supervisor for the City of Colorado Springs at the Martin Drake
Station provided in reviewing all the vast amount of information on the
baghouse operation and maintenance which allowed us to reconstruct its history
to allow a more accurate paper to be written.  We also acknowledge the help of
Environmental Consultant Company (ECC) for their effort in reconstructing the
bag testing history for this installation.

                                   REFERENCES

1.   Ronald L. Ostop and Larry A. Thaxton.  "Optimization Of Material, Design
     And  Operational Parameters Associated With A Full-Scale 400,000 ACFM
     Fabric Filter Baghouse On The City Of Colorado Springs' Martin Drake
     Generating Unit No. 6." 40th Annual Meeting Of The American Power
     Conference, April 26, 1978, Sponsored By The Illinois Institute Of
     Technology, Chicago, 1978.

2.   Ronald L. Ostop and John M. Urich, Jr.. "Start-up And Initial Operational
     Experience On A 400,000 ACFM Baghouse On City Of Colorado Springs' Martin
     Drake Unit No. 6."  Second Symposium On The Transfer And Utilization Of
     Particulate Control Technology."  July 23-27, 1979, Denver, Colorado.

3.   Ronald L. Ostop.  "Baghouse Research & Development And S02 Removal."
     Utility Air Pollution Control Seminar, Ponte Verva, Florida.  September
     29-30, 1978.

4    David A. Single.  "Internal Buell Report, Colorado Springs Baghouse
     Initial Operation Summary."  October 16, 1978.

5.   Environmental Consultant Company.  "Filter Bags Test Report On Four (4)
     Used Bags From The City Of Colorado Springs, Martin Drake No. 6 Baghouse."
     January 30, 1990.

6.   "Buell Internal Baghouse Performance Test Report, Martin Drake Unit #6."
     July 31, 1979.  Enviro-Test, Ltd.
                                       1-10

-------
                               COAL ANALYSIS

PROXIMATE ANALYSIS (%)             As RECEIVED              DRY BASIS
     MOISTURE                           9.91
     ASH                                5.42                    6.01
     VOLATILE MATTER                   35.90                   39.85
     FIXED CARBON                      48.77                   54,14
                                      100.00                  100.00

CALORIFIC VALUE (BTU/LB)           11,420.00                12,676.00
SULFUR (%)                              0.43                     0.48
MINERAL MATTER FREE (BTU/LB)            -                   13,568.00

ULTIMATE ANALYSIS (%)
     MOISTURE                           9.91
     CARBON                            65.73                    72.96
     HYDROGEN                           4.43                     4.92
     NITROGEN                           1.52                     1.69
     SULFUR                             0.43                     0.48
     ASH                                5.42                     6.01
     OXYGEN (DIFFERENCE)               12.56                    13.94
                                      100.00                   100.00

                         ELEMENTAL ANALYSIS OF ASH

                         OXIDE               ASSAY %
Si02
AL203
Ti02
FE203
CAO
MGO
NA20
K20
P205
S03
64.80
24.04
0.66
3.51
3.56
0.92
1.15
1.55
0.80
0.51
          % Loss ON IGNITION = 0.92

                                TABLE 1

                                  1-11

-------
     SYSTEM  DESIGN PARAMETERS
GAS FLOW RATE:               400,000 ACFM
GAS TEMPERATURES:
     DESIGN
     RANGE

GAS-TO-CLOTH RATIOS:
     GROSS
     NiT-1 W/0 REV.GAS VOL
     NET-2 W/O REV.GAS VOL

MAX INLET DUST LOADING:

MAX OUTLET DUST LOADING:

DESIGN PRESSURE DROP:
315 DEG. F
250-550 DEG. F
1.85 TO 1
2.02 TO 1
2.22 TO 1

5.5 GR/ACF

0.005 GR/ACF (99.86%)

6.0 IN. W.C.
                    TABLE 2
- - -
MARTIN DRAKE UNIT #B BAGHOUSE
EMISSION TEST RESULTS
Date Of Test [NLET OUTLET EFFICIENCY
GR/ACF GR/ACF ttlWM ETU %
05-24-79 1.32 0.001 0.004
05-25-79 2.15 0.004 0.004
05-30-79 1.50 0.001 0.005
05-31-79 1.42 0.003 * 0.009
06-01-79 2.81 0.001 0.004
* BOILER
99.90
99.80
99.91
99.80
99.95
UPSET DURING TEST
                    TABLE 3
                      1-12

-------

CENTER STRUCTURALLY FIXED EXPANSION POINT
Expansion ^11 X |M^ Expansion
1
2
3
4
5
6

7
8
9
10
11
12
1
Expansion

                      FIGURE 1
c
12000
10000
8000
0
ITY OF COLORADO SPRINGS MARTIN DRAKE #6
FILTER BAG STRENGTH LOSS VS TIME
MIT FLEX MIT FLEX
(WARP) (FILL)
WARP














v
\
\

X\
A
\ '
\

























' 	 — .
"~"~-----_












•-Trr""""''1-— ~.
***•••

FILL










—^^rss^













1000
800
600
400
200
0
1978       1980       1985       1986      1987
                YEAR FABRIC TESTED
1990
                      FIGURE 2
                         1-13

-------
BAGHOUSE HOPPER/BAG ASH PARTICLE SIZE DISTRIBUTIONS







Bag: 8.3 Micron*
1
13.7 Microns

2
14.5 Micron*

3
13.8 Micron*

4
12.1 Micron*

5
13.0 Micron*
Bag: 7.1 Micron*
6
11.7 Micron*
HI^GAS FLOW

7
13.7 Micron*

8
14.0 Micron*

9
10.9 Mlcrone

10
13.5 Micron*

11
16.0 Micron*
B*g: 6.4 Micron*
12
10.9 Micron*








                            FIGURE 3
   BAGHOUSE  PARTICLE  SIZE   DISTRIBUTION
                INLET VS OUTLET SAMPLING
 _
o
Q^
O
        10  20  30  40  50  60  70  80  9O  1 OO

            %  LESS  THAN STATED SIZE
                                                TEST  CONDUCTED


                                                	 TEST A INLET


                                                	TEST B INLET


                                                	 TEST C INLET


                                                	 TEST A OUTLET


                                                	 TEST B OUTLET

                                                	 TEST C OUTLET
                                               CASCADE IMPACTOR TESTING
                            FIGURE 4
                               1-14

-------
              1990 UPDATE,  OPERATING HISTORY,  AND CURRENT STATUS OF
                     FABRIC FILTERS IN THE UTILITY INDUSTRY
                               Kenneth M.  Cushing
                                Randy L.  Merritt
                           Southern Research Institute
                                 P.O. Box 55305
                            Birmingham,  Alabama 35255
                                 Ramsay L. Chang
                        Electric Power Research Institute
                                 P.O. Box 10412
                           Palo Alto, California 94303
ABSTRACT
In order to keep its member utilities apprised of the status of fabric filtration
applied to utility coal-fired boilers, the Electric Power Research Institute,
since 1978, has been conducting surveys of the operating and maintenance experi-
ence at utility baghouses.  This paper presents results from the latest baghouse
survey of Canadian and U.S. utilities conducted during 1989.  The previous survey
was conducted in 1985.  Since that time baghouses have been placed in service on
approximately 5,250 MW of new utility generating capacity (out of a total  of
21,047 MW generating capacity with fabric filtration).  Fabric filtration is in
use on 4,902 MW of generating capacity where there is some type of flue gas
desulfurization system (spray drying, dry injection) upstream of the baghouse.
Also, 270 MW of generating capacity have recently come on line where baghouses  are
used downstream of fluidized bed combustion boilers.

While this recent survey updated previously acquired information on the design  of
these fabric filter units, the main emphasis was on the operating and maintenance
experiences at each of the 101 operating baghouses.  These data were acquired
through the use of questionnaires mailed to the utilities, telephone inquiries,
and plant visits.  Data were collected in the following areas:  baghouse startup
procedures, baghouse shutdown procedures, baghouse operating data (air-to-cloth
ratio, pressure drop,  temperature, opacity, inlet and outlet dust loading, bag
cleaning procedures),  bag service life and bag failure data, experiences with
sonic horns, and baghouse and ash removal system problems.  Tables and graphs
summarize and compare the operating and maintenance data collected from this
survey.

                                        2-1

-------
              1990 UPDATE,  OPERATING HISTORY, AND CURRENT STATUS OF
                     FABRIC FILTERS IN THE UTILITY INDUSTRY
INTRODUCTION
The electric utility industry continues to be faced with the requirement of highly
efficient particulate control devices downstream of their coal-fired boilers.
This requirement is a consequence of the United States EPA's new source particle
emission limit of 0.03 Ib/MBtu and the current attention placed on the control of
fine particles (PM10 regulations) and stack opacity.

The utility industry, through the Electric Power Research Institute, has responded
to these constraints by conducting a number of research programs to improve
existing technologies and develop cost-effective alternatives which will provide
highly efficient particulate control.  One of these programs, RP1401, "Reliability
Assessment of Particulate Control Systems," developed operating, maintenance, and
design databases for both fabric filters and electrostatic precipitators.  Summary
articles on the fabric filter databases have been reported previously (1,Z,3).
This paper summarizes the results of the most recent updating of this survey  (the
last survey update was in 1985).  Data have been collected by visiting selected
plants and from questionnaires sent to baghouse-operating utilities in the United
States and Canada.   Utility  fabric filters have been in operation for the past
seventeen years on  stoker-fired, cyclone-fired, and pulverized coal-fired boilers
of various designs.   In the  past five years the number of baghouses operating
downstream of pulverized-coal boilers followed by dry FGD systems has increased
dramatically.  There  are also several fluidized bed combustion boilers using
fabric filtration  for final  flue gas cleanup.

Several  trends  are  appearing in  the evolution of the fabric  filter  industry.
First, a number of  utilities have selected to retrofit sonic horns  to assist
reverse-gas cleaning  at their baghouses.  These retrofits have met  with  success,
and  it appears  that  reverse-gas  cleaning with sonic assistance  is the cleaning
method of choice for full-scale,  low-ratio utility  baghouses on pulverized  coal-
fired boilers.  Second, for  the  past  six years the  bag failure  rate has  stabilized
                                         2-2

-------
at about 1% per year of the installed bag capacity.  Third, there is little, if
any, reduction in boiler availability as a direct result of baghouse malfunction.
Fourth, baghouses are being successfully employed downstream of dry flue gas
desulfurization systems and fluidized bed combustion boilers.

As of December 1989, there were 101 baghouses in operation on utility boilers,
representing 21,047 MU of generating capacity (only 2 in Canada, both pulse-jet
units).  Another four units with a combined generating capacity of 1,150 MW are on
order or under construction.  Eight units, Bullock 1 and 2 (Colorado Ute Electric
Association), Kramer 1, 2, 3, and 4 (Nebraska Public Power District), and North
Broadway 2 and 3 (Rochester Public Utility Department),  totaling 202 MW, have been
permanently retired from service.  Several units indicated as planned in 1985 are
still in the planning stage or have been canceled.  These include Irvington 1,  2,
and 3 (Tucson Electric Company), and Coronado 3 (Salt River Project).  Survey data
were previously reported for two EPRI fabric filter test facilities.  The one at
Public Service Company of Colorado's Arapahoe plant was  decommissioned in 1988.
The pilot fabric filter at Gulf Power Company's Scholz plant was reconfigured from
reverse-gas cleaning to pulse-jet cleaning in 1989.  EPRI also has a small
reverse-gas cleaning fabric filter pilot plant at its High Sulfur Test Center
located at New York State Electric and Gas Company's Somerset Station.

Utility plants using fabric filters range in size from 6 MW at Marshall  Unit 4  to
860 MW at Sherburne Unit 3.  The largest operating baghouses (19,872 bags per
unit) are the two 739 MW units at Four Corners.  The oldest operating installa-
tion, Sunbury, was commissioned in 1973.

Figures 1 and 2 present, respectively, new utility fabric filter installations  by
year and the cumulative megawatt capacity of operating units for 1973 through 1989
with the projection of new capacity through 1992 based on units in design or under
construction.  Data in Figure 2 are shown separately for low-ratio baghouses
downstream from pulverized-coal boilers and for low-ratio baghouses downstream
from pulverized-coal boilers utilizing dry FGD processes because of their large
percentage (23%) of total megawatt capacity (4,902 MW out of 21,047 MW).  The
sharp rise in baghouse utilization beginning in 1978 reflects the impact of NSPS
legislation in 1979.

Coals burned in boilers equipped with baghouses are:  anthracite, eastern and
western bituminous, western subbituminous, and Texas and North Dakota lignites.
Sulfur contents for these coals average 1% and range between 0.3% and 3.5%.
                                         2-3

-------
Figure 3 shows the megawatt capacity of fabric filter installations in each state
of the continental United States.   The greatest concentration of fabric filters is
clearly in the southwest.  Although Colorado has the greatest number of installa-
tions, 16, Texas has the most installed capacity, 5341 MW.

FABRIC FILTER DESIGN DATA
Design data for utility fabric filters in operation, under construction, or in the
engineering phase have been updated.  Of the 101 baghouses now operating, 59 are
reverse gas, 30 are reverse gas with sonic assistance, 9 are shake/deflate, and 3
are pulse jet cleaning designs.

The largest baghouses (Four Corners 4 and 5, Intermountain 1 and 2, and Sherburne
3) have 48 compartments.  The largest number of bags per compartment is 648 at
North Valmy 1.  Bag sizes are commonly 8" diameter by 22' to 24' in length or 12"
diameter  by 30' to 35' in length.   Figure 4 shows the distribution of fabric
filters by baghouse manufacturer based on installed MW capacity.

Bag material  in the installations surveyed is mostly woven glass fiber.  Fabric
coatings  include  Teflon  B, silicon graphite, and proprietary acid-resistant
materials.  Bags  are  attached to the tubesheet by means of thimbles, typically of
a  length  identical to the bag diameter.  The thimbles protect the bottom of the
bags  from erosion by  fly  ash where  its flow is most turbulent.  Bags are attached
to the  thimbles by means  of  slip rings sewn in the bottom cuff or by screw-
tightened clamps.

Most  units  are  designed  for  easy access to allow for  inspection and bag replace-
ment.   For  most installations, the  bag replacement time is 15 to 30 minutes per
bag  for two men.  Most  units have  insulation between  compartments, as well as
ventilation  systems to  cool  compartments quickly and  permit personnel to work
comfortably  and safely  while the rest  of the baghouse remains in service.

FABRIC  FILTER OPERATION  AND  MAINTENANCE DATA
During  1989  questionnaires were sent  to all domestic  and  Canadian  utilities with
operating baghouses to  obtain  performance,  operation, and maintenance data.  To
support this  effort,  special  site  visits were made  to fifteen of the  larger
low-ratio baghouses  (all  in  the U.S.), representing  10,627 MW of generating
capacity.  There  were special  areas  of interest  including bag service  life,  bag
                                         2-4

-------
failure rates, coal and ash data, pressure drop histories, emissions data,
maintenance problems,  and major component failures.  Some utilities have had
several years of operating experience while others have had only a few months,
thus the comments and problems listed in the completed questionnaires were
indicative of both short- and long-term operation and maintenance concerns.

Efficiency, Emissions, and Opacity
Design efficiencies for baghouses range from 98.00% to 99.91%.  Most units meet or
exceed their design efficiencies.  A number of plants are reporting outlet mass
emission rates significantly less than the NSPS of 0.03 Ib/MBtu.  These data are
shown in Figure 5.  None of the plants that have retrofitted sonic horns to assist
reverse gas cleaning report any deterioration in baghouse efficiency, when the
sonic horns are used properly.  All plants visited reported stack opacities of 1
to 5%, well within the legal limits.

Pressure Drop
Pressure drop continues to be an area of major interest to utilities since fan
power is costly.  High pressure drop can also contribute to shorter bag life due
to the additional cleaning cycles that the fabric must endure.  The highest
reported flange-to-flange pressure drops for 1989 were 11 to 12 inches H20 (Texas
lignite coal, shake/deflate cleaning, 1.7 acfm/ft2 air-to-cloth ratio) and 6.0 to
8.0 inches H20 (western subbituminous coal, shake/deflate cleaning, 3.2 acfm/ft2
air-to-cloth ratio).  The lowest value was 3.5 inches H20 (anthracite coal and
petroleum coke, reverse gas cleaning with sonic assistance, 1.9 acfm/ft2 air-to-
cloth ratio).  The flange-to-flange pressure drop generally falls between 4 and
8 inches H20.  At most plants the difference between flange-to-flange pressure
drop and tubesheet pressure drop was 1 to 2 inches H20.  At one plant a difference
of 3 inches H20 or greater was observed.

According to the latest survey results, there is an even split between installa-
tions using intermittent cleaning and continuous cleaning.  Intermittent cleaning
is defined as time-dependent cleaning where the compartments are cleaned sequen-
tially with a short dwell period between cleaning one compartment and the next.  A
longer dwell period between cleaning the last compartment and the start of a new
cycle can be overridden when baghouse differential pressure becomes greater than
some limit.  Continuous cleaning can be initiated on either time or pressure drop
set point; it cleans one compartment after the other sequentially, without dwell.
                                         2-5

-------
In the report on the previous survey data (3), weighted average pressure drop*
data were reported for the years 1980 through 1984.  This same analysis has been
performed for pressure drop data reported in 1989.  Data from 33 plants
representing 9,833 MW of generating capacity were used.  The weighted average was
6.3 inches H20.   These data are combined with the previous data in Figure 6.

Sonic Horns
Seventeen baghouses since the last survey in 1984 have had sonic horns installed
to help maintain a lower baghouse pressure drop.  Reported pressure drop reduc-
tions are from 1 to 3 inches H20.  The data indicate that several plants use the
horns only to recover from high pressure drop excursions; one plant even uses one
as a mobile device to be carried from compartment to compartment.  One plant uses
sonic assist only intermittently, once or twice a day, and not during every
cleaning cycle.   Based on the responses from the plants, it appears that each
plant develops it own criteria for frequency of horn use.  One plant may require
assistance with each cleaning cycle, while another may need sonic assist only in
times of high pressure drop excursions.

Bag  Failures
Approximately 70% of the reporting units had fewer than 50 bag failures in  1989.
Reported causes of bag failures are abrasion, acid attack, poor manufacture, and
improper tensioning.  Most installations report bag failure at random locations
within each baghouse compartment and within the baghouse as a whole.  However, bag
failures are largely limited to the lower half of the bag, often near the thimble.

Bag  failure rates have been fairly steady at about 1% of installed bags per year
for  the  last six years, as shown in Figure 7.  In 1989, 1528 failures were
reported out of a total of 132,772 bags, for a failure rate of 1.2%.
 ^Weighted  average pressure drop is defined as:
       N
       z   (Number of bags),- x  (pressure drop),-
               2   (Number of bags);
               i = l
where  i runs over  each unit reporting pressure drop and N is the total  number of
baghouses reporting pressure drop  in each year.
                                         2-6

-------
Bag Service Life
Data obtained from the survey questionnaires and the site visits to selected
plants have allowed the preparation of graphs showing bag service life history
(years in service before replacement) for low-ratio baghouses filtering flue gas
from utility pulverized-coal-fired boilers.  Figure 8 shows the service life for
bags downstream from low-sulfur and high-sulfur coal boilers.  The data represent
information for 95.0% of the low-ratio baghouses installed on utility pulverized-
coal-fired boilers (based on megawatt generating capacity).  The mean service life
for bags on low-sulfur coal boilers is 5.2 years.  The mean service life for bags
on high-sulfur coal boilers is 3.3 years.  These data represent bag service life;
since many utilities rebag when cumulative failures reach 10%, it is likely that
many bags are removed from service prior to actual  failure and could have remained
in service for an unknown number of additional years.

Coal Characteristics
Figure 9 presents the breakdown of coal types burned at plants utilizing fabric
filters.  As can be seen, the most common coal type by far is western subbitumi-
nous, used in 62.3% (13,112 MW) of the total generating capacity of utility
boilers using baghouses.  This is a natural result of the large fraction of
baghouses located in the western United States.

Baqhouse Maintenance Experience
The types of problems reported in 1989 are similar to those obtained from earlier
surveys.  Some baghouses continue to experience stratification of dust loading
from the front to the back of the baghouse, as well as within specific compart-
ments.  Generally the heavier loading occurs in the compartments near the back of
the baghouse.  One plant noted severe stratification of dust loading within
compartments.

Many of the plants continue to have problems with poppets, valves, and their
operators.  Typical responses include failed limit switches, seal leakage, spider
guide breakage, shear pin failure, improper seating of poppets, corrosion, ash
buildup on poppets, valve stem guide wear, sticky or sluggish operation in cold
weather.

There were fewer reported problems with reverse gas fans in 1989.  Two plants
reported that their installed fans provided marginal capacity, while another plant
reported severe corrosion with their ductwork and fans.
                                         2-7

-------
Baghouses designed in 1979 and later are generally equipped with microprocessor
controls for baghouse operation.  These controls perform very reliably.  Most of
the questionnaires received in 1989 reported few problems with their systems.
Some plants reported problems due to condensation in control air lines.

A number of plants reported problems with expansion joints, the highest incidence
occurring with the reverse gas system.  This problem is probably due to cooler
temperatures leading to corrosion and the increased flexing that these systems
must endure.

Corrosion continues to be a problem at several plants, especially in the reverse
gas system.  Another typical site for corrosion is compartment doors where
inleakage occurs as a result of improper or worn gasket material.

Several  plants reported problems with their bag tensioning mechanisms.  The  most
common  problem was with improperly sized springs that collapse under moderate bag
weight  and  allow the bags to cuff at the thimble.  Most of these plants have
replaced their springs.

The variety of maintenance  problems points out one of the strong impressions from
 the plant visits  conducted  during this survey.  Baghouses that were performing
well  were usually  located at plants where considerable effort was taken in
 monitoring  baghouse  operation.   If problems arose, fixes were quickly  implemented.
 In contrast,  problem  baghouses  appeared to have little attention paid  to them.
 Corrections of  problems occurred only during  normal outage  periods  or  if plant
 availability was  threatened.

 Baghouse Availability
 Most  installations report little,  if  any, reduction in boiler availability  as a
 direct result of baghouse malfunction,  indicating  that problems  are such that they
 can be corrected in service or  during outages for  other  plant equipment.

 Operation and Maintenance Costs
 Few of the questionnaires returned  in 1989  contain  complete  operation  and  mainte-
 nance cost data.   Respondents generally  do  not isolate fabric filter system costs
 from those of related systems.
                                          2-8

-------
ASH HANDLING SYSTEMS
Most baghouses use a vacuum system combined with dry storage.  The remainder are
either combination vacuum/pressure systems or are unidentified.  Most units have
hopper level detectors of either nuclear or capacitance probe design.  Of the 82
units for which data are available, 52 (63%) have hopper heaters.  Of the 73 units
for which data are available, 21 (29%) have hopper vibrators and warm air fluid-
izers.

In 1989 few problems were reported with the vacuum producer or blower and the
level detector and hopper fluidizer-  One plant reported that their entire ash
system is at the end of its useful life and will have to be replaced.  Minor
problems were reported with line and hopper plugging.  In general, typical
problems with ash removal systems were due to normal wear and usage.

SUMMARY AND CONCLUSIONS
Data have been collected on 113 fabric filter installations.  These are distrib-
uted as follows:
     --97 in operation
     -- 4 on auxiliary status
     -- 8 retired
     -- 4 in design or construction

Of the 97 fabric filters in operation, 63 units are low-ratio units downstream
from pulverized coal-fired boilers using reverse-gas cleaning or reverse-gas
cleaning with sonic assistance, 5 are low-ratio baghouses downstream from
pulverized-coal fired boilers using shake/deflate cleaning, 2 are pulse-jet
cleaned baghouses downstream from pulverized-coal fired boilers  (Canadian only),
15 are low-ratio baghouses downstream from dry FGD systems on pulverized-coal
fired boilers, 2 are low-ratio baghouses downstream from fluidized bed combustion
boilers, 9 are low-ratio baghouses downstream from stoker-fired  boilers,  and 1  is
a pulse-jet cleaned baghouse downstream from a stoker-fired boiler.

There has been a significant increase in the number of installations retrofitted
with or designed with sonic assistance for reverse-gas cleaning.  Thirty  plants
report sonic horns installed in their baghouses.  Reported  improvement in pressure
drop from the use of sonic assistance ranges from 1 to 3 inches  of water.
                                         2-9

-------
Most installations report very high collection efficiencies for their baghouses,
generally greater than 99%.  Of the reporting plants, 85% had mass emission rates
of 0.03 Ib/MBtu or less.  At the plants visited during this survey operating stack
opacities were in the range of 1 to 5%, significantly better than legal require-
ments.

Flange-to-flange pressure drop ranged from a high of 11 to 12 inches of water to a
low of 3.5 inches of water   Typically the range was from 4 to 8  inches of water
for the plants reporting data.  The weighted average pressure drop for 1989 was
6.3 inches of water, similar in magnitude to those values previously reported for
the years 1980 through  1983.

Approximately 70% of the reporting units had less than 50 bag failures in 1989.
The average bag  failure rate has remained steady at about 1% of installed bag
capacity for the last six years.  Typical causes for bag failure  continue to
include abrasion, improper tensioning, poor manufacture, and acid attack.

Bag service life is generally impressive considering the concern  about bag life
when  fabric filters were first being  installed on utility boilers.  The mean
service  life of  bags installed in low-ratio baghouses on low-sulfur pulverized
coal  boilers has risen  to  5.2 years.  The mean service life of bags in low-ratio
baghouses on high-sulfur pulverized-coal boilers is 3.3 years.

Operating and maintenance  problems are not confined to any specific area.  The
most  common problem areas  in 1989 were valves and valve operators, bag tensioning
mechanisms, expansion joints, and reverse gas fan systems.  Problems associated
with  ash handling systems  appeared to be associated with normal wear on the compo-
nents .

Baghouses have continued to maintain  high availability factors and were seldom,  if
ever,  a  cause  for boiler down time.   Baghouses continue to demonstrate that they
are efficient, reliable, and cost effective particulate control devices for the
uti1ity  industry
                                         2-10

-------
ACKNOWLEDGMENTS

The authors wish to thank the many utilities that responded to the 1989 survey

questionnaire, providing accurate and up-to-date information on baghouse operation

and maintenance.  The authors are especially grateful to the following plants that

allowed on-site visits to obtain survey data:  Bonanza, Brunner Island, Cherokee,

Four Corners, Harrington, Hunter, Intermountain, Monticello, North Valmy, Nucla,

Parish, Reid Gardner, Sammis, and Tolk.  This work was performed under EPRI

Contract RP1129-8.


REFERENCES

1.   W. Piulle, R. Carr, and P. Goldbrunner.  "Operating History and Current
     Status of Fabric Filters in the Utility Industry."  In Proceedings:  First
     Conference on Fabric Filter Technology for Coal-Fired Power Plants.  CS-2238.
     Palo Alto, California:  Electric Power Research Institute, February 1982,
     p. 1-1.

2.   W. Piulle, R. Carr, and P. Goldbrunner.  "1983 Update, Operating History and
     Current Status of Fabric Filters in the Utility Industry."  Proceedings:
     Second Conference on Fabric Filter Technology for Coal-Fired Power Plants.
     CS-3257.  Palo Alto, California:  Electric Power Research Institute,
     November 1983, p. 1-1.

3.   W. Piulle.  "1985 Update, Operating History and Current Status of Fabric
     Filters in the Utility  Industry."  Proceedings:  Third Conference on Fabric
     Filter Technology for Coal-Fired Power Plants.  Scottsdale, Arizona:
     Electric Power Research Institute, November 19-21, 1985, p. 1-1.
                                         2-11

-------
       26
       24  -


       22


       20  -


       18  -


       16  -


       14  -


       12  -


       10  -


         8  -


         6  -


         4  -


         2
                                 Utility  Fabric  Filter  Installations

                                             All Active Saghouses
                                       i'  ' '  f  ' '  i'  ' '  i     r     i     i     r
              73   74  75   76   77   78   79   80   81   82   83   84   85   86   87   88   89
                       Figure 1.  The number of new utility baghouses placed in service
                       during each year beginning in 1973.
       22  -
       16  -
oo    14
li    12
                                 Cumulative  Megawatt  Capacity

                                       Current and Planned Baghouses
                                           ALL BAGHOUSES
                                                    PC w/FGD, LOW RATIO BAGHOUSES
                        —I—

                         76
                                                                  88
                                                                                92
                       e 2.  Chronological  history of cumulative megawatt capacity of utility
                       iuses.  Data  are presented for all baghouses  and separately for low-ratio
                  uaynuuses downstream from pulverized-coal boilers and low-ratio baghouses
                  downstream of dry FGD systems on pulverized-coal  boilers.   Data for baghouses
                  in design or construction are shown beyond 1989.
Figure 2.
bagho
bagho
                                               2-12

-------
                                  Total  Megawatt  Rating of Stations  with
                                    Fabric Filters  in the United  States
       j 1 - 500

11| || || || || | 501 - 1000


i:::i:i&;:i&l 1001  " 200°

|%///d 2000 +
                    Figure  3.   Distribution of utility  baghouses currently in service  by state
                    (based  on  total megawatt generating capacity).
                                                                                                        'MARYLAND
                                                                                                          390
                                                                                                             MASSACHUSETTS
                                                                                                                 0
                                                                                                             HHODE ISLAND
                                                                                                                0
                                                                                                           'CONNECTICUT
                                                                                                               o

-------
                            Fabric  Filter  Manufacturers
                                 Baghouses Currently In  Service
6  -
5  -
2  -
                                                      Pulverized  Coal and Stoker  Boilers
                                                      Pulverized Coal Boilers with FGD
                                                      and Fluidized Bed Boilers
80
70  -
40  -
30  -
                                         Manufacturer

           Figure 4.  Distribution of currently installed utility fabric filters in the
           United States and  Canada by manufacturer (based on total megawatt  generating
           capacity).
                            Utility  Fabric  Filter Emissions

                              Reporting  Basis:  6669 MW (19 plants)
      0.0-0.01    0.01-0.02   0.02-0.03   0.03-0.04    0.04-0.05   0.05-0.06      >0.1

                                     Emission Rate, Ib/MBtu

               Figure  5.  Percentage  of plants (based on megawatt generating capacity)
               reporting mass emission rates in the  indicated ranges.
                                       2-14

-------
10



 9
                                 Pressure Drop  History

                                  Weighted by Bags in Service
 2  -
         1980
                       1981
                                      1982
                                                    1983
                                                                   1984
                                                                                 1989
                                             Year
10



 9



 8



 7



 6



 5  -.,



 4



 3



 2



 1



 0
              Figure 6.  Weighted average flange-to-flange pressure drop for  utility
              baghouses for six years during the 1980's.  Weighted average is based
              on the number of bags In each baghouse having indicated pressure drop.
              No data are available for 1985 through 1988.
                               Yearly Bag  Failure Rate

                                EPRI Utility Baghouse 0&.M Survey
\
      78
              79
                     80
                            81
                                   82
                                          83
                                                 84
                                                        85
                                                               86
                                                                      87
                                                                              1—

                                                                              88
                                                                                     89
                                             Year
                Figure 7.  Yearly bag failure rate based on  data developed for the
                EPRI Utility Baghouse OiM Survey  (low-ratio  baghouses only).
                                      2-15

-------
   26


   24


   22


   20


   18


   16


   14


   12


   10


    8


    6


    4


    2


    0
                                 History of  Used  Bags

                                95.0% of PC Low-Ratio Boghouses
               Low-Sulfur Coal


               High-Sulfur Coal
\
         0-1
                       2-3
                               3-4
                                             5-6
                                                    6-7
                                                            7-8
                                                                   3-9    9-10   10-11
                               Years in Service Before Replacement
                Figure 8.  History  of used bags  on low-ratio baghouses downstream from
                utility pulverized-coal  boilers  in the United States.  Data are shown
                individually for low- and high-sulfur coal boiler installations.
                             Types  of  Coal  Used
                           Percentage Based  on MW Rating
                                                  N DAK LIG (8.0%)


                                                             TEX LIG (5.6%)


                                                                 ANTHRA (1.9%)
                                                                       E BIT (16.7%)
W SUBBIT (62.3%)
                                                                  W BIT (5.5%)
      Figure 9.  Types of coal used in utility boilers using fabric filters
      Percentages based on megawatt generating capacity.
                                        2-16

-------
               AUSTRALIAN EXPERIENCE WITH HIGH RATIO FABRIC FILTERS
                                ON UTILITY BOILERS
                                  Peter R. Heeley
                            Engineer/Flue Gas Cleaning
                   The Electricity Commission of New South Wales
                                   GPO Box 5257
                           SYDNEY  NSW  2001  AUSTRALIA
                                  Colin Robertson
                              Manager/Power Projects
                                Flakt Australia Ltd
                                     PO Box 42
                         ST LEONARDS  NSW  2065  AUSTRALIA
ABSTRACT
The  Electricity  Commission  of  New  South  Wales first began development work on
fabric  filters  on  utility  boilers in 1966,  with pilot plant trials of low ratio
shaker  style  plants  at  Tallawarra Power Station.   As a result of the success of
these trials, shaker filters were installed on all of the boilers at Tallawarra and
on the older half of Wangi Power Station between 1972 and 1976.

High  ratio  fabric  filters  were installed on the newer half of Wangi in 1975/76.
Due  to  the relatively short bag life and high emissions experienced on Wangi 'B',
the  Electricity Commission of New South Wales elected to install low ratio filters
on  the  ten  new  660  MW units to be installed at Eraring,  Bayswater and Mt Piper
Power Stations.

Since  that time, high ratio filters have been retrofitted to boilers at White Bay,
Pyrmont,   Tallawarra, and most recently Munmorah Power Stations in New South Wales,
and  at  Bulimba, Tennyson and Callide Power Stations in Queensland.  The operation
of  these units is discussed, as are the reasons for calling tenders for high ratio
filters  for  four  500  MW  boilers at Liddell Power Station.  The extensive pilot
plant  investigations  carried out into the technology are reviewed, and pricing of
high  ratio filters received as an alternative to precipitators for four new 350 MW
units  at  Stanwell  Power Station in Queensland are compared with low ratio filter
prices.
                                        3-1

-------
               AUSTRALIAN  EXPERIENCE  WITH  HIGH RATIO FABRIC FILTERS
                                ON  UTILITY BOILERS
INTRODUCTION
The  needs of the Electricity Commission of New South Wales (the Commission)  in  the
application  of  fabric  filter  technology to utility boilers were born out  of  the
particular characteristics of New South Wales coals,  namely low sulphur content  and
high  ash.    These  characteristics  meant  that  the  conventional  electrostatic
precipitator  (cold  side)  technology which was generally acceptable in the  United
States  and  Europe in the 1960's and early 1970's,  did not apply well to New South
Wales  power  stations.   Even with the relatively high allowable emissions at that
time  (0.4  g/Nm ,  equivalent to 0.32 Ib/MBtu for a typical New South Wales coal or
0.17  gr/scf),  conventional  gas  cleaning  technologies were unable to perform to
expectations  when new and their performance generally deteriorated with age.  This
was accelerated by the need for frequent water washing.

The  extensive  experience obtained by the Commission in electrostatic precipitator
design  and  operation gained both through its pilot  plant programmes and full size
operating  plant  gave  the  organisation  the  confidence required to undertake an
initial   fabric filter pilot testing programme at Tallawarra Power Station in 1966.
By  1972,  the  performance  of  the  existing  Centicell  mechanical collectors at
Tallawarra  had  deteriorated  to  such  a degree that a full scale trial plant  was
installed  on  one half of a pulverised coal fired 30 MW boiler.  The fabric  filter
was  a  mechanical  shaker  design  using  bags  made  of a woven synthetic fabric.
Because   of  the  low  back  end  temperature  a  homopolymer  acrylic  (Dralon  T
manufactured by Bayer) became the standard after a short trial period.

With  the  success  of  the  trials,  a  range of progressively larger boilers were
subsequently  retrofitted  with both mechanical shaker cleaned (low ratio) plant as
well  as  reverse pulse air (high ratio) plant.  Installations were, where possible,
carried   out  in  the  existing  dust collection plant casings with the result that
filtration  velocity  was  often  significantly  higher than the perceived optimum.
This  was  particularly  significant until the development of bags of six metres or
more in length allowed the effective reuse of even undersized precipitator casings.

Electrostatic  precipitators  were  however  still  considered  to have an economic
advantage  where  new  plant was being constructed.  It was not until 1978 when  the
Commission  made  the  commitment  to  construct  a new 4 x 660 MW power station at
Eraring,  with  a  requirement  for  clear  chimney  stacks, that a major base load
station   was  planned  to have fabric filtration from  the outset.  The four Eraring
units,  four  more  at  Bayswater  and  two  at Mt Piper (due to be commissioned in
1992/93)  result  in a total of 6,600 MW of new base load mechanical shaker cleaned
units constructed since that time.
                                        3-2

-------
OPERATING EXPERIENCE AT RETROFITTED STATIONS

Retrofits  of  fabric  filter plant have been necessary on many of the Commission's
older  power stations due to the inadequate performance of mechanical collectors  or
undersized  electrostatic  precipitators.    Most  of these boiler units were small
(less  than  100  MW)  and in two cases totalling eight boilers were located in  the
middle  of  the  Sydney metropolitan area (Pyrmont and White Bay).  In these cases,
the  high  level of emissions produced by the stations was particularly undesirable
and causing significant public concern.

With  the  large  planned  expansion  in  generating  capacity  expected   to become
available  during  the  1980's  these  small  units  were  due  to be progressively
decommissioned.    A  higher  than  expected  rate of load growth at the end of  the
1970's  combined  with  the  electrical  failure of three out of  four of the 500  MW
alternators at Liddell meant that all of these small units were required to operate
at high loads for a significantly longer time than had been planned.  This extended
usage required a low cost rapidly fitted solution to the emission problems of these
units.    All  of  these early retrofit units have now been decommissioned.  Recent
work  has been undertaken at Munmorah (4 x 350 MW) to retrofit pulse jet filters  on
Units  3  and  4 where the electrostatic precipitators were performing very poorly.
Munmorah's  retrofit  has  a planned lifetime of 15 years during which time it will
operate  as  an  intermediate  load  station.  Table 1 indicates  the details of  the
filters installed and currently on order.

Major  problems  occurred  initially  with the mechanical shaker  filters at Eraring
with  both high pressure drops and short bag life.  These problems were overcome  by
changes  in  the  bag configuration from tab-top to cap-top and developments in  the
filter bag material.  While further improvements in the filter material continue  to
be  developed, pressure drops well below guarantee and three year plus bag life  are
already being achieved (see references 1, 2 and 3).

The  Commission's  first  application  of  pulse  jet  filters  was at Wangi during
1975/76.    In this case the existing concrete precipitator casings associated with
three  60  MW  pulverised  coal  fired  units  were  far  too  small  to   allow  the
installation of shaker filters and the decision was taken to use  pulse jets.

The extra cleaning energy associated with pulse cleaning meant that the necessarily
high  filtration velocity could be used.  A Ducon-Micropul plant  was installed with
a  filtration  velocity  of 0.0344 m/s (6.7 ft/min) using 4,032 bags on each of  the
three  boilers.   The limitations of the technology at that time  with regard to  bag
length  and  the  small  size  of the precipitators, meant that this extremely high
filtration velocity was unavoidable.  The acrylic Dralon T bags were 3.05  m  (10  ft)
long, 116 mm (4-1/2 in) diameter and were constructed from a needle felt.

The  plant used the standard Ducon-Micropul high pulse cleaning pressure of  700  kPa
(100  psi).   This plant suffered from a high DP and excessive emissions due to  the
subsequent  high  pulse  frequency  and  leakage of the joint between the  cell  tube
plate  and the concrete casing.  Bag life varied from 9,000 to 15,000 service hours
with  bags  suffering mechanical failure in the top 600 to 800 mm.  Dust puffs were
visible  from  the  stack  during  pulsing  and  the high differential pressure  was
sufficient to cause load limitation.

Following  on from the Wangi experience it was necessary to improve the performance
of  the  four  50  MW  units at Pyrmont and the four 250 MW units at White Bay.   In
1982,  Ducon-Micropul  completed  the  installation  in  each  of the four units at
                                        3-3

-------
Pyrmont,   of  3,432   bags   of   needle  felted  fibreglass.  Fibreglass was  required
because  of  the  high back end temperature at Pyrmont (200 C) and in  this  case  the
bags were 3.76 m (12'4")  long  and 116 mm (4-1/2") in diameter.

The  larger  precipitator   casings  meant  that a much more conservative  filtration
velocity  was  possible  (0.02 m/s or 4 ft/min).   The installation at  White Bay  was
similar  to  that at Pyrmont except for the fractionally higher filtration  velocity
(0.023  m/s  or  4.6 ft/min),  and the provision of new casings.  The performance of
units  at  both stations  was excellent with clear stacks at all times  and emissions
of  0.004 to 0.02 g/m  (0.002  to 0.01 gr/scf) at  full load.  Unfortunately  reliable
bag life data was not able to  be recorded as the  units were decommissioned  within a
year  of  completing  the  installation.  Pulsing frequency however, was  increasing
steadily  at White Bay suggesting possible problems if operation had continued much
longer.


RETROFITS AT MUNMORAH

As  a result of high maintenance requirements and  particularly poor performance from
the  350  MW  Units 3 and 4 precipitators at Munmorah Power Station, a decision  was
made   to  replace  the plant.   These units were originally commissioned  in  1969  and
1970.    A  review  of  the  available  options to provide satisfactory  performance
indicated   that  the  most cost effective solution was to install fabric  filters of
the  pulse  jet  cleaning  style  inside  the  existing  precipitator  casings (see
Reference 4).

Due  to the  more  intense cleaning system allowing the use of felted filter media  and
the  ability   to  clean  on  line,  pulse  jet  filters  are  able   to  operate   at
approximately  double  the  filtering  velocity  of shaker or reverse  flow  filters.
This   coupled  with   the  use  of bags of six metres or more in length meant that a
pulse  jet   filter  could  comfortably  be  installed ~in  the casing,,of  even these
significantly  undersized precipitators (SCA of 77 m /m /sec or 390 ft  /kcfm).

A  specification  was issued for Munmorah in 1985 for fabric filters of  the pulse  jet
type   with   a  maximum  filtration  velocity  of 0.02 m/s (4 ft/min) at  the maximum
expected   boiler  gas  flow  of 480 m /s at 140°C.  New I.D. fans and  fan inlet  and
outlet silencers were also included in the extent of supply.

Following  an  examination of the tenders and a review of the similar  plants  supplied
around   the  world, it was decided to split the contract and purchase  a  Flakt plant
for Unit 4  and a Howden plant  for Unit 3.

The  two   plants  vary  significantly  in  a  number of key areas, with  Flakt using
"medium"   pressure  cleaning air at 280 kPa, with round acrylic felt  bags  7.2 m long
and  Howden  using  "low" pressure cleaning air at around 100 kPa, with oval acrylic
felt  bags  6 m long.

A   full  description  of the installation and operation of the Flakt plant  is being
presented  in  another  paper  today but  it is briefly described here for  completeness.


UNIT  4 - MUNMORAH - FLAKT

Unit   4  was   built by Flakt and commissioned in 1988, and to date has accrued  some
8,500  operating  hours,  with  very  low  emissions  (0.01  g/mj or 0.005  gr/scf),
acceptable  pressure drop and no bag  failures.
                                         3-4

-------
The filter is split into eight compartments, each individually isolated  for  on-line
maintenance.    Each compartment contains 1,140 bags, 130 mm diameter and  7.2 m  long
(5"  diameter  23'7"  long).   The filter bags are fabricated from 500 g/m   acrylic
felt singed on both sides, supplied by E W Andrews and fabricated by Filtertex  Ltd,
in  Sydney.   Incoming gas is controlled to a maximum temperature of 125  C by means
of  attemperating  air admitted in the inlet ducts by eight individually  controlled
dampers.    The  layout  of  the  plant  is  shown  on Figure 1.  Gas entering  each
compartment  is  distributed horizontally across the upper half of the filter bags,
causing  the  flow  on the dirty side of the bags to be generally downwards  towards
the  hoppers.    This  arrangement  minimises  re-entrainment  of  dust  pulsed  from
adjacent   filter  bags.  A full walk in clean side plenum is provided to  facilitate
bag  inspection  and  changing,  with bag changing being carried out from the clean
side.    Two piece bag cages are used to minimise the height of the plenum,  with  no
tools or special skills being required to change filter bags.  The operation of the
plant  can  be viewed on-line through viewing windows, to assist in the  location  of
leaking filter bags.

Cleaning air is distributed from external manifolds and pulse valves to  the  bags  by
internal   pulse  tubes running across the tops of the bags in the clean  gas  plenum,
each valve and pulse tube servicing a row of 20 bags.

The plant  is designed for maximum emission of 0.05 g/Nm  and a bag pressure  drop  of
1.2  kPa   with  an overall plant pressure drop of 2.5 kPa at the maximum  design gas
flow.  A bag life guarantee of 27,000 hours has been given.


UNIT 3 - MUNMORAH - HOWDEN

The  Howden  plant  uses  filters of "Carter Day" design (Carter Day America is now
owned by Howden) and fans and silencers are of Howden supply.  The plant  is  divided
into  eight  compartments,  each compartment containing 1,608 singed acrylic felted
filter bags manufactured by Albany International at Gosford, New South Wales.  Each
bag  is  oval  in shape, with a circumference of 400 mm and a length of  6 m  (14.75"
circumference  and 19'8" long).  The layout of the plant is shown in Figure  2.  The
layout  differs  significantly  from  the  Flakt  plant  in that the plant has  been
arranged   with  four  front  and four rear compartments to suit the original hopper
layout.    This layout has enabled fuller utilisation of the original plant  ducting
and  casings.    A  duct  is  provided  through  the  hopper  of  each of the front
compartments to distribute gas to the rear compartments.

As  in  the Flakt plant, attemperating air is used to cool incoming flue gas.  Also
the  plant  is  provided with full walk in clean gas plenums for bag inspection and
changing.  Split cages are again used to minimise the plenum height.

The  Howden  cleaning system uses a rotating manifold to distribute cleaning air  to
the  filter  bags,  with  six  large  pulse  valves  and  manifolds  fitted  to  each
compartment,  each  servicing  268  bags.  The pulse valves and manifold drives are
located  on  the roof of the clean gas plenum to allow maintenance without  the  need
to isolate compartments.

The space  efficient oval bag arrangement enabled additional bags to be installed  at
minimal cost, providing a maximum filtering velocity of 0.017 m/s at  the design gas
flow.
                                        3-5

-------
Installation of the Unit  3 fabric filter is virtually complete with  the  Unit  due to
be returned to service in March,  1990.


RETROFITS IN QUEENSLAND

The  Queensland  Electricity  Commission carried out similar  retrofits  to New South
Wales  to  24 small spreader stoker boilers in the Brisbane suburbs  at  Tennyson and
Bulimba.     Four pulverised coal fired 30 MW boilers were also upgraded at Callide.
High back end temperatures resulted in the use of Teflon filter  bags which were 6 m
long  and  130 mm O.D.  Filtration velocities were initially  0.022 m/s  (4.3 ft/min)
with  all  passes  in service and 0.026 m/s (5.1 ft/min) with one gas pass isolated
for  maintenance.    The  design  pressure  drop  was 1.37 kPa (5.5"wg) in order to
utilise  the  existing  ID  fans  with  only  the  addition   of  new fan impellers.
Excessive  pressure  drops occurred shortly after commissioning.  The high pressure
drop caused an extremely short cleaning cycle and thus led to excessive emissions.

The  problems  at  Callide  required additional bag area in the  plant to reduce the
filtration  velocity  to  0.017  m/s  (3.35  ft/min).    The  plant   has  performed
satisfactorily since being extended.


LIDDELL  RETROFIT

The  Electricity Commission of New South Wales 4 x 500 MW base to intermediate load
power  station at  Liddell was commissioned between 1971 and 1974 and included large
electrostatic  precipitators  from  new.   Historically, because of  the higher than
design inlet  dust  burdens (typically 30 g/Nm  or 15 gr/scf compared  to  the 18 g/Nm
maximum  originally expected) and the high resistivity of the  fly ash, the plant has
always   operated   near the Clean Air Act emissions limit (0.4 g/Nm   or  0.2 gr/scf).
If   an   electrical   problem  occurred  within  any  zone,  emissions  could only be
controlled  by  reducing boiler load.

The  very high  sensitivity of fly ash resistivity to changes in gas  temperature made
special  demands   on  the  plant  designers.  Peak resistivity of  the order of 10
ohm-cm   occurred   at  135 C with rapid decreases occurring at both  lower and higher
 temperatures.    This  temperature would have been the normal back end temperature at
Liddell  for   a  mixture  of  the discharged gas from the primary and secondary air
heaters.

Gas   flow   emitting   from both primary and secondary air heaters was therefore kept
separate  and   fed   through,  different  primary  and secondary precipitator casings
 (SCA'si     Primary   101   m /m /sec  or  511  ft /kcfm;  Secondary   64  m /m /sec or
324  ft /kcfm).

Several  problems existed  with the plant including:

 (a)  Corrosion   of  secondary  casings  and  collecting electrodes caused by low gas
     temperature  (107  C or 225 F) and regular water washing.

 (b)  Poor rapping performance caused by sticky dust,  function  of  (a)  and also of  the
     ammonia   flue  gas conditioning (which was added to improve  performance).  This
     performance  improvement was  realised but leads to very sticky deposits.
                                         3-6

-------
(c)  High  inlet  dust  burdens  meant  there  was  no ability to maintain the plant
    without reducing boiler load or exceeding emissions licence levels.

(d)  Large   variations   in   precipitator   performance   with   changes  in  coal
    characteristics.

The  level  of  corrosion  was very high and generally whole gas paths of the plant
required replacement of collecting electrodes every seven years.

This  combined  with  a  long  term  need to operate the^boilers at reduced load to
maintain  emissions  at  the  licence  limit  (0.4  g/Nm   or  0.17  gr/scf) was art
unreasonably high cost.

Trials  had been carried out using pulsed energisation and ammonia conditioning was
operated  on  the plant from soon after initial commissioning.  Further development
work  with  these  technologies was not seen as providing sufficient improvement to
justify continuation with that course of action.

A  decision  was made in May, 1989, that it was economically and technically viable
and  socially desirable to replace the precipitators with pulse jet fabric filters,
fitted into the existing precipitator casings.

A  specification  was  prepared  and  tenders invited in July, 1989.  The plant was
specified  in a similar way to Munmorah, with a maximum filtration velocity of 0.02
m/s  (4  ft/min) at the maximum expected boiler gas flow of 848 m /s at 135 C.  New
ID fans and fan inlet silencers were also included in the extent of supply.

After  review  of  the  tenders, a contract was established with Flakt Australia in
December,  1989, to retrofit pulse jet filters to all four of the 500 MW boilers at
Liddell.

The  offer  accepted is based on the use of 8.0 m long, 130 mm diameter (26'3" long
and  5"  diameter) singed acrylic bags using the medium pressure cleaning regime as
at Munmorah.
DETAILS OF PROPOSED LIDDELL PLANT

The  first  unit is due to be converted at Liddell during late 1990 with subsequent
units at yearly intervals.

The  filter  utilises  the  existing casings, five for each boiler.  Each casing  is
able  to be independently isolated for on-line maintenance.  Figure 3 indicates the
general layout of the plant.  Gas is taken from the previously separate primary and
secondary  air  heaters  and  mixed  in  new mixing chambers.  Attemperating  air  is
admitted  to  the  inlet  ducts  by  ten  individually  controlled  dampers.    The
attemperating  air  enters just after the mixing chambers and controls the  incoming
mixed gas to a maximum temperature of 125 C.

Most  parts  of  the plant are identical to Munmorah Unit 4 with  the main exception
being that the former primary casings at Liddell are dimensionally smaller  than the
secondaries,  and  will  have  a  lesser  number  of  bags  (primaries  2,652 bags,
secondaries 3,120 bags).
                                       3-7

-------
The  worst maintenance situation for the plant is when one of the secondary  casings
is  off line for maintenance at which time the filtration velocity will  increase  to
0.0253  m/sec  or  5.06  ft/min.  This is not seen as presenting any  problem in  the
light of operating experience at Munmorah, and pilot plant tests at Liddell.
Because  of  the  high  inlet  dust  burden,  the  guaranteed bag  life  is  less
Munmorah,  at  17,520  hours whilst emission is guaranteed not  to  exceed 0.08 g/Nm
(0.04  gr/scf).    This  emission  level was a requirement of the  specification  and
represents  the  new licence level for the plant.  Actual anticipated emissions  are
much  less  than  this  at 0.015 g/Nm 70.007 gr/scf) and should ensure  a completely
clear stack.


CONTINUING RESEARCH AND DEVELOPMENT

The  use  of   fabric  filters  on utility boilers is relatively new, with  the  first
commercial  unit  going  into service around the world in the early  1970's.  At  the
time of  commissioning in 1982   1984 the Eraring fabric filter  installation  was  the
largest  in the world.  The pulse jet fabric filters installed  on  Munmorah are also
the  largest   of  this  type  in  the world, each being over double  the size of  the
previous   largest  plant  at  Milner  Power  Station  in  Alberta,  Canada.     The
installation   at   Liddell,   being  larger  again  than  Munmorah,  confirms   the
Commission's commitment to continue to develop this technology.

Because  of  the immaturity of these types of plants, the Commission has considered
it  appropriate  to carry out extensive research and development work in this  area.
The   Commission   had   been  involved  in  development  work  with  electrostatic
precipi tators  and  had  been involved in pilot plant investigations with  equipment
suppliers prior to the installation of precipi tators since the  early sixties.

Pilot  plant   work  by  the  Commission  to  date has led to major developments  and
improvements   in  the  shaker fabric filter plants, in particular  in the filter  bag
area.    The eight compartment plant at Eraring is continuing in operation  to screen
potential replacement filter bags for Eraring, Bayswater and Mount Piper and at  the
same   time  provide  an  invaluable development tool for filter bag  suppliers.   The
Commission  has a policy of only purchasing replacement filter  bags  which  have been
both  tested in this plant and also trialed in a single cell of  tha main plant.

In   the  area  of the newer pulse jet fabric filters, it is probable that  even more
stands   to  be gained from research and development work.  The Commission recently
participated in a joint research project with the Electric Power Research  Institute
of  the USA  (EPRI) into the applicability of different pulse jet technologies to  the
utility  industry.  This project involved funding James Howden  to  operate  two  pulse
jet  pilot  plants at Eraring.  At the same time the Commission has  participated in
pilot  plant trials with Flakt and Howden at Munmorah Power Station.

Three  pilot   plants were installed at Liddell by Ducon Micropul,  Flakt and  Howdens
in  order  to gain the required site specific experience prior to submitting  tenders
for  the Liddell  fabric   filter  project.   These pilot plants gained substantial
experience  for the companies, in the two years or so that they  were  installed  prior
to  tenders  being invited.

The  operation of   these   plants  has  proven  the viability of this  technology to
utility  applications  encompassing  a  broad  range  of boilers and burning a wide

-------
variety  of  coals.   It is expected that this and future work with such plants will
enable the use of smaller and cheaper pulse jet filters on utility applications.


PRICING FOR STANVELL

The  Queensland  Electricity  Commission, having also gained significant experience
with  fabric  filtration  considered its use when they recently invited tenders for
flue  gas  cleaning  plant  for four new 350 MW boilers at Stanwell.  Plant offered
included  either  electrostatic precipitators to achieve emissions of 0.23 g/Nm  or
fabric filters to achieve 0.05 g/m  (0.13 and 0.02 gr/scf respectively).

The  plant  selected  was  an  electrostatic precipitator with an equivalent 300 mm
(12")  Plate spacing SCA of 140 m im /sec (714 ft /kcfm) at a cost of $43M ($31/kW)
not including concrete or foundation works.

Actual  plate  spacing  being used is 400 mm (15.75").  By way of comparison, pulse
jet  fabric  filter  tenders  were received at $25/kW from two suppliers whilst the
cost  of  a  suitable  mechanical shaker style plant is estimated at $53/kW.  (Note
that  at  time  of  submission  of tenders in November, 1988, the exchange rate was
$1AUD:$0.88USD.)

Figure  4  indicates   the relative differences in life cycle costs on an Australian
dollar/installed  kW   basis  for  the  various  types of plant that would have been
suitable  for  this  application  (i.e. electrostatic precipitators - EP1, 2 and 3;
pulse  jet filters   FF1, 2, 3, 4 and 5; and low ratio shaker type fabric filters  -
LR1 and 2).  For the purpose of this calculation a 20 year lifetime was assumed.

Additional  operating  costs  are  obviously experienced throughout the life of the
plant  with  a fabric  filter.  However, the increased fan power consumption and bag
replacement  costs  are  offset  by the potential for clear stack operation and the
substantial community  benefits that attracts, the relative insensitivity to changes
in coal characteristics, and the simplicity of operation and maintenance.

It  is  the  relative  value placed on these items that influences the evaluation of
the relative economic  benefits of these different technologies.


SUMMARY

The Electricity Commission of New South Wales has operated pulse jet fabric filters
on  utility  boilers for some 15 years.  All installations have proven  satisfactory
in  achieving  low  emission levels.  The application of the technology to large PF
utility  boilers  was  initially undertaken to give short term extension to the life
of  quite  old units,  but in recent installations has become a long term option for
base load units.  Maintenance requirements were at first onerous, but have now been
reduced to acceptable  levels.

Design  developments   which  have  occurred  in recent years have enabled pulse jet
fabric filters to be developed which can be fitted inside the casings of relatively
small  precipitators,  making  retrofitting  of fabric filters a viable solution  to
emission  problems  caused  by old or undersized precipitators.  This technology  is
now  considered  to  be well proven on large scale plant in Australia,  backed up  by
operating  experience  both  here  and  overseas on small plant and extensive pilot
plant studies carried  out here.
                                         3-9

-------
It  is  expected  that   research  and  development work now underway will  result  in
fabric  filters  using   pulse  jet technology continuing to become smaller and  more
economical in future years.


ACKNOWLEDGEMENTS

The  assistance  of  the Electricity Commission of New South Wales Power Production
and  Power  Plant  Engineering  Group  Management and staff who are  involved  in the
various  projects  and   have  contributed  to  the  preparation  of  this   paper  is
thankfully acknowledged.


REFERENCES

1.  Messrs.  G.J.  Floyd  and A.Th.M. Vandewalle "Australian Experience with  Fabric
    Filters  on  Power   Boilers",   EPRI  Conference on Fabric Filter Technology for
    Coal-Fired Power Plant,  July,  1981.  EPRI CS 2238 Published Proceedings.

2.  Messrs.  F.H.  Walker  and  G.J.  Floyd "Operating Experience in Australia  with
    Fabric  Filters on  Power Boilers".   EPRI Conference on Fabric Filter Technology
    for   Coal-Fired  Power  Plants,  November,  1983.    EPRI  CS   3257   Published
    Proceedings.

3.  C.  Robertson  "Australian Experience with Fabric Filters on Power Boilers  -  An
    Update  for  1985",  Third  EPRI  Conference  on  Fabric  Filter Technology for
    Coal-Fired Power Plants, November,  1985.

A.  C.  Robertson  "Flue  Gas  Cleaning  Using  Fabric Filters in New South Wales
    Current  Experience  and Recent Developments", 7th Conference on Electric Power
    Supply Industry, Brisbane, Australia, October, 1988.
                                        3-10

-------
                                    TABLE 1
ECNSW FABRIC FILTER INSTALLATION LIST

Station Name
Boiler No.
Type
Tallawarra 'A'
No. 4A
PF
Tallavarra 'A'
Nos. 1 to 4
PF
Tallavarra 'B'
Nos. 5 and 6
PF
Wangi 'A'
Nos. 1A and IB
Nos. 2A and 2B
Nos. 3A and 3B
SS
Wangi 'B'
Nos. 4 to 6
PE
Tallawarra 'B'
Eraring
Prototype
No. 5 Cell 7
Eraring
Nos. 1 to 4
PF
White Bay
Nos. 1 and 2
PF
Nos. 3 and 4
Tallavarra 'A'
No. 1

Rating Filter Year in
MW Type Service
1 x 30 MS 1972
4 x 30 MS 1974
Combined
Plant
2 x 100 MS (b) 1975
(c) 1976
6 x 25 MS 1976
3 x 60 PJ (4) 1975
(5) 1976
(6) 1976
MS 1978
(Cell 7)
4 x 660 MS (1) Mar 82
(2) Nov 82
(3) Jul 83
(4) Apr 84
2 x 25 PJ Jun 82
2 x 25 PJ Mar 82
1 x 30 P Jul 82

Nominal
Gas-Flow Number
m /s of Bags
Per Per Unit
Boiler
38 1,320
250 7,200
170 5,760
123 3,760
124 4,032
18 592
990 47,360
64 2,022
64
78 1,296

Gas to
Cloth
Ratio,
in /s/m,
(cfm/ft )
.011
(2.1:1)
.013
(2.6:1)
.011
(2.1:1)
.018
(3.5:1)
.034
(6.7:1)
.011
(2.1:1)
.0076
(1.5:1)
.023
(4.6:1)
.023
(4.6:1)
.024
(4.7:1)
PF
                                       3-11

-------
                                 TABLE 1 (CONTD.)



Station Name
Boiler No.
Type

Pyrmont
Nos. 1 to 4
PF
Bayswater
Nos. 1 to 4
PF

Mount Piper
Nos. 1 and 2
PF
Munmorah
No. 4
PF
Munmorah
No. 3
PF
Liddell
Nos. 1 to 4
PF

ECNSW FABRIC FILTER INSTALLATION LIST

Nominal
„ , „., „ . Gas0Flow Number
Rating Filter Year in m3
MW Type Service per per ^
Boiler
4 x 50 PJ (1) 1982 120 3,432


4 x 660 MS (1) Dec 85 1,030 49,920
(2) Jun 85
(3) Jun 86
(4) Dec 86
2 x 660 MS (1) 1993 1,090 52,600
(2) 1992

1 X 350 P 1988 480 9,120


1 x 350 HC 1990 480 12,864


4 x 500 P (1) Dec 90 848 14,664
(2) Dec 91
(3) Nov 92
(4) Nov 93


Gas to
Cloth
§ati°2
m /s/m_
(cfm/ft )
.020
(4.0:1)

.0076
(1.5:1)


.0076
(1.5:1)

.020
(4.0:1)

.017
(3.4:1)

.020
(4.0:1)


LEGEND :
PF - Pulverised Fuel
SS - Spreader Stoker
MS - Mechanical Shaker
P  - Optipulse Flakt Medium Pressure Pulse Jet
PJ   Ducon Micropul High Pressure Pulse Jet
HC   Howden Carter Day Low Pressure Pulse Jet
                                        3-12

-------
           \
                   FILTER BAGS

                  -DIRTY GAS
                   FLOW
                //4-INLET
                /A MANIFOLD

                   -BAFFLE
                   -OUTLET
                    PLENUM
C ROSS SEC TION
THROUGH TWO CELLS
HAL F  PLAN VIEW
    Figure 1 - ARRANGEMENT  OF  FLAKT  FABRIC   FILTER, UNIT  4 ,  MUNMORAH

-------
CO
I
I—I
-C.
                                               PERFORATED
                                                 PLATES
                                                _ - ISOFLEX
                                                   DAMPER
                                             	 REAR CELL
                                                  FEEDER DUCT
                                                 -FILTER MODULE
                                                    DIAPHRAGM  VALVE
                                                                                    CELL
REVERSE  PULSE
   MANIFOLD
-BAG  CHANGE-OUT
     AREA
                                                                                                   CELL
                        Figure  2 - ARRANGEMENT  OF HOWDEN  FABRIC  FILTER. UNIT 3.MUNMORAH

-------
CO

t—'
U1
                                OUTLET
                                PLENUM
                               - FILTER BAGS
                 CROSS SECTIOM
 PRIMARY AIR
 HEATER OUTLET

SECONDARr AIR
HEATER OUTLET

GAS MIXING
                                                                HAL F PLAN VIEW
               Figure 3 - ARRANGEMENT OF FLAKT FABRIC  FILTER  -  LIDDELL  1-4

-------
CO


en
                              GAS CLEANING PLANT
                                  4 x 350 MW UNITS
             80
             60
             40
             20
               Life Cycle Cost ($/Kw)
                EP(1)   EP(2)  EP(3)   FF(1)   FF(2)   FF(3)   FF(4)   FF(5)   LR(1)   LR(2)
                                        Plant Type
                   Capital

                   Extra ID Power
Power Consumption

Rebagging
Normal Maintenance
                                         Figure 4.

-------
                         A TEN YEAR REVIEW OF
             PULSE JET BAGHOUSE OPERATION AND MAINTENANCE
                AT THE H.  R.  MILNER GENERATING STATION
                             B.  R.  Thicke
                        Alberta Power Limited
                          10035  -  105 Street
                     Edmonton, Alberta   T5J 2V6
ABSTRACT
Alberta Power Limited operates one of the world's largest utility
pulse-jet baghouses  at  the H. R. Milner Generating  Station near
Grande Cache, Alberta.   The baghouse is now  10  years  old.   This
paper  presents  a   review   of   the   operating  and  maintenance
experience gained over the years.  Comparisons are made of design
data  and  current   operating conditions   in  order  to  document
typical changes that can occur over  the life  of  a baghouse.  The
effect  of  these  changes   on   cost  effective   performance  is
estimated.     Troubleshooting   techniques   are  discussed   to
illustrate how component  failures  can be detected.  An  O&M cost
breakdown is given showing  that  bag  cost/life considerations are
most important.  An  historical  record of  bag  lives  is presented
together with lab test results on failed bags.  Conclusions drawn
from this  analysis  indicate desirable felt properties that have
potential  to  increase  bag  life.    Finally,  present  and  future
directions in baghouse  performance  enhancement  at  Alberta Power
are discussed.
                                4-1

-------
                      A TEN YEAR REVIEW OF
          PULSE JET BAGHOUSE OPERATION AND MAINTENANCE
             AT THE H. R. MILNER GENERATING STATION
1.0  INTRODUCTION


Alberta  Power  Limited's  H. R. Milner   Generating  Station  is
situated in  the  foothills  of  the  Canadian  Rockies.   It  is  a
single unit  generating station  burning  pulverized  coal  rejects
from a nearby coal mine.  150 MW of electricity are produced by a
Babcock & Wilcox  double downshot  boiler  and  an  Hitachi  Steam
Turbine/Generator set.  The  emission  control  equipment presently
consists of  a  Research Cotrell multi-clone mechanical separator
upstream of a Flakt pulse-jet baghouse.  This arrangement permits
the plant  to meet  the local environmental regulations allowing
particulate  emission  rates  of  .2 Ibs  per 1000 Ibs.  of flue gas
with violations  occurring less  than  50 hours  per  year.    Up to
60 TPH  of  ash  are  collected  in  the mechanical   separator  and
20 TPH in the baghouse.

The design data and description of  the unit  have been previously
presented.   A  brief review of  the  significant design parameters
is  given here  for completeness.   Table 1  shows  the pertinent
design  and  operating  data  for  the  Flakt   LKP type pulse-jet
baghouse.  Figure  1 shows a  cassette  layout.   Table 2 shows fuel
and   ash  analysis.    Table  3  shows  typical   particle  size
distribution of ash that is filtered by the baghouse.

The  fabric  filter  was manufactured  by  Flakt  Canada Ltd.   It
consists of  12 compartments that can be individually  isolated for
maintenance  purposes.    Two   cassettes   are  located  in  each
compartment.   Each  cassette contains 216 filter  bags  that are
19'6"  long and 5  3/16" in  diameter.   The  bags  are supported on 10
or  20  wire cages.   The air to cloth ratio of  the baghouse during
operation  varies  between  approximately   4:1  and  7:1, depending
upon  load.   Actual inlet dust  loading to the  baghouse has been
measured  at  20 grains  per standard cubic foot.   As an average
value  approximately  700,000  ACFM  at 320°F  passes  through  the
baghouse.    The  normal pressure  drop  is 4.7 inches of  water
 (flange  to flange).

The  baghouse  cleaning  system consists of  three  pulse  air
compressors  rated  at  125 Hp  each.   These  compressors  can be
staged up  using  one  of the  four pistons per  stage.   The pulse
control  system monitors  the differential  pressure  between the
inlet  and  outlet  flanges to  the baghouse.    As the  differential
pressure increases  above 4.7 inches of water another  stage of the
compressors  is  turned on.   This provides more compressed air to
the pulse  air  manifolds.   The  bags are  pulsed when  the pressure
in  the pulse air manifolds reach 22 psi.  The more  air provided
by  the   compressors,  the  faster  the   frequency   of  pulses.
Eighteen bags  from each of the  12  compartments are pulsed at one
time.   As  the differential  pressure  across   the  baghouse drops
                                4-2

-------
below 4.7  inches  of water one  compressor stage  is  removed from
service.   In this  way  the bags  are cleaned  less  often.    As  a
result of this control strategy, the differential pressure across
the baghouse is relatively constant at 4.7 inches of water.
2.0  OPERATING EXPERIENCE
2.1  Unit Load Versus Operating Cost

Figure   2   shows  the   approximate   operating  costs   for  the
H.  R. Milner Generating Station baghouse at different loads.  The
operating cost is made up of several factors:

1.    Additional  ID  fan  horsepower  required  to  overcome  the
     pressure drop  through  the baghouse is  approximately 600 Hp
     at 60 MW and 1600 Hp at 150 MW.

2.    The  energy  requirements  of  the  pulse  air compressors  is
     about 25 Hp at 60 MW and 375 Hp at 150 MW.

3.    Operator manhours are required to:

     a)  check for the compartment contributing the most to stack
         opacity (weekly);

     b)  isolate  and  start-up  compartments  before  and  after
         maintenance;

     c)  walk down the entire baghouse on daily rounds;

     d)  check pulse  air solenoid  valves  for  correct operation
         (bi-weekly).

     Approximately 1 manhour  per day is required  to successfuly
     operate the baghouse.

4.    Miscellaneous operating  costs include air  leakage  to flues
     through the  baghouse access  hatches,  oil  for  compressors,
     instrument air, lighting,  crane cost.   These  are estimated
     to cost 5% of the total operating cost at most.

Ash removal costs are not included here because these costs would
be incurred irregardless of flyash collection method used.

2.2  Flow Averaging

It is considered desirable to have  equal air flow  through all of
the compartments  in the baghouse.   At the  H.  R.  Milner Station
there is no way  of achieving this because gas  flow through each
compartment cannot be measured.   A differential pressure gauge is
installed across the tube-sheet  on  each compartment.   There is a
temptation to use  this  measurement as an  indication  of  gas flow
rate through the compartment.  This  is, however, much more of an
                               4-3

-------
indication of  the differential  pressure between  the  inlet and
outlet flues than of  flow  through the bags.   This  is especially
true if the inlet and outlet ducting  from the compartments has a
very low resistance in comparison with that  of the  bags (the way
all baghouses should be designed).

Assuming  that  the  baghouse has  been  designed  for equal  flow
through each of  the cassettes  it is likely that the  flow from
cassette  to  cassette  will  not  vary  greatly.    Suppose  one
compartment has  bags that  are  blinding.   This would  mean that
less flow  of  dirty gas would enter  the  compartment.   Thus less
dust would be deposited on the bag,  tending to reduce the effects
of  the  blinding  process.    As  well,  more gas would  flow  to the
non-blinded bags permitting more  ash to land  on the non-blinded
bags, decreasing their capacity to allow free flow.

This  situation has  arisen at  the  H. R. Milner baghouse  where
crude flow  indications  from a pitot  tube  show  similar  flows to
all compartments.  Yet blinding is suspected due to higher-than-
design  pressure  drop  through  the  baghouse.    Laboratory  tests
(Table 4)   of  bags removed  from  service indicated  that certain
bags exhibited a 35% lower air permeability than other bags.

2 . 3  Opacity History

Alberta Power  Limited  is  required  to   submit  two   stack  survey
tests per year to the environmental  regulating authorities.  Our
emission  limit is  .2 Ibs.  of dust  per 1000  Ibs.  of  flue gas,
corrected to 50%  excess air.    Figure 4  shows the result  of the
stack  surveys  for  the   last  10 years.     Note  the  improved
performance after dilution air usage  was  eliminated in 1984.

The indicator of flue gas particulate emissions from our stack is
a  Lear  Siegler  opacity  meter.    A  relatively constant and low
level of  particulate emission  from  the  stack is normal  at the
H. R. Milner   Station.     These  emissions  would   normally  be
invisible  against  the  sky and within  our  environmental  limit.
One  failed  bag  in  the baghouse  is  enough  to cause an opacity
spike on  the  opacity meter immediately after the  bag  has been
pulse-cleaned.      Background   levels  of   particulates   go  up
noticeably when several bags have failed.  As many as five or six
ripped  bags  can be  tolerated before opacity  violations  occur.
This is dependent on the size  of the  rip  of course.

Laboratory test  (Figure 5) results  have shown  that 15%  to 20%
more  dust  passes   through  the  felt   material  in  the  first
30 seconds after a pulse  than  during the rest of the pulse.  It
is the author's opinion that most of  the dust passes through the
bags in the  first  5  seconds after the pulse.   Signs of this are
very  visible  in  the  baghouse  in  terms  of  the opacity  meter
indications and  in terms  of  physical observations  through view
ports.   Lowering pulse  frequency is therefore  a key  factor in
lowering emission levels.
                               4-4

-------
2.4  Pulse Pressure

The original design called for a  pulse  pressure  of 35 psi.   This
has been lowered to 22 psi through plant  tests.   The impetus for
lowering  the pulse  pressure  is  based  on  the  assumption  that
higher  pulse pressures  stress  the  bags  more  on  each  pulse,
leading  to  premature  fatigue failure  of  the  bag.   The  pulse
pressure  was   lowered   in  stages,  a   few  psi   at   a   time.
Observations  were  made  of  the  pulse  frequency  required  to
maintain the  4.7"  W.C.  differential pressure and  cleanliness of
the bag.  Below 22 psi  a build-up was observed  on the bottom of
the bag  and the pulse  frequency  increased slightly.   Many  bags
cleaned  at  22 psi  have  been  sent to a  lab.  Results indicated
that  the top,  middle  and bottom  of the  bags  have  similar  air
permeability.
3.0  MAINTENANCE EXPERIENCE
3.1  Unit Load Versus Maintenance Cost

Figure  3  shows a  graph of maintenance  costs versus unit  load.
Maintenance costs  are  not  constant each year.   A  large purchase
of bags will create an unusually high cashflow requirement.

The largest contributor to maintenance  costs  is  bag replacement.
Maintenance costs  vary  widely depending on the  service life and
cost of bags.    The effect of  unit  load  is  also  a  significant
factor.  This is due to the effect on bag life from more frequent
pulse  cleaning  cycles  and  slightly higher  flue  gas temperatures
at  high load.    Both  these   factors  particularly   affect  Nomex™
bags.

Another factor that affects maintenance  costs is the  labour cost
for troubleshooting and changing  bags.   Troubleshooting requires
approximately  16 man-hours per  week.   Because there is  a  spare
cassette, bags  are usually changed out one  cassette  at  a time
rather  than individually.   This also allows  more  flexibility in
scheduling maintenance  work.   It takes  two labourers one eight-
hour shift to change out 216 bags in one cassette.

General  baghouse  clean-up, material  clean-up and recordkeeping
are also performed by labourers  in  the  baghouse.   On  average one
man working continuously is required  to  maintain the  baghouse in
peak operating condition.


3.2  Troubleshooting

Troubleshooting of baghouses  is  a  fairly difficult matter.   This
is no exception at the H.  R.  Milner Generating Station.  The root
cause  of troubleshooting  problems is  that  baghouses  typically
have  very  limited  and rudimentary  instrumentation.    At  the
                                4-5

-------
H. R. Milner  Generating  Station  poor  baghouse  performance  is
indicated by high emission levels on the  stack opacity meter.   A
difficulty  arises  in   attempting  to  distinguish  between  high
opacity  caused  by  any  one of  the following:   one  of  24 inlet
ducts plugged; one  of  24  cassette seals  leaking;  several of the
288 pulse diaphragms failed;  several of the 5,184 bags ripped; or
one  or  two  compartments  of  bags  blinded.    Even  if   it  were
possible  to  distinguish  between  these  causes  of high opacity
there  is still  the problem  of  determining  exactly where  the
problem lies.  Such as, which of the bags failed or which of the
pulse diaphragms failed.

Many  problems  can  be  isolated to  a single  compartment  by  the
isolating-the-compartment  method,   or  the  sampling-the-outlet-
flue-gas  method.    In  the former  method  the  compartments  are
removed from service sequentially,  one at a time.  When the stack
opacity  reading  drops,  then  that  compartment  is removed  from
service  and  inspected  more closely.   The  latter method involves
sampling  the  gas  from  each of  the compartments  and  determining
which  compartment has  the dirtiest  outlet  gas.    This  can  be
accomplished  relatively easily by  use  of  a  vacuum  cleaner  and
filter paper.  The vacuum draws a  sample from the compartment for
a  fixed  period  of time (for example one  minute).  The  color of
the  filter paper  is used  as  an indication  of  the amount of dust
passing through the compartment.

The two methods each have their own advantages and disadvantages.
Basically, the isolating-the-compartment  method  is easy to carry
out.   But often  the change in opacity is very small, especially
if  more  than  one  compartment  is  responsible  for  the opacity
excursion.  The sample-the-outlet-gas method has the advantage of
pinpointing  the  problem  compartment  more  precisely   (to  the
cassette  level).   As  well there is a positive identification of
all  compartment failures.   The main disadvantage  of  this method
is that  it is the more time-consuming and labour intensive.

Once  the  source  of  high  opacity  has  been  located  to  the
compartment,  or  better still,   cassette  level,  it  is relatively
easy to  identify the cause of the  problem.  Usually a compartment
can  be  isolated on-line and the cassette  cover removed to expose
the  entire  clean  side  of  the  cassette.    Drifts  of dust build up
around  the  leakage points.  Blinded  bags are very  difficult to
detect.   They  have been  identified  by  a light  layer  of  dust
spread   consistently  throughout  the  cassette.     The  use  of
fluorescent powders or  other markers is not necessary because the
clean side of the cassettes are readily accessible.

Other failure modes are not so easily determined by inspection of
the  clean side of  the compartment.   Plugged inlet  ducts (from
poor  ash distributuion  or prior maintenance)   can  be  detected by
observing the differential pressure gauge across the tube-sheet.
Zero  pressure  differential indicates a  plugged inlet  flue    It
does  not  necessarily  follow,  however,  that the  inlets are clear
if there  is  a differential pressure across the  tube-sheet.   The
foolproof method  of checking for plugged  inlets  is to attempt to
close the inlet damper  manually.   If it  cannot be closed then it
                               4-6

-------
can be assumed that the inlet  is  at  least  partially plugged with
ash.

Another  significant  problem  area   is  the  pulse  diaphragms.
Failure modes  can  be detected by observing a pressure  gauge  on
the  pulse manifold.    After  each  pulse   the  pressure  in  the
manifold drops  approximately 10 psi.   If  the pressure  does  not
drop then the pulse valve is not  opening.   On the other hand,  if
the pulse pressure never gets up to its set point before the next
pulse comes, a  leaking  diaphragm valve is  suspected.   The  exact
location  of  the  leaking   diaphragm can  be  found  during  the
inspection of  the  clean  side of the compartment  (air will  be
leaving the pulse header).
3.3  Bags

The  most  significant  cost   of   maintaining  the  H. R. Milner
Generating Station  baghouse  is the  cost  of replacement  bags.
There  is  a   continual   search for   new  and  better  bags.    A
definition is in  order at  this   point.    A "good"  bag may  be
described as one that:

a)    allows a large quantity of gas to  pass through the bag (has
     a high permeability);

b)    allows  a minimum  of  dust  to pass  through  the bag  (high
     efficiency);

c)    lasts for a long period of time  (durability).

By this definition all presently manufactured bags are eliminated
from the  category  of "ideally  suited"  to the conditions  at  the
H.  R. Milner Generating Station.

A history  of H.  R.  Milner  Station experience with different  bag
materials is shown in Table 5.   The unpredictable life expectancy
of Nomex™  bags is attributed to variations  in the manufacture of
the bags  (for example,  quality control,  acid resistance treating
of the  bags  (does not  always  improve life.) and  changes in  the
baghouse  operation  (high  loads   mean  higher  temperatures  and
pulsing rates).   The variation  in  the life  of the  Ryton™ bags is
a result of lack of ability on the Plant's part to determine when
the  Ryton™  bags  actually  fail.    Roughly  12,000 Nomex™  bags,
3,500 Ryton™  bags,  1,000 Apyiel™ bags and  less than 500 bags of
the  several   other  types  have  been used   at  the  H. R. Milner
Station.

Nomex™  bags  have been used most  often  because of  cost  and life
considerations.    The Nomex™ bags  typically  lose  their strength
from chemical attack  from  water vapour  and  SO  in the flue gas.
The most  common  failure of  100%   Nomex™  bags are  long vertical
rips along the vertical cage wires.  The service life before this
failure occurs varies depending on the number of  pulses and the
flue gas temperature.   It  is difficult to  establish the life of
the  Nomex™  bags  in  months  due  to  these  variations.    One
                                4-7

-------
manufacturer's bags lasted 34 months and 208,000 pulses while the
unit was running at a 40% average capacity and only 18 months and
240,000 pulses at a 50% average capacity factor.  It would appear
that the number  of  pulses is a better parameter  to monitor life
expectancy of Nomex  bags.

Using a Rastex™  scrim  with Nomex™ bags  improves  the strength of
the bags and consequently  the  bag  life.   However, comparing bags
from the  same manufacturer  installed in  our baghouse  over the
same period of time, the Nomex™ bags  with  a  Rastex™ scrim lasted
42 months  compared  to  34  months  for 100% Nomex™.    The failure
mode of the  Nomex™ on  Rastex™  scrim  was  delamination of the
Nomex™  from  the scrim.   The  Rastex™ appeared to be in near-new
condition,  retaining  its  strength   at  the  end  of  42 months,
whereas the Nomex™  had lost  all  its  strength and  was being blown
off the outside surface of the scrim by  each pulse.

Ryton™  bags  appear to  have the  ability to  last  a  long  time
without  a Rastex™  scrim.   Ryton™  bags  fail  by  a process  of
blinding and eventual  bleed-through of the dust.   It is believed
that the   dust  that enters  the  felt past  the  outside surface
fibres cannot be removed from the bag by pulsing.

This dust  must either pass  completely  through the  bag, thereby
showing up as  emissions,  or  remain trapped  in the felt, reducing
the permeability of the  filter bags.  Both  of these effects have
been noted in the baghouse.

A comparison  of  the properties of used  Nomex™ and  Ryton™  felts
is  given  in Table 4.  Analysis  of this  data shows  that  Nomex™
provides  superior   filtering  properties  but  tends  to   lose its
strength.   On the  other hand Ryton™ retains its  strength  well
but  tends to  trap  dust  particles  in the  felt  and blind.    An
obvious improvement could be made by  combining Nomex™ and  Ryton™
fibres into a felt that would capitalize on the good qualities of
both  Nomex™  and Ryton™  (ie. the  strength  of  Ryton™ and the
filtering properties of Nomex™) .

Laboratory  test  results  shown in  Figure 5  show the   filtering
efficiency of  Ryton™,  Nomex™  and  a  blend of  Ryton™ and  Nomex™
fibres.    The  blended  felts  contain approximately  50%  Nomex™
fibres  and  50%  Ryton™  fibres.    These  results  show   that the
blended felt  resulted  in a filtering efficiency  between that of
Nomex™  and Ryton™.   This did not meet the objective of the test
since  we  were  wanting to  achieve the  filtering efficiency  of
Nomex™.

P84™   felt was tested  along with the Nomex™  and  Ryton™ samples.
The  results  indicate  that  P84™    passes  about one-half as  much
dust  as  Nomex ,  making  it  a  most efficient  filter  medium.
Unfortunately P84™   is a rather expensive fibre, as well  it has
an unknown life in our flue gas conditions.  For these reasons an
asymmetrical  felt  bag is being tested  in  the  baghouse.   This
asymmetrical  bag consists  of P84™    fibres on  one  side   of  a
Ryton   scrim  and Ryton™  fibres  on the  other side  of the scrim
The  bags  are made  in  such a  way  that  the  P84™   is  facing the
                               4-8

-------
dirty flue  gas and  the  Ryton™  is  facing the  cage wires.   The
intention  is  that  the  P84™     surface  layer  provides  high
filtering efficiency.  The  Ryton™ scrim  and  inside batt provide
strength and abrasion resistance.  Although no further laboratory
tests have been carried out to prove it, this is expected to be a
better way of  combining  the fibres than  blending  them as tested
with the Nomex™ and Ryton™  blended felt.
3 . 4  Cassette Seals

One  very  nice  feature  of  the  H. R. Milner  baghouse  is  the
removable cassette.  A cassette consists of a steel tube-sheet on
a  tubular  frame  around  216 bags.    An   overhead  crane  allows
removal  of  the  entire  cassette  from  a  compartment  of  the
baghouse.   Several  design  of seals  between  the  cassette  and
baghouse  have been  tried.    The  current  tadpole  gasket  works
reasonably well  but  high  pressure excursions can result  in many
seal  leaks  throughout the  baghouse.     These  seal  leaks  are
repaired using high temperature silicone sealant.   Hold-down lugs
have been  added  to prevent movement of the  cassette  during high
differential pressure  transients.   A new  seal  design using high
temp silicone  foam backing  on a hard  silicone sealing  face  is
being tested at present.


3.5  Pulse Diaphragms

Both  the   original  pulse  diaphragms   and  the  third  party
replacement diaphragms  that  have  been  used at the  H. R. Milner
Station have  provided  good service.   An  estimated  500,000  pulse
life from  a  diaphragm is not  uncommon.   The pulse air  from the
compressors  is  cooled to   190°F  maximum temperature.    Recent
attempts to increase this  temperature  have shortened  the life of
some diaphragms.   Increased  pulse air temperature  is thought to
be  desirable   in  order to  reduce acid  attack on  Nomex™  bags,
which may  be  aggravated by  the cooling effect of  the expanding
pulse air.  The  pulse diaphragms fail around the bleed  holes in
the  form  of a  tear.    When  this tear  becomes  large  enough the
pilot solenoid valve cannot relieve  the  air pressure above the
diaphragm  to  allow the pulse  valve  to open.   Often  the tearing
will cause  the disk to become cocked in  the seat,  allowing the
air  pressure   to  escape  down  the pulse   header.    This  affects
proper cleaning of all  of  the  bags  in one compartment.  Changing
out the diaphragm is a simple matter, handled by labourers.
4.0  PERFORMANCE ENHANCEMENTS
4.1  Ongoing Bag Testing Program

The high cost and  short  life  of bags in the H. R. Milner Station
baghouse has prompted an ongoing testing program.  New felt bags
of various  materials  are being  tested  on a regular  basis.   The
strategy for testing new felt material is as follows:
                                4-9

-------
Step 1 The physical  properties  of the  proposed felt  fibres are
       solicited from  the supplier,  along with  any  other test
       data available.   These properties  are compared  with the
       properties of felts that have previously been tested with
       a view to predicting any obvious  failure modes.  Assuming
       the  felt has  a  reasonable  chance  of  surviving  in the
       baghouse and  is  not  excessively expensive we progress to
       Step 2.

Step 2 A minimum of two bags are requested  from the supplier for
       installation in the baghouse.  These two bags are watched
       for signs of deterioration or failure  for  a  period of six
       months to one year.

       In parallel,  or  occasionally  in lieu  of,  the in-baghouse
       testing, laboratory tests may be  carried out to determine
       filtering efficiency  and pressure  drop through  the new
       felt sample in comparison with other felts.

Step 3 When the new  felt  has  shown itself to withstand  the flue
       gas conditions reasonably well,  purchase is  then made of
       approximately 450 filter bags.  Installing this many bags
       (one  compartment)  allows   a   reasonable    test   of  the
       filtering efficiency  of the  new felt.

The testing  program of  new  bags  is  hampered in  Step 3  by the
limited instrumentation available.   There is  no way of measuring
flow through  the compartment of  new bags  purchased on  a trial
basis.  It is  only possible to observe the  amount of ash passing
through the bags by  the dust  accumulations on the  clean side of
the  cassette  (usually  very  minimal)   and  by  waiting   for  bag
failures to occur.   The effectiveness of this  bag  testing program
is therefore somewhat limited.
4.2  Baghouse Performance Monitor

The  lack of  instrumentation available  on  the  baghouse  limits
troubleshooting and new bag testing.   Preliminary work to develop
a baghouse performance monitor has been undertaken.   The idea of
such  a  performance  monitor  would  be  to  pinpoint  equipment
failures.  Two  important parameters need  to be monitored  to do
this opacity and gas flow.

Correlating stack opacity  and  pulse  sequencing could  be used to
indicate ripped  bags.   Gas  flow  rate  could be used  to indicate
plugged  inlet  flues,  blinded  bags  or poor pulse  performance.
Measurement  of  gas  flow   rate  proves   to  be   difficult  and
expensive.
                               4-10

-------
5.0  CONCLUSIONS

The  operating  and  maintenance  experience  at  the  H. R. Milner
Generating  Station  pulse  jet baghouse  leads  to the  following
conclusions:

1.   It appears  difficult to  operate  a pulse-jet baghouse  cost
     effectively  for  extended periods  of time  at  an  air/cloth
     ratio greater than 5:1.

2.   Fan  horsepower  requirements  are  the  largest  component  of
     operating expenses.

3.   Replacement  filter  bags   are  the   largest  component  of
     maintenance expenses.

4.   Short  bag life  is  responsible  for  the  high  cost of  bag
     replacement.

5.   It takes  several  years of  testing  to determine if  new bag
     types are good.

6.   Troubleshooting and  new  bag  testing could  be  enhanced  by
     better instrumentation or the use  of  a  baghouse performance
     monitor.

7.   The  outlet  emissions  from  this  pulse  jet  baghouse  can.
     readily meet stringent environmental limits.

8.   Most of the dust passing through a pulse jet baghouse passes
     immediately after a pulse.  High pulse rate should therefore
     be avoided.

9.   Pulse  pressure  can  be  field  set  to  a  minimum value  by
     observing the dust accumulation on the  bag after pulsing at
     different pulse pressures.

10.  Variations in ash distribution to the baghouse cassettes can
     result  in both operating and  maintenance  problems.   Care
     should be taken to avoid this situation.

11.  Removable cassettes and a spare cassette allows changing out
     bags more  quickly  and in a scheduled manner.   Seal leakage
     problems can be overcome  with new  elastomeric materials now
     available.

12.  Blinded   bags   are   very  difficult   to   detect   without
     instrumentation to indicate flow rate through the bags.

13.  If most of the pressure  drop  through  the baghouse is across
     the  bags,  tube-sheet  differential  pressure  is not  a  good
     indication of gas flow through the compartment.
                                4-11

-------
14.   Pulse diaphragm  failures  occur  regularly  but are  readily
     detectable by careful observation of a pressure gauge on the
     compressed air manifold.

15.   Combining  different  fibres  together  to   form   one  felt
     material with  superior  properties  to  the  constituents  is
     possible but requires some laboratory tests.
                               4-12

-------
                            FIGURE 1
                         CASSETTE LAYOUT
OUTLET DAMPER
ACTUATOR
OUTLET
DAMPER
CLEAN GAS
OUTLET FLUE
DIRTY GAS
INLET FLUE
                            PULSE MANIFOLD
INLET DAMPER
                                            ACCESS COVER


                                            PULSE HEADER
                                            CASSETTE
                                            BAGS 53/t6Mx20'LG.
                                            FLYASH
                       TO FLYASH DISPOSAL

-------
0
p
E
R  $160,000.00
A
                          FIGURE  2
               OPERATING COST VS UNIT LOAD
N
G

C
0
S
T
Y
R
   $120,000.00  I
   $80,000.00  4-
   $40,000.00
        $0.00
             60
                        90          120

                          UNIT LOAD Mw
MISC

MAN-MRS

PULSE AIR

DIFF. PRESSURE
                                                   150
M
A
I
N
T
E
N
A
N
C
E

C
0
S
T
 Y
 R
    $500,000.00
    $400,000.00
                                FIGURE 3
                   MAINTENANCE COST VS UNIT LOAD
              60
                           90             120
                             UNIT LOAD Mw
                                                         150
                                     4-14

-------
                          FIGURE 4
           STACK SURVEY RESULTS 1980 TO 1989
                                                             T 7.00
                                                               6.00
                                                               5.00
                                                             I 4.00
                                                               3.00
                                                               2.00
                                                               1.00
                                                               0.00
80  81  81  82  82  83  84  84  85  85  86  86  87  87  88  88  89
                            YEAR
           PARTICULATES LB/1000LB • AIR / CLOTH RATIO
0.07 -
% 0.06 -
L 0.05 H
E 0.04
A
K 0.03
A 0.02 •
G
E 0.01
0


RY1
FIGURE 5
FABRIC PERFORMANCE
NORMAL CONDITION

•ON Nor

HEX NOMEX

/ RYTON Pf
H % LEAKAGE AFTER PULSE H % LEAKAGE AVERAGE


J4

                            4-15

-------
                          TABLE 1
     DESIGN AND OPERATING DATA FOR FLAKT LKP BAGHOUSE
Number of compartments
Number of cassettes per compartment
Number of bags per compartment
Total number of bags
Bag size
Gas volume at 250 deg F
Air to cloth ratio
Inlet grain loading (design)
Inlet grain loading (measured)
Inlet grain loading (design)
Inlet grain loading (design)
Dust size (90% of population count)
Operating pressure drop flange to flange
Cleaning mode
Cleaning interval (150Mw / 60Mw)
Normal temperature (150Mw / 60 Mw)
                     12
                      2
                    216
                   5184
       5-3/16DiaX19'-€"
           860,000 ACFM
                   6.44
                3gr/scf
               20 gr/scf
            0.0186gr/scf
            0.0614 gr/scf
         < 1.956 microns
                4.7" w.c.
                 on line
        1.2 min. / 10 min.
    340 deg F / 280 deg F
                               TABLE 2

                         FUEL AND ASH ANALYSIS

COAL PROXIMATE ANALYSIS:
Coal Source

Fixed Carbon %
Volatile Matter %
Ash%
Moisture %
HHV Btu/lb

ASH ANALYSIS:
ROM
Tailings
Mix
56.7
17.4
20.2
5.7
11015
37.6
13.9
38.9
9.6
7290
23
15.1
54.2
7.7
4673
SiO2
AI203
Fe203
TiO2
P2O5

53.2
28.3
3.8
1.7
1.3

CaO
MgO
SO3
Na2O
K2O
UNDT
4.2
0.4
3
1.1
0.2
2.8
                                4-16

-------
                                   TABLE 3
                            PARTICLE SIZE  SUMMARY
                                                              microns
              10% of the volume /
              20% of the volume /
              30% of the volume /
              40% of the volume /
              50% of the volume /
              60% of the volume /
              70% of the volume/
              80% of the volume /
              90% of the volume /
       population
       population
       population
       population
       population
       population
       population
       population
       population
     is below
     is below
     is below
     is below
     is below
     is below
     is below
     is below
     is below
            2.968 / 0.342
            5.391 / 0.442
            9.394 / 0.539
            15.51 / 0.642
             22.7 / 0.758
            30.62 / 0.900
            39.92/1.091
            49.74/1.383
            67.17/1.956
                                     TABLE 4

                                USED BAG ANALYSIS

               Material                               Ryton      Nomex

               Service Life - months                     28          29
               Material weight (as received) - oz/yd       29          24
               Material weight (washed) - oz/yd          16          16.5
               Air Perm, (as received) - cfm @ 0.5" we     3.6          5.5
               Air Perm, (washed) - cfm @ 0.5" we       28.2         20.7
               Mullen Burst Strength - psi               273          97
               Tensile Strength - Ib/in.                 205          141
                                   TABLE 5
                                BAG LIFE HISTORY
TYPE OF BAG
  LIFE
MONTHS
  LIFE
PULSES
FAILURE MODE
Fiberglass (woven)
Acid resistant Nomex
Untreated Nomex
100% Ryton
Nomex / Rastex scrim
Ryton / Rastex scrim
Apyiel
Used Nomex
0.5
12 to 34
5 to 30
7 to 51
30 to 42
28 to 42
6 to 12
2 to 4
-6,300
< 240,000
< 200,000
< 450,000
< 350,000
< 350,000
< 120,000
< 30,000
                       Many small rips, fiber wear / fatigue
                       Vertical rips along cage wire closest to seam
                       Vertical rips along cage wire closest to seam
                       Blinding and bleed through of dust (no rips)
                       Nomex delaminated from rastex scrim
                       Blinding and bleed through of dust (no rips)
                       Vertical rips and bleed through of dust
                       Vertical rips along cage wire closest to seam
                                         4-17

-------
                      DESIGN AND PERFORMANCE EVALUATION
                         OF A 350 MW UTILITY BOILER
                           PULSE-JET FABRIC FILTER
                             Peter W.R. Funnell
                             Flakt Australia Ltd
                  124 Pacific Highway, St.Leonards, N.S.W.
                                  AUSTRALIA
                               Peter R. Heeley
                  Electricity Commission of New South Wales
                       Cnr. Park & Elizabeth Streets,
                               Sydney, N.S.W.
                                  AUSTRALIA
                               Stig Strangert
                             Flakt Australia Ltd
                  124 Pacific Highway, St.Leonards, N.S.W.
                                  AUSTRALIA
ABSTRACT

The ESP to pulse-jet fabric filter conversion carried out by Flakt on a 350MW
unit at Munmorah Power Station in Australia has been in successful operation
since August 1988 with an average operating time of about 6000 hours per year.
This is currently the world's largest utility pulse-jet filter.

The unit burns coal  with the low sulphur and high ash content which is
characteristic for Australian coals.  Acrylic needlefelt filter bags are used
with attemperation of the flue gas to control gas temperature.

A brief description  is given of the major modifications undertaken to convert
the ESP plant to a pulse-jet fabric filter without compromise of design
criteria.

Details are presented on the performance of the plant from startup, including
maximum design conditions and boiler upsets.

Performance and maintenance aspects of the filter bags, which at 23'7" (7.2m)
are the longest so far in commercial operation, are also discussed.
                                         5-1

-------
                      DESIGN  AND  PERFORMANCE  EVALUATION
                         OF A 350 MW  UTILITY  BOILER
                          PULSE-JET  FABRIC FILTER
INTRODUCTION


Munmorah Power Station  (4x350MW)  located  on  the Central  Coast of New South
Wales, Australia,  was commissioned  from the  middle to the late 1960's.   After
twenty years of service with  the  plant nearing  the end of its economic  life
the power utility    ECNSW - made  the decision  to carry out major refurbishment
and life extension work,  initially  on Unit 4 then on Unit 3.   More recently
the further decision has  been made  to carry  out similar refurbishment work on
Units 1 and 2.

The original Air Pollution Control  plant  on  Units 3 & 4 consisted of two twin
chamber three field Research  Cottrell cold side electrostatic precipitators
designed for an emission  of 400mg/Nm3.  Deterioration of the  plant with time
and changes in coal and combustion  conditions  resulted in loss of availability
and difficulty in  achieving  the statutory emission limit.  The emission level
was also far more  than  that  necessary to  satisfy the ECNSW clear stack  policy
adopted for all new power stations  which  is  equivalent to an  emission level of
not more than 50mg/Nm3.

Economic and technical  feasibility  studies were carried out on upgrading the
APC plants to achieve enhanced availability  and clear stack emission. This
included investigations into  extending the ESP's advanced ESP energisation
techniques of milli-and micro-second pulsing,  installation of new low ratio
fabric filters as  well  as retrofitting of high ratio (pulse-jet) fabric
filters into the existing ESP casings.

The ECNSW had sixteen years  of experience from the operation  of various types
of fabric filters  and had either operating or on order over 6.600MW of shaker
cleaned low ratio bag houses.  Significant successful experience had also been
gained with three small ECNSW power stations (Wangi, White Bay and Tallawarra)
with  similar operating conditions which had  been retrofitted  with pulse-jet
fabric filters.

The economies were strongly in favour of the pulse-jet retrofit concept which
would utilise much of the existing  structure and casings, would fit into the
station layout and require no more  land area and which could  be constructed
with minimal plant downtime.
                                         5-2

-------
Successful pulse-jet pilot plant tests for more than one year at Munmorah
Power Station provided the necessary confirmation of the technical feasibility
of the retrofit concept.


PILOT PLANT TESTS
Pilot plant testing to predict full scale plant operation is a well
established technique with the ECNSW and Flakt. For almost two years prior to
the tendering of the Munmorah upgrading, Flakt in cooperation with the client
operated an OPTIPULSE pulse-jet high ratio filter pilot plant on boilers at
Munmorah. The pilot plant was a twin cell, 18 bag, self contained and fully
remotely controlled and monitored unit. It replicated essential features of a
full scale plant, pulse cleaning system and filter element design, leaving
only few parameters such as the dust migration pattern in the bag matrix to
scaling up. The test results were used to verify relationships between filter
specific load (filtering velocity), filter pressure loss, intensity of filter
cleaning and dust emission. The comprehensive data acquisition and processing
facilities also enabled up to six different filter media to be monitored
simultaneously for screening tests.

Extended duration testing was carried out with the pilot plant to verify long
term bag cleaning effectiveness and to ensure the effectiveness of cage and
bag removability.

PULSE CLEANING CONCEPT
Pulse-jet fabric filters operate with the dust laden gas approaching the
filter elements from the outside, depositing the particles on the fibres of a
depth filtering medium. The cleaned gas leaves the open end of the element,
typically of a tubular design with a diameter of 120   150 mm, (4 3/4 - Gin)
fitted in a plate separating the cleaned gas from the dust laden gas. An
internal wire cage supports the filter element against the pressure caused by
the gas when flowing through the filter medium. Periodically the dust cake is
cleaned off the element by rapidly expanding it with a pulse of air. The
inertial force acting on the dust cake when the filter element is rapidly
stopped in its expansion overcomes the bond to the filter medium and the
removed dust is transported towards the dust hopper beneath by gravity and by
the gas flowing through the matrix of filter elements .

The effectiveness of the cleaning is clearly dependant on the rise speed and
magnitude of the pressure pulse inside the filter element. The OPTIPULSEW
produces a forceful pulse by an optimised geometry of the pneumatic system
that delivers the pulse.  Key features are:

    The pulse air is injected into the filter element without dissipation of
    the kinetic energy into a large volume of entrained secondary gas by
    optimum selection of injection nozzle size relative to the filter element
    size and.
                                         5-3

-------
    The area of all  nozzles on the header serving one row of filter elements
    is matched to the area of a large pilot operated, fast opening supply
    valve.

The high force pulse enables large filter elements - long filter bags   to be
used.  This  feature of the OPTIPULSE concept is of particular value in the
retrofit of fabric filters into the existing structure of mechanical
collectors  or electrostatic precipitators,  where sufficient height is often
available but ground area is limited.

PLANT DESIGN SPECIFICATION


The specification called for a plant to meet the following essential
requirements.

    Capable of intermittent operation,  with frequent on-line starts, over a
    continuous boiler load spectrum 150   300 MW and with flue gas
    temperatures below the acid dewpoint during extended boiler starts.

    To be based on the suppliers experience with similar plant.

The design operating conditions were specified as

    Gas flow from airheaters
                           design                 480tn3/s
                           maximum                550m3/s
    Gas temperature at airheater                  140 °C
      with spot temp variation                    + 15°C
    Inlet burden
                           typical                25g/m3NTP
                           maximum                40g/m3NTP
    Median ash particle size                      5 - 30micrometer

Measurements during pilot plant testing had indicated an inlet burden of up to
20 g/m3NTP and a median particle size of 11 - 12 micrometer.

The main plant performance parameters to be met were

    Pressure drop between plant terminal points not to exceed 2.5 kPa.
    Filter velocity at gas flow of 528m3/s  not to exceed 0.02m/s (gas to cloth
    ratio 4:1).
    Emission of particulate matter not to exceed 0.05 g/m3NTP dry, 12 % C02-

Main  properties of the coal supplied from a range of sources were given as
(proximate analysis):

    Ash content                                   15   29%
    Volatile matter                               22 - 30%
    Gross specific energy                         21.5 - 27.6 MJ/kg
    Total sulphur content                         0.28   0.5%
                                        5-4

-------
Typical content of the main species of the ash are:

    Si02                                          61%
    A1203                                         24%
    Fe203                                          6%

PLANT DESIGN
The design option selected based on economic considerations and experience
from full scale and pilot plant operation, was a retrofitted Flakt OPTIPULSE
filter using acrylic felt filter bags.

Filter media made from homopolymer acrylonitrile fibre (Bayer Dralon T) had in
fact been used in all except two of the fabric filters in service at other
ECNSW power stations at the time and had proven their suitability for the
particular flue gas conditions resulting from burning pulverised coal with the
low sulphur content and high ash content characteristic for the local supply.
Dralon T media permits continuous operation at temperatures up to 125 °C
(257°F) with some latitude for surges up to 150 °C (302°F)

The Dralon T needlefelt has a nominal weight of 500 g/m2 with batts of 2.4
dTex staple fibre and a combination spun staple/multifilament scrim, a
permeability of 0.14 m^/m2,s at 0.12 kPa and a minimum burst strength of 3000
kPa, a singed surface finish and is heat set for a maximum shrinkage
(unrestrained) of 3 % at 150 °C for 24 hours.

The filter bags have a diameter of 130 mm and a length of 7.2 m. The filter
bags are provided with double layer reinforced cuffs at the top and bottom
ends. The top cuff has a grooved profile and a steel snap band that fits into
punched holes in the bag plate. The bottom cuff has a closed end for sealing
and a sewn-in stainless steel band supporting a stainless steel cup which is
inserted at installation of the bags. The supporting bag cages are made from 4
mm dia mild steel wire, designed with ten longitudinal wires spotwelded to the
tips of star shaped spacer rings. This configuration allows the bag, with its
necessary looseness relative to the cage, to fold along the longitudinal wires
only. If instead circular spacer rings had been used, the bags would have
folded in two directions at the crossings of wires and rings and would thus
have been subject to fatigue and shortened life due to the sharp creasing at
these points. Further the permanent set would have made withdrawal of the cage
from the bag difficult, substantially increasing rebagging costs.  The cages
are supported by four of their wires being formed into hooks resting on a
steel ring around the top cuff of the bag. The remaining six wires finish
about 100 mm from the top to allow deformation of the bag top cuff for un-
snapping it with the cage still in place. The bottom of the cages is open and
fits inside the bag bottom cup. The cages are joined in the centre so that the
headroom requirement for lifting out the cages from the bags is halved. The
joint is designed as a split ring with a locking latch between adjacent wires
in one of the cage halves, which is fitted and rigidly secured into a groove
in a spun collar in the other half.
                                        5-5

-------
A considerable amount of development and testing of the seemingly simple
design of the filter bags and cages to ensure that they could be easily
installed and reliably and quickly removed without the use of tools. This
requirement is understandable in view of the large number, 9120 elements, to
be handled in as hygienic a manner as possible during replacement. The normal
method of removing filter elements in larger quantities during scheduled bag
replacement is to pull out the cages and drop the bags into the hoppers for
collection via the hopper access doors.  Should the bag catch on the cage
during its withdrawal, the bag can be held stretched by lowering a weight into
the bottom cup in the bag. In the worst  case the bag cuff can be un-snapped
from the bag plate and the bag and cage  pulled out together.

With the boiler air heater outlet temperature in the range 110 to 150 °C, the
maximum allowed continuous operating temperature of 125°C for the filter bags
is controlled by flue gas attemperation.  Ambient air is admitted into the
individual cell inlet ducts through four vertically  distributed intake ports
to achieve efficient mixing of flue gas  and air. The flow to the ports is
regulated by a hydraulically actuated damper, with feed back control from a
thermocouple grid at the cell inlets. The system has the capacity to control
flue gas temperatures up to 170°C, however if the temperature after
attemperation exceeds 150°C a back-up emergency water spray system is
activated and the boiler tripped. An additional thermocouple grid is installed
for monitoring purposes upstream of the  cooling equipment.

In order to give the required redundancy for servicing of part of the plant it
was laid out with eight fully isolatable cells by dividing each of the four
precipitator gas passages into two with  a solid wall, for which the design of
the existing dust hoppers was also suited. Each cell contains 1140 filter
bags, arranged in a bag nest of 57 rows, each 20 bags deep. The installed
filter area gives a gas design velocity  of 0.02 m/s through the filter medium
with all cells on line and the boiler running at maximum continuous rating.

The precipitator inlet funnels and part  of inlet ducts were replaced by two
bifurcated ducts, each with an isolating damper, diverting the flue gas to the
outsides of the casings. The gas then flows sideways-downwards into the bag
nest from a plenum stretching along the  casing. The lower half of the bag nest
is covered by a solid baffle and the upper half by a mesh screen which
protects the bags adjacent to the plenum from abrasion by ash particles. This
so called side entry is a leading principle in Flakt pulse filter design and
limits the re-entrainment of dust into the gas to be filtered by creating a
short migration route for dust removed from the bags at pulse cleaning.

The roofs of the precipitator casings were removed and clean gas plena, each
complete with bag attachment plate and the pulse cleaning system, were
installed. The height of the gas plenums equals half the length of the filter
elements to allow handling of the cages.
                                        5-6

-------
The pulse air tanks, five individual tanks flexibly interconnected, with their
integrated OPTIPULSE valves run along the length of the plenums. The pulse
valves are mounted on the upper face of the square tube tanks and the pulse
air is piped into the plenum via a 90° bend. The pulse pipes are located with
a simple slip fit on the incoming pipe and with a pin and bracket arrangement
at the other end. The nozzles in the pulse pipe are accurately located on the
centreline of each bag and are also provided with directing spigots to ensure
that the pulse jet is properly directed into the centre of each bag for
maximum efficiency and minimum wear of the fabric at the bag opening.

For aerodynamic reasons and considering the structural condition of the
ducting between the former precipitator outlets and the ID fans, it was
completely replaced by individual cell outlet ducts with dampers, connected to
the fans through two T-shaped outlet manifolds. The cell outlet dampers as
well as the inlet dampers have electric actuators.

The increase in pressure drop across the plant of about 1.5 kPa (6 in WG)
compared to precipitator operation, required replacement of the ID fans and
motors. The centrifugal fans are of double inlet, double width type with inlet
guide vane control. Through the use of high efficiency air foil blade
impellers run at increased speed - from previously 740 RPM to 980 RPM - the
new fans could be installed on the existing foundations. The inlet guide vanes
are actuated hydraulically and a hydraulic supply unit also serving the
attemperation air and fan outlet damper actuators was provided.

To protect the system upstream of the ID fans from the higher suction pressure
that could be reached with the new fans, implosion protection dampers were
installed in each of the cell casings. These are double poppet dampers on
common shaft actuated by a pneumatic cylinder which open up to  atmosphere
when automatically activated. Further protection is provided by fast acting
hydraulically operated fan outlet dampers.

The flow of collected fly ash from the existing hoppers has been improved by
replacing the lower portions with stainless steel conical outlet sections and
lining the remainder of the hoppers with stainless steel. Hopper heaters
mounted in an air jacket around the bottom portion for even temperature
distribution further facilitate the dust discharge.

Great emphasis has been placed on facilities for monitoring and servicing of
the plant. For the monitoring, each filter cell is provided with sensors for
inlet temperature, gas flow, bag pressure differential and optical density of
filtered gas. These data together with plant overall pressure drop are
continuously logged and instantaneously accessible on a dedicated computer.

The clean gas plena have many viewing ports for inspection and visual
detection of bag leaks during operation. The existing access around the
precipitator casings was revised to enable these and other regular inspections
and to allow for rebagging.  Major access areas at the filter clean side
plenum, hopper level and at inlet and outlet dampers are reached by lift and
covered walkways.
                                        5-7

-------
A cell  cooling and forced ventilation system, enabling access to a particular
cell within 30 minutes after isolation,  further improves the working
conditions during maintenance work inside the cells. Air is drawn from the
access  doors through the clean side plenums by a duct system connecting to
central exhaust duty/standby fans.

It was  a requirement that individual  cells could be serviced with the
remainder of the cells in operation.  Due to the high sound level resulting
from the increased speed and pressure capacity of the ID fans, it was
necessary to install inlet silencers  on  the ID fans to give acceptable sound
levels  in the clean side plenum of an isolated cell. Outlet side silencers
were installed to reduce the sound emission from the ducting leading up to the
stack entry. The silencers are of the compact reactive type.

Compressed air for the bag cleaning is supplied by two rotary screw
compressors, each with 100% capacity, in a duty-standby arrangement. The
compressors are housed together with  the fabric filter plant electric
switchboards in the previous precipitator control room. The compressor
discharge pressure is 275 kPa for a pulse tank pressure of 260 kPa. The air
passes  through a fan forced after-cooler and is distributed to the pulse tanks
through stainless steel pipes. The duty  compressor runs continuously and
excess  air is vented through two pneumatically pilot operated unloading valves
at the  extremities of the distribution piping. The unloading valves are set to
open at the correct pulse pressure in a  lead-lag arrangement. When venting,
flaps on the openings of the discharge silencers trigger switches which in
turn enables pulsing of the next pulse valve in the cleaning sequence. This
system circumvents any problems with  drift in set points for separate controls
of compressor unloading and pulse activation. The surfaces in the compressed
air system are maintained about 30 °C above ambient temperature to avoid
condensation. This is achieved by a corresponding set point for the compressor
after-cooler, uninterrupted flow through the piping system and installing the
pulse tanks in an enclosure partially heated by the filter outlet plenum.

The existing precipitator structure was  assessed prior to the conversion work
in view of the new duty as a fabric filter with higher gas suction pressure
and ash load and against changes in structural design codes.  With implosion
damper control of the pressure in the casings and inlet ductwork, these could
be largely reused with only minor additional bracing for wind loadings. The
existing precipitator roof beams were retained and relevelled for the new bag
plate and outlet plenum. Some critical areas of the hoppers and support
structure were shown to need reinforcement, and the supports were rock bolted
for increased wind loads. With careful pre-planning the work outlined could
proceed during the construction phase with little or no delay from unforseen
circumstances.

PLANT PERFORMANCE
The rehabilitated unit went back into operation in August 1988.

-------
During an extended period of boiler recommissioning with oil firing only, the
flue gas was passed through two filter cells which had not been bagged out -
there is no filter bypass duct installed in accordance with current
legislation and combustion products from oil firing may irreversibly blind the
filter material.

Before the filter was put on line for a subsequent extended period of boiler
operation at low load with combined oil and intermittent coal firing, the bags
in three selected cells were pre-coated with fly ash in order to absorb any
condensibles in the flue gas. The ash was supplied at an amount of about 0.5
tonne per cell  and was pneumatically injected into the cell inlet ducts from a
bulk carrier. The filter cleaning was turned off until a certain pressure drop
was reached and the pulsing was activated, whereupon the pre-coat was renewed.
This procedure was repeated throughout the mill commissioning period.

After completed commissioning of the boiler firing plant all cells were
successively brought on line as required while key operating data were logged
continuously using the performance monitoring system.  As predicted from the
pilot tests, the plant reached a plateau in terms of pressure drop and
cleaning frequency within a short period of time, less than 1000 operating
hours.

The filter cleaning strategy influences the filter emission, the filter bag
life and the boiler draught control. The principle adopted at Munmorah was to
initiate and continue filter cleaning until the plant differential pressure
had been reduced below a set point (1.55 kPa) by a small amount. A timer
controlled cycle is interlaced with the pressure controlled sequence to ensure
that the entire filter is cleaned once per two hours during low boiler load
operation. This limits the amount of fly ash that is allowed to accumulate on
the bags so that over-shooting of the plant pressure drop and unstable boiler
draught control is avoided during rapid boiler load rise. - A third, manually
initiated, rapid pulsing facility is also provided to be used for instance
prior to filter bag replacement.

At normal full  boiler load of 300 MW the filter plant maintains the overall
pressure differential of 1.55 kPa at a total cleaning cycle time of about 40
minutes with all  cells in service. The cycle time increases with reducing load
until the timer control takes over for loads below about 200 MW. At continuous
operation at half rated boiler load, 150 MW, the filter plant differential is
about 1.0 kPa.

The plant overall performance has to date been tested on two occasions, in
April 1989 and in October 1989, the latter being an acceptance test after an
operating time of about 7000 hours for the cells in longest service. Data from
the two tests are given in Table 1 together with the expected performance
according to the contract.

A bag leak test using fluorescent tracer was carried out on the two filter
cells that gave the highest cleaned gas optical density prior to the
acceptance test.  As no noticeable leak was detected it appears that the
difference in emission between the cells is due to normal variation in the
properties of the particulate and the filter material.
                                         5-9

-------
The predicted full  scale filter emission based on pilot plant data and using
an emission flux model  (where the particle emission is regarded as occurring
in pulses solely in conjunction with pulse cleaning) gives 5 and 15 mg/m NTP
for the April and October tests respectively.  Whilst the model accurately
predicted the effect of varying the pilot plant operating parameters, the base
value obtained from the pilot test overestimated the full scale emission by a
factor of up to 2.

The increase in power absorbed by the higher duty ID fan and by the cleaning
air compressor at full  boiler load is about 1700 kW. The power absorbed by the
continuously operating, duty pulse air compressor is about 70 kW, or only
about 4 % of the fan power corresponding to the filter plant pressure drop.
(The compressed air flow for cleaning is a mere 0.05 % of the filter flow at
design operation.) The capitalised cost over the remaining plant life using
the client's evaluation criteria is about 1.3 million USD. On a generated
energy basis, the upgraded plant requires an additional 0.5 % of the unit
output,  representing a cost of about 0.0004 USD/kWh.

The filter  plant has required very limited operator attention and maintenance
and has  not caused any unforseen limitations or interactions with the
operation  of the boiler. A recent incidence demonstrates the tolerance to
upset  operating  conditions:  During firing up of the boiler with  light fuel
oil,  about  10 000  litres of  unburnt oil was inadvertently deposited  on the
filter bags,  resulting  in a  sticky dust cake and drastically  increased
pressure drop.  The plant was purged with hot air,  allowing the oil to
evaporate  to avoid a potential explosion, and after a few days of normal
operation  the pressure  drop  had  completely recovered. In this case the
combination of  an  absorbing  ash  layer on the bags  and the forceful cleaning
pulse prevented  any permanent  effect.

Filter bags have been  regularly  inspected and tested  for any  signs of
deterioration.   To date  no bags  have  failed in  operation.

 CONCLUSION


 At the time of  finalising  this paper  the  plant  had logged nearly 10,000  hours
 of operation at loads  of up  to 360MW.   The  plant  pressure drop  and  the  pulse
 cleaning intervals have stabilised,   Transits  through acid  dew point and oil
 contamination have not affected  stability.

 The plant  is maintaining a clear stack  at all  times including during pulse
 cleaning.  Tests have  confirmed that the emission level  is  maintained below
 10mg/Nm3 except during rapid load increases.

 The 350mw  Munmorah Unit 4 base load plant is  the largest operating utility
 pulse-jet  fabric filter plant in the world.   Its successful,operation has
 extended this filter technology into the arena of large utility boilers.

 Pulse-jet  filters provide a cost effective solution to upgrading existing ESP
 plant performance and will challenge low ratio bag houses for use on new
 installations in the future.

 REFERENCES     1. K. Mascord and S. Strangert "The Ascendency of
                   the Pulse-Jet Filter", Third C.S.I.R.O. Conference
                    on Gas Cleaning, August 1988.
                                         5-10

-------
                                DESIGN     TEST      TEST
                                           APRIL    OCTOBER
Load                 (MW)         300       240       300
Gas Flow             (m3/s)       547       411       475
Inlet Temp.          (°C)         125       106       111
Filter Velocity      (m/s)       0.02      0.018     0.018
Pressure Drop        (kPa)        2.0       1.5       1.9
Cleaning Cycle       (s)           -       6840      2400
Inlet Burden         (g/m3NTP)    <45      14-20     12-19
Outlet Burden        (mg/m3NTP)   <30       5.1       7.5
                     TABLE 1 - PLANT PERFORMANCE
                                    5-11

-------
en

i—'
INO
                                OUTLET
                                PLENUM
                               FILTER BAGS

                               DIRTY GAS
                               FLOW

                            [//S-INLET
                            /A MANIFOLD

                               BAFFLE
            C ROSS SECTION
            THROUGH TWO CELLS
HAL F  PLAN VIEW
                      Figure J  - ARRANGEMENT  OF  FLAKT  FABRIC  FILTER. UNIT /f,MUNMORAH

-------
                     A SURVEY OF THE PERFORMANCE OF
                          PULSE-JET BAGHOUSES FOR
                    APPLICATION TO COAL-FIRED BOILERS,
                                   WORLDWIDE
                                 Victor H. Belba, P.E.
                                     Consultant
                                  4758 Edison Lane
                              Boulder, Colorado  80301

                                    Theron Grubb
                         Grubb Filtration Testing Services, Inc.
                                8006 Route 130 North
                              Delran, New Jersey 08075

                                   Ramsay Chang
                           Electric Power Research Institute
                                3412 Hillview Avenue
                              Palo Alto, California  94304
ABSTRACT
Pulse-jet fabric filters (PJFFs) are widely used in U.S. industrial boiler applications and in
utility and industrial boilers abroad.  Their smaller size and reduced cost relative to more
conventional baghouses make PJFFs appear to be a particularly attractive particulate control
option for utility and industrial boilers: for both new plants as well as retrofits and for the range
of boiler type from stoker- to pulverized-fired boilers and fluidized bed combustors.

This paper summarizes and presents preliminary results of a survey funded by the Electric
Power Research Institute and the Canadian Electric  Association  to  characterize the
performance of and operating experiences with PJFFs applied to coal-fired  boilers.   The
survey involved site visits to interview technical and plant personnel involved in the design,
installation and  day-to-day operation  of  PJFFs.  In this way,  actual field experiences
worldwide with PJFF performance in terms of outlet emissions and pressure drop, different
types of pulse-jet cleaning methods and fabrics, startup/shutdown and O&M procedures, and
the myriad design details were compared.
                                          6-1

-------
                     A SURVEY OF THE PERFORMANCE OF
                         PULSE-JET  BAGHOUSES FOR
                    APPLICATION TO COAL-FIRED BOILERS,
                                   WORLDWIDE
INTRODUCTION

Pulse-jet fabric filters (PJFFs) are widely used in U.S. industrial boiler applications and in
utility and industrial boilers abroad.  Due to their smaller size and reduced cost relative to
more conventional fabric filters, pulse-jet baghouses appear to be  a new generation of
alternative particulate control options for U.S. utility and industrial boilers; for both new plants
as well as retrofits and for the range of boiler types from small stoker-fired to pulverized coal-
fired boilers,  fluidized bed combustors  and other advanced power generation systems.
Pulse-jet baghouses have been applied downstream of spray dryer absorbers and dry
sorbent injection as well. Also, despite the use of gas-to-cloth ratios which are typically larger
for PJFFs relative to more conventional reverse-gas fabric filters, PJFFs have demonstrated
similar  operating pressure drops.

The first pulse-jet fabric filter was installed over thirty years ago,  in the mid-1950s.  The first
PJFFs  served primarily as  process equipment to collect valuable product  from pulverizing
mills (1).  It wasn't until the early to mid-1970's, twenty years later, that pulse-jet technology
was applied to the collection of flyash from coal-fired boilers.  Since that time, the use of and
design details incorporated  in PJFFs used to collect flyash have evolved in distinctly different
directions throughout the world.  In the United States,  PJFFs have been  perceived primarily
as a "low-cost"  alternative for smaller industrial boilers as opposed to  the reverse-gas fabric
filters more conventionally applied to utility boilers.  Whereas, in Canada, Europe, Australia
and Japan, PJFFs have seen a wider application to both small  industrial boilers as  well as
larger-sized utility boilers.

The primary evolutionary differences involve the type of pulse-cleaning method used and the
types of fabrics used.  Also, other design details vary from continent to  continent.  What
differences could such trends make  with respect to the applicability  and  maintainability of
PJFFs for boiler applications?  Particularly of interest is whether  these  design  details
                                          6-2

-------
significantly affect baghouse performance in terms of pressure drop, outlet emissions,  and
bag life.

Thus, the Electric Power Research Institute (EPRI) and the  Canadian  Electric Association
(CEA) have funded this survey to obtain information on the performance and operating  and
maintenance histories of PJFF installations worldwide.  Actual  PJFF installations were
identified and contacted in the U.S., Canada, Australia,  Japan and Europe to allow direct site
visits and interviews with operating plant personnel. Thereby, the performance of PJFFs in
terms of outlet emissions, pressure drop and bag life and any significant issues and problem
areas could be identified and/or verified  by up-to-date information.

Visited were sites which represent the gamut of boiler type in use today at utility and larger
industrial boiler installations, primarily: pulverized coal, stoker  and fluidized  bed combustors.
Also, sites were selected to represent the range of cleaning methods and the types  of fabrics
from woven fiberglass to the various felts in use today.  Additionally, PJFFs downstream of
spray dryer absorbers for SO2 control,  catalytic deNox systems and in use with dry sorbent
injection were visited to verify the  viability of pulse-jet technology in combination with such
technologies.

There are well over 300 installations of pulse-jet baghouses applied to coal-fired  boilers
worldwide (2). This survey involved direct site visits and/or interviews with personnel directly
involved with PJFF installations at over thirty utility and industrial plants.  The information  and
data gathered represents PJFFs installed on over 70 separate units.

This paper presents preliminary findings of this Pulse-Jet Baghouse  User's Survey.  More
detailed results will be available under the final report for this EPRI project, RP1129-21.

MAJOR ISSUES AND TRENDS

The  major issues and trends facing  the application of pulse-jet baghouses  today  can be
distilled into the following general categories:
           cleaning method and suitability for long bags
           fabrics and suitability of woven fiberglass versus the various synthetic
           felts available today
     •     off-line versus on-line cleaning
     •     maintainability and suitability of  pulse-jet baghouses for large boiler
           installations
                                          6-3

-------
          air-to-cloth ratio
          performance in terms of outlet emissions, pressure drop and bag life.

Most of these variables are interrelated and it is difficult to lump them into distinct categories
without consideration of the other variables. Therefore, the discussion that follows is only
loosely organized into the above categories and may jump around.

Cleaning Method and Long Bags

Pulse-jet cleaning  methods have evolved into three basic types which  can be generally
characterized in terms of the pressure and volume of the pulse air used.  These methods are
high-pressure/low-volume (HP/LV), medium pressure and volume (MP/MV)  and low-
pressure/high-volume (LP/HV) pulsing.   The details of these  cleaning  methods and the
purported advantages of each are well documented in the literature.  However, for general
classification purposes, the ranges of pulse pressures used by each method are summarized
in the following:

                                            Typical Pulse Pressure
      Cleaning Method                       psiq                bar
          HP/LV                          50 to 100         3.4 to 6.9
          MP/MV                         20 to 30          1.4 to 2.1
          LP/HV                          7.5 to 10          0.5 to 0.7

The original  PJFF design used the HP/LV cleaning method; and traditionally, most pulse-jet
applications in  the United States  have been primarily of the  HP/LV type.  Whereas,  in
Canada, Australia and Europe, the predominant cleaning modes on larger  boilers have been
the MP/MV and LP/HV.

The LP/HV  and MP/MV pulsing modes have  been more recently developed and their
designers claim some advantages over the original HP/LV mode.  The LP/HV and MP/MV
modes are said to consume less energy, yet are able to deliver  more  of that energy into the
pulse.  The alleged stronger pulse is said to penetrate further  into  the bags before
dissipating; and thus, the perception that longer bags of up to 20 feet or more can be applied
effectively on LP/HV and IP/IV collector.  Whereas, some believe that  bag  lengths should be
limited to 12  to 14 feet on HP/LV pulse-jets.

Is such a perceived disadvantage of the HP/LV indeed true? This is a critical question for
larger PJFF  installations since longer bags mean fewer bags to provide the same air-to-cloth
ratio.  Fewer bags mean fewer parts such as cages,  pulse pipes, diaphragm and solenoid
                                        6-4

-------
valves and  bags that require replacement.   Additionally, and  particularly important for
retrofits, the  prospect of smaller plan areas if "long" bags are viable is appealing.

So far in our analysis of the data gathered on the site visits, we are not able to say definitively
that one cleaning method is superior to the others.  This is due to the numerous variables
which  are interdependent.   Attempts are still  underway to normalize the variables  to
determine if  it can be proven with this database as to whether or not one method is superior.
We can say that we have visited three sites which use HP/LV  pulse jets and bags with
lengths of approximately 19 and 20 feet.  The operators of these baghouses are pleased with
them, have  achieved reasonable bag lives and  pressure drops and, in one  case,  have
maintained exceptionally low outlet emission levels.

This suggests then that so far all three cleaning methods appear to  be  viable  contenders.
However, in  any case, anyone contemplating the installation of a PJFF with long  bags should
ensure that  the bidders  and manufacturer finally selected have current field experience
and/or at  least extensive pilot investigations  to  prove the  applicability of their  cleaning
method to long bags with the type of fabrics anticipated and treating dust of similar properties
and difficulty.  In other words, successful experience with long bags collecting flyash from, for
example,  a stoker-fired  boiler  may  not be sufficient experience to justify  that OEM's
experience on the more-difficult ashes typical of PC-fired boilers or fluidized bed combustors.
In general, felted fabrics have been preferred for pulse-jet applications since typically the
fabric is relied upon more for filtering than the filter cake, as in the case of more conventional,
less-energetic cleaning  methods such as reverse-gas and shake-and-deflate fabric filters.
For early applications of PJFFs in the United States, today's felts which can withstand flue
gas conditions were not available.  This prompted an evolution  in the U.S.  toward woven
fiberglass bags, primarily in HP/LV units.  Off-line  cleaning, in  which compartments  are
isolated from the flue gas stream for cleaning, was developed in part to accommodate the
more fragile woven fiberglass fabric; and perhaps this evolution was accompanied somewhat
by a trend towards more conservative air-to-cloth ratios.

Fiberglass fabrics were  appealing due to their ability to withstand gas temperatures of up to
500° which are approached by many industrial boilers in the U.S. and more forgiving of upset
temperatures for larger boilers employing regenerative-type air heaters.  Note that one of the
original  advantages of pulse-jet cleaning over the more conventional fabric filter designs is
the ability to pulse and thus clean rows of bags in a compartment while the remainder of bags
in that compartment are  still "on-line," operating in the filtering mode.
                                         6-5

-------
Since the early applications of PJFFs, Nomex™ and Teflon™ fabrics have become available.
Also,  more recently, felts made of homopolymer acrylic (Dralon-T™) and  polyphenylene
sulfide (PPS or Ryton™) have been used successfully in coal-fired boiler applications.

Experience  in Canada,  Australia and Europe has been primarily with these various felted
fabrics. Homopolymer acrylic has proved to be a very good fabric for applications at less
than 285°F;  such as, downstream of spray dryers or of boilers firing low sulfur coal and which
have no risk of high excursion temperatures. The Ryton™, Daytex™ and Nomex™ felts must
be operated at temperatures of less than about 400°F and felted Teflon™ can operate at
somewhat higher temperatures, perhaps of up to 475°F.

Fiberglass Versus The Felts.  Recently in the United States, and paralleling the success of
felts in other countries, there has been a trend away from fiberglass towards felted fabrics for
PJFF applications.  Many installations that initially experienced difficulties with fiberglass, for
numerous reasons, changed to felts.  Also, it seems that many new installations of PJFFs on
industrial boilers are more frequently being specified with felts as the preferred fabric.  Is such
a trend justified?

Qualitatively, the results of this study  indicate that observing proper design and construction
tolerances, woven fiberglass bags can work and provide reasonable bag lives and pressure
drops while still maintaining very low outlet emissions comparable  with those achieved by
felts.  However, it is generally conceded in the industry that fiberglass is less forgiving than
the synthetic felts.  Care must be taken in ensuring proper bag and cage fit and exceptional
care  must  be exercised  in  installing the  filter  bags.   Also, fiberglass  bags  are more
susceptible to abrasion and related design and construction problems which may exacerbate
abrasion such as  pulse pipe misalignment, flue gas maldistribution and hopper dust removal
problems.   Nevertheless, and as in the case of selecting any fabric, the proper design
parameters  must be considered in selecting woven glass or any felt.

For example, Nomex™ felt has been applied with varying degrees of success. Barring major
installation and fabrication errors, failure of Nomex™ often can be traced to not observing this
fiber's lack of tolerance of moisture and acid attack at high temperature. However, it certainly
seems a suitable fabric for application to low sulfur flue gas, such  as after fluidized bed
boilers and even spray dryers.

An interesting alternative not pursued in general on this continent is homopolymer acrylic or
Dralon-T™.  With great success, this low temperature felt is almost  exclusively  applied in
Australia.  This fabric has been useful by virtue of the low sulfur Australian coals and the use
of air tempering systems which admit  ambient air to cool the flue gas.  Filter bags made of 16
                                         6-6

-------
oz./sq yd Dralon-T™ can cost as little as half that of bags made  of 22 oz. fiberglass (1).
Therefore, this would appear to be an interesting fabric to pursue in cases where acid content
and other properties of the flue gas are such to allow low flue gas operating temperatures.  Of
course, the economics of cooling the flue gas either via a tempering/dilution system or a gas-
to-air heat exchanger must be weighed against the  savings in bag cost.

Felts of Ryton™ or PPS  exhibit exceptional resistance  to high gas temperatures and
corrosive gases; and thus are finding a wider application to coal-fired boilers despite a cost
difference of at least 50%  over that of woven glass bags for a  16  oz. Ryton™ cloth.  Note
heavier weights of  Ryton™ of 22 to 27 oz. are often applied and may be necessary on
domestic felts to ensure the achieving of low emission levels.

These fabrics  and their  effects  on PJFF performance in  terms of outlet emission, pressure
drop and bag life are discussed further below.

Maintainability/Suitability.  Many contemplating the application of a PJFF to a large utility or
industrial  boiler  often  have  expressed concern about the  prospect  of exceptional
maintenance  problems due to the greater number of parts in a PJFF compared to more-
conventional  fabric filters.  The excessive and  constant failure of diaphragm  valves,
solenoids,  and cages has been  a major fear; and thus, an area of investigation for this
survey.

The general concensus of  this survey is that pulse-jet baghouses are suitable particulate
control collectors for the range of coal-fired boilers  from small industrial boilers to large  utility
boilers.  Currently, there exist numerous, very large installations handling flue gas volumes of
up to 1.1 x 106 acfm.  Few, if any, installations visited reported excessive diaphragm valve
and solenoid failures. In the few installations where the failure rate was perceived to be a
problem, diaphragm failure was usually resolved by substituting materials.  In many cases
with an excessive failure rate of diaphragms, the cause could be traced to either insufficient
cooling of the pulse air  or excessive condensation  of  moisture  in  excessively  long
compressed air lines from a remote plant-air system.

Another possible cause  suspected in one installation was acid attack from flue gas backing
up into the pulse pipes  during lengthy  periods in  which  cleaning was not  taking place.  It
appears that the installations  in which special care was taken to individually design and
supply pulse  air  systems which  are devoted to  the baghouse did  not suffer excessive
diaphragm failures.  Provided future systems are  adequately designed, massive failures of
diaphragms should not be a problem even in large  installations.

Several years ago,  it was commonly believed that one  could expect to throw away one's
                                         6-7

-------
complement of cages almost with every bag changeout. In fact, in early cost comparisons of
pulse-jet baghouses versus reverse gas or shake-and-deflate baghouses, many would
assume that the cages were replaced with every bag changeout.  This survey indicates that
as long as high quality cages are used, excessive loss should not be expected. This concern
perhaps stems from earlier days in which cheaper cages of thin gauge wire commonly were
used. Also, in the  early days  of using woven fiberglass, a wire  mesh cage was common.
Such early cages of thin gauge wire or mesh were easily deformed and broken and difficult
to reuse.  However, it was found that well-made cages, of heavy-gauge wire of from 3 to 4
mm, have been quite serviceable and exceptionally reusable.

Materials of construction of the cages must be considered  and made compatible  with the flue
gas  conditions.   Often, plain carbon  steel cages have proven suitable; whereas,  an
exceptionally moist or corrosive flue  gas may require the  use of a protective coating on the
cages.  Epoxy-type coatings  have been effective at  lower flue  gas temperatures typical
downstream of spray dryer absorbers.  Special  note should  be  taken that  galvanized
coatings may be ineffective and perhaps even counterproductive in flue gases which are
corrosive and with  high chloride contents. Many horror-stories of premature and excessive
cage failure from the past could have been avoided by use of proper materials.

Thus, it can be seen that the fears of a constant maintenance headache in tracking down and
replacing continually-failing components such as solenoids and diaphragms as  well as
damaged cages are unfounded.  Relatively large utility installations of PJFFs are currently in
place and have not demonstrated such problems; provided proper components are used and
design details are observed.

Currently, many U.S. utilities are contemplating plant-life extensions of older plants which
use small ESPs.  Also, coal switching brought about by potential acid rain legislation or other
reasons  and PM10 regulations may make smaller,  existing ESPs unsuitable for future use.
Also, hot-side  ESPs are notorious for their difficulties in collecting ash from low-sulfur, low-
sodium  coals.   PJFFs have been retrofitted into existing cold-side ESP casings in  utility
boilers abroad and on industrial process applications in this country.  In  general, these
retrofits have  performed quite well and promise  to be  a retrofit  option where sufficient
precipitator plan area  is available  and the extensive outage times  would not prove
uneconomical.

The  Electricity  Commission of  New South Wales pioneered the retrofit of PJFFs  into existing
ESP casings at their Wangi Station in 1976.  This innovative installation was undertaken out
of necessity to  temporarily extend the lives of three aging units while awaiting  new capacity to
come on line during a short fall of generating capacity on their system. This HP/LV retrofit

-------
was, in general, successful at extending the lives  of these  plants; however, perceived
problems with higher pressure drop, shorter component life, and shorter bag life compared to
adjacent retrofits of shaker baghouses led  the  ECNSW to depend more on that type  of
technology for the bulk of their installations in which there was sufficient room.

Since then, successful retrofits at the Kyndby Power Station in Denmark, at the installation of
the Ostereichische Draukraftwerke in St.  Andra, Austria as well as the quite famous and
innovative  retrofits of PJFFs into the Munmorah  Station, Units 3 and 4 of  the ECNSW
demonstrate the viability  of such retrofits.  The major concerns always are availability  of
sufficient plant area to allow retrofit of adequate air-to-cloth ratios  and the cost of the lengthy
outage times required. Also, usually long bags are essential to allow such retrofits to contain
sufficient cloth area.
Air-to-Cloth Ratio

An early perceived benefit of pulse-jet collectors, due to their higher energy cleaning method,
was the use of exceptionally high air-to-cloth ratios of from 5  to 7 fpm.  Since these early
days, however,  there  has been a healthy  trend towards lower air-to-cloth ratios to ensure
reasonable pressure drops,  less frequent cleaning and thus longer bag lives and lower outlet
emissions.  Figure 1  is a plot of design air-to-cloth ratio versus startup date for the population
of PJFF sites visited.  With a little imagination, a trend towards lower air-to-cloth  ratios can be
identified on this plot.  In  general, the larger air-to-cloth ratios still being installed as  of late
involve felted applications in Europe.  Note  that the general trend for a conservative design is
towards air-to-cloth  ratios of about 4 or less fpm for felted  bags. Currently, it is generally
conceded that air-to-cloth ratios for woven fiberglass applications should be closer to 3 fpm
or less  to allow off-line cleaning, less frequent cleaning, to ensure long bag life and  to
accommodate the higher pressure drops which appear to be typical of glass bags.
PERFORMANCE

Ultimately,  how well  PJFFs  perform in terms of pressure drop,  outlet emissions and bag life
determine whether or not this technology will be suitable for general application to large
industrial and utility  boilers. The following summarizes preliminary findings of this survey
with respect to PJFF  performance.
Pressure Drop

Consistent with previous literature, this survey confirms that pressure drop characteristics of
well designed and built PJFFs are reasonable. Figure 2 is a plot of flange-to-flange pressure
drop versus air-to-cloth ratio.  In general, this plot indicates reasonable total flange-to-flange
pressure drops which can be tolerated by boiler installations.  Note first of  all that in most
                                          6-9

-------
cases the lower air-to-cloth ratios on the Figure are indicative of those plants operating at
lower loads than design at the time of the visit.  Note also that these pressure drops represent
actual APs observed at the time of the site visit. The air-to-cloth ratios are based on actual
observed flue  gas volumes which  were  estimated  stoichiometrically based  on boiler
operating parameters and flue gas conditions at the time of the visit.

Looking  at Figure  2 with a little  imagination,  one  can  picture a line drawn from  the origin
through the maximum values of pressure drop. It is interesting to  note that most of these
points  represent PJFFs which are installed on fluidized bed boilers  and/or using woven
glass.  This is consistent with general opinions in the industry that the use of heavier woven
glass bags can result in somewhat higher pressure drops.  Also, general observations  are
that fluidized bed boilers can generate a particulate which is difficult to remove from the cloth
and thus results in  higher pressure drops.

Coincidentally,  many of these  PJFFs also are of  the HP/LV cleaning  method.  This might
suggest  an inferiority of this cleaning  method;  however, these data points are joined closely
by  numerous points representing the other two cleaning methods.   Also, indicated on the
Figure are points  A and B which help define the lower range  of  the plot.  These points
represent  HP/LV  installations which incidentally use synthetic felts.  Although, recent
discussions with personnel at  plant A indicate that some nine months after our site visit,
pressure drop  and opacity levels have been rising alarmingly  at  this PJFF bagged with
Ryton™ felt.

This problem raises the issue of maintainability of synthetic felts.  Several of the installations
which were visited have washed felted bags with success.  One facility has had experience
with both Nomex™ and Ryton™  Their opinion is that the Ryton™ fabric perhaps survived the
cleaning ordeal with somewhat  less stress than  the Nomex™  cloth.   Nevertheless, their
technique  of using a water lance was a  successful procedure in restoring reasonable
pressure drops after episodes of blinded bags due to tubeleaks.  The lances were supplied
by  normal water from the mains  and  sprayed  water from the insides of the bags in a radial
pattern.

Also, another facility has successfully washed Ryton™ bags by means of a gentle washing
cycle in industrial washing machines.  These bags subsequently were hung to dry.  Testing of
the washed  fabric revealed an  increase  in  permeability  and  a  25%  increase in outlet
emissions; although, the original emissions were quite  low and the result was that the plant
was still  in compliance with its limitations.  Reasonable pressure  drops  were  restored to  this
facility.
                                         6-10

-------
Teflon™ felt has proven to be an incredibly durable fabric and can withstand numerous
washings.  One  installation has washed their Teflon™ bags up to six times to recover from
episodes of excessive blinding and bleedthrough.  However, their calculations indicated that
they would have to achieve many more washings and years of life for the bags to justify
purchasing  more Teflon™ due to its excessive cost relative to other applicable  fabrics.
Nevertheless, Teflon™ should still find application to unique difficult applications  and,  of
course, will  continue to play a major role as a treatment for fibers to prevent abrasion and
corrosion and will be applied as membranes.

Also, another site has successfully hand-cleaned woven glass bags in place  by use of a
compressed air lance. As a result of this method, this PJFF was restored to original pressure
drops after experiencing bag blinding due to excessive slip of ammonia past their catalytic
deNox system upstream of the baghouse.  Despite fiberglass's noted fragility, these bags
achieved a  minimum life of 1-1/2 years with up to two years expected.  The bags were
changed out after 1-1/2 years due only to the requirements of a scheduled maintenance
outage.
Tubesheet  Pressure Drop

Figure 3 is  a plot of tubesheet pressure drop versus air-to-cloth ratio for the  sites visited.  As
in the case of Figure 2 above which represents flange-to-flange  pressure drops, these data
points also  represent actual observed tubesheet pressure drops  plotted against actual
operating air-to-cloth ratio.  Likewise, the air-to-cloth ratios were calculated based on the
cloth in service at the time  and flue gas volume calculated stoichiometrically.

These points display a pattern consistent with the flange-to-flange pressure drops on Figure
2.  The same grouping of PJFFs installed on fluidized  bed boilers and/or using woven
fiberglass still applies to the points on Figure 3.

Note that Figure 4 is essentially the same plot of tubesheet pressure  drop versus air-to-cloth
ratio; however, superimposed are shaded areas which represent ranges of pressure drop
and air-to-cloth ratio for full-scale and pilot reverse-gas and shake-and-deflate baghouses.
Also presented  is a projection of pressure drops  versus  air-to-cloth ratios for pulse-jet
baghouses.  Note that the data gathered by this survey tends to fall within the originally
projected performance regime for pulse-jet baghouses but does spread over and border  on
the regime  thought to be  exclusive to shake-and-deflate and reverse-gas  baghouses.
However, note that the PJFFs whose performance does approach that of the reverse-gas and
shake-and-deflate baghouses tend to be the applications to fluidized bed boilers which are
thought to be more difficult.
                                         6-11

-------
Particulate Emissions

Figure 5 summarizes the range of particulate emissions levels measured  by the various
facilities visited and groups them according to basic fabric type.  Note that the category of
"woven glass" represents lighter-weight woven fiberglass weighing 16 oz./sq. yd. or less.
Note that this bar represents the range of stack tests for only four installations; however, this
data tends to corroborate current  knowledge that  lighterweight woven glass is  not a
particularly  suitable fabric for a facility which must consistently achieve  New Source
Performance Standards or lower.  On the other  hand, the heavier weight woven fiberglass
bags of 22 oz./sq. yd. or higher are quite capable of achieving lower emission  levels.

Once again, a small population represents this range  of data, four sites.  However, most of
the data values hover towards the low end of the scale and one of these lower data  points
represent a test after one year's worth of service.  Incidentally, as of the date  of that test, the
particular facility had not lost one filter bag.  Presumably, the other two sites at the low  end of
the range were tested early in their lives.  One of the sites easily achieved  1-1/2 years of bag
life or more with a failure of only two  bags.  This site used woven glass with a weight of 27
oz./sq. yd.

By contrast,  the third site at this low emission level is required to remain at this emission level
and zero visible emissions.  This plant has been faced with a constant changeout weekly of
at least a few bags to ensure maintaining compliance.  This bag changeout requirement, was
exacerbated by a severe abrasion of the bags at their bottoms.  Apparently, this abrasion has
been traced to primarily poor flow distribution of  gas and  dust into the hopper entry of each
compartment.  Direct impingement of the dust on the bags results in severe abrasion and
forming of large wear holes just above the bags' "bumper" cuffs. Additional  abrasion might
be due to filling of  the  hoppers with ash and  subsequent recirculation  of  ash and
impingement on the bag bottoms due to temporarily disturbed flow patterns.

Thus, as can be seen from Figure 5, heavy weight woven fiberglass can be competitive with
felts in terms of outlet emission levels.  However, as always, the cheaper cost of the fabric
must be weighed  against other mitigating factors such as lower air-to-cloth ratios which may
be required, the relative fragility of the fabric and its  exceptional sensitivity  to mishandling
and abrasion.

A disappointment of the survey was the inability to schedule  within budget and schedule
constraints visits to sites currently using felted fiberglass, Tefaire™,  P84, fabrics with
GoreTex™ membrane or  other such exotic fabrics.  As can  be seen,  Figure 5 contains
emission data from one of two sites which previously had been early users of  felted
                                         6-12

-------
fiberglass.  As can be seen, emissions from this vintage installation were rather high.
Bag Life

As can be  seen,  bag life can be dependent upon many factors including outlet emission
limitation and the design  details incorporated in the installation.  Ultra  low emission
limitations  can require continual changeout of bags,  especially if design and operating
problems accelerate bag wear and failure.

As indicated up above, some fiberglass installations have achieved 1-1/2 to 2 years of bag
life. One of the installations using 22 oz. woven glass had experienced a 5-1/2 year bag life
to date with only two small failures. This is remarkable and rivals the bag lives hoped for on
some of the better installations of felt.  Part of this success may be due to the use of 40-wire
cages to support the woven glass bags and an exceptionally low air-to-cloth ratio of close to
2fpm.

In general, however, felts  are considered to exhibit longer bag lives than most fiberglass
installations.   Synthetic felts are much more resilient and  capable of withstanding the
constant flexing during pulse cleaning and perhaps  are  less  sensitive to abrasion than
woven glass.  Indeed,  Ryton™  and similar felts under  different names can exhibit
exceptionally  long bag lives. One installation had experienced lives of 6 or 7 years on its first
complement of bags made  of felted PPS. However, subsequent bag lives after this first set
were  considerably reduced  and now 3-1/2 years may be more typical. This is due to several
reasons, not the least of which is the continual operation of this plant at higher loads and thus
exceptionally high air-to-cloth ratios.  This  baghouse  now is  in a continual cleaning cycle
which has taken its toll on bag life. Also, control of procurement of replacement filter bags
was turned over from the engineering and operating personnel to the purchasing function of
this company. This,  along with less attention to maintenance and bag installation, also may
have contributed to shorter bag life.

Nevertheless, it may not be unreasonable to expect Ryton™ to maintain its mechanical
properties  and survive in a flue gas environment for four years  or more  based on  many
plants' experiences.  However, there is a concern about the ability of Ryton™ felts in this
country to  withstand bleedthrough of the particulate and subsequent blinding and excess
emissions.

And, in general, comments  from several sites visits indicate that the quality of domestic felts
may be sub-par compared to foreign  products.  This points at a minimum to  a  need for
increased efforts at specification  and quality control  during the bag  and felt  fabrication
process.
                                         6-13

-------
Given the proper flue gas conditions; namely low enough temperatures and sulfur content of
the flue gas, Nomex™ has proven to be a suitable fabric for PJFF applications. Nomex™ felts
have  survived in numerous plants for up to  1-1/2  years with less than  ideal and
recommended flue gas conditions. Two-and-one-half year bag lives and more, possibly up
to 4 years, have been documented downstream of spray dryers and fluidized bed boilers
where sulfur dioxide and trioxide levels are consistently low.  Note that some fluidized bed
boilers have experienced difficulties  of excessive  levels  of SOs  concentrations during
slumped-bed operation.   If such operational upsets are contemplated or  experienced by a
fluidized bed application, effective use of Nomex™ may be precluded.
CONCLUSIONS

This survey has verified that pulse-jet baghouses have been  applied successfully to coal-
fired boilers worldwide.  Although once feared to present a maintenance headache for large
utility and industrial boiler applications, well designed  and built PJFF  installations  have
proven such concerns to be unfounded.

However, to ensure adequate performance and long life, design, fabrication and construction
details cannot be ignored.  Poor gas flow distribution into compartments, misaligned and
poorly made pulse pipes, cheap construction and components, and misapplication of fabrics
are the contributing factors to those  installations  which  have exhibited  disappointing
performance in terms of pressure drop, outlet emissions and short bag life.

Pulse-jet baghouses should not be thought of as a cheap alternative particulate collector for
smaller boilers.  However, an installation which works well due to high quality design and
construction still holds the promise of cost savings due to their smaller size relative to more
conventional collectors.   To achieve today's more stringent emission requirements, generally
more conservative air-to-cloth ratios than applied in the past are required. Whereas, at one
time,  an air-to-cloth ratio of 5 to 7 fpm was considered adequate for a PJFF; these days, air-
to-cloth ratios approaching 4 fpm or less should be considered for a new installation  using
felted cloth.  Lower air-to-cloth ratios in the range of 3 fpm or lower should be considered for
those installations which plan to use woven fiberglass.

ACKNOWLEDGMENTS

This paper was prepared based on preliminary findings of work funded by the Electric Power
Research Institute and the Canadian Electric Association.

We heartily thank the Electricity Commission of New South Wales and the Queensland
                                         6-14

-------
Electricity Commission in Australia for their sharing of valuable expertise and time.  We thank
Mr. Brian Thicke of Alberta Power for his providing invaluable information, suggestions and
guidance.  We thank the many original  equipment manufacturers who provided updated
installation lists and suggestions. We gratefully acknowledge the exceptional assistance
provided by the Fla'kt  and  Howden  companies  worldwide,  Environmental  Elements
Corporation and Wheelabrator Air Pollution Control. These companies generously devoted
time and resources to provide assistance in identifying and providing introductions to the
appropriate people for seeking approvals for plant site visits, making contacts and even in
setting up some site visits. And, finally, we graciously thank the owners and operators of the
plants which allowed site  visits and their  innovative plant operating personnel who
contributed considerable amounts of time, effort and expertise.

REFERENCES
1.    William Gregg, "Pulse-Jet Dust Collectors for Utility Boiler Emission Control." Paper
      presented at the Joint ASME/IEEE Power Generation Conference, Dallas, Texas,
      October 22-26, 1989.
2.    "Pulse-Jet Fabric Filters for Coal-Fired Utility and Industrial Boilers." EPRI CS-5396s;
      Research Project 1129-8.  Prepared by Southern Research Institute.
                                         6-15

-------
     8.00
     7.00
  E
  £•  6.00
<  5-°°
tr

H  4.00
O
_j
O
6  3.00
h-
cc

<  2.00
     1.00
           0.02 m/sec
        1975      1977      1980       1983      1986

                               YEAR OF START UP
                                                        1988
1991
           Figure 1.  Design Air-to-CIoth Ratio Vs. Start-Up Date
O
 10.00


  9.00


  8.00


  7.00


  6.00


  5.00


  4.00


  3.00


  2.00


  1.00


  0.00
              KPa
                                              B
                                       0.02 m/sec
       °-°°      1-00     2.00     3.00      4.00      5.00

                            AIR-TO-CLOTH  RATIO  (fpm)
                                                          6.00
                                                                   7.00
         Figure 2.  Flange-to-Flange AP Vs Air-to-CIoth Ratio
                                    6-16

-------



5
c
D.
<



1 U.UU
9.00
8.00
7.00
6.00
5.00
4.00
3.00
2.00
1.00
0.00
.
•
•
•
•
•
•I.OKPa
• •
•'-".-•
B
•
0.02 m/sec
  10
   9
   8
__  7
9  6
Z
c  5
D.  4
   3
   2
   1
   0
         123456
                  AIR-TO-CLOTH RATIO  (fpm)

     Figure 3. Tubesheet AP Vs Air-to-Cloth Ratio
  REVERSE-GAS
SONIC ASSIST
                                    SHAKE/DEFLATE
                                              PULSE-JET
     I         2        3        4        5        I
                 AIR-TO-CLOTH RATIO  (fpm)

      Figure 4. Tubesheet AP Vs Air-to-Cloth Ratio
                          6-17

-------
m
2
2

m
_i

CO

O
(/)
C/)

LU
  0.1
                                    O.O3  LB/MMBtu
                                                 i
0.001
        Ryton   Nomex Dralon  Teflon  Woven  Woven  Felted
                                       Glass   Glass   glass
                                       >22oz

              Figure 5. Particulate Emissions Vs Fabric
                              6-18

-------
                           RETROFIT OF FABRIC FILTERS
                                TO POWER BOILERS
                                  H.F.  Johnson
                        Howden Environmental Systems  Inc
                             One Westinghouse Plaza
                              Hyde Park,  Ma. 02136
ABSTRACT
With increasing frequency, coal  fired  utility  boilers are being load  limited  due
to unacceptably  high  particulate  emissions caused  by  either  poorly  performing
electrostatic  precipitators  (ESP)  which  are  no  longer able  to  meet  original
guaranteed emission levels or  more  stringent particulate emission limits  imposed
by regulatory authorities on existing or "life  extended"  plant.

Rather than demolish and replace existing  particulate control  equipment or provide
supplementary  ESP,  growing  interest has  been  shown  in retrofitting  pulse  jet
fabric filters into  existing ESP casings.  Modern low pressure  pulse  jet  fabric
                                                                                2
filters (FF)  operate at air to cloth ratios in  the order  of  0.02 m/s  (4 acfm/ft  )
for coal  fired boiler applications and are capable of meeting  NSPS  requirements  of
0.03 Ibm/MBTU.

Designs prepared  for completed contracts and current  bids have shown  that  for  ESP
with a specific  collecting  area  above  50  (250)*  a  pulse  jet  fabric  filter  can
generally fit inside the ESP casing.

This paper will  give  operating  and design  data on  retrofit  FF  for p.c.  fired
boilers and detail  some  of  the  space  and  cost  savings  that can  be  realised  by
adopting   this  approach.  Operating   and  design  data  for  new  pulse   jet  FF  on
fluidized bed boilers  up to 60 MW will also be  given.
                    23        2
'•Units for SCA are m /m /s or ft /1000 acfm
                                        7-1

-------
INTRODUCTION





With increasing frequency, coal fired  utility  boilers are being  load  limited  due


to unacceptably high particulate  emissions  caused by poorly performing  ESP  which


are no longer able to meet  original  guaranteed emission levels or  more  stringent


particulate emissions limits imposed by  regulatory authorities on  existing  plant


or "life extended" plant.





Rather than  demolish and  replace  existing particulate control equipment  and flue-


gas conditioning equipment or provide supplementary ESP,  growing  interest has been


shown  in  retrofitting  pulse  jet  fabric   filters  into the  existing ESP  casings.


Modern low  pressure  pulse  jet  FF  operate at air  to  cloth  ratios  in the  order  of


0.02 - 0.023 m/s (4 - 4.5  acfm/ft  )  for coal fired boiler applications and  occupy


approximately 50% of the  plan  area  of  more traditional  low  ratio reverse air  or


shaker-deflate type  filters. Several designs  studies carried  out by Howden  have


generally shown that for ESPs with a  specific  collecting  area (SCA)  above 50(250)*


that  the  pulse jet  FF  can  generally  fit  inside the  existing  ESP  casing  with


acceptable air to cloth  ratios.





Electric utilities in Australia,  Africa and Canada have replaced  or  are  currently


replacing approximately  4000 MW of poorly performing  ESP  with pulse  jet FF.





The attractions of retrofitting pulse  jet FF  rather  than the  alternatives  is  the


relative low cost of the  conversion, the considerable savings in outage  time  for


the  power  generating   plant,  the  ability  of  the   retrofitted   FF  to  achieve


comparable  reliability  and collecting  efficiency  to  new  low ratio  FF  designs  and


the possibility of combining particulate  and SO. removal  with the  FF by installing


induct dry  sorbent injection.
MUNMORAH POWER STATION 350MW FABRIC FILTER RETROFIT





Following  earlier  successful retrofits The  Electricity Commission  of New  South


Wales  (Elcom)  decided  to  apply  pulse jet FF  technology to  2  x 350MW pulverized


coal fired boilers as part of a  life extension program at Munmorah  Power Station


in Australia.  Following extensive  pilot  plant trials the contract  for Unit  No  3


was awarded to Howden.




                     23         2
* Units for SCA are m /m /s or ft /1000 acfm.
                                        7-2

-------
The conversion of Unit No  3  ESP will be based  on  a  design gas flow of  528  am /s
(1,118,620  acfm)  with a maximum  temperature of  155 C  (310 F) .  Air  infiltration
will be used to control the flue gas temperature to 130  C (266 F)  to allow the use
of  acrylic  bags.  The  inlet  dust  burden is  in  the range  of 25-40g/Nm   (11-17
      3
gr/Nft )  with  a  median  particle  size of  5-30  microns.  Allowable  pressure  drop
across the fabric filter  is 2.5 kPa (10 in.Wg) with an air to cloth ratio  of  0.02
               2                                                 3
tn/s (4 acfm/ft ). The specified emission guarantee  is  50 mg/Nm  (0.05  Ibm/MBTU)
although much lower  emissions and a clear stack were  indicated by the  pilot  plant
results.
The Howden  design is  based on  minimum modification  of  existing  equipment  and
structures.   Essentially,  the  ESP  casings,   hoppers,   structures,   ducting  and
insulation will be re-used. An internal division wall,  transverse to  the  gas  flow,
will  be  added  to the  casings  to  provide  8  compartments  per  boiler  so   that
maintenance can be carried out in any compartment without  impeding boiler  output.
Normal bag cleaning will be in the on-line  mode. The general arrangement is  shown
in Figure 1.

The upper ESP casing will be modified to incorporate the tube sheet and  clean  gas
outlet which doubles  up  as a penthouse for bag  inspection and  replacement.   Each
compartment will contain 6 modules with 267 bags per module.  Bags will  be 6.1m (20
ft) long and each module is cleaned  by  a  rotating manifold to which cleaning  air
of  80  kPa (12  psig)  is supplied  by a positive  displacement blower.  Inlet  and
outlet ESP  evases will  be modified to cater for  the  new  8  compartment  filter
configuration,  and to incorporate gate type dampers for  compartment  isolation and
protection of the casings against excessive pressure excursions.  Aerodynamic  model
tests were carried out  to  ensure that  flue gas  entering the casings  and  flowing
upwards into the  bag  arrays is distributed evenly  to  avoid bag abrasion  and  to
minimize re-entrainment during on-line cleaning (1).

Design details  of the Munmorah filter are  given in Table 1.

To meet load requirements  the  FF retrofit has  been carried  out  in such a manner
that 1/2 the unit (4 compartments) can be  returned to service prior  to completion
of the retrofit.  The  first 1/2 boiler  and FF is  scheduled to be commissioned  in
February 1990 with the complete unit operational 6 months later.
                                        7-3

-------
ROOIWAL POWER STATION 2x60MW FABRIC FILTER RETROFIT

Deteriorating performance and rapidly increasing maintenance costs of the existing
ESP on this  base  load  plant prompted the  decision to  retrofit  pulse  jet  fabric
filters into the existing ESP casing.

The 4 boilers at Rooiwal are 60 MW p.c.  fired with a design gas flow of 61.2 Nm /s
(130,000  Ncfm)  and  a   design  gas  temperature  of 155 C   (310 F)  with  frequent
temperature surges to 190 C (375 F).  Coals fired were  typically 0.4%S  with  a  high
(  15%) ash content.

                                                     3            3
Inlet burden to the FF was specified to  be 14-24 g/Nm   (6-llgr/Nft ).

Allowable pressure drop across the  FF  was  1.5  kPa  (6  ins wg) .   Dust  collection
efficiency was required to exceed 99.3%  with a corresponding outlet  mass emission
of 100 mg/Nm3 (0.1 Ibm/MBTU).

Guaranteed bag life of 2 years was specified.

Following award of the  contract to Howden  for  design,  supply  and erection  of  a
retrofit FF  for Unit  1 in 1987 a design was  selected  with  12 bundles  of 232,  5.8
metre  (19 foot) bags  to give a maximum air to cloth ratio of 0.02  m/s
 (4 acfrn/ft2).
The existing ESP were of a rigid frame European design with a concrete casing. The
2  existing  isolatable  gaspaths  compartments  were retained with  the  capability of
60% design gas flow through one compartment should one compartment be taken out of
service for maintenance  or  rebagging.  The general arrangement of  the  retrofit FF
is shown in Figure 2.

Ryton    /Ryton    filter bags  with  a  cloth  weight  of  600g/m   (18oz/yd  )  were
selected.
Minimum modifications were made to the existing ESP to  complete  the  conversion to
a  FF.  The  existing concrete casing,   hoppers,  support  structure,  insulation,  ash
disposal system, I.D. fans and inlet and outlet ducting were reused.

 R Ryton - Phillips Petroleum
                                        7-4

-------
Following successful performance of the FF on Unit 1  a second contract  was  awarded

for Unit 2 in 1988.




Performance test results for both Units 1 and 2 carried out  by an  independent  test

contractor on  behalf  of  the  owners are  compared to  the  guaranteed  performance

requirements in Table 2.




Even though  inlet  burden  significantly exceeded specified levels  and  was  in  the
                 3            3
order of  28 g/Nm   (13  gr/Nft  )  collection  efficiencies  were 99.9%  and outlet

emissions in the order of 15 mg/Nm  (0.015 Ibm/MBTU).




Bag life on Unit 1 exceeds 2 years at this stage with the final bag life  yet  to  be

determined. Bags on Unit  2  appear to be well  on  the  way to exceeding the 2  year

guaranteed bag life.
PULSE JET FABRIC FILTER ON FLUIDIZED BED BOILERS




Following an agreement with Mitsubishi Heavy Industries reached in 1987, 5 Howden

RF  design  pulse jet  FF  have been  or are  in  the process  of  being  installed  on

fluidized bed boilers in Japan.




The boilers are  of  Mitsubishi  circulating fluidized bed boiler design  using  sand

as the bed material. Details of  the boilers and associated FF  are  given  in  Table3 .




Although all of  the boilers are nominally  coal  fired significant (up  to  40%  of

heat  input)  amounts  of   oil  firing  and  incineration  of  waste  products  occur.

Several of the  installation use Calcium  Carbonate  for flue gas  desulfurization.

There  have  been  no  noticeable  detrimental  effects  of   these  practices  on  FF


performance over extended periods.




Conservative maximum  air to cloth ratios  between  0.015  and 0.02  m/s  (3   to  4
       2
acfm/ft ) have been selected to  allow for required fuel and operating  flexibility

and the high inlet burdens of up to llOg /Mm  (50 gr/Nft ).



       n                                                             R            R
Nomex     filter bags have been used on one installation with  Ryton      /Ryton

filter bags on the remaining 4 installations.

 R Ryton -  Phillips Petroleum

 R Nomex - Du Pont
                                        7-5

-------
Test results collected by an  independent  test contractor for  the  longest  running


FF at Ide are compared to the  guaranteed performance requirements in Table  4.





Outlet emissions were measured to be 0.007 mg/Nm3 (0.008 Ibm/MBTU)  with collection


efficiency exceeding 99.9%.





Bag  life  at  Ide  commissioned  in early  1988 has  exceeded 2  years  at  the  current


time .
COST, SPACE AND TIME SAVINGS REALISED





Total  capital  requirements*  for  retrofitting  pulse  jet  fabric  filters  to  ESP


casings have been  found  to be  in  the  order  of US  $25-30/KW (1990  dollars)  for

                                                   R                  R
boilers in the range of 300 - 500MW where Dralon T   acrylic or Ryton   filter bag


can be utilised .
Supplementary ESP's for one 500 MW boiler  using  low sulphur coal  (less  than 1.5%


sulphur) show a total capital  requirement of more than US $45/kW (1990 dollars)  to


approach an outlet emission of 50 mg/Nrn  (0.05 Ibm/MBTU) and require an additional


70% of the original ESP plan area.





If emission levels below 30 mg/Nm  (0.03 Ibm/MBTU) were required the total capital


requirement for supplementary  ESPs  would increase dramatically.





Several  studies,  most  notably  that  by  Belba  and  Carr  (2),  have  shown  that


levelized  operating  costs  for fabric  filters  are  comparable  or  lower  at  2.8


mills/kWh than for ESPs at 2.8 - 4.4 rnills/kWh.





Design studies carried out by  Howden for 300  - 500 MW boilers have  shown that  by


sequentially converting each  ESP gas  path during  operation  that  outage  time  for


retrofitting pulse jet  fabric filters  into the  ESP  casing  can be limited  to  the


outage  time  required for  installation  of  supplementary ESPs,  typically  in  the


order of 10 to 12 weeks.






* Total capital requirement here is as  used by Belba, Carr et al (2)


 R Dralon 'T'  - Bayer


 R Ryton - Phillips Petroleum
                                        7-6

-------
PERFORMANCE IMPROVEMENTS WITH RETROFIT FABRIC FILTERS

After retrofit of  pulse jet fabric filters  the  emission  levels  achieved are  the
same as  those  available from  new  fabric filter  installations  and are  typically
well below 30 mg/Nm  (0.03 Ibm/MBTU) and provide a clear stack.

More typically,  a  conversion  carried  out by Howden on a  60  MW pulverized  fuel
fired boiler  burning coal  with  1% sulphur  and  257, ash  has  consistently  shown
                                            3
outlet emissions  in the  order of  15  mg/Nm  (0.015 Ibm/MBTU)   over  an  extended
period and well  within  US 1979  New Source  Performance  Standards  (NSPS) of  0.03
Ibm/MBTU and other international new source codes.
SUMMARY
Growing interest has been  shown  in  retrofitting modern pulse jet cleaned FF  into
existing  ESP  casings   to  upgrade  particulate  emission  control  on  coal  fired
boilers.

Modern pulse jet designs occupy 50% of the plan area of more traditional  low ratio
FF and generally fit within the existing ESP casing.

Approximately 4000 MW of poorly performing ESPs have been  or are  being  replaced  by
Australian, African and Canadian utilities.

The Howden design concept of minimum modification to existing plant can  result  in
cost savings of  more  than 50% over the  cost  of alternative solutions  with lower
levelized operating costs. Construction and outage time are  also  minimized.

Performance of retrofit FF is the same as that available from new FF and achieves
                                            3
an  emission  typically well  below 30  mg/Nm   (0.03 Ibm/MBTU) providing a clear
stack.

Control of S0_ emission is also possible using in-duct dry sorbent injection.
                                        7-7

-------
REFERENCES







   1.      "Australian Experience with Fabric Filters  on Power Boilers,  an  Update



           for 1987",  A  Th M  Vandewalle,  H F  Johnson.  Proceedings  of  Seventh



           Symposium  on   Transfer    and   Utilization   of   Particulate   Control



           Technology,  1988.







   2.      "Economics of Electrostatic Precipitators  and Fabric  Filters",  Victor



           Belba,  F A Homey,  R  C Carr,  W Piulle.  Proceedings of Fifth  Symposium



           on  the  Transfer and  Utilization  of  Particulate  Control  Technology,



           Volume  I, February  1986.
                                       7-8

-------
                                                                 Revtrst Pultt
                                                                 Monllold
                                                                 Bog Chongi Oul
                                                                 Arto
               SECTIONAL ELEVATION
                                Filler Modult
              (£)0i0fr;j/
,

,.
—
J- -4-
•t- -f

— J
>vt
— 4" I
Division Wall- 	
SECTIONAL PLAN
2X- 	 ^xiphrogm Votve

L
                                                   KEY PLAN OF CELLS
Figure 1.   Arrangement of Howden  Pulse Jet Fabric  Filter Modules
            Retrofitted into existing Electrostatic  Precipitator Casing
            Munmorah Unit 3.
                                     7-9

-------
                       Bag Change-out
                       Area
                                            Cleaning
                                            Air
                          Reverse
                          Pulse
                          Manifolds
                             Gasflow
Tank "j
T [
^**


~*.
J D

^^\

xV
                        Existing Convete
                        ESP  Casing shown
                        as
 Compartment


SECTION  A'A'
SECTIONAL  ELEVATION
                  Filter Bag Module
                   /
                                                                 BOILER
SECTIONAL PLAN
                                          KEY PLAN OF COMPARTMENTS
  Figure  2.  General Arrangement of Howden Pulse Jet Fabric
  Filter  Modules retrofitted  into existing  Electrostatic
  Precipitator Casing Rooiwal  Unit 1.
                            7-10

-------
Boiler Design
                                      Table 1
                          DESIGN DETAILS FOR 350MW BOILER
                            AND ASSOCIATED  FABRIC FILTER
                          MUNMORAH POWER STATION AUSTRALIA
Manufacturer/Design
Nominal Rating,  MW
Commissioning Date
ICAL(CE)/p.c.  fired
350
1971
Fabric Filter Design
Design Gas Volume, a m /s (acfm)
Max Gas Temperature, °C (°F)
Inlet Burden, g/Nm  (gr/Nft )
Configuration
                   2    2
Total Cloth Area, m  (ft )
Maximum Air to
                         2
Cloth Ratio, m/s (acfm/ft )
Commissioning Date
528 (1,118,620)
155 (310) Air Dilution used to lower  to
130(266)
25-40 (11-17)
8 compartments
6 bundles of 267 bags per compartment
30,300 (326,000)

0.02(4)
First 1/2 unit February,  1990
Complete unit August, 1990
                                      7-11

-------
                                      Table 2

                            PERFORMANCE  TEST RESULTS  FOR
                              PULSE JET FABRIC FILTER
                         ROOIWAL POWER STATION UNITS 1 & 2
                          Specified/Guaranteed    Unit  1           Unit  2

FF Inlet Gas

Volume,  Nm3/s (Ncfm)            61.2  (130,000)    79.6  (169,000)   64.4 (136,000)

Gas Temperature,  °C (°F)        155  (310)         150 (302)        151  (303)

Inlet Burden                   14-24             25              29

g/Nm3 (gr/Nft3)                 (6-11)            (11)             (13)

Outlet Emission

mg/Nm3 (Ibm/MBTU)               100  (0.1)         15 (0.015)        15  (0.015)

Collection Efficiency  %        99.3              99.9             99.9

Pressure Drop, kPa  (ins wg)     1.5  (6)           1.34  (5.4)       1.45  (5.8)
Source:  Test Report  Environmental  Science  Services  CC
August 23,  1989
                                    7-12

-------
                                      Table 3
                   JAPANESE CIRCULATING FLUIDIZED BED BOILERS AND
                        PULSE JET FABRIC FILTER DESCRIPTIONS
Site
                         Ide
                                  Marazumi
Kobe
Yokkaichi
Saiki
Boiler Design
Manufacturer
Nominal Rating,  MW
Bed Material
Flue Gas Desulfurization NA
MHI
25
Sand
NA
MHI
25
Sand
NA
MHI
50
Sand
CaCO^
MHI
60
Sand
NA
MHI
20
Sand
CaCO
Fabric Filter Design
Gas Volume,Nm /s(Ncfm)  17(36,000) 18(38,000) 40(85,000) 60(127,000)  17(36,000)
                        170(340)   205(400)   155(310)   180(355)     190(375)
Max Gas Temp.  °C(°F)
Inlet Burden,
g/Nm3(gr/N£t3)          110(50)
No of Bags              760
Filter Bag Material     Ryton
Maximum Air to Cloth
ratio (m/s (acfm/ft2)   0.015(3)
Commissioning Date      1/88
                                   40(18)
                                   760
                                   Ryton
100(45)
1680
Nomex
  17(8)
  2700
  Ryton
                                   0.018(3.6)  0.017(3.5)  0.02(4)
                                   4/88       5/89       9/90
60(27)
750
Ryton

0.015(3)
1/90
                                      7-13

-------
                                      Table 4
                         PERFORMANCE TEST RESULTS FOR PULSE
                        JET FABRIC  FILTER - IDE WORKS UNIT 1

                                  Specified/Guaranteed     Test  Results
FF Inlet Gas
Volume, Nm3/s (Ncfm)                    17 (36,000)             16  (34,000)
Gas Temperature,  °C (°F)               170 (340)              174 (345)
Inlet Burden, g/Nm3 (gr/Nft3)           110 (50)               100 (estimated)(45)
Outlet Emission,  mg/Nm3  (Ibm/MBTU)      50 (0.05)              7  (0.008)
Collection Efficiency, %               99.9                   99.9
Pressure Drop,  kPa                     1.5 (6)                 1.25 (5)

Source: MHI Performance  Test Report 15/5/88
                                     7-14

-------
                           THE EPRI PILOT PULSE-JET BAGHOUSE
                                 FACILITY AT PLANT SCHOLZ
                                           K. J. Mills
                                   Southern Company Services
                                         P. O. Box 2625
                                     Birmingham, Al. 35202
                                          R. F. Heaphy
                                    Southern Research Institute
                                        2000 9th Ave. S.
                                     Birmingham, Al. 35255
ABSTRACT

The Electric Power Research Institute is currently directing their fabric filtration research efforts toward
pulse-jet baghouses.  Pulse-jet baghouses have the potential to be 40-50 percent smaller and 30-40 percent
less expensive than conventional reverse gas baghouses for new and retrofit applications.  Southern
Company Services Inc. developed the project concept of converting three compartments of the pilot
baghouse at Plant Scholz to pulse-jet operation. The pulse-jet baghouse systems under investigation were
procured from General Electric Environmental Services, Flakt Inc., and Howden Environmental Systems
Inc., and represent three different cleaning concepts associated with pulse-jet technology.  Compartments
with high, medium, and low pressure pulsing will be operated side-by-side along with a reverse gas sonic
assist compartment. A comprehensive test program conducted with Southern Research Institute is
underway to investigate parameters such as pulsing energies, bag length, fabrics, and cleaning modes.
                                              M

-------
                           THE EPRI PILOT PULSE-JET BAGHOUSE
                                FACILITY AT PLANT SCHOLZ


INTRODUCTION

Maintaining low emissions concurrent with low flow resistance and low capital and maintenance costs
dictates baghouse economics. Reverse-gas-cleaned (RG) baghouses which have low air-to-cloth ratios
(low ratio) have been good systems for achieving low particulate emissions. However, baghouses which
impart more cleaning energy to the bags operate at lower levels of flow resistance. Therefore, they can be
built smaller and operated less expensively, offering economic advantages over RG cleaned units. The
effects that high energy cleaning techniques have on baglife, equipment maintenance, and energy
consumption remain unknown.

Pulse-jet (PJ) cleaned fabric filters with high air-to-cloth ratios (high ratio) have recently become an
attractive paniculate-control option for U. S. utilities. PI fabric filters also incorporate high energy
cleaning techniques. Interest in pulse-jet cleaning followed improvements in fabrics and systems with
enhanced reliability. These advances offer the possibility of significantly reducing the costs of particulate-
controls while maintaining low emissions and operating costs. Pulse-jet cleaning results in lower
resistance to gas flow than reverse-gas, reverse-gas-sonic-assist (RG/S), or shake/deflate (S/D) cleaning,
thereby allowing smaller baghouses to filter the same volume of flue gas.  Although PI cleaned baghouses
are used extensively with industrial boilers and a number of utility power plants in Europe and Australia,
they are not presently used by the U.S. electric utility industry due to emissions and bag life concerns (1).

In 1982 Electric Power Research Institute (EPRI) sponsored the construction and operation of a fabric
screening and high sulfur bag house pilot facility (HSFP) utilizing a 10-MWe gas slipstream at the Gulf
Power Company Plant Scholz site. Plant Scholz Steam Plant consists of two 1953 vintage 40-MW,
Babcock and Wilcox, front-fired pulverized-coal boilers with General Electric turbine generators. The
boilers are rated at 475 x 106 Btu/h maximum heat input with 850 psig steam at 900°F. The 10-MWe
slipstream originates downstream of the air heater and returns to Plant Scholz Unit 2 immediately upstream
of the electrostatic precipitator.

The high-sulfur coal fabric filter research program at Plant Scholz was initiated to quantify fabric filter
design parameters for a low-ratio baghouse, and to develop operating guidelines that would address the
problems encountered in the application of the technology. This research, completed in December 1988
included conventional reverse gas, shake/deflate and sonic assist reverse gas cleaning method
optimizations, and evaluations of exposed surface texturization, flue gas conditioning, and experimental
fabrics. EPRI report CS-6061 describes this work.

Southern Company Services Inc. (SCS), developed a project concept of converting three of the five
compartments of the Plant Scholz pilot facility to pulse-jet operation to address U.S.  electric utility
industry concerns with emissions and bag life. During the latter  half of 1989, R. N. Pyle Construction
Co. of Pensacola, FL, made the required plant modifications engineered by Baskerville-Donovan, also of
Pensacola, FL. These modifications where conducted under the supervision of SCS & Southern Research
Institute (SRI).
                                              8-2

-------
PROJECT SCOPE

The research conducted at the EPRIHSFP is designed to address utility concerns with high ratio fabric
filter. Comparisons of PJ with RG/S cleaning will be made to evaluate the relative merits of low-ratio and
high-ratio fabric filters.

Since several different PJ cleaning designs are commercially available, the three most widely marketed
technologies where chosen to be compared (2).  These three technologies are: low-pressure/high-volume,
intermediate-pressure/intermediate-volume, high-pressure/high-volume. The comparison of these
technologies will produce design and operating information for full-scale utility application of PJ cleaning.
The initial comparison tests will be conducted while the boiler is fired with an eastern, sub-bituminous
coal, containing 3% sulfur over a period of one year. The second phase of testing, if approved, will be
based on low sulfur fuel.

This research seeks to determine if PJ cleaning meets utility needs in terms of initial capital expenditures
and operating and maintenance costs. A further goal is to establish the paniculate collection performance
of the PJ cleaned fabric filters and to demonstrate whether paniculate emissions rates either meet or
surpass current and anticipated performance standards.


LOW RATIO VS. HIGH RATIO FABRIC FILTERS

Most utility baghouses currently being operated are low ratio fabric filters which employ some method of
reverse gas cleaning. With RG cleaning, residual dustcakes tend to build over a period of several months.
The dustcakes can become undesirably thick (up to 1 in.) and heavy (up to 1.5 lb/ft2 or 150 Ib/bag) in
extreme cases.  RG/S bag cleaning, utilized in Compartment #1 of the Plant Scholz pilot facility, utilizes a
sonic horn to add energy to the cleaning process as illustrated in Figure 1. Due to the low-energy cleaning
methods used with low ratio or inside-collectors, bags must always be isolated from the flue gas flow to
achieve effective cleaning (off-line cleaning).

In PJ systems or high ratio fabric filters, a valving and manifold arrangement directs a compressed air
pulse down one or more rows of bags within a compartment to remove the dustcake.  PJ cleaned
baghouses collect ash on the outside of the bags and incorporate metal cages  on the inside to prevent bag
collapse. This is why high ratio baghouses are typically called outside collectors.  The air pulse
propagates rapidly as a pressure wave down the length of the bag, causing the bag to accelerate away from
the cage and then to immediately collapse back onto the cage.  This rapid flexing of the fabric dislodges the
dust, which falls into the ash collection hopper below. Figure 2 illustrates PJ cleaning during filtering,
purging prior to shutdown or maintenance, and cleaning. The three major PJ technologies illustrated in
Figure 3 are: low-pressure/high-volume, intermediate-pressure/intermediate-volume, high-pressure/high-
volume.

Most baghouse types are compartmentalized, but only PJ systems are typically designed to clean the bags
on-line without isolating compartments from the flue gas flow. This is accomplished by sequentially
pulsing selected bags while the remainder continue to filter flue gas. On-line  cleaning greatly simplifies
overall fabric filter operation.  However, since some of the dislodged dust is immediately reentrained back
onto the bags, it may be necessary, in some designs, to use off-line cleaning for certain fuels to maintain
pressure drop and outlet emissions within acceptable bounds.

In either low ratio or high ratio fabric filters, compartmentalized design permits maintenance to be
performed without requiring an outage of the entire fabric filter. Compartmentalization is achieved with
inlet and outlet dampers for isolation purposes.  Current experience on operating units indicates that on-
line cleaning is feasible and highly successful.
                                               8-3

-------
Figure 1. Schematic of a reverse-gas baghouse at Plant Scholz.
                        OUTVET VALVE
OLTTLET
MANIFOLD      COMPRESSOR
           OR BLOWER
  DIRTY FLUE
  GAS INLET
                       FILTERING    /   PURGING

               INLET MANIFOLD       CAGE
                                                                        ASH
                                                                        DISPOSAL
                                          INLET
                                          VALVE
Figure 2. Schematic of compartments in a pulse-jet-cleaned baghouse filtering flue gas,
purging (or ventilating) prior to maintenance, cleaning bags, and filtering flue gas again (2).
                                           8-4

-------
                       HIGH PRESSURE,
                       LOW VOLUME
INTERMEDIATE PRESSURE
    AND VOLUME
LOW PRESSURE,
 HIGH VOLUME
       Figure 3.  Comparison of typical operating characteristics for high-pressure, intermediated-
       pressure, and low-pressure pulse-jet cleaning (3).
DESCRIPTION OF THE PILOT PLANT

Four of the five original HSFP baghouse compartments are in use for the evaluation of PJ cleaning.
Compartment #1 uses RG/S cleaning at a filtering air-to-cloth ratio (A/C) of 2.5 acfm/ft2. Compartment
#2 has been converted to low pressure, high volume, PJ cleaning (LPHV) using equipment suppled by
Howden Environmental Systems, Inc.  Compartment #3 has been converted to intermediate pressure,
intermediate volume, PJ cleaning (IPIV) using a system supplied by Flakt, Inc. Finally, Compartment #4
has been converted to high pressure, low volume, PJ cleaning (HPLV) using equipment supplied by
General Electric Environmental Services, Inc. All three of the PJ compartments will be operated at an A/C
of 4.0 acfm/ft2. Details  of each technology are presented in Table 1.


Reverse Gas/Sonic - Compartment #1.

With reverse gas cleaning, an auxiliary fan forces a relatively gentle flow of filtered flue gas (1-2 ft/min at
the fabric surface) backwards through the compartment and the bags to be cleaned. This causes the bags
to partially collapse inward and dislodge the dustcake.  This dust falls through the bags, the thimble, and
the tubesheet into the hopper. Also,  metal rings sewn into the bags about 4-5  ft. apart along their length
prevent complete collapse (2).

Reverse-gas/sonic (RG/S) bag cleaning incorporates low-frequency pneumatic horns, sounded
simultaneously with the  normal reverse-gas flow to add energy to the cleaning process. Sonic
enhancement has been shown to reduce gas flow resistance and dustcake weights by as much as 50%
below those achieved by RG cleaning alone, without increasing emissions or decreasing bag life. Fligh
sound pressures (80-160 Pa) in the low frequency range (below 250-300 Hz)  have proven most effective
for dustcake removal.
                                               8-5

-------
                                           Table 1

               SUMMARY OF DESIGN AND OPERATION SPECIFICATIONS (£>
                                      RG/S
LPHV
IPIV
HPLV
Location
Manufacturer
Number of bags
Bag diameter (nominal), in.
Bag length (nominal), ft
Measured filter area, ft2
Tubesheet design


Cleaning air header volume, ft3
Cleaning air header pressure, psi
Available air volume ft3/bag/pulse
Filtering air-to-cloth ratio, acfm/ft2
Comp.#l
G. E.
28
12
35
2,915
Rectangular
7 rows
6 bags/row*
N/A
N/A
N/A
2.5
Comp.#2
Howden
72
4.9
20
1,861
Circular
4 circles
24, 20, 16, 12
22
11
2.1
4
Comp.#3
Flakt
70
5
20
1,901
Rectangular
7 rows
10 bags/row
13.6
28**
1.4
4
Comp.#4
G. E.
56
6.0
20
1,807
Rectangular
8 rows
7 bags/row
8.4
40
0.3
4
 *Two rows are capped off.
**Header pressure is =23 psi on Comp.#2 after first pulse in each clean cycle


Low Pressure High Volume - Compartment #2.

LPHV pulsing utilizes a novel, circular arrangement of oval-shaped bags, with nozzles attached to a
rotating cleaning manifold located above the bags as illustrated in Figure 4. This design, developed by
Howden, Inc., utilizes a high volume pulse of only 10 psig. The high volume low pressure pulse permits
the use of blowers rather than compressors to supply the pulse cleaning air. This allows bag lengths up to
20 ft., reduced cleaning energy, fewer pulse distribution valves, on-line cleaning, and simplified
configurations and components.

The tubesheet in the LPHV compartment contains 72 holes for bag installation.  The holes are 2-3/4 in. by
3-5/s in. rectangles with semi-circles of l-3/g in. radius at each end. These tubesheet holes have the same
area as a circle with a diameter of 5.05 in. The tubesheet holes are arranged in four concentric circles to
allow cleaning by a single rotating blowpipe. The bags are cleaned by compressed air pulses  at 10 to 15
psi.  The blowpipe contains four nozzles that are aligned over the circumferential rows. The  diameters of
the four circumferential rows measure 61.5 in., 52.5 in., 43.5 in., and 34.5 in., respectively.
                                              8-6

-------
                           SOLENOID.
                           VALVE
                          DRIVE MOTOR
                         FOR REVERSE PULSE
                        AIR CLEANING SYSTEM
                         FILTER TUBE SHEET
                         FOR HOLDING FILTER
                        TUBE CAGE ASSEMBLY
                                                           VWLK-IN COMPARTMENT
                                                         •FOR TOP ACCESS TUBE REMOVAL
                                                            AIR RESERVOIR TANK
                                                            'FOR REVERSE PULSE
                                                            AIR CLEANING SUPPLY
                                                             •HOPPER ACCESS
                                                             HOLE NOT SHOWN
                                  COLLECTED
                                  PAHTICULATE'
                                   OUTLET
       Figure 4. Schematic of one compartment of a low-pressure/high-volume, pulse-jet fabric filter
       supplied by Howden Environmental Systems. Oval bags are arranged in concentric circles,
       with up to 484 bags per compartment.  The bags are cleaned by pulses from the manifold arm
       which rotates at one revolution per minute. Cleaning air, supplied to the manifold by blowers,
       is introduced to the manifold through a central dump valve. The cleaning air is injected into the
       bags via nozzles in the pulse air manifold.
A production LPHV fabric filter system incorporates blowers to fill the low pressure air storage tank.
However, the Plant Scholz pilot facility incorporates high pressure air, reduced to 12 psi by a regulator, to
simulate the blower.


Intermediate Pressure Intermediate Volume - Compartment #3.

Flakt, Inc. developed the IPIV design, installed at the pilot facility, for large, custom, high temperature,
metallurgical fume and fly ash applications. A look at the gas path and distribution will indicate that, in
effect, there is an integral pre-collector within the unit.  The gas entering the inlet plenum turns vertically
up and is evenly distributed across the side of the compartment at very low velocities and then continues in
the horizontal and/or downward direction depositing paniculate on the bags.  This side entry method
                                                  8-7

-------
minimizes reentrainment, fabric abrasion, and pressure drop that is otherwise prevalent with designs
allowing introduction of dust laden gas into hoppers below the bags (4).

The IPIV design typically uses between 15 and 30 psi compressed air. A unique pulsing valve eliminates
the necessity of the tubesheet venturi required in high pressure PJ systems. The primary air expands
directly into the filter bag minimizing the induced air to one or two times the primary air effecting a short,
quick but deep pulse. Figure 5-a illustrates the pulse valve mounted integral to the pressure tank. The
valve consists of two parts, the solenoid pilot valve and the main pulse valve.  In this figure, the pulse
valve is in the open position with compressed air passing through the distribution pipe and into the bags
for cleaning. A main feature of this design is the large diameter of the pulse valve (2" diameter) which
minimizes pressure drop to the distributing pipe.

Figure 5-b shows the complete rPIV fabric filter cleaning system, general location of the pressure tank and
pulse valve from which compressed air is directed through the lateral pipe through nozzles into the bags.

The tubesheet in the rPIV compartment is made up of a 7 by 10 array of 5.31 in. (135 mm) diameter
circular holes.  The rPIV bags are cleaned by pulses of compressed air at 20 to 30 psi that are directed
down into ten bags at a time from one of the seven blow pipes in the compartment..
                         - Solenoid Pilot Valve
                          -— Pressure Tank
                                        Lateral Pipe
Compressed Air

   Supply


        ir«  »
   Sound
   Insulation

Outlet Damper
     In
Open Position
                     (a)
                                                             (b)
       Figure 5.  IPIV design pulse valve mounted integral to the pressure tank (a).  IPIV fabric filter
       cleaning system, general location of the pressure tank (b) (4).
High Pressure Low Volume - Compartment #4.

During the normal filtering mode of the HPLV fabric filter system, the particulate laden flue gas enters the
compartment (1,200 fpm), energy efficient inlet opening located in the straight side of the casing
approximately sixteen inches below the filter bag bottoms. Strategically positioned inlet duct turning vanes
produce uniform flow distribution as the gas enters the module. The flue gas passes through the bags and
exits the filtering area at the top of each bag at the tubesheet level. The cleaned gas enters an open plenum

-------
above the tubesheet and exits the module through the outlet poppet damper. The discharge from each
module enters the outlet manifold which terminates at a single outlet ductwork connection (5_).

The HPLV fabric filter system is intended for off-line cleaning.  During the cleaning mode, one module at
a time is disengaged from normal filtering service by closing the outlet poppet damper, thus isolating the
module from ID fan suction.  The compressed air pulsing system is then activated to sequentially clean
each row of bags in the isolated module. This pulsing system consists of a compressed air inlet
accumulator manifold and cylindrical distribution headers over each row of bags.  Each distribution header
directs a pulse of compressed air (70-90 psig) through the venturi located in the opening at the top of the
bag.  The venturi extends 4" into the bag, and the top is flush with the bag cage to permit access. The
venturi is believed to enhance the uniformity of the pulse pressure wave as it travels the length of the bag,
thereby improving cleaning and reducing the deceleration of the bag against the cage.

After all bags in a module have been cleaned, the module remains off-line for an adjustable period of time
to allow settling of the paniculate into the hopper.  The outlet poppet damper is then opened, and the
cleaned module returns to the normal filtering mode. The next module is sequentially selected and the
process repeated.

The HPLV tubesheet consists of 56 6-in. holes arranged in eight rows of seven holes each.  HPLV
cleaning  uses compressed air at 40 to 90 psi. Table 1 summarizes the design specifications of the fabric
filter cleaning systems that are being tested at the HSFP (6_).


OPERATION

Three SRI employees conduct the studies at the Plant Scholz pilot facility.  These  studies include
monitoring and updating the data acquisition systems, manual testing, and overall onsite management of
the project. Three R. N. Pyle employees maintain the equipment and controls, and operate the facility 160
hrs/wk. The facility is equipped with its own control room where  the automatic data acquisition system is
housed.

The unit is equipped with a General Electric Series Six Programmable Controller (PC).  This PC is
programmed to automatically control cleaning, flow, alarms, and trips.  This gives the facility operators
the ability select cleaning modes (time, Ap,  manual/automatic, on/off-line), air-to-cloth ratios, and pulsing
characteristics for each compartment. Each compartment, excluding Compartment #1-RG/S, is set to
initialize a clean when the tubesheet pressure drop reaches 5.0"wtr.  Compartment #1-RB/S is set to be
cleaned every 90 minutes regardless of tubesheet pressure drop. Table 2 shows other operational
characteristics of each compartment cleaning cycle.

All instrumentation has been calibrated prior to the initial filtering of hot flue gas on December 6,1989. At
vendor request, the bags have been precoated with a substance marketed as Neutralite, sufficient to coat
each square foot of filter with =0.07 Ib. by while under a air load.  It is estimated  that from 25% to 50% of
the pre-coat material actually adhered to the bags. The precoat is expected to improve filtering and provide
some bag protection. Two gallons of fluorescent powder have also been injected  into the system with the
Neutralite. This powder, marketed as Visolite,is normally used for leak detection, but in this instance was
also used as a tracer to determine the fate of the precoat.
                                              8-9

-------
                                           Table 2

                        SUMMARY OF NORMAL OPERATING MODES


      Compartment   Clean mode    Clean Init.   Automatic Cleaning Logic

      #1  RG/S        Off-line     90 min. time  Reduce baghouse flow by =7,000 acfm,
                                   initialized,   close compartment poppet damper, RG &
                                               sonic horn initiated for 1 minute, shut-off
                                               RG & sonic horn, open compartment
                                               isolation poppet damper, ramp baghouse
                                               flow back to set point.

      #2 - LPHV       On-line       5.0"wtr.    Initiate pulse cleaning and maintain until
                                               tubesheet Ap drops below 3"wtr.

      #3 - rPIV        On-line       5.0"wtr.    Initiate pulse cleaning and maintain until
                                               tubesheet Ap drops below 3"wtr.

      #4-HPLV       Off-line       5.0"wtr.    Reduce baghouse flow by =7,000 acfm,
                                               close compartment poppet damper, initiate
                                               pulse cleaning, after all solenoids fire, open
                                               compartment isolation poppet damper,
                                               ramp baghouse flow back to set point.
The facility is being operated continuously during the first phase of testing. Facility outages will
obviously follow Plant Scholz Unit # 2 scheduled and forced outages.  The 10 MWe slip stream remains
constant regardless of plant unit load.  However, unit load swings affect inlet temperature and other
temperature related operating characteristics.


DATA ACQUISTTON & REPORTING

Both SCS &  SRI issue monthly status reports to EPRI and the participating vendors. These reports
include performance, operating, and maintenance updates and any insights into the outcome of the
research. Pulse-jet operating problems are discussed with then- respective vendors as soon as possible in
an attempt to optimize the corresponding PJ technologies.

Table 3 shows the methods the operating variable data are being collected at the facility. The data is
collected and recorded automatically every one to five minutes. The data is then archived once per day.
The archived data will be used in the overall performance evaluation of each PJ system compared to RG/S.
                                            8-10

-------
                                        Table 3
                 METHODS OF MONITORING OPERATING VARIABLES
Parameter
Method
Flow rate
Temperature
System pressure
System pressure drop
Tube sheet pressure drop
Pulse-Jet Header Pressure
Cleaning Activity

Inlet particle concentration
Inlet particle size distribution
Outlet particle concentration
Outlet particle size distribution
Opacity
Flyash unburned carbon
Sulfur dioxide, SO2
Nitrogen oxides, NOx
Sulfur trioxide,  SOs

Dew point
Data acquisition
Total flow venturi & inlet damper microprocessor
  controlled system booster fan; Compartments #2-#4
  each have flow Venturis & outlet butterfly damper
  microprocessor flow controls.
Thermocouples.
Electronic transducers.
Electronic transducers.
Electronic transducers.
Electronic transducers.
Digital signals linked to data acquisition software to
  quantitatively determine cleaning activity.
Mass train; Optical Monitor (ESC Model P-5A).
Cyclone; cascade impactors.
Mass train; Optical Monitor (Tribaflo).
Cascade impactors.
Specially developed system.
Loss of weight after 4 hr. temperature burn.
Continuous commercial monitor.
Continuous commercial monitor.
Controlled condensation train; Continuous Monitor
  (Severn Sciences, Ltd.).
Wet bulb thermometer, Land Acid Dew Point Meter.
IBM PC-AT computer
                                        8-11

-------
As a backup, the following data are logged manually each hour:

           Ambient Temperature (°)F

           Barometric Pressure ("Hg. abs.)

           Air Heater Outlet Temperature (°F)

           Baghouse Inlet Temperature (°F)

     •     Baghouse Outlet Temperature (°F)

     «     Compartment Operating Mode

     •     Baghouse Pressure Drop

           Baghouse Outlet Static Pressure ("wtr.g.)

     •     Compartment Tubesheet Pressure Drop ("wtr.)

           Baghouse Flowrate (acfm)

     •     Compartment Flowrate (acfm)

     •     Reverse Gas Flowrate (acfm)

           Reverse Gas Temperature (°F)

           Reverse Gas Static Pressure ("wtr.)
After compiling all the operating and performance data, SRI will conduct an economic analysis of the
evaluated fabric filter technologies.  This economic analysis will include:

           Capital Cost

           Operating Cost; air hp / clean (compressor hp + reverse gas hp) x number of cleans

     «     Maintenance Cost; life cycle cost of bag replacement + replacement of other parts due to
           corrosion or mechanical failures.

This report is expected to be completed after the second phase of testing.  The research completion date is
dependent upon bag life determination.

-------
REFENCES
     1.     A. H. Dean, K. M. Gushing, W. B. Smith.  "Pulse-Jet Fabric Filters for Coal Fired Utility
           and Industrial Boilers." Electric Power Research Institute, December, 1986.  CS-5386.

     2.     Southern Research Institute, Electric Power Technologies, Inc. "Bag Cleaning." Fabric
           Filters for the Electric Utility Industry, Volume 1 Electric Power Research Institute, 1988.
           CS-5161.

     3.     R. C. Carr, A. T. Vandewall, W. B. Smith.  "Pulse-Jet Fabric Filters For The U.S. Utility
           Industry."

     4.     A. Wiktorsson, Flakt, Inc.  "Current Status: Future Potential For HRFF Technology Applied
           to Utility Coal Fired Boilers."  July, 1981.

     5.     February 8, 1990, Correspondence with T. Lugar of General Electric Environmental
           Services, Inc.

     6.     R. F. Heaphy, K.  M. Gushing. Monthly Report to EPRI. January, 1990.
                                             8-13

-------
   PILOT  DEMONSTRATION OF A PULSE-JET FABRIC FILTER FOR PARTICULATE
            MATTER CONTROL AT A COAL-FIRED UTILITY BOILER
                              R.  C.  Carr
                  Electric Power Technologies, Inc.
                            P. 0. Box 307
                         Menlo Park,  CA  94025
                            C. J. Bustard
                        ADA Technologies,  Inc.
                  304  Inverness  Way  South,  Suite  110
                           Denver, CO 80112
ABSTRACT

A consortium of electric utility companies and the Electric Power
Research Institute sponsored a 1 MWe  (equivalent) pilot
demonstration project of pulse-jet fabric filters at the Martin
Drake station of the Colorado Springs Department of Utilities.
Project objectives were to obtain particulate matter emissions and
pressure drop data that utility companies can use to design full-
scale systems.

Pulse-jet fabric filter  (PJFF) performance was evaluated at an air-
to-cloth ratio of 4 acfm/ft2 using felted Daytex bags for 2800
service hours,  and felted Ryton/Ryton bags for approximately 4600
service hours.   Results show that on-line cleaning at a pulse
pressure of approximately 12 psig easily maintained pressure drop
below 4.3 in. H2O for both the Daytex and Ryton/Ryton bags.  Outlet
particulate matter emissions were measured to be 0.0005 Ib/MBtu for
both the Daytex bags and Ryton/Ryton bags, with outlet opacity less
than 1% (clear stack plume).  The pilot plant operated without bag
failures.
                                  9-1

-------
  PILOT DEMONSTRATION OF A PULSE-JET FABRIC FILTER  FOR  PARTICULATE
            MATTER CONTROL  AT A COAL-FIRED UTILITY BOILER

INTRODUCTION

For utility companies facing requirements to upgrade existing
particulate matter (fly ash)  controls on coal-fired boilers, but
where space is restricted,  the smaller size and modular design of
pulse-jet fabric filters (baghouses)  is very appealing.   Experience
with pulse-jet baghouses in coal-fired power plants outside the U.S.
suggests that pulse jets can operate at sizes 50% smaller and at 30-
40% lower cost than conventional reverse-gas fabric filters, yet
still achieve comparable reliability and particulate matter
collection efficiency.   Consequently, several U.S. utility companies
are evaluating pulse jets as a retrofit solution for poorly
performing hot- and cold-side electrostatic precipitators.   Further,
U.S. utility companies  have expressed interest in using pulse jets
in conjunction with dry SO2 control processes.

Since the majority of pulse-jet experience on coal-fired utility
boilers is nondomestic,  a pilot demonstration project was conducted
on a U.S. coal-fired boiler to evaluate the technology in a utility
setting.  The project was conducted at the 85 MW Martin Drake Unit  6
of the Colorado Springs Department of Utilities, and was sponsored
by the Electric Power Research Institute, a consortium of electric
utility companies, and an equipment supplier.   Overall project goals
were to document particulate matter emission control and pressure
drop performance of a 5000 acfm (1 MWe) pilot unit that is of
similar geometry to one module of a full-scale, pulse-jet fabric
filter.  Specific objectives included confirmation that pulse jets
can achieve:   (1)  a clear stack plume;   (2) high particulate matter
collection efficiency (99.9%) and outlet mass emissions less than
0.02 Ib/MBtu; and  (3) average tubesheet pressure drop of 5  in. H20
with on-line cleaning at an air-to-cloth ratio of 4 acfm/ft2.

                                  9-2

-------
The pulse-jet fabric filter  (PJFF) pilot plant project consisted  of
an 18 month, three phase test program.  Phase I evaluated  felted
Daytex bags  (Ryton fiber with a Rastex scrim), Phase II  felted
Ryton/Ryton bags  (Ryton fiber with a Ryton scrim), and Phase  III
woven fiberglass bags.  This paper presents  final results  for the
Daytex and Ryton/Ryton tests in Phases I and II, respectively.
Results presented for Phase  I supplement information reported
previously  (1).

PILOT PLANT DESCRIPTION AND  TEST CONDITIONS

Pilot Plant Description

The pulse-jet pilot plant was located between the Martin Drake
boiler house and the stack,  and filtered a 5000 acfm flue  gas
slipstream withdrawn immediately downstream  of the air heater.
Martin Drake Unit 6 is a Babcock & Wilcox, pulverized-coal-fired
boiler that burns western, low-sulfur, subbituminous coal.  The coal
has an average higher heating value of 11,000 Btu/lb, and  contains
approximately 0.4% sulfur and 5-10% ash.

Figure 1 shows a schematic of the PJFF, which is a low-pressure and
high-volume design (2.,i) .  The unit contains 48 oval-shaped bags
arranged in three concentric circles.  A single cleaning manifold
located above the bags is equipped with nozzles that align with each
bag row.  The manifold continuously rotates  at 1 rpm.  During
cleaning,  pulse cleaning air supplied to the manifold through a
central dump valve is injected through the nozzles into the bags.
The air is supplied from a reservoir that is maintained at a
pressure of approximately 12 psig by a positive displacement blower.
Due to the circular bag arrangement and the  rotating cleaning
manifold,  bag cleaning occurs on a somewhat  random basis since a
pulse nozzle will not always be perfectly aligned with a bag.
However,  on a time-averaged basis all bags are cleaned equivalently.

The pilot  plant  was equipped with both top and bottom flue gas inlet
duct  configurations.   Testing was conducted  exclusively with the top
inlet duct located immediately below the tubesheet.   Two sets of
                                 9-3

-------
             PULSE-JET FABRIC FILTER PILOT PLANT
                     48 BAGS, 5000 acfm
      CLEAN GAS
      OUTLET
      ROTATING
      PULSE MANIFOLD
      ASH COLLECTION
      HOPPER   	
 AIR RESERVOIR
^-J
                                                 DIRTY GAS
                                                 INLET
  •*-TUBESHEET
                                                        1
FIGURE 1.  Schematic diagram of the pulse-jet fabric filter  (PJFF)
pilot plant.  The unit is a low-pressure/high-volume pulse jet
designed to filter 5000 acfm.  The PJFF uses 48 oval-shaped bags
that are arranged in three concentric circles.  Each bag is 20  ft
long and 15.5 inches in circumference.  The rotating manifold is
supplied with air from a reservoir that is filled by a positive
displacement blower to approximately 12 psig.  Nozzles in the
manifold are aligned with each bag row and direct the cleaning  pulse
to the bags.
                                  9-4

-------
baffle plates were located in the inlet transition duct to
distribute flue gas to the bags.  As shown in Figure 1, dirty flue
gas entered the fabric filter through the inlet plenum and flowed
from the outside of the bags to the inside, with the particulate
matter collected on the outsides of the bags.  The bags were hung
from the tubesheet, were closed at the bottom, and were open at the
top where the clean gas exited the bags.   To support the bags during
filtration, wire cages were inserted within the bags.  Clean flue
gas exited the fabric filter via the outlet plenum located just
above the tubesheet.  Flue gas flow through the pilot plant was
controlled by a damper that was located between a flow measurement
venturi and the PJFF induced draft  (ID) fan.  The collected fly ash
was discharged into the outlet duct downstream of the ID fan through
a rotary valve at the bottom of the PJFF ash hopper.  The ash laden
gas was then transported to the Martin Drake main duct approximately
20 feet downstream of the pilot plant inlet scoop.

Figure 2 provides a photograph of the tubesheet and the rotating
manifold at the top of the fabric filter.  The view is from the top
of the pulse jet, looking down on the clean sides of the bags.
Clean gas exits the top of the bags through the openings in the
tubesheet.  The clean sides of the bags,  and the cages that support
the bags during filtration, are visible through these openings.

Pilot Plant Operating Conditions

The felted Daytex bags evaluated during Phase I were constructed of
23 oz/yd2 fabric with an active filtering area of approximately 26
ft2/bag.   The bag dimensions were 20 feet in length and a
circumference of 15.5 inches.  The bags were supported by carbon-
steel cages,  each with 14 vertical wires.  The felted Ryton/Ryton
bags evaluated during Phase II were constructed of 19 oz/yd2 fabric
with the same dimensions as the Daytex bags.  Ryton/Ryton was
selected for evaluation because it represented an evolutionary
improvement over Daytex at approximately 40% lower cost.  The
primary difference between Ryton/Ryton and Daytex is the substi-
tution of Ryton for the Rastex scrim used in Daytex.   (The scrim is
the woven support structure that provides strength and reinforcement
for the felt).   The lower cost for the Ryton/Ryton bags results

                                  9-5

-------
FIGURE 2.   Photograph of the tubesheet and cleaning manifold.  The
view is from the top of the unit,  looking down on the clean sides of
the bags.   The cages that support  the bags during filtration are
visible through the openings in the tubesheet.  During operation,
the cleaning manifold rotates at 1 RPM.   Cleaning was initiated when
pressure drop reached 5 in. H20, and continued approximately every 5
seconds until pressure drop was reduced to 3 in. H20.
                                  9-6

-------
from:  (1) Ryton scrim instead of a Rastex scrim, and  (2) lighter
weight fabric  (19 oz/yd^ for Ryton vs. 24 oz/yd^ for Daytex).

The outsides of the Ryton/Ryton bags were singed/glazed to provide a
smooth, "non-fibrous" filtration surface, thus minimizing the
potential for formation of thick or nodular residual dustcakes.  The
bags were supported by carbon-steel cages, each with 8 vertical
wires.  It was anticipated that pulse cleaning might be more
effective with the Ryton/Ryton bags because of:  (1) the glazed
filtering surface, and  (2) the fewer number of vertical cage wires
which, when coupled with the lighter weight Ryton/Ryton fabric,
should allow greater flexing of the bags during cleaning.

To simplify data interpretation, the volumetric flow rate to the PJFF
was maintained constant at 5000 acfm, which is equivalent to an air-
to-cloth ratio (A/C)  of 4 acfm/ft2.   Pulse cleaning was performed on-
line using a pressure drop initiate/terminate sequence.  With this
method, pulse cleaning was initiated when pressure drop  (AP) reached
5 in. H2O, and pulsing continued until AP was reduced to 3 in. H20.
The pulse cleaning rate required to maintain pressure drop within
this range was used to compare PJFF performance for the two bag sets.

Tubesheet pressure drop, flue gas volumetric flow rate, pulse
pressure, boiler load, excess oxygen, number of cleaning pulses,
outlet opacity, and operating temperature were continuously
monitored and recorded by a Campbell Scientific data logger system.
Periodic manual measurements included inlet and outlet particulate
matter mass concentrations, inlet and outlet excess oxygen, gas
volume flow rate, fly ash size distributions, and bag integrity.

PULSE-JET FABRIC FILTER PERFORMANCE

Pulse Cleaning Rate and Pressure Drop

As a consequence of the AP initiate/terminate cleaning procedure
used, pressure drop was maintained nearly constant throughout the
project at approximately 4.3 in. H20.  Therefore,  pulse cleaning rate
was used to characterize PJFF performance with respect to pressure
                                  9-7

-------
drop.   Pulse rate data will be presented in units of pulses/h and
pulses/cleaning cycle.

The pilot plant was started up on flue gas in November 1988, and
accumulated 2800 service hours with Daytex bags through April 1989.
Ryton/Ryton bags were then installed and operated in the PJFF for
approximately 4600 service hours through October 1989.  Figures 3
and 4  compare pulse rate in terms of pulses/h and pulses/cleaning
cycle,  respectively,as a function of cumulative hours in service for
the Daytex and Ryton/Ryton bags.  (Note that the vertical scales are
offset for clarity).   Figure 3 shows that the pulse rate for the two
bag sets increased steadily up to approximately 1800 service hours,
and then appeared to level off.  This behavior is normal and
reflects "seasoning" of the bags, which results from development of
a thin,  residual dustcake on the bags.  Similar time periods to
achieve stable operation with respect to pulse rate and pressure
drop have been reported for commercial pulse-jet fabric filters
operating on coal-fired utility boilers overseas (.4J •   Between 1800
and approximately 3000 service hours, pulse rate fluctuated between
8-18 pulses/h for both bag types.  No significant difference in
pulse rate was evident between the Daytex and Ryton/Ryton bags.  The
sharp increase in pulse rate between 3000 and 4200 service hours for
the Ryton/Ryton bags is a direct result of extended operation of the
host boiler at full load, as will be discussed below.   During this
time interval, pulse rate fluctuated between 16 and 23 pulses/h,
with one surge to 35 pulses/h.  After 4200 service hours, boiler
load and pulse rate returned to levels observed prior to 3000 hours.

Figure 4 shows that approximately 2 to 3 pulses/cleaning cycle were
required to reduce AP from 5 to 3 in. H2O,  despite significant
fluctuations in pulse rate (pulses/hour)  caused by variable boiler
and baghouse operating conditions.  This result suggests that,
although the baghouse cleaned more frequently under conditions where
the dustcake deposition rate on the bags was higher, the number of
pulses required to remove the collected dust was nearly constant.

The short-term pulse rate fluctuations evident in Figures 3 and 4
are typical, and are caused by:   (1) the random nature of the pulse

-------
              MARTIN DRAKE PULSE RATE OPERATING HISTORY
             COMPARISON  OF DAYTEX AND RYTON/RYTON FABRICS
                   600
1200  1800  2400  3000  3600  4200
CUMULATIVE HOURS IN SERVICE
FIGURE 3.   Pulse cleaning history in  terms  of  pulses/h  for the PJFF
with Daytex and Ryton/Ryton bags.   The  data shown  represent  12-24
hour averages.   The PJFF was operated at  an A/C  of  4 acfm/ft2 with
on-line cleaning.   The high pulse rate  for  Ryton/Ryton  bags  at
startup was due to volumetric flow excursions  caused by problems
with the flow controller.   The pilot  plant  was shut down briefly at
4200 hours  because the host boiler burned natural  gas fuel.
                                  9-9

-------
             MARTIN DRAKE PULSE RATE  OPERATING HISTORY
            COMPARISON OF DAYTEX AND  RYTON/RYTON FABRICS
s< °-u .
LiJ
>-
< 7.0 -
Q
^"^
UJ e n "
_1 6.0 -
O
O
„ 5.0 :
o
z
2
£§ 4-° :
_i
\ 3.0 :
CO
UJ „ n -
co 2.0 -
,— J "~
n
Q.
1.0 -
u_ .
0
ffi o.o-
m
^
n
z

OPERATION AT A/C = 4 ocfm/ft2











«
J
/ »«M DAYTEX FABRIC
k 1 »X fh A 1 f 1
JN SiiiTVlLW w\
^ V \p ^ w«i / I
1 i n
ii /,* v ^, i!?
. I , 'l j*1 IL f ^,\/ ^ A . i 'l il \
J^ llWil f*T 1 ' / A I
*i ,i ('1 i, " i ** ' r/*-*\a-A-*^ l/kV* J1* V
, V ' i 'i A/ v ^ ' i
l^ '^ X V ***** RYTON FABRIC '[
't
I,
'1

Z
C

tD
-8.0 5
'
o
-7.0 "^
; TI
•• n~"
-6.0 w
n
LO
:5.0 \
I 0
:4-° |
2
-3.0 °
o
-<
-2.0 P
m
:i.o|
o
:0.0 ^
              0    600   1200  1800  2400  3000  3600 4200
                        CUMULATIVE HOURS IN SERVICE

FIGURE 4.   Pulse  cleaning history in terms of pulses/cleaning cycle
for the PJFF with Daytex and Ryton/Ryton bags.  The data shown
represent  12-24 hour  averages.  The PJFF was operated at an A/C of 4
acfm/ft2 with on-line cleaning.
                                 9-10

-------
cleaning system,  (2) the sensitivity of the control system used for
the pulse cleaning, and  (3) variable operation of the host boiler.

Sensitivity of the control system for pulse cleaning is a direct
result of using on-line cleaning and a pressure drop initiate and
terminate cleaning procedure.   Although this procedure was appro-
priate for the pilot plant program because it simplified pulse rate
comparisons for different test conditions, momentary fluctuations in
AP occasionally triggered or terminated cleaning cycles prematurely.
For example, when the maximum and minimum AP set points were 5.0 and
3.0 in. H2O, respectively,  pulse cleaning occasionally began at an
average pressure drop well below the set point due to a near
instantaneous AP spike to 5.0 in. H20.   Similarly,  pulse cleaning
occasionally stopped prematurely at average AP values significantly
higher than the 3.0 in. H2O minimum set point due to similar AP
fluctuations.   Consequently, under some conditions this cleaning
method can inadvertently lead to short filtering times and variable
pulse rates.  A more appropriate cleaning approach for commercial
applications might be to filter until a predetermined pressure drop
is reached, and then to pulse for a fixed time interval or at a
given rate until pressure drop is reduced to a desired value.

The primary impact of variable boiler operation was increased
particulate matter concentration to the PJFF at higher boiler load.
Since the rate that pressure drop increases is a strong function of
the rate that solids are collected on the bags, the frequency of bag
cleaning will increase as the solids deposition rate on the bags
increases.  This effect can be seen in Figure 5, which shows pulse
rate as a function of boiler load for the Daytex and Ryton/Ryton
bags.   The data show that pulse rate approximately doubled for both
bag types as boiler load increased from one-half to full load.  The
primary cause for this behavior was increased fly ash inlet
concentration at higher boiler load (recall that volumetric flow
rate was maintained constant), an effect that has been observed by
the authors in previous pilot  plant tests of reverse-gas fabric
filters.  As shown in Figure 3,  the highest average pulse rate of
approximately 16 pulses/hr for the Daytex bags occurred between 1600
and 1800 service hours, which coincided with one period of boiler
operation at predominantly full load.   Between 1900 and 2100 hours,

                                 9-11

-------
           20.0
                     MARTIN DRAKE  OPERATING  HISTORY
                   EFFECT  OF BOILER LOAD  ON PULSE RATE
           18.0 -
         o:
         ID
         O
         I
           16.0 -
         in
         QJ
         §14.0 H
         CL
         U.
         O
         UJ
         QQ
           12.0 -
           10.0 -
            8.0 -
            6.0
                          OPERATION  AT A/C = 4 ACFM/FT2
                       RYTON/RYTON FABRIC
                                     ***** DAYTEX FABRIC
                30     40     50      60     70
                                BOILER LOAD  (MW)
80
90
FIGURE 5.  Effect  of  boiler load on pulse rate  for  the  PJFF with
both Daytex and Ryton/Ryton bags.   The data shown represent 12-24
hour averages at each operating condition.  The  increase in pulse
rate at higher boiler load is indicative of higher  fly  ash inlet
concentrations.  The  PJFF  was operated at an A/C of 4 acfm/ft2 with
on-line cleaning.
                                   9-12

-------
another period with considerable full-load operation was experi-
enced, during which time the pulse rate averaged approximately 14
pulses/hr.  For the Ryton/Ryton bags, an extended period of full-
load operation occurred between 3000 and 4200 service hours, and
pulse rate increased to 16-23 pulses/h.  However, even during the
high boiler load operating periods, the pulse rates measured for
both bag types were still quite low.

Although the pulse rate varied as shown in Figures 3 and 4, pressure
drop for both bag types was easily maintained at an average value of
approximately 4.3 in. H2O.  Moreover, neither pressure drop or pulse
rate were significantly affected by boiler outages or other off-line
periods.

The pulse rates measured at Martin Drake are 50-70% lower than
values reported for similar pulse jets operating on coal-fired
boilers in Australia  (4.)•  Although the fly ash concentration at
Martin Drake is typical of many U.S. coal-fired boilers, it is as
much as a factor of five lower than the "normal" fly ash
concentrations experienced in Australia.  Consequently, since pulse
jets have delivered good performance overseas with exceptionally
high fly ash inlet concentrations, the Martin Drake results suggest
that the low pulse rates required to maintain pressure drop below 5
in. H20 at an A/C of 4 acfm/ft2 are reasonable.
Particulate Matter Emissions

Particulate matter emissions tests were performed on several
occasions during both the Daytex and Ryton/Ryton test periods.
Averages of the test data are summarized below.

Fabric

Daytex
Daytex
Ryton/Ryton
Ryton/Ryton
Ryton/Ryton
Service
Hours
1700
2800
2900
3700
4600
Inlet Cone.
gr/dscf
1.91
2.40
2.41
2 .42
2.27
Outlet
gr/dscf
0.0003
0.0002
0.0014
0.0022
0.0003
Cone . 1
Ib/MBtu
0.0005
0.0004
0.0025
0.0042
0.0005
3ffic
\
99
99
99
99
99
ienc1
5
.98
. 99
.94
.91
. 99
                                  9-13

-------
Results show that high particulate matter collection efficiency was
achieved for both bag types throughout the test program.  Further,
outlet opacity remained below 1% (clear stack plume).   The first two
measurements with the Ryton/Ryton bags showed reduced collection
efficiency due to fly ash leaks around the collars of the bags at
the tubesheet.   Following correction of the problem, outlet
emissions measurements were similar to results obtained during the
Daytex bags tests — approximately 0.0005 Ib/MBtu.   The outlet
emissions and opacity values measured are comparable to levels
reported for reverse-gas fabric filters (.5.) .

The fly ash leaks around the bag collars were caused by loosening of
bolts used to secure a few of the cages to the tubesheet.  The
problem was corrected by retightening the bolts per the recom-
mendation of the manufacturer.  The cage bolting system installed in
the Martin Drake pilot plant has been used extensively in industrial
applications containing a relatively small number of bags.  The
system is not designed for repeated removal and installation of
cages as performed at Martin Drake for bag inspections and
measurement of bag weights.  For utility pulse-jet  fabric filters,
which require a significantly larger number of bags than industrial
units, a much simpler system is used that does not  require fastening
of cages to the tubesheet.  Instead,  the seal between the bag and
tubesheet is accomplished by installation of bags that are equipped
with  "snap-in" top cuffs.  The cages are then inserted into the bags
and rest on the tubesheet with no fastening required.   Consequently,
the outlet particulate matter concentrations measured for the
Ryton/Ryton bags at 4600 hours (after tightening of the cage bolts)
are most representative of the performance capability of this bag
fabric in utility applications.

Operation and Maintenance

The pulse-jet pilot plant operated without bag failures.  Start-up
of the unit with new Daytex and Ryton/Ryton bags was performed
without precoat.  Further, the pilot plant frequently operated below
the sulfuric acid dewpoint, and was shut down on numerous occasions
without purging flue gas from the unit.  These operating excursions
appeared to have no adverse impact on pulse-jet performance.
                                  9-14

-------
Measurements conducted at the conclusion of each test period showed
average residual dustcake weights of 0.21 lb/ft2 on the Daytex bags,
and 0.12 lb/ft2 on the Ryton/Ryton bags.  The lower residual dustcake
weight on the Ryton/Ryton bags may be due to use of 19 oz/yd2 glazed
fabric and 10-wire cages as discussed previously.

SUMMARY

Evaluation of a 1 MWe pulse-jet fabric filter pilot plant was
performed at the Martin Drake station of the Colorado Springs
Department of Utilities.  The pilot plant was operated at an air-to-
cloth ratio of 4 acfm/ft2 using felted Daytex bags for 2800 service
hours, and felted Ryton/Ryton bags for approximately 4600 service
hours.  Results show that on-line cleaning at a pulse pressure of
approximately 12 psig easily maintained tubesheet pressure drop
below 4.3 in. H20 for both the Daytex and Ryton/Ryton bags.  The
pulse rate for the two bag fabrics increased steadily up to approx-
imately 1800 service hours, and then appeared to level off between
8-18 pulses/h.  Outlet emissions were measured to be 0.0005 Ib/MBtu,
and outlet opacity less than 1% (clear stack plume), for both the
Daytex and Ryton/Ryton bags.  The pilot plant operated without bag
failures.  The results indicate that pulse-jet fabric filters offer
utility companies an alternative particulate matter control option
that is approximately one-half the size of conventional reverse-gas
fabric filters, yet can deliver comparable performance.
ACKNOWLEDGEMENT

The authors acknowledge the technical and financial support of the
project sponsors.   The sponsors include the Electric Power Research
Institute  (EPRI),  Colorado Springs Department of Utilities,
Consolidated Edison Company,  Empire State Electric Energy Research
Corporation (ESEERCO), Houston Lighting & Power Company, Southern
California Edison Company, Public Service Company of Colorado, and
Howden Canada.
                                  9-15

-------
REFERENCES

1.  R.  C.  Carr and C.  J. Bustard.  "Pilot Demonstration of a Pulse-
Jet Fabric Filter on a Coal-Fired Utility Boiler."  Presented at the
Fifteenth Biennial Low-Rank Fuels Symposium, St. Paul, MN, May 1989.

2.  R.  C.  Carr,  A. Th.  Vandewalle, and W. B. Smith.  "Pulse-Jet
Fabric Filters for the U.S. Electric Utility Industry."  Presented
at the Seventh Symposium on the Transfer and Utilization of
Particulate Control Technology, Nashville,  TN,  March 1988.

3.  R.  C.  Carr.   "Pulse-Jet Fabric Filters  Vie  for Utility Service."
Power,  Vol. 132,  No.12,  New York, NY,  December 1988.

4. Personal communication with representatives of the Electricity
Commission of New South Wales,  Sydney,  Australia.

5.  W.  B.  Smith  and R.  C. Carr.  "Reverse-Gas Baghouses for the
Collection of Fly Ash."   Presented at  the Third Conference on Fabric
Filter Technology for Coal-Fired Power Plants,  Scottsdale, AZ,
November 1985.
                                 9-16

-------
                  ACID  RAIN  REGULATIONS  IN  GERMANY
                         AND THEIR EFFECTS

                          DR.  PETER  DAVIDS
                          President  of  the
               State  Agency  for  Air  Pollution  Control
                        and Noise Abatement

                               Essen
                    Federal Republic of Germany
Acid Rain

Today,  "acid  rain"  is  a  term  used for  the  description  of several
different  and  rather  complicated  phenomena   in  the  environment
(Fig. 1) . In  general,  it  characterizes  the  problems  of  long range
and  transboundary  transport of  air pollutants.  Environmental  ef-
fects attributed to  acid  rain are  in  particular  the acidification
of lakes and  the novel  forest decline.

In Central Europe  the public discussion  is  focusing  on  the causes
and  consequences of  forest decline.  Today,  we know  that  a lot of
mechanisms is affecting  this  phenomenon. The  impact of  air  pol-
lutants  must  be regarded as a  main  factor.  But only  under  this
aspect  the characterization  of  the  relevant parameters  is already
difficult. Two main impacts seem to be decisive:

           - deposition of acid forming  substances
           - effect of photochemical oxidants.

Countermeasures must  focus on  the  emission reduction of  all  air
pollutants contributing to these factors.

Under this aspect  the  most relevant compounds are  sulfur oxides,
nitrogen oxides and  volatile  organic  compounds  (VOC).  Sulfur oxi-
des and nitrogen oxides are acid  forming substances.  Nitrogen oxi-
des  and  VOC's are precursor  substances, from  which photochemical
                                Pl-1

-------
oxidants are formed in the atmosphere  in  the  presence of solar ra-
diation. Principally,  also  particulates  could have  an  impact due
to catalytic  effects  of  certain  particulate  compounds  on conver-
sion reactions  in the  atmosphere;  but  quantitative data  are not
available.

Thus,  the solution of  the acid rain problem  requires a comprehen-
sive strategy for a simultaneous  emission reduction of  several air
pollutants .
Sulfur Oxides

Reduction measures for  sulfur  dioxide   (Fig. 2)  were  already ini-
tiated a long time before the acid rain discussion arose.

Caused  by  unacceptable  levels  of  ambient  air  concentrations  of
sulfur dioxide in  the  sixties,  reduction measures  were initiated.
Due to  the dominating  contribution of the  burning  of  fossil fuels
to  the.  sulfur dioxide  emission,  different  kinds of  limit values
     been set and were stepwise reduced.
Fig. 3 shows  the  sulfur  content  limit for light  fuel  oil and Die-
sel fuel.  The sulfur content was  reduced from 0,7 %  by  weight in
the     sixties to 0,2 % at present.  In  spite  of  a consumption in-
crease of  light  fuel oil and  Diesel fuel, SOa emission  from this
source has been reduced for about 70 %.

In the past,  the  main source  for  sulfur  oxide  emissions were large
combustion plants,  in particular  power  plants.  Fig. 4  shows  the
S02 emission  standards for  large  combustion  plants.  Coal- and oil-
fired plants  with  a  thermal capacity  of  more  than 300 MW are re-
quired to  apply flue  gas desulf ur ization with  a removal efficiency
of at least  85 %,  corresponding  to  a  maximum  sulfur  emission rate
of 15 %,  and  to meet an  emission standard of  400 rng/m3 .  The emis-
sion concentration is corrected to  a certain  oxygen content in the
flue gas,  depending on the fuel type.

Plants with a thermal capacity of less  than  300 MW  have to meet  a
standard equivalent to a sulfur  content of 1 % in  coal  or oil and
                               Pl-2

-------
additionally a  sulfur emission  rate based  on the  application of
lime injection techniques.

In  an  ordinance of  1983, application  of  these  standards  is  not
only required  for  new plants,  but  also  for existing  plants.  The
retrofit program was completed in 1988.

By  all  these measures,  SO2   emissions  in  the  FRG  are  decreasing
since the  seventies  (Fig. 5). Most  effective  was the  retrofit of
existing large  combustion plants  in the last  years.  In  the  mid-
ninties we shall achieve  an  overall  reduction  rate of  75 %, compa-
red to the early seventies.

The impact on ambient  air concentrations  is significant,  as demon-
strated in  Fig.  6.  It shows the development  of  SO2  ground level
concentrations in the  Rhine-Ruhr-area,  the  most industrialized and
most densily  populated area  in  Germany,  where  about  50 %  of- the
fossil  fuel  based  electric  power  generating capacity  is  located.
The biggest  percental  reduction  was achieved  since  the completion
of  the flue gas desulfurization retrofit program in 1988.
Particulates

Particulate control  comprises different  aspects (Fig. 7).   In  the
FRG, as  in the  US,   the  application of  highly  efficient  control
equipment has a  long tradition,  Emission  standards  below 50 mg/m3
are usual. Fig.  8 shows the standards for large combustion plants.

Particulate emissions  are decreasing  continuously since  the six-
ties (Fig. 9).  Three  decades ago,  particulate  control  was  the
starting point of systematic  air pollution control.  In view of the
achieved  low  emission level  the  potential for  further reductions
is limited.

The impact  of control  measures  on  ambient air  concentrations  is
significant  (Fig. 10).  Dust  Fall   has  decreased  more  than 65 %;
Concentration of TSP  also about 65  %  (Fig. 11). The  current con-
centrations are  far  below the ambient air quality standards.
                                Pl-3

-------
Today,  control  activities  have  to  focus on  particulate compounds
with high effect  risks.  Thus,  in addition  to  the general particu-
late emission standards  specific  standards  for certain particulate
compounds have been set. Divided  into  three classes,  the standards
range between  0,2 and  5 mg/m3 .  On top  of  that,  for carcinogenic
and  extremly  toxic  substances  a  minimizing strategy  is required.
An example are dibenzodioxines.

Emission  reduction  of particulate  compounds  means more stringent
standards  for  total   particulate   emission,  because  a  selective
removal of particulate  compounds  is not  possible.  Usually,  appli-
cation of extremly  efficient control  technologies  is  required,  in
particular fabric filters.

Due  to the control  measures initiated two  decades  ago,  a signifi-
cant progress  in  reducing  the  ambient  air  concentration  of heavy
metals has been achieved. The  current  figures  are far below criti-
cal  levels.  Examples   are  the lead  concentration and  the cadmium
concentration (Fig.  12 and  13).

A not  fully  identified  problem might  be the  environmental  pollu-
tion by  highly toxic  organic  particulates,  like dibenzodioxines.
The  knowledge on  emission  sources,   ambient  air  concentrations,
soil contaminations and  transfer  mechanisms in  the  environment is
still very limited.
Nitrogen Oxides

Compared to  sulfur  oxides  and particulates,  the  public discussion
on  nitrogen  oxides  arose  in Europe  relatively late,  namely when
air  pollution  was identified  as  a main  cause for  forest decline
(Fig. 14).  Due to  the  two  contributing factors, acid formation and
oxidant precursor, control activities  for NOX are most  important.

For  stationary sources  reduction  programs are  close to completion
or  in  progress.   Fig.  15   shows  the  emission  standards  for  large
combustion plants, set  in  1983.  They  are based on the application
of low NOx  burners  at  plants with a  thermal  capacity of less than
300 MW and  flue   gas  denitrification  above  300 MW.  The standards
                               Pl-4

-------
also apply  to  existing  combustion plants  with more  than 300 MW.
The  retrofit  program  for  flue  gas denitrif ication  at  all  power
plants will be  completed  in  1990.  Selective catalytic reduction is
the  preferred  technology.  NOX  emission  from  powerplants  will  be
reduced by about 70 %.

For other industrial  processes  specific standards  have been set in
1986 (Fig. 16). They  also  require  application  of flue gas denitri-
fication  to major  sources. Due to  limited  technological experien-
cies  completion of  the  retrofit  program  for  existing  plants  is
terminated until 1994.

For  small combustion  plants,  in particular  for  domestic heatings,
emissions standards  can  only  be set within  the  European Community
and  require the unanimous  approval by  all  member  states. Alterna-
tively, in the  FRG national  guidelines  have been set for different
burner types,  which  are  based on low NO*  combustion.  Burners  mee-
ting those  recommendations can  get a  special  environmental  seal,
which  can be  used  for advertising  and  marketing.  This possibility
is widely used by burner  manufactures  and has  appeared  as  a  very
effective instrument.

Similar problems  are existing  with motor  vehicles.  Standards  can
only be  set  by a  common  decision  of  the  European  Community.  Be-
cause  of  different interests  of the individual  member  states,  for
more than a decade no progress  has  been achieved.  Initiated by the
FRG  stringent  emission  standards,  depending on  the  engine volume,
were set  stepwise,  since  the  mid-eighties.  The  final  step  for
small  engines  will  come into force in  1992.  Then,  catalytic  con-
verters will  be obligatory  for  all new  cars  with  spark ignition
engines.

Already prior  to  these  European regulations the Government of the
FRG  has  initiated an national   impletation program  for catalytic
converters on  a voluntary  basis.  The instruments  are  tax reducti-
ons  for  cars   which  already meet  standards equivalent   to  the  US
standards and  tax  penalties for cars  which do  that  not. The  pro-
gram is very successful  but  nevertheless  in Europe the application
of catalytic converters is a decade behind  the US.
                               Pl-5

-------
The unsatisfactory  development with  motor vehicles  is  the reason
for steadily  increasing  NOX   emissions  in the  FRG,   because  more
than 50 %  of  the total emissions  are generated  by motor vehicles
(Fig.  17). A  stabilisation  has been  achieved end  of  the eighties
by  the  measures  at  stationary  sources.  A  significant reduction
will be achieved in  the  second half of  the nineties,  when most of
the presently existing motor  vehicles  will  be  replaced  by  cars
with catalytic converters.  Than,  the  remaining  problem will be NO*
reduction  from duty  vehicles,  which are usually  powered by Diesel
engines .

The development  of  ambient  air concentrations  is corresponding to
the emission  situation  (Fig.  18).  Concentrations remain at  a con-
stant level since the beginning of the eighties.

A  comparison  between two monitoring  stations  close  to  frequently
used roads  and  the  average of all stations  in the industrialized
Rhine-Ruhr-area  indicates the  outstanding impact of  motor vehicle
emissions  to  ambient air  concentrations  (Fig.  19) .  The  average is
set to  100 %. The  measurement data  indicate  a  rather  comparable
situation  for S02 , TSP and  NO2 . However,  for  nitric oxide (NO) and
CO  the situation is  completely different.  In  the vicinity of roads
ambient  air  concentrations  of  these  two compounds  are  several
times higher  than  the  average.  This  underlines the  necessity of
emission reduction measures with motor vehicles.
Volatile Organic Compounds  (VOC)

The term  "Volatile  Organic  Compounds" covers a  wide  range of sub-
stances with  extremely  different effect  risks,  e.g.  dibenzodioxi-
nes on  the  one  hand and methane on the other  hand (Fig. 20). From
the viewpoint  of  long  range  air  pollution  and the  formation of
photochemical oxidants  reactivity with  nitrogen oxides in the pre-
sence of solar radiation is the outstanding criterion.

In total,  industrial  processes are a emission  source of minor  im-
portance  (Fig. 21) .  Emission  standards for  organic compounds have
already been  set  in 1974 and were reduced  in  1986. Application of
                               Pl-6

-------
control technologies  is  usual.  The  reduction  potential  by further
measures is limited.  Activities  must focus specifically on certain
compounds with a high effect risk.

A  much larger  amount of  total  VOC's  is  emitted from  motor ve-
hicles. The situation is comparable  to  nitrogen oxides. Emissions
have increased  due  to the unsatisfactory  regulations  of the Euro-
pean Community in the past.  As  a  result  of the standards set mean-
while  a halving  of  the  emission  from this  sector  is  estimated un-
til the end of the nineties.

Other  important  sources  are  small  plants, in  which  organic com-
pounds  are used  as  solvents,   and  solvent   containing products.
Examples are dry cleaning  and the use of solvents  in  paints. Emis-
sions  are  approximately  constant  since   two  decades.  Reduction
measures are available,  e.g.  the  replacement  of solvents in paints
and the use of water  soluble paints.

But, in the European  Community  we are faced with  the  same problem
as  in  the field  of  motor vehicles.  Setting  of product standards
requires an unaminous agreement.  Alternatively, in the  FRG natio-
nal recommendations have  been  set for low  solvent containing pro-
ducts.  Those  products  can  be  marked by  a special  environmental
seal for public  information  (Fig. 22 and  23).  Within  the next de-
cade a halving of the emission from  this source type is  expected.
Photochemical Oxidants

Photochemical oxidants  seem  to be a major  factor  countributing to
novel forest decline  (Fig.  24).  Usually, ozone  is  used  as indica-
tor for  these  substances.  It is formed  in  the  atmosphere from ni-
trogen oxides and hydrocarbons in the presence of solar radiation.

Corresponding  to  the emission situation with  the  precursor sub-
stances  ambient air  concentrations  show an  increase  rather  than a
decreased
A comparison between  monitoring stations in  urban  and rural areas
indicates that in contrast  to all  other compounds ozone concentra-
                                Pl-7

-------
tions in rural areas are higher  than  in urban areas  (Fig. 25). The
average of the concentrations  in urban areas  is  set  to 100 %. The
effect of  the long  retention time  of the  precursors in  the at-
mosphere and the reaction  during long range transport are obvious.
That's also  an explanation  for   the  serious novel  forest decline
observed in rural areas far away from industrial regions.

A control  strategy  for photochemical  oxidants must  include  a si-
multaneous reduction  of  nitrogen  oxides  and  hydrocarbons.  As far
as we know, a  significant  reduction of  the  ozone level requires an
emission reduction of  the  precursors  of at  least  50  %.  In Europe,
this can be achieved only by internationally concerted activities.
Summary

Summarizing my presentation it can be  concluded  that  a lot of pro-
gress in reducing problems attributed  to  the  so  called "acid rain"
has been achieved.  But  nevertheless,  several problems  still await
a  solution.  In  particular,  the  international   dimension  requires
comparable measures in  all  European  states, a deficite  whose eli-
mination is  also  a  big challenge  in  conjunction with  the current
dramatic political changes in Europe.
                               Pl-8

-------
               "Acid Rain"
aspects




-  long range and transboundary transport of air pollutants




-  acid deposition




-  formation of photochemical oxidants
compounds




-   sulfur dioxide




-   particulates




-   nitrogen oxides




-   volatile organic compounds (VOC)




-   photochemical oxidants (ozone)
                        Pl-9

-------
                 "Acid Rain"
aspects




   long range and transboundary transport of air pollutants




-  acid deposition




-  formation of photochemical oxidants
compounds
   sulfur dioxide
   participates




   nitrogen oxides





   volatile organic compounds (VOC)




   photochemical oxidants (ozone)
                        Pl-10

-------
      fD
        1. 0 -
        0, 8-
      c
      o


      8
      •i-i
      E
      QJ
      cu
      O
      Ul
      -100
                                                          ra
          c
          o
          a.
          E
          D
          w

          o
          u
                                                        0
          1965
89
           — consumption of light fuel  oil  and diesel fuel
           ==sulfur content limit
           ===== S02-emission
L  1  S
                      862 Emission from
               Li
           D

-------
              sulfur emission  rate
                 for coal  and oil
              <50%
        <40%
  SO;
        2000
  mg/m  1800  H
        1700
        1600
        1400

        1200

        1000

         BOO

         600

         400

         200

           35
                     coal
       oil
02 - correction:
oil/gas       : 3%
 coal:
— wet bottom
— dry bottom
— stoker fired
5%
6%
7%
          gas
                   coal/oil
              50  100           300 MW 400
                  thermal capacity
L  IS
           SC>2 Emission Standards for

            Large Combustion Plants
                         D
                      90-23
                      Pl-12

-------
106t/a
              motor
              vehicles
              domestic
              heatings
              power plants
              and industry
       1966  70
78
82
86
1998
       Sulfur Oxide Emissions (as 802)

                   in the FRG

-------
|ag/m3  250
 5-4

 OJ

        200-
150-
        100-
         50-
          0
Z7

 I
 1
                                   Z7
                                    .
-------
                "Acid Rain"
aspects




- long range and transboundary transport of air pollutants




- acid deposition




- formation of photochemical oxidants
compounds



- sulfur dioxide
   particulates
-  nitrogen oxides




-  volatile organic compounds (VOC)




-  photochemical oxidants (ozone)
                        Pl-15

-------
Particu-
lates
mg/m3
          75 -
          50
          25
           5
           0
                       coal/oil
02 - correction:
oil/gas
coal:
—  wet bottom
_  dry bottom
—  stoker fired

        gas
3%


5%
6%
7%
             50 100           300  MW 400
                 thermal  capacity
          Participate Standards for

          Large Combustion Plants
                     D
                   90-25
                    Pl-16

-------
    106t/a
          3-
          2-
               motor
               vehicles
               domestic
               heatings
               power plants
               and  industry
             1966   70
78   82
86
1998
L  I SI
                   Particulate Emissions
                         in the FRG

-------
00
        g-nr^d'1  0.5
                0.4-
          cd
               o.o
                   O^a^O^O^O^O^a^O^O^O^O^O^O^a^a^O^C^\a^O^O^O^a^O^O^O^
                      Ambient  Air  Concentration of
                                  Dust  Fall
  210

02/90

-------
Mg/m3  200-1
 CD
 b
 cti
 t-i
 (D
 ctf
      150-
      100-
      50^
            Ambient Air  Concentration of
                Suspended  Participates
                         (TSP)
 210


02/90

-------

Mg/m:
 CD
 bJ)

Ambient Air Concentration
        Lead in  TSP
                                                   i
                                                   i
                                             210
                                            01/90

-------
                 B^^

ng/m-
12-





10-





 8-





 6-




 4-





 2-
       0
                        Z_7
                         ss
                           7~7I
          O^O^O^O^O^O^O^O^a^O^a^O^O^O^O^a^a^O^a^
            Ambient Air Concentration of

                    Cadmium in TSP
                                                 210



                                                01/90
                                           aa^^^

-------
                11 Acid Rain"
aspects




-   long range and transboundary transport of air pollutants




   acid deposition




   formation of photochemical oxidants
compounds




-  sulfur dioxide




   particulates
   nitrogen oxides
   volatile organic compounds (VOC)




   photochemical oxidants (ozone)
                       Pl-22

-------
           500 -
  NOX (N02)
  mg/m3
           400
           300 4-


           200

           150

           100
             0
                      coal
                      oil
     gas
    •^^•^^^^•^^H
02 - correction:
oil/gas : 3%
coal:
—  wet bottom    5%
—  dry bottom    6%
—  stoker fined  7%
coal
Oil
• ^™ ^™
gas
               50  100            300 MW 400
                   thermal  capacity
L  /  S
            NOY Emission Standards for
               A

             Large Combustion Plants
                        D
                     90-24
                    Pl-23

-------
  Emission Reduction Technologies
          for Nitrogen Oxides
small combustion plants:
   low NO  burners

large combustion plants:
   flue gas denitrification

industrial processes (e.g. melting):
   low NO  combustion or flue gas denitrification,
   depending on plant size

 motor vehicles:
   •  spark ignition engines: catalytic converter
   «  diesel engines: low NO combustion
                          A.
                   Pl-24

-------
     4 i
106t/a
     3
     2-
     0
                    motor
                    vehicles
                    domest ic
                    heatings
                    power  plants
                    and industry
        1966   70
74
78
82
86
1998
      Nitrogen Oxide Emissions (as NO9
                                           A*

-------
Mg/m3   50
  CD
          1981    1982   1983   1984   1985   1986   1987   1988   1989
             Ambient  Air  Concentration  of
  210



02/90

-------
jag/m3  2001
  o
  •i-H
  -4— •
  cti
  5-H
  -j— >
  a
  
  o
  c
  o
  U
  (D
150-
100-
       50-
                       May - December 1989
               Rliein-Rulir
             ._ Essen-Ost
             LJ D-Morsenbroich
                               TSP
                                        NO
                                                     /  7
             Comparison between ambient        210
         concentrations near frequently  used
           roads  and  in the Rhine-Ruhr-area    03/90

-------
                "Acid Rain"
aspects




-   long range and transboundary transport of air pollutants




-   acid deposition




-   formation of photochemical oxidants
compounds




   sulfur dioxide




-  particulates




   nitrogen oxides
   volatile organic compounds (VOC)
   photochemical oxidants (ozone)
                       Pl-28

-------
     4i
10bt/a
     3-
     2 -
     1 -
     0

                          motor
                          vehicles

                          solvents

                          domest ic
                          heatings
                          power plants
                          and industry
         1966   70
74
78
82
86
1998
          Volatile Organic Compound Emissions (VOC)

                         in the FRG

-------
Pl-30

-------
                "Acid Rain"
aspects




-  long range and transboundary transport of air pollutants




-  acid deposition




   formation of photochemical oxidants
compounds




   sulfur dioxide




-   particulates




-   nitrogen oxides




-   volatile organic compounds (VOC)
   photochemical oxidants (ozone)
                       Pl-31

-------
          )ug/m3  35
GO

IV)
           cd
           5~i
           
-------
CO
CO
                       Rhem-Ruhr
                       Nettetal
                       Eggegebirge
                       Eifel
                       Rothaargebirge
                          Comparison  between             210
                 ambient  concentrations  at stations in
                  rural areas and the Rhine-Ruhr-area    03/90

-------
          PARTICULATE  EMISSIONS  CONTROL  AND  ITS  IMPACTS  ON  THE
             CONTROL OF OTHER AIR POLLUTANT EMISSIONS FROM
                       MUNICIPAL WASTE COMBUSTORS
                 Theodore G. Brna and James D. Kilgroe
                 U. S. Environmental Protection Agency
             Air and Energy  Engineering  Research  Laboratory
                 Research Triangle  Park,  North Carolina
ABSTRACT

     On December  20,  1989,  the  Environmental Protection  Agency  (EPA)
proposed revised  new source performance  standards for  new municipal
waste combustion  (MWC) units  and  guidelines  for  existing sources.  The
proposed national  regulations require  tighter  particulate matter  control
and address pre-combustion,  combustion, and  post-combustion controls,
the latter  two depending  on capacity and age of the  facility.

     The air pollutants  of  concern  when municipal solid waste (MSW)  is
burned will be discussed.  Generally,  particulate  control is  an inherent
part of the systems  used  to limit  the  emissions of these air  pollutants.
The relationships  between MWC air  emissions  (acid  gases, trace organics,
and  trace   heavy  metals)   control  and particulate  control will  be
discussed.   Test  results to quantify air pollutant emissions from MWC
units and their control will be  presented and  compared  with the proposed
regulations.


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.
                                   P2-1

-------
          PARTICULATE EMISSIONS CONTROL AND ITS IMPACTS ON THE
             CONTROL OF OTHER AIR POLLUTANT EMISSIONS  FROM
                       MUNICIPAL WASTE COMBUSTORS
INTRODUCTION

     The combustion of municipal  solid waste  (MSW)  has  nearly doubled
over the  past 3  years  in  the  U.S.,  with  waste-to-energy  conversion
complementing the  volume reduction of waste to be landfilled.  Now about
15% of the  160 million tons of MSW  generated  annually is being burned,
with about  63,000  tonnes/day  (70,000 tons/day)1  being  combusted in
energy  recovery  facilities.   While municipal  waste combustion  (MWC)
ranks after waste reduction and material  recovery  (recycling)  in EPA's
agenda for solving the MSW  crisis, it precedes land disposal—a disposal
option  now  facing higher  costs,  rapidly  shrinking  existing capacity,
tighter  siting  constraints,  and  declining  availability  of  land.
However,  along with  the  benefits  of MWC,  there  are  environmental
concerns regarding air  pollutant  emissions,  the safe disposal of ash or
residues,  and the  NIMBY  ("not  in my  backyard") syndrome.

     On December  20,  1989, EPA  proposed more  stringent  new  rules to
control pollutant  emission  from MWC  facilities.2  New Source Performance
Standards  (NSPS)  were proposed under Section 111(b) while guidelines for
existing  sources  were  proposed under  Section  111(d) of  the  Clean  Air
Act.    The  proposed  rules  require  all  facilities  to  use  three
complementary  methods  to  control  air polluting  emissions:   material
separation  (pre-combustion fuel  cleaning),  combustion  control  [good
combustion practice  (GCP)],  and post-combustion  flue  gas  cleaning  (FGC).

     Whereas the  current rule  for municipal  waste  incinerators applies
only  to  particulate  matter  (PM),   the   proposed  rules  address   "MWC
emissions,"  NOX,  and CO.   "MWC emissions" are  defined  to include:"MWC
metals" as  measured by  total PM, "MWC organics"  as  measured by total
polychlorinated dibenzo-p-dioxins (CDD) and  dibenzofurans (CDF)   (tetra
through octa  isomers) ,  and  "MWC  acid gas"  as measured by  hydrogen
chloride  (HC1) and  sulfur dioxide (S02) .   As  shown in  Table  1,  the
proposed  rules   incorporate   emission  limits  which  depend  on  the
facility's  size  (capacity)  and age  (existing or new).   All facilities
must comply with  PM, CDD/CDF,  and CO  emission limits.  All facilities,
except  small  existing  sources [<225 tonnes/day  (<250 tons/day)],  must
also comply with acid gas  emission limits.  Only large new  sources [>225
tonnes/day  (>250 tons/day)] must comply with NOX  emission limits.  The
new rules  also include  facility operating  and monitoring  requirements to
ensure compliance  with the  proposed  emission limits.
                                  P2-2

-------
     Most  advanced  air  pollution  control  techniques  used  at  MWC
facilities  are  designed  to  convert metals, organics, and acid gases to a
solid  form which  can  be  collected  as PM.    Thus,   strategies  for
controlling organic emissions,  metals emissions,  and acid gas emissions
are directly or indirectly related to the techniques for controlling PM
emissions.    The  focus  of this  paper  will  be  a discussion  of  the
relationship between  combustion and FGC process  conditions which affect
collection  of organics,  metals,  and acid gas reaction  products  as  PM.
The potential effects of material separation and  NOX  emission  control
technologies on PM emissions are outside the main  focus  of  this  paper
and will not be discussed.   Similarly,  the treatment or disposal of  ash
or solid residue  will  not be discussed as this waste is subject to solid
waste rules .

COMBUSTION  CONTROL  OF  AIR POLLUTANTS

Good Combustion Practice (GCP)

     GCP is to be employed on all combustors to  limit the formation  and
emission  of organic  emissions.   Total  CDD/CDF  was selected  as  a
surrogate  for  "MWC  organics"  because  of  concern  regarding  their
potential health effects.   Also, more is known about  the  emission  and
control of these compounds than  other  organics.   The  GCP  provisions of
the  proposed  MWC  rules  include  requirements  related  to  combustor
operating  conditions  and flue gas  CO  concentrations.   Operator
certification and MWC  personnel  training  are also required.

     Organic emissions  may originate  from  compounds contained  in  the
waste  which  are  not  destroyed  during  combustion,  waste  thermal
decomposition products which are not completely  destroyed,  and chemical
reactions which occur at relatively  low  temperatures  downstream of  the
combustor.   Three goals  of  GCP are  to:    (1)   maximize  in-furnace
destruction of organics, (2)  minimize entrainment  and carry-over of  the
PM from the furnace,  and (3)  minimize the occurrence of low temperature
reactions which form  CDD/CDF.

     The concentration  of  CO  in MWC flue gases is  a good  indicator of
combustion  conditions  associated with  the  destruction  of  organics:
appropriate waste feed conditions; adequate combustion temperatures;  and
the  proper amount, distribution,  and mixing of  combustion air.   To
ensure  continuous  furnace destruction of organics,  the  proposed  MWC
rules specify CO emission limits  ranging  from  50 to  150 ppm depending on
the type of combustor  (Table 2).

     Low temperature  reactions downstream of  the  combustor lead to  the
de novo  synthesis  of  CDD/CDF  on the  surface  of  flyash.3'4  Design  and
operation of the combustor at  conditions which minimize  downstream PM
surface  area  are  deemed to be  key factors  in  limiting  post-furnace
formation of CDD/CDF.   Figure  1 shows  that CDD concentrations  at  the
electrostatic  precipitator  (ESP)  are highly correlated  (correlation
coefficient,  r2  =  0.899)  with the  total PM carried from  two types of
combustors.5  High  levels of PM  carryover are generally related to flue
gas volumetric  flow rates above  design  conditions and improper overfire-
to-underfire air ratios.   The proposed MWC  rules  prohibit operation in
excess of 100% rated  load  as measured  by steam  load to avoid excessive
PM carryover.
                                  P2-3

-------
     There is  substantial evidence that   de novo  synthesis of CDD/CDF
occurs on the  surface  of flyash at temperatures  ranging from less than
250°C  (480°F)  to  more  than  400°C  (750°F) .    Maximum net  rates  of
formation occur at approximately 300°C  (570°F).   MWC facilities equipped
only with an ESP for PM control exhibit higher CDD/CDF concentrations at
the outlet than at the inlet  for ESP  operating temperatures higher than
approximately  230°C  (450°F),  an indicator  that  PM  control devices can
operate  as  reactors which generate  CDD/CDF.    Thus,  the  proposed MWC
rules specify a maximum temperature of  230°C (450°F) at the  inlet to the
PM control device to limit CDD/CDF formation.

Acid Gas Control

     While  in-furnace  sorbent  injection  is being  used  in  several MWCs
to  control  HC1  and  SC>2,  the degree  of  control  is estimated because
uncontrolled  emissions  are  not  measured  directly when  this  control
option  is used.   Tests on an Alexandria, VA,  mass burn waterwall unit
equipped with  an  ESP gave average estimated HC1  and SC>2 removals of 67
and  82%, respectively, for average  outlet HC1  and  SC>2 concentrations of
166  and 37 ppmv, both referenced to dry gas with  7% 02.6  Hydrated lime,
supplied at  68 kg/hr (150 Ib/hr)  and  corresponding to a stoichiometric
ratio of 0.9,  was  injected into the combustor  with  the overfire air for
each of the three test  runs.   Other ESP outlet  (controlled) emissions
during  the  1987 tests were 55  ng/dscm*  (correlated  to 7%  02)  for CDD/CDF
and  54 mg/dscm  (0.024  gr/dscf)  for  PM versus  a  permit   limit  of 69
mg/dscm (0.03  gr/dscf),  with  the  PM data being referenced  to  12%  CC>2 in
dry  gas.*   Since  the  PM emissions for the two  other  Alexandria units
without sorbent  injection averaged  61  mg/dscm   (0.027 gr/dscf),  the
injection of  hydrated lime into  the  furnace did  not  degrade the ESP' s
performance during  the  tests.  However,  PCDD/PCDF data for these two
units were  not reported.

      The 295   tonne/day   (325  ton/day)   mass  burn  unit  tested  at
Alexandria  would meet  the proposed air emission guidelines  for organic
and PM emissions,  assuming  the  unit  performed  as tested in December
1987.    It  would  also meet  the  least stringent   acid  gas guidelines
because over  50%  removal of  both the  inlet  HC1  and  SC>2 was attained,
although the  outlet concentration of  both  HC1  and SO2  exceeded the
alternative guideline concentrations.   No data were  available on the
control of  metal emissions.

      Fluidized  bed combustors operating  in  Japan  control  acid gas
emissions by adding sorbent,   such as  limestone  or  dolomite, to the bed
material, which  is frequently sand.   With lime powder injected into  the
combustor,  the performance data  reported  for the Machida  plant,  which
has three  150 tonne/day  (165 ton/day)  units,  ranged from  15  to  26
mg/Nm3
-------
and 3 to 5 ppmv for SO2 versus its standard of 20 ppmv.7  All  emissions
data were  collected downstream  of the  ESP following  each combustor.
Results for the February  1984 tests at  the  Fujisawa City plant, which
has  three  fluidized  bed  combustors  rated  at  130  tonnes/day   (145
tons/day)   and  adds dolomite  to  the  revolving sand bed for  acid gas
control,  ranged  from 5 to  10 mg/Nm3  (0.002  to  0.004 gr/dscf)  for PM
compared with its  standard of  30  mg/Nm3 (0.013  gr/dscf),  non-detected to
120 ng/Nm3  (42 ppmv)  for  SC>2,  and 62  to  95  ppmv  for HC1 compared  with
its standard of 100 ppmv.8  All  emission measurements were made at the
ESP outlet, and the  Japanese  emissions  are  referenced to dry gas  with
12% 02.


PARTICULATE MATTER CONTROL

     For both the proposed  NSPS  and the  emission guidelines, PM  control
is used as a surrogate for controlling heavy metals  [e.g., arsenic  (As),
cadmium (Cd), chromium  (Cr), lead  (Pb),  mercury  (Hg)]  emissions.  Since
a  PM emission  limit of 34  ng/dscm (0.015  gr/dscf)  corresponds to 99%
removal of all heavy metals of concern  (except Hg) and a  PM limit of 69
ng/dscm  (0.03 gr/dscf)   corresponds  to  97%  removal  of heavy metals
 (except Hg), PM collection downstream  of the combustor  will  be  needed to
meet the  proposed rules.6  The  discussion on  the  proposed  rules notes
that Hg removals  of  70%  with a  lime  spray dryer absorber  (SDA)/fabric
filter  (FF) system and 30%  with   duct  injection of lime/ESP system  can
be achieved  by  the proposed PM  rules.  However,  these correlations are
being studied by EPA because other  results show  inconsistent Hg  control
at several plants.

POST-COMBUSTION CONTROL OF AIR POLLUTANTS

Acid Gas Control

     Dry  scrubbing systems  continue to be  preferred  for  controlling air
pollutant  emissions  from  MWC  units in the U.S.   While  several  plants,
generally  with  combustors  rated at  180 tonnes/day  (200 tons/day) or
less, use in-duct (or duct)  injection  of powdered  hydrated lime followed
by a FF or baghouse,  most  recent  plants and those  being planned use  lime
SDAs followed by FFs or ESPs.   Figures 2 and 3 are  schematic diagrams of
these  systems.   The  duct  sorbent  injection  (DSI)  system is an  all-dry
system, while an alkali slurry is injected into the  SDA.   Dry  solids and
unsaturated  clean flue gas are  discharged  from  the SDA/PM collection
system.  Thus,  the latter  system is often called  a  semi-dry, rather  than
a dry scrubber.

     While acid gas  scrubbers in MWC plants are intended primarily for
controlling  HC1  and SO2,  they  also  remove  organic and  heavy metal
pollutants.  Several factors appear to be responsible for removal of the
organics and metals.  First, the flue  gas  entering the dry  scrubber has
been cooled, by heat recovery  components or humidification following the
furnace/boiler system, to about  150°C(300°F)  for DSI and to about  200°C
 (400°F)  at  the  SDA  inlet.     Thus,  most   heavy   metals  which  were
volatilized from  the  MSW  feed during combustion are condensed onto ash
or adsorbed onto  the surface of  other  flyash.  These particles are  then
removed  with PM  partly  in the SDA,   if  present,  but  mostly in the
following  FF or  ESP.   While  chemical  reactions leading  to solid-phase
                                  P2-5

-------
products of  Ca,  Cl,  and  S  are responsible  for  removing HC1  and SC>2,
chemical reactions  converting metals  to  solid-phase compounds  in the
scrubber are not believed to be significant in controlling heavy metals
emissions.

     Flue gas cleaning techniques now  used  in  the U.S.  can achieve the
proposed PM emission limits and adequately  control the  metal emissions
of concern,  except possibly Eg2.  Hg  in MWC flue gas has been reported
to be  predominantly mercuric  chloride and  elemental  Hg,9'10  with the
former representing about  70%  of the total Hg.  At  flue gas temperatures
near 150-200°C  (300-400°F) ,  a  reduction of  oxidized  Hg  to elemental Hg
has  been observed.11  One  theory proposed  for  the reduction  is that
steel corrosion  (on duct  walls or  other iron  surfaces) induced by HC1 at
these temperatures results  in an  activated Fe surface  on which the Hg
(II)   compounds  (i.e., mercuric  chloride  and mercuric  oxide)  may be
reduced.11   This  theory  may  be  important in  explaining  the different
ratios  of elemental  Hg to total  Hg present in ESPs and  FFs,  with the
latter often showing lower ratios  and  greater Hg  removal.

     Another  factor that may impact   Hg  control  is  the C  content of
flyash.  RDF combustors,  which typically have higher C content in flyash
than some very combustion-efficient  (as determined  by CO content of flue
gas)  mass burn combustors,  followed by  SDA/FF systems have shown high Hg
removal.  Very  combustion-efficient  mass  burn units  with similar flue
gas cleaning systems have shown little  or  no Hg control.12  In one case,
the  addition  of activated C into  the   flue  gas entering  a lime SDA/ESP
system  improved Hg removal  to about 90%,  with an  emission of 30 (ig/m3
 (referenced to dry gas with  11% 02)-13

     Another  additive used to enhance  Hg  removal from flue  gas is
sodium  sulfide.     A  spokesman  for Flakt,   Inc.,  reported  to  an
EPA/Industry meeting on Hg control issues  on February 7,  1990, that the
addition of this  additive  to flue gas  entering  a fluid  bed reactor
preceded by humidification (see Figure  2)  limited the Hg emission to 100
|ig/Nm3,  an  objective that  was  not  attained with  the  original design.
The  design of this system in Vancouver, Canada, was based  on  pilot-scale
data  from the Quebec City tests,  which achieved Hg removals exceeding
90%  when the pilot unit was operated with a FF inlet gas  temperature of
140°C  (285°F)  or less.14

      Since  the vapor  pressure curves for  elemental  Hg  and mercuric
chloride indicate vapor  pressures  significantly higher  than partial
pressures   for  these  substances in  MWC  flue  gas  at  temperatures
corresponding to dry FGC  system operation,  condensation does  not  explain
their   removal.    Mercuric  chloride  is   normally  a  solid at  these
temperatures and,   if present,  would be expected to be removed in the PM
collector.  Cooling the flue gas  in the acid  gas removal process would
also  be expected  to  enhance  the adsorption  of  Hg  onto  ash,  reaction
products,  and  lime  particles.    It   has  also  been  observed  that   a
noticeable  oxidation  of  elemental Hg  occurs even in the absence of HC1
when activated  C is added to the flue  gas.11  Reference  11 further  noted
that activated  C acts  as  a  catalyst for the  formation of  mercuric  oxide
and  may do  so  even at elevated  temperatures, while the oxidation of
elemental   Hg   to  mercuric  oxide  normally  occurs  in  the  flue gas
temperature  range  of  300-500°C  (570-930°F)  .    Thus,   conversion of
elemental Hg, which  passes  through  an ESP  or FF  as a fine particulate,
                                  P2-6

-------
to  solid-phase compounds would enhance Hg  removal  in  either an ESP or
FF.

     CDD/CDF removal  in  an acid gas  scrubber parallels that  of heavy
metals,  but appears to be aided by the presence of lime-based sorbents.
The  scant  vapor  pressure  data  on  chlorinated  dioxins  suggest  that
condensation of  these compounds  is  not the  sole  removal  mechanism.
Reduced  flue  gas  temperatures,  however,  are  believed  to  promote
adsorption  of CDDs, CDFs,  and other organics onto fine particles having
relatively  large  surface  areas.  The C content of  the particles seems to
affect the  capture of these compounds.  The addition  of activated C to
flue gas has been  reported  to  enhance CDD/CDF  and Hg  capture in a lime
DSI/FF system.15   With  the injection of C  additive,  powdered hydrated
lime, and  filter-aid material,  CDD/CDF  capture exceeded  93%  and Hg
capture  was over  96%,  both  significantly greater  than  without  C
addition.  However, the  individual effects of C  and lime could not be
determined  from the test  conditions.

     Acid gas  control test  results  from nine different facilities with
dry scrubbers are shown in Tables 3 and 4.   These results indicate that
the currently used technologies can control  HC1 and SO2 to levels needed
to  meet  the proposed  NSPS and  emission  guidelines.   However,  the
Dutchess County  [225  tonnes/day  (250 tons/day)]  units in Table  3  and
Marion County  [249 tonnes/day (275 tons/day)]  and the Millbury  [680
tonnes/day  (750 tons/day)]  units in Table  4 would require upgrading to
meet  the proposed emission guidelines, if  implemented and  the  units
performed  as  shown.   HC1  removal  data would also be needed  for  the
Dutchess County and SEMASS  units.

Particulate Matter

     As discussed above,  the  collection of PM matter appears  to be  a
major factor  in   the  control  of CDD/CDF  (semi-volatile organics)  and
heavy metals.  Test  results for CDD/CDF and heavy metals are given in
Tables 5 and 6.  Inlet concentrations  were not normally measured for the
DSI/FF systems  (Table 5) .   Higher  CDD/CDF  control (>95%)  with the lime
SDA/FF system than with the lime SDA/ESP combination (64%)  is indicated
by the limited data.   The Hg  outlet emissions of the  Millbury SDA/ESP
systems approximate  the inlet  values of the  SDA/FF  systems  shown in
Table 6; thus,  little, if  any,  Hg  control  is apparent  for  this SDA/ESP
system.   The  SEMASS  units,  however,  have  outlet Hg  emissions values
suggesting  some  control,  but  probably less  than the  Mid-Connecticut
values.   Without  inlet  Hg concentrations  for  the  SEMASS   and Marion
County units, quantitative  comparisons are not possible.  The control of
Cr and Pb by the  SDA/FF  systems  is similar  to PM control (>99%). While
the PM emissions  with the  three-field ESPs  at  Millbury were  similar to
those with  the  SDA/FF  systems  shown in Table 6, the metal concentrations
across the  ESPs were higher than those  for  units with FFs.  Noting that
the ESP inlet  temperature  at  Millbury was  about  125°C  (255°F) compared
with about  140°C   (285°F)  at the FF inlet and the suspected  affinity of
both organics  and metals  for  fine particles, it  is suspected that the
FFs showed  higher control  of CDD/CDF and metals   because of their higher
capture  of  fine particles  relative  to  the ESP.

     An ESP is the only air pollution  control device  on many existing
MWC units.   Under  the  proposed emission  guidelines,  facilities with
                                  P2-7

-------
capacities greater than 225 tonnes/day (250 tons/day)  would be required
to add acid gas controls and might be required to upgrade or replace PM
collectors  as well  as  reduce CDD/CDF  emissions.    One approach  to
reducing  CDD/CDF  emissions  is   to  improve  combustion.    Combustion
modifications at Quebec  City  led  to reduced PM  emissions and improved
CDD/CDF  and  CO  control.17    Figure  4  shows  the  effect of  improved
combustion  (as  shown  by  reduced  CO emission)  on  reducing  CDD/CDF
emissions.  The addition of effective  acid  gas  control and improved PM
removal would also be  expected to reduce both CDD/CDF  and heavy metal
emissions.

SUMMARY

     The proposed NSPS  for MWC facilities  concern material separation,
good combustion practice, MWC  emissions  (acid gases, chlorinated dioxins
and furans,  heavy metals, and  particulate matter), and NOX control.  For
existing sources, emission guidelines  have  been  proposed which include
material  separation, good  combustion  practice,  and the  control  of MWC
emissions.  Effective particulate  control is essential to minimizing air
pollutant emissions  such as CDD/CDF and heavy metals.  Well-designed and
properly  operated  combustors augmented   by  multipollutant  control
systems, such as the lime SDA/FF system,  can meet the proposed NSPS for
HC1, SO2,  CDD/CDF,  and  PM.  Although some existing facilities now comply
with most proposed guidelines,  many will  require equipment modifications
or additions to comply  with the proposed  GCP and FGC requirements.

REFERENCES

1.   Waste-to-Energy Report,  McGraw-Hill,  Inc.,  New York, NY, September
     20, 1989, p. 7.

2.   Air Pollution Standards of Performance for New Stationary Sources;
     Rule and Proposed Rules  (40 CFR Parts  60,  51,  and  52),  Federal
     Register,  Vol.  54, No.   243,  Wednesday,  December  20,   1989,  pp.
     52188-52304.

3.   Hagenmaier, H., H.  Brunner,  R.  Haag,   and M.  Kraft, Environmental
     Science and Technology,  21(11):1085, 1987.

4.   Stieglitz,   L.  and  H.   Vogg,  Formation   and Decomposition  of
     Polychlorinated Dibenzo-dioxins  and  -furans in  Municipal  Waste,
     Report No. KfK4379, Laboratorium fur Isotopentechnik, Institut fur
     Heisse Chemie,  Kernforschungszentrum Karlsruhe, FRG,  Februar  1988.

5.   Schindler, P., "Combustion Retrofit Strategies  for Control of Air
     Emissions  from  Municipal   Waste  Combustors,"   International
     Conference  on Municipal  Waste Combustion,  Hollywood,  FL,  April
     1989.

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.   IHI Engineering Review,  Vol.  16,  No.  2,  April 1983, Ishikawajima-
     Harima Heavy Industries  Co.,  Ltd.,  Tokyo, Japan.
                                  P2-8

-------
8.    Ebara Corporation, Tokyo, Japan,  Fluidized-Bed  Incineration  Plant
     for  Municipal  Refuse   (Fujisawa  City,   Ishinagaka  Center),
     unnumbered and undated report.

9.    Bergstrom,  J.G.T.,  Waste  Management and Research, 4:57, 1986

10.   Vogg, H.,  H.  Braun, M. Metzger,  and J. Schneider,  Waste Management
     and Research,  4:65, 1986.

11.   Hall, B.,   O.  Lindqvist,  and E.  Ljungstrom,  Environmental  Science
     and Technology,  24(1): 108,  1990.

12.   Brna,  T.G.,   M.G.  Johnston,  C.E.   Riley,  and  C.C.  Masser,
     "Performance  of  Emissions  Control   Systems on  Municipal  Waste
     Combustors,"   Paper  No.  89-109.2,   Air  and  Waste  Management
     Association Annual Meeting,  Anaheim, CA,  June 1989.

13.   City of Zurich,  Switzerland  (Waste  Disposal Department), "For  a
     Clean  Zurich,"  undated  brochure  reporting  October  1986   test
     results.

14.   Environment Canada,  National Incinerator  Testing and  Evaluation
     Program:    Air Pollution Control Technology,  Report EPS  3/UP/2,
     September  1986.

15.   Teller, A.J., J.Y.  Hsieh,  P.  Koch,  and A.  Astrand,  "Emission
     Control:   Hospital  Waste  Incineration,  Cottrell  Environmental
     Sciences,  Inc.,  Somerville,  NJ,  October  1988.

16.   Brna,  T.G.  and R.K.  Klicius,  "Characterization  of  a Lime  Spray
     Dryer  Absorber/Baghouse on  a Refuse-Derived  Fuel  Combustor,"
     International Conference on Municipal Waste Combustion,  Hollywood,
     FL, April  1989.

17.   Environment Canada,  National Incinerator  Testing and  Evaluation
     Program:     Environmental   Characterization  of  Mass  Burning
     Incinerator Technology at Quebec City, Summary Report, EPA 3/UP/5,
     June 1988.

18.   McClanahan, D., A. Licata, and J. Buschmann,  "Operating Experience
     with Three APC  Designs on Municipal  Incinerators,"  International
     Conference  on Municipal  Waste  Combustion,  Hollywood,  FL,  April
     1989.

19.   SEMASS Waste-to-Energy  Resource  Recovery  Facility:   Compliance
     Test Report.   Prepared by Eastmount Engineering, Inc., Wapole, MA,
     for SEMASS Partnership, Rochester,  MA,  August 1989.
                                  P2-9

-------
    10
Q
Q
O
00
z
M 	
X "O
O 
Q **->
i     6
Cu <1>
I  C

CNl O  5
W
PQ
Q O
W l-i
M U
M
Di
o
J
a
o
     7 -
      4 -
             iCD
                                                      r  - 0.899
                                                    •  RDF
                                                    D  Mass Burn Uateruall
                       12      16       20      24       28      32      J6

                   PARTICULATE  MATTER  (kg/tonne  feed)
        Figure 1.  Dependency of chlorinated  dibenzo-p-dioxin
                    emissions on particulate matter.5
      LIME
     AIR
   FLUE GAS

      WATER
    (optional)
                                                      1. LIME SILO
                                                      2. REACTOR
                                                      3. CYCLONE
                                                      4. OUST COLLECTOR
                                                      5. STACK
                                                      6. WASTE SILO
                                              DRY WASTE
     Figure  2   Dry  sorbent  injection  into  fluid bed  reactor
                or  flue gas  duct  (dashed line).
                                 P2-10

-------
                             LIME reeoen
                             LIME SLAKCR
                             FEED TANK
                             HEAD TANK (opt lonal)
                             SPRAY ABSORBER
                             OUST COLLECTOR
                         - PARTICLE RECYCLE
                           (opt ional)
                                       DRY WASTE
  Figure 3.  Spray dryer absorption  (semi-dry)  process
      1400
           0             100            200           300
                CARBON MONOXIDE  (ppmv  @  7%  Q2r  dry gas)
Figure 4.  Total CDD/CDF  emissions as  a function  of  carbon
           monoxide concentration, both measured  at  the
           outlet of  the ESP of the Quebec  City MWC unit.17
                          P2-11

-------
                             TABLE  1   PROPOSED MUNICIPAL WASTE COMBUSTION EMISSION  STANDARDS3'2
rsi
i—i
ro

CAPACITY, tonnes/day
(tons/day)

METAL EMISSIONS
•Particulate Matter, mg/dscm
(gr/dscf )
•Opacity, %
ORGANIC EMISSIONS
•Chlorinated Dibenzo-p-dioxins
& Dibenzofurans (CDD/CDF) ,
ng/dscm

ACID GAS CONTROL, %, OR
EMISSIONS (ppmv)
•HC1
•S02
.NOX
New Source Performance
Standards (NSPS)
<225
(<250)


34
(0.015)
10



75
250C


80 (25)
50 (30)
None
>225
(>250)


34
(0.015)
10



5-30b



95 (25)
85 (30)
(120-200)b
Emission Guidelines
<225
(<250)


69
(0.030)
10



500
1000C


None
None
None
>225,<2000
(>250,
<2200)

69
(0.030)
10



125
250°


50 (25)
50 (30)
None
>2000
(>2200)


34
(0.015)
10



5-30b
250°


95 (25)
85 (30)
None
                aAll emission limits are referenced to dry gas with 7% 02 concentration.
                ^Single value, probably in this range, will be supplied at promulgation  of  rules.
                cValue applies only to refuse-derived fuel (RDF) combustors in the  capacity (size)
                category shown.

-------
                   TABLE  2.   PROPOSED  CO  LIMITS ACCORDING TO COMBUSTOR TYPE
                                FOR GOOD COMBUSTION PRACTICE2
              Combustor  Type
Maximum CO, ppmva
MASS BURN
•Modular
•Waterwall
•Refractory
•Rotary, Waterwall

REFUSE-DERIVED FUEL  (RDF) SPREADER STOKER

FLUIDIZED BED

COAL/RDF CO-FIREDb
        50
       100
       100
       150

       150

       100

       150
aValue is referenced to dry gas with 7% 02 concentration and CO measurement by a  continuous
emission monitor.
bpuel is more than 50% municipal solid waste  (mixture of paper, wood, yard wastes,  plastics,
rubber,  and other combustible materials, plus noncombustible materials  such as glass,  metal,
rock, and soil).   Municipal solid waste includes household wastes, municipal-type wastes
from commercial,  industrial (but not process wastes),  and institutional  (but not  medical
waste)  sources,  and processed solid waste such as refuse-derived fuel  (RDF).

-------
                     TABLE 3.  ACID GAS CONTROL WITH DUCT LIME INJECTION/FABRIC FILTER  SYSTEM^
ro

i—"
-F»
Average Concentration, ppmv @
HC1
Location and Test Date
•Claremont, NH
Unit 1, 5/87
Unit 2, 5/87
'Springfield, MAb
7/88
•St. Croix, WI
10/88
•Dutchess County, NY
Unit 1, 2/89
Unit 2, 2/89
inlet

788
642

503

743

NR
NR
outlet

104
36.6

31

NDC

30
183
7% 02, dry
S0?
inlet

NAa
NA

129

99

121
138
outlet

231
60.1

22

28

105
123


Removal, %
HC1

86.8
94 .3

94

100

NR
NR
S02

NA
NA

83

71.7

16.4
10.2
           aNot  available or not measured.
           ^All  concentrations are referenced to dry gas with 12% CO2•
           GNot  detected.

-------
                       TABLE 4.   ACID GAS CONTROL WITH LIME SPRAY DRYER
                          ABSORBER/PARTICULATE COLLECTOR  (PC)6'12'16
Average Concentration, ppmv @
7% 02, dry
HC1 SO?
Location and Test
Date
PC
inlet
outlet inlet
outlet



Removal, %
HC1
SO?
•Marion County, OR
Unit 1,
•Biddef ord,
Unit A,
6/87
ME
12/87
FF

FF
646

582
48 0 333

5.84 101
151

22. 6
92 .5

99.0
54

77
.7

. 6
• Mid-Connecticut a
Unit 11


•Millbury ,
Unit 1,
Unit 2,
•SEMASS
Unit 1,
Unit 2,

7/88
1/89C
MA
2/88
2/88

3/89
4/89
FF



ESP
ESP

ESP
ESP

451
367

770
697

NA
NA

4 . 2 NAb
15 165

23.3 205
6.08 296

NA 154
NA 162

NA
11

53. 9
61.5

67
55

99.1
95.9

97.0
99.1

NA
NA


NA
93

73
79

56.
65.
.3

.7
.2

.6
,0
aAll values for this location are referenced to dry gas with  12%  CO2.
bNot available or not measured.
cThese are average of characterization test runs  12 and 12a which were  for normal unit
operation.

-------
-a
IX)
                   TABLE  5.   CONTROL OF PARTICULATE MATTER (PM)  CHLORINATED DIBENZO-P-DIOXINS  (CDD) AND
                               DIBENZOFURANS  (CDF) AND SELECTED HEAVY METALS WITH DRY LIME
                                          INJECTION/FABRIC FILTER  SYSTEMS6'16'18
Average PM
Concentration3
qr/dscf (<
Location and Test
Date
•Claremont, NH
Unit 1, 5/87
7/87
Unit 2, 5/87
7/87
•St. Croix, WI
5/88
6/88
10/88
•Springfield, MAd
7/88
•Dutchess County,
NY
Unit 1, 2/89
Unit 2, 2/89
3/89
5/89

inlet

NAb
NA
NA
NA

NA
NA
NA

0.090


NA
NA
NA
NA
! 12% C02

outlet

0.011
NA
0.0043
NA

0.015
0.015C
0.012

0.0016


0.0097
0.035
0.011
0.0079
Average Total
CDD/CDF
Concentration
Average Concentration, (Ig/dscm
ng/dscm @ 7% 0?

outlet

NA
37. 6

32 .3

NA

NA

0.15e


4.83
17.9
NA
NA
chromium
outlet

NA
NA
NA
NA

NA

NA

10


8.27
6.48
NA
NA
@ 7% 02
lead
outlet

NA
NA
NA
NA

NA
18


21


38.9
49.1
NA
NA

mercury
outlet

NA
NA
NA
NA

NA
35


300


1080
84 .7
NA
NA
             aMultiply  gr/dscf by 2288 to obtain mg/dscm.
             bNot available  or not  measured.
             GNot measured simultaneously with metal,  but is value from 5/88 tests.
             ^All concentrations  are referenced to dry gas with 12% CO2.
             eDioxin  at outlet reported as 2,3,7,8 tetrachlorinated dibenzodioxin  equivalent  (EPA
             method).

-------
-o
INJ
                                        TABLE 6. CONTROL OF PARTICULATE MATTER (PM), CHLORINATED DIBENZO-P-DIOXINS (CDD)
                                                  AND DIBENZOFURANS (CDF), AND SELECTED HEAVY METALS WITH LIME
                                                           SPRAY DRYER ABSORBER (SDA)/FABRIC FILTER (FF) OR
                                                           ELECTROSTATIC PRECIPITATOR (ESP) SYSTEMS6^2'19

                                                   Average Total
                                     Average PM     CDD/CDF
                                    Concentration3  Concentration
                                    gr/dscf@12%   ng/dscm@7%             Average Concentration
                                        CO2            O?       	u.g/dscm @ 7% O2
chromium (Cr)
Location, Control
System and Test
Date
•Marion County,
OR
Unit 1, SDA/FF
9/86
•Biddeford, ME
Unit A, SDA/FF
12/87
•Mid-Conn.
Unit 11, SDA/FF
7/88
2/8^
•Millbury, MA
Unit 1, SDA/ESP
2/88
Unit 2, SDA/ESP
2/88
• SEMASS
Unit 1, SDA/ESP
3/89
Unit 2, SDA/ESP
4/89


inlet



0.881


320


2.41
1.78


NA

NA

4.280


3.860


outlet



0.0023


0.014


0.0040
0.0018


0.0018

0.0083

0.008


0.012


inlet



43.0


903


996
747


NA

170

NA


NA


outlet inlet



126 4.22


4.38 2,745


0.646 NA
0.368 NA


NA NA

59.2 NA

9.3 NA


311h NA


outlet



0.17


NDC


NA
NA


98.7

47.7

386S


15.6
lead


inlet



20,500


27,352


NA
NA


NA

NA

NA


NA
(Pb)


outlet



19


159


NA
NA


278

330

300


235
mercury (Hg)


inlet



NAb


389


845
657


NA

NA

NA


NA


outlet



239


ND


48.6
8.8


565f

954

59.3


105


PM



99.7


995


99.8
99.9


NA

NA

99.8


99.6
Removal, %

CDD/
CDF



95.7


99.4


99.9
99.9


NA

64.3

NA


NA


Cr



99.9


NA


NA
NA


NA

NA

NA


NA


Pb



99.9


NA


NA
NA


NA

NA

NA


NA


Hg



NA


100


942
98.7


NA

NA

NA


NA
                   aMultiply gr/dscf by 2288 to obtain mg/dscm.
                   b Not available or not measured.
                   cNot detected.
                      l concenttrations are for dry gas referenced to 12% CO2.
eValues shown are averages for normal SDA/FF temperature
  operation (performance tests 6,8,12,13, and 14).
fFrom May 1988 test.
SAverage values of 4.0,1148, and 7.3.
"Average value of 18.0, 6.6, and 907.

-------
                      BAGHOUSE DESIGN CONSIDERATION UNIQUE TO
                               FLUIDIZED BED BOILERS
                              Joseph B.  Landwehr,  P.E.
                               Fred W.  Campbell,  P.E.
                                 J. Gary Weis,  P.E.

                    Burns & McDonnell Engineering Company,  Inc.
                                4800 E.  63rd Street
                            Kansas  City, Missouri 64130
ABSTRACT

Baghouses represent state of the art technology for control of particulates from coal
fired boilers.   Fluidized bed boiler  installations  to date have utilized baghouses as
the predominate choice for  particulate control.   This paper  examines the application
of baghouse technology to  the  unique characteristics of fluidized bed combustion.

The  types  of  baghouses  that  are  commercially  available  are  reviewed  and  the
applicability of each addressed  for  technical compatibility with CFB boilers.   The
fundamental design parameters  for baghouse sizing  are  reviewed with emphasis  on the
gas and ash properties of fluidized bed combustion.  Recommended baghouse construction
features and bag materials  for CFB applications are provided.
                                        10-1

-------
                     BAGHOUSE DESIGN  CONSIDERATIONS  UNIQUE TO
                                FLUIDIZED  BED  BOILERS
BACKGROUND

During the last decade fabric filters have gained increasingly widespread acceptance
for gas  cleaning on coal-fired boilers. Therefore, it is not a surprise that baghouses
have been the predominate choice for particulate removal for circulating fluidized bed
(CFB) boilers.   Of the 50 or more CFB boilers installed or under construction in the
United States,  all but 6 are equipped with baghouses.   Of the remaining CFB boilers,
5 use electrostatic precipitators (ESP's) and one uses  a mechanical collector followed
by a wet scrubber for control of particulate emissions.

Increased interest in CFB boilers over the last  5 years is primarily due to the ability
to effectively burn  low grade  coal and control S02 emissions.   In the past,  design
parameters for  baghouses were based on ash quantities,  not ash characteristics.  These
early baghouses experienced varying degrees of success.  A baghouse can be effective
when boiler design includes a wide range  of  fuels.  However, ash characteristics do
play an important part in a successful baghouse design.

ESP's can be equally  effective for controlling fly-ash  emissions  if designed properly.
The design and subsequent performance  of  ESP's  hinges  on the electrical resistivity
of the fly ash particles.   CFB  ash particles  are  generally higher in resistivity as
a  result  of  the higher   percentage of alkali  oxides   and  sulfates  and  lower
concentrations  of sulfur  trioxide  to condition  the particles.    To date  little
precipltator research has been performed on ash  from CFB boilers and only limited pilot
and  full scale  field  data is  available in  the U.S.    European experience  with
precipitators at  CFB  installations  is more extensive, but published data is generally
unavailable.    Precipitator design for CFB boilers,  therefore, has  not  progressed
rapidly enough to gain the whole-hearted confidence of owners and designers.

Other more traditional methods of particulate control  (mechanical collectors and wet
scrubbers) have hardly been considered.  Mechanical collectors,  typically multi-clone
separators,  are not  efficient enough for  current  new  source regulations,  especially
since the enactment of federal regulation (40 CFR 60 Subpart Db) for  smaller boilers
in the range of 70,000 to 180,000 pounds per hour.

Many wet particulate  scrubbers have the capability to meet  the new standards but have
been  labeled, perhaps unjustly, as requiring high pressure drops,  costly corrosion-
resistant materials and high maintenance.


BAGHOUSE SELECTION

Baghouses have been  the predominant  choice  for two  significant reasons:   1)  High
removal efficiency without  dependence  on  fly ash  characteristics which is desirable
for burning a range  of  fuels, and 2)  Lower capital  investment than ESP's   the next
logical choice.  The  supporting evidence for these primary reasons are  addressed below
along with other  consideration for baghouses  servicing CFB boilers.

Basically, three  types  of  baghouses   are  available  in sizes  to  match available CFB
units:

 1) Pulse-jet
 2) Reverse  air
 3) Shake and deflate


                                        10-2

-------
Most CFB installations  in  the  U.S.  to  date utilize pulse-jet type filters, probably
more from a lowest capital cost than any other  factor.  Pulse-jet baghouses typically
operate at higher air to cloth  ratios and, therefore, can be physically smaller.  More
importantly,  shop-fabricated  pulse-jets  compartments  are available  in  the  sizes
required for typical  CFB units constructed to date.  Therefore, pre- engineered baghouse
compartments are available with the cost saving advantage of complete shop fabrication
and installation  of  all components  except bags and cages.   Pre-engineering not only
saves the cost of custom design engineering, it provides a complete bill of material
to enable the manufacturers  to provide an accurate and competitive price during the
bidding stage.

The largest U.S. market for baghouses  for flue gas cleaning is the utility industry.
Over 80 percent of the baghouses installed on coal-fired utility boilers are reverse
air type units using  woven fiberglass bags.  Fiberglass is ideally suited  for flue gas
cleaning because it has  a continuous temperature rating of 500°F   However,  it has poor
resistance to acids and  only fair resistance to  flexing and abrasion.  Woven fiberglass
has performed successfully on utility boilers because most installations are on boilers
burning low sulfur coal and using reverse air cleaning.   Reverse air cleaning has the
advantage of being a gentle cleaning method which provides  an acceptable bag life for
fiberglass bags.  Reverse air units  also use low  air-to-cloth ratios which allow high
particulate removal efficiency with woven fabrics,  but  increases the physical size of
the baghouse.  These  baghouses  are generally field erected,  but pre-engineered modules
are available for smaller units.

The shake and deflate baghouses are similar in size and application  to  the reverse air
baghouses, but  have  not been  as popular.   The shake portion of the  cleaning  cycle
increases flexing of the bags  and can  shorten bag life.


BAGHOUSE DESIGN

A properly designed  and sized  baghouse should  not,  of  course,  limit  the  capacity of
the boiler under any operating condition.  During the design phase of a project,  the
baghouse design criteria should be established to meet these requirements.  Additional
design requirements may include the following:

 1.   Providing  a spare unit  or  compartment   to  facilitate some degree  of  on-line
      maintenance.

 2.   Minimizing  the capital  costs of  the baghouse  equipment including  avoiding
      oversizing the baghouse.

 3.   Considering (or minimizing) the  long term operating costs, especially bag cost
      and replacement interval.


The basic  baghouse  design  parameters  are  the  gas  flow rate,  inlet  and  outlet
particulate emissions, pressure drop,  and air-to-cloth ratio.   Outlet emissions are
normally limited by New Source Performance Standards (NSPS).  However, emissions may
be  required to  meet Best Available  Control  Technology   (BACT)  or   stack  opacity
requirements.  Design inlet dust loadings, in addition to  loading during soot blowing,
should be based on worst-case conditions of highest ash content per unit heating value
of the fuel, highest  alkali feed rate and the  highest ratio of fly ash to bottom ash
and should include additional loading during soot blowing.

The starting point for  the design gas  flow should be the calculated volumetric flow
with the design fuel  and  the  rated  boiler fuel and air  rate.   The design flow rate
should also  include the  volume  of  carbon dioxide  (C02) liberated from the alkali feed
                                         10-3

-------
(primarily CaC03 and MgC03) less  the  sulfur dioxide (S02) absorbed.  Safety factors
should be added for the additional gas volume  due  to higher than design excess air,
air inleakage and  gas  temperatures.   In most  cases, a minimum  safety  factor of 10
percent is recommended.  Where  appropriate,  additional  safety  factors  should be added
to account for worst case  fuels  if the overall boiler  capacity is much greater than
the nameplate rating, even if such rating is only a peak rating for a couple of hours.

In order  to  determine  the required filter  cloth  area an attempt  should  be  made to
predict the  relationship  of gas  flow versus  pressure  drop.   Ideally,  the baghouse
pressure  at  the  I.D.  fan  test  block  gas flow  should  be used  to  determine  maximum
allowable pressure  drop.   Baghouse pressure  drop  and total  cloth area,  therefore,
should be  selected to  match the available  I.D. fan capacity.    Such procedures to
estimate baghouse and boiler pressure drops  are very site specific and, thus, should
include careful consideration of operating data at similar installations and burning
similar fuel.

Baghouse  sizing  for CFB boilers  should  be  more conservative  than for conventional
stoker and pulverized coal-fired boilers.   Fly  ash particles  are much finer  and ash
loadings  much heavier  for identical fuels.   The  primary reason for finer  fly ash
particles in a CFB boiler is the use  of hot  cyclone(s)  immediately downstream of the
combustion chamber.  Higher ash loadings are  due primarily to the  addition of limestone
and the formation of sulfates from the capture of sulfur dioxide.  In addition, many
CFB units burn low-grade fuels, intrinsically high in ash.

Much research has  been done  in the  U.S. and Australia  on  fly ash  from fluidized
combustion.  This research has  shown that characteristics of  fly ash from fluidized
bed combustion  differs  considerably  from that  of  pulverized  (PC)  and stoker-fired
combustion.  Among  other things,  fly ash from fluidized  bed combustion  have been shown
to exhibit a smaller particle size and jagged,  irregular shape.   The jagged shape is
in contrast to flyash from  a PC combustor  which has  a generally spherical shape.  Such
irregular  shapes are  characteristics  of low  combustor temperatures  which keep the
particles below the ash fusion  temperature.   The rounded shapes characteristic of a
PC boiler are caused by the ash particle  actually melting or fusing due to the heat.

According to current baghouse theory,  smaller particles and irregular shapes  produce
a higher  ash cohesivity.   High  ash cohesivity  generally  produces high porosity dust
cakes which result  in low filtering drag and  low emissions.  High cohesivity, however,
also requires highly energetic  cleaning methods such as reverse-pulse cleaning.

Non-cohesive particles, on the  other hand,  cause  low dust  cake porosity which is
usually associated with high drag and seepage of particles through the dust cake and
filter media.  When these problems occur in CFB baghouses,  the  most likely cause is
fine, non-cohesive particles of calcium oxide/calcium sulfate.

Properly designed pulse-jet units offer very  effective bag cleaning for the heavy dust
loadings typical of CFB boilers.  However, care should be taken not to overclean the
bags.  Overcleaning can cause initial  bleed-through after the  cleaning cycle.

In  general,  high  baghouse  pressure  drop  and high  particulate  penetration  (low
collection efficiency)  often occur simultaneously,  assuming that the baghouse has been
sized properly.   The high pressure drop is caused by small  particles or particles with
low cohesivity (or  both) which pack tightly together to  form a  low porosity dust cake.
Seepage (or bleed through) of particles through the dust  cake  and through the filter
bags increases in inverse  proportion  to the  size and cohesivity of the particles.
                                         10-4

-------
During  the bag  cleaning,  non-cohesive  particles  also  tend  to break  into smaller
agglomerates  which settle into  the  hopper at a  slower rate.   The slower settling
velocity allows more redeposition of dust on adjacent bags.  Redeposition reduces the
cleaning effectiveness and causes higher pressure drops.


RECOMMENDATIONS FOR CFB BOILER APPLICATIONS

The design of the baghouse should consider all variables which can affect the size and
type  of  fly ash particles.    The important variables  include  the  specific fuel and
alkali materials,  the  type of fuel  and alkali preparation and the combustor design.


Ideally, the  baghouse design  and  selection of fabric should be  based on baghouse
performance data at similar  installations operating  with similar fuel and alkali.  If
a  similar  installation  does  not exist,  a  conservative  baghouse design  should be
selected.  When operating data from similar installations indicates  that  high pressure
drops may  occur,  the  baghouse (and I.D. fan)  size  should reflect additional safety
factors to permit more frequent  cleaning cycles and higher pressure operation.

A  conservatively-designed reverse  air  baghouse  should  include  the following,  as
necessary to minimize dust penetration and pressure drop:

 1.   Reduced air-to-cloth ratio.
 2.   Use of texturized  (woven)  fabrics.
 3.   Extra long null periods in the cleaning cycle to allow the dust to settle into
      the hoppers.
 4.   Provisions for adding sonic horns to increase the cleaning energy,  if necessary.


A conservatively-sized pulse-jet baghouse, designed to minimize dust penetration and
pressure drop, should include the following:

 1.   Reduced air-to-cloth ratio.
 2.   Felted fabrics.
 3.   Extra long null  periods  if  off-line cleaning is used or additional  space between
      bags if on-line cleaning is used.


A number of felted fabrics  are available for the range  of most CFB boiler exit gas
temperatures.    These  include the following  in approximate order  of  increasing bag
costs:

 1.   Nomex®   aramid
 2.   Ryton® - Polyphenylene sulfide
 3.   P-84   polyamide
 4.   Huyglas®   glass felt
 5.   Tefaire®   Blend of fiberglass and teflon
 6.   Teflon®
 7.   Gore-tex®   PTFE membrane on woven fiberglass  or felted fabric


The lowest price bag,  Nomex,  has performed successfully on CFB boiler  applications,
but is not recommended as the  initial bag  material.   Nomex has poor acid resistance
and deteriorates rapidly in the  presence of  sulfur  trioxide  and water vapor in flue
gases.  During  boiler startup  and  initial  operation,  a malfunction of  the  alkali
handling and feed system could easily occur causing complete failure of the bags.
                                         10-5

-------
For initial operation,  one of the higher-priced fabrics is recommended to reduce the
risk of total bag failure.   Ryton has been used successfully  at  several coal-fired
installations and should be seriously considered as the  initial bag material.   Each
installation,  of  course,   has  specific  requirements  for  gas   composition,   gas
temperatures and dust  characteristics which should be carefully  considered before
selecting the fabric.
                                       10-6

-------
       FABRIC FILTER MONITORING AT THE CUEA NUCLA AFBC DEMONSTRATION PLANT
                               Kenneth M. Cushing
                           Southern Research Institute
                                 P.O. Box 55305
                            Birmingham, Alabama 35255
                                Thomas J. Heller
                     Consolidated Edison Company of New York
                                 4 Irving Place
                            New York, New York 10003
                                 Ralph F. Altman
                                 Thomas J. Boyd
                                Mike A. Friedman
                                 Ramsay L. Chang
                        Electric Power Research Institute
                                 P.O. Box 10412
                           Palo Alto, California 94303
ABSTRACT
Baghouse performance was assessed at Colorado Lite Electric Association's Nucla
110-MW AFBC (circulating bed) Demonstration Plant.  Air pollution control was
upgraded at this plant by the addition of a twelve-compartment Research-Cottrel 1
baghouse to supplement three existing, six-compartment, Wheelabrator-Frye
baghouses that were installed in 1974.  All four baghouses use shake/deflate
cleaning.  The baghouses have been in service since mid-1987 when the demonstra-
tion plant was placed in service.

Several types of data have been collected on a regular basis to monitor the opera
tion of the Nucla baghouses.  These data include flange-to-flange and tubesheet
pressure drops, filtering air-to-cloth ratios, baghouse operating temperatures,
bag weights, baghouse inlet and outlet particle mass concentrations, baghouse
outlet opacity, and a record of the baghouse operating and maintenance history.

Although AFBCs, such as Nucla, generally produce a finer particle size distribu-
tion and dustcakes with higher flow resistance compared to pulverized-coal
systems, the pressure drop for an AFBC baghouse can be maintained at reasonable
levels with more vigorous cleaning.  With shake/deflate cleaning, the baghouse
system at Nucla has been able to provide tubesheet pressure drop in a range
normally expected for pulverized-coal boilers equipped with similar baghouses.

                                      11-1

-------
       FABRIC FILTER MONITORING AT THE CUBA NUCLA AFBC DEMONSTRATION PLANT
INTRODUCTION
Fabric filter monitoring at the Colorado Ute Electric Association's Nucla 110-MW
AFBC (circulating bed) Demonstration Plant is part of a larger EPRI research
program to develop a comprehensive FBC fabric filter database.  The creation of
this database was begun in 1984 under EPRI RP1179-19 (1) which undertook the
monitoring of the fabric filter at TVA's 20-MW AFBC (bubbling bed) Pilot Plant at
Paducah,  Kentucky. Under EPRI RP2303-21, this database is being expanded by incor-
porating data from the Nucla Demonstration Plant and the 160-MW AFBC  (bubbling
bed) Demonstration Plant at TVA's Shawnee Power Plant near Paducah, KY.  The
fabric filter monitoring program has been coordinated and integrated with the test
programs planned for each facility.  Data are being collected on various operating
parameters including flange-to-flange pressure drop, tubesheet pressure drop,
filtering air-to-cloth ratios, inlet and outlet emissions, bag weights, bag
service life, operating and maintenance problems,  and bag cleaning methods.  This
paper discusses recent baghouse operating data from the Nucla Demonstration Plant.

DESCRIPTION OF THE FABRIC FILTER SYSTEM
The CUEA Nucla Station originally consisted of three 12-MW stoker-fired boilers.
In  1974, in order to meet more stringent emission regulations, a baghouse was
installed downstream from each boiler.  These three six-compartment,  shake/deflate
cleaned baghouses (Unit #1, Unit #2, Unit #3) were built by Wheelabrator-Frye.   In
1984 it was decided to retrofit the Nucla Station with an AFBC boiler(2); the
three stoker boilers were subsequently removed from service.  To accommodate the
additional gas flow generated by the new 110-MW AFBC (circulating  bed) boiler,
Research-Cottrell was contracted to build a new baghouse to supplement the three
older baghouses.  The new twelve-compartment baghouse (Unit #4A) also  uses
shake/deflate cleaning.  Figure 1 shows the general layout of the  four baghouses
in  relationship to the boiler.  Table 1 presents the design data for  the fabric
filter system.
                                      11-2

-------
For cleaning purposes the three older baghouses are treated as a single eighteen-
compartment baghouse.  Thus, there is an eighteen-compartment baghouse and a
twelve-compartment baghouse.  The cleaning logic for each baghouse is independent.
At a flange-to-flange pressure drop of 6 in.  H20 the Unit #4A baghouse begins its
cleaning cycle in the "slow" mode, 360 s between compartment cleaning sequences.
In the slow mode the Unit #4A baghouse requires about 90 min to cycle through the
12 compartments.  If the pressure drop reaches 7 in. H20, the cleaning cycle
reverts to a "fast" mode which allows only 10 s between compartment cleanings.  In
the fast mode the cycle through the 12 compartments requires about 19 min.  If the
flange-to-flange pressure drop across the Unit #4A baghouse reaches 8 in.  H20 a
high pressure drop alarm will sound.  If the  pressure drop reaches 9 in.  H20, the
Unit #4A baghouse will trip to bypass.

The three older baghouses (Unit #1, Unit #2,  Unit #3) also have slow and  fast
modes for cleaning, although their set points are different.  At 5 in.  H20 the
three older baghouses begin their cleaning cycle in the slow mode (25 s between
compartments).  To clean all 18 compartments  in this mode requires about  33
minutes.  At a flange-to-flange pressure drop across any of the three baghouses of
6 in. H20, the fast cleaning mode  (10 seconds between compartments) will  commence.
In the fast mode a complete cleaning cycle requires about 28 min.  At a flange-to-
flange pressure drop of 7 in. H20 the high pressure drop alarm will sound.  There
is no provision for the three older baghouses to trip to bypass since there are no
bypass ducts.

The cleaning cycle logic is diagrammed in figure 2.  It should be noted that, for
this fabric filter system,  bag shaking occurs during deflation and not following
deflation as in most other large shake/deflate cleaned baghouses.  Reference to
this fact will be made later.

EARLY OPERATION OF THE FABRIC FILTER SYSTEM
The baghouses were placed in service in July  1987 and have accumulated over
9700 hours in service.  Two sources of coal have been used during this operating
period.  Considering the new boiler technology on which they are applied,  the
Nucla baghouses have had a very good performance record, comparable to other  full-
scale utility baghouses downstream from pulverized-coal boilers.  The original
cleaning set points are still used, and there have been no periods of uncontrolla-
ble high pressure drops.  The shake/deflate cleaning system, in general,   has
                                      11-3

-------
worked well  in cleaning the bags, maintaining low bag weights, and controlling
pressure drop.  A bag failure episode that commenced in October 1988 will be
described later.

DETAILED BAGHOUSE PERFORMANCE EVALUATION
A detailed baghouse performance evaluation took place at Nucla during the third
quarter of 1989.  This evaluation was performed in order to expand the existing
AFBC database on fabric filter operation.  A variety of measurements were per-
formed, mostly at the Unit #4A baghouse.  These included measurements of the
baghouse inlet and outlet mass concentration and particle size distribution.
During the evaluation, the Unit #4A baghouse flange-to-flange pressure drop and
the tubesheet pressure drop on compartments A, D,  E, H, J, and M were monitored,
as well as the total mass flow through the baghouse as measured by the airfoil
located downstream of the Unit #4A baghouse.  Bag weights were measured in com-
partment E of the Unit #4A baghouse.  In addition, Individual Bag Flow Monitors
(IBFM) installed in compartments P and Q of the Unit #2 baghouse were used to
compare the behavior of two different bag fabrics.  Bag weights were also measured
in these two compartments.  The boiler was operated under normal coal and sorbent
feed rates at nearly full-load conditions.  Table 2 presents the daily average
(8:00 a.m. to 4:00 p.m.) for several boiler and baghouse operating parameters.

The following sections summarize the data developed for each portion of the
performance evaluation.

Unit #4A Baghouse Inlet and Outlet Mass Concentration
Measurements of the inlet and outlet mass concentration were conducted at the Unit
#4A baghouse.  Two simultaneous 96-minute tests were conducted each day at each
location.  Glass fiber thimble filters were used at the inlet and 47mm glass fiber
filter disks were used at the outlet.  The sampling protocol was based on the EPA
Method 17 procedure.  Table 3 summarizes the results for each of the tests.  The
average inlet mass concentration was 8.9 gr/dscf.  The average outlet mass concen-
tration was 0.0037 gr/dscf.  The average emission rate was 0.0072 Ib/MBtu, less
than one-fourth of the New Source Performance Standard (0.03 Ib/MBtu).  The
average mass collection efficiency for the Unit #4A baghouse was 99.959%, the
average mass penetration was 0.041%.  During the test period, the stack opacity at
the plant averaged slightly less than 10%.  The outlet opacity is based on the
combined emissions from the four baghouses at the plant.  Subsequent to the tests,
                                      11-4

-------
a number of bag leaks (described later) were found in the other three baghouses at
the plant.   This fact explains the large disparity between the low mass emission
rate for the Unit #4A baghouse and the high stack opacity.

Unit #4A Baghouse Inlet and Outlet Particle Size Distribution
Particle size distributions were measured at the Unit #4A baghouse inlet and
outlet.  The inlet particle size distribution was determined from laboratory
sizing (BAHCO) of the ash samples collected during the mass concentration measure-
ments.  (Cascade impactors could not be successfully used at this location due to
high mass concentration, high gas velocity, and short sampling time.)  At the
baghouse outlet seven-stage University of Uashington Mark III Source Test Cascade
Impactors were used.  The mass median diameters at the baghouse inlet and outlet
were essentially the same, being 7.1 and 8.0 micrometers, respectively.

Unit #4A Baghouse Flow Rate and Pressure Drop Data
The Unit #4A baghouse is unique in that the mass flow rate through this baghouse
is measured with an airfoil mounted in the baghouse outlet duct.  The baghouse
flange-to-flange pressure drop and the tubesheet pressure drop for six of the
twelve compartments are also monitored with pressure transducers.  The plant's
computerized data acquisition system records these values for later retrieval and
analysis.  There is no measurement of the flow to the other three Nucla baghouses,
either individually or collectively.

Measurements (pitot traverses) of the flow rate through the Unit #4A baghouse were
made as part of the mass concentration and particle size distribution tests.
During each test the baghouse outlet mass flow rate as measured by the airfoil was
determined.  The value  (klb/h) was then converted to actual cubic feet per minute
of gas flow using gas density, pressure, and temperature data.  Figure 3 shows the
relationship between the baghouse flow rate (acfm) measured by the airfoil versus
the flow rate determined with the pitot traverses.  These data can be used for
future baghouse performance analysis at other operating conditions.

During the  performance evaluation at stable boiler load conditions, the baghouse
cleaning cycles were almost exclusively operating in the slow mode.  In this mode
of operation there is a six-minute pause between compartment cleanings on the Unit
#4A baghouse and a 25-second pause between compartment cleanings on the cleaning
cycle incorporating the three other baghouses.  The Unit #4A baghouse initiates
                                      11-5

-------
cleaning in the slow mode when its flange-to-flange pressure drop exceeds
6 in.  H20.   The cleaning of the eighteen compartments in the other three baghouses
is initiated when the flange-to-flange pressure drop across any of the three
baghouses exceeds 5 in.  H20.  Occasionally the Unit #4A baghouse would jump into
its fast cleaning mode between the cleaning of compartments A and B.  This would
occur because of the momentary excursion of the baghouse flange-to-flange pressure
drop above 7 in. H20 when compartment cleaning began on the other baghouse system.
In general  it was observed that when the Unit #4A baghouse began cleaning and
compartment A was removed from service,  the pressure drop across the other three
baghouses would rise to  the point of initiating their cleaning cycle.  When the
first of their compartments (IT) was removed from service, the flange-to-flange
pressure drop across the Unit #4A baghouse would briefly exceed 7 in. H20.  The
Unit #4A baghouse would  revert to the slow cleaning mode after compartment B was
cleaned.

An analysis of the baghouse operating data during the performance evaluation shows
that the filtering air-to-cloth ratio for the Unit #4A baghouse ranged from 2.4 to
2.9 acfm/ft2, the tubesheet pressure drop ranged from 3.7 to 5.2 in. H20, and the
flange-to-flange pressure drop ranged from 5.0 to 6.5 in. H20.  The corresponding
range of drag (tubesheet pressure drop divided by air-to-cloth ratio) was 1.5 to
1.8 in. H20/(ft/min).  Figure 4 presents the tubesheet pressure drop data versus
filtering air-to-cloth ratio for the Unit #4A baghouse.  For comparison, data from
the TVA 20-MW AFBC (bubbling bed) Pilot  Plant baghouse with two different cleaning
methods, the TVA 160-MW  AFBC (bubbling bed) Demonstration Plant baghouse, and
representative full-scale baghouses downstream from pulverized-coal boilers are
shown.  Although AFBCs generally produce a finer particle size distribution and
dustcakes with higher flow resistance compared to pulverized coal (PC) system
(3,4), the pressure drop for an AFBC baghouse can be maintained at reasonable
levels with more vigorous cleaning.  With sonic-assisted reverse-gas and
shake/deflate cleaning,  the tubesheet pressure drops for the AFBCs are in the
range normally expected  for PC boilers equipped with similar baghouses.

Bag Weights
The bags at Nucla had been in service long enough for mature, stable dustcakes to
form.  Bag weights were  measured in compartment E of the Unit #4A baghouse and
compartments P and Q of the Unit #2 baghouse.  In the Unit #2 baghouse compart-
ments a comparison of two different bag  fabrics is in progress.  In compartment  P
                                       11-6

-------
bags are installed that are similar to all other bags in the entire baghouse
system, except for compartment Q of the Unit #2 baghouse.  Compartment P has bags
manufactured with the fabric oriented in the warp-out configuration.  This means
that the texturized side of the fabric is facing the dirty gas stream; for this
3 x 1 twill weave fabric, 75% of the texturized fill yarns face the dirty flue
gas.  The bag is spoken of as having a 75% exposed surface texturization.  On the
other hand, the bags in compartment Q of the Unit #2 baghouse were manufactured
"inside out."  The fabric facing the dirty flue gas has a 25% exposed surface
texturization.  Previous testing at EPRI's Arapahoe Test Facility with reverse-gas
cleaning had indicated that a lower surface texturization could result in lower
residual dustcake weights, providing the possibility for lower drag without
compromising emissions for pulverized-coal fly ash.

Due to problems with the upper walkway in compartment P, it was possible to weigh
only three bags adjacent to the door.  The bag weights were 12, 11, and 16 Ib.   In
compartment E of the Unit #4A baghouse four bags of the same type were weighed.
The bag weights were 15, 14, 16, and 17 Ib.  Thus the average bag weight for the
seven warp-out bags was 14.4 Ib.  In compartment Q six bags were weighed.  The bag
weights were 8, 6, 7, 6, 6, and 6 Ib.  The average bag weight for the six warp-in
bags was 6.5 Ib.  Since a new, unused bag weighs 4 Ib, the net average residual
dustcake weight was 10.4 Ib for the warp-out bags and 2.5 Ib for the warp-in bags.
Since the bag filtering area is approximately 44.3 ft2, the average residual
dustcake areal density was 0.23 lb/ft2 for the warp-out bags and 0.06 lb/ft2  for
the warp-in bags.  For comparison, residual dustcake areal densities for warp-out
bags at the TVA 20-MW AFBC (bubbling bed) Pilot Plant baghouse were 0.45 lb/ft2
during reverse-gas cleaning and 0.30 lb/ft2 during reverse-gas cleaning with sonic
assistance.  Residual dustcake areal densities at baghouses downstream from
pulverized-coal boilers range from 0.2 to 0.9 lb/ft2, depending on coal type, bag
cleaning method, and baghouse operating history.

Comparison of Flow and Pressure Drop Behavior of Warp-In and Warp-Out Bags
As mentioned above two different types of bag fabric are installed in compartments
P and Q of the Unit #2 baghouse.  Six Individual Bag Flow Monitors (IBFM) are
mounted in each compartment.  These are used to measure the flow and pressure drop
characteristics of individual bags.  During the tests data were taken on the flow
and pressure drop behavior of the bags during several filtration cycles.  The
following average values were measured.
                                      11-7

-------
                                             Comp. Q           Comp. P
                                             Warp-In           Warp-Out
                                             25% EST           75% EST
Air-to-Cloth Ratio, acfm/ft2                    1.7               1.6
Pressure Drop,  in. H20                          5.5               5.1
Drag, in. H20/(ft/min)                          3.3               3.1
Residual Drag,  in. H20/(ft/min)                 2.3               2.2
Drag Coefficient (K2), in. H20.min.ft/lb       13.7              13.9

Considering the range of values used in developing these averages, these compart-
ments are essentially behaving the same, and there is no distinct benefit indi-
cated for the warp-in (25% EST) fabric.  The explanation for this result stems
from the fact that the residual drag for each type of fabric is the same, even
though the residual dustcake areal density is four times higher in the warp-out
case (0.23 lb/ft2) versus the warp-in case (0.06 lb/ft2).  The more texturized
warp-out fabric has created a residual  dustcake external to the fabric yarn
interstices.  The warp-in fabric, with less surface texturization, has allowed the
ash to penetrate into the interior of the fabric, clogging the gas passages.
Thus, even with less residual dustcake, the warp-in fabric still has the same
residual drag created by the clogged fabric.  As expected, the pressure drop due
to the new dust deposited during a filtration cycle is similar in both cases, as
is the drag coefficient, K2, a measure of the gas flow resistance of the ash.

BAG  FAILURES
Although the mass emission rate of the Unit #4A baghouse was only about
0.007 Ib/MBtu,  the stack opacity during the test was averaging about 10%.  Bag
failures in the other three baghouses were suspected.  An inspection of these
baghouses was conducted the week following the testing at Nucla.  Forty failed
bags were replaced.  These bag failures are part of a series, predominately on the
three older baghouses, that first occurred in late 1988.

Since October 1988 the baghouses at Nucla have experienced 326 bag failures  (out
of a total of 4,176 bags).  The majority of these bag failures have been due to
ash  abrasion on the lower two feet of bag fabric, adjacent to and above the  snap
ring bag attachment.  Figure 5 shows a typical abrasion  pattern and the point of
failure within the abraded area.  Ash abrasion at this location occurs because the
bag  sits inside the thimble  (due to the snap ring attachment) and is exposed to
direct  impingement from the ash-laden flue gas as it passes into the bag.  The
problem is compounded by ash that accumulates between the opening in the tubesheet

                                      11-8

-------
and the outside surface of the bag and by the fact that bag shaking occurs during
deflation, exaggerating stress on weakened areas of the bag.  Of the 326 bag
failures through the end of 1989, 13 have occurred in the Unit #1 baghouse, 269 in
the Unit #2 baghouse, 7 in the Unit #3 baghouse, and 37 in the Unit #4A baghouse.
Of these bag failures, about 83% have occurred in the Unit #2 baghouse. It has
been determined that the large number of failures in the Unit #2 baghouse were
exacerbated by overdeflation (up to 10 in. H20) and by the fact that most of the
shaker mechanisms in this baghouse were working properly.  Overdeflation was also
occurring in the Unit #1 and Unit #3 baghouses, but not to so great an extent.
Inspection of the baghouse shaker systems in early 1989 revealed that 210 of the
672 bags in the Unit #1 baghouse, 126 of the 672 bags in the Unit #2 baghouse, and
308 of the 672 bags in the Unit #3 baghouse were not being shaken during the
cleaning cycle.

In March 1989 a maintenance program was begun to correct the problems with over-
deflation and non-working shaker systems.  The deflation flowrate was adjusted to
a lower setting (allowing about 1 in. H20 pressure drop at full load) in each of
the three older baghouses. Because it had the most shakers not working, all of the
shakers were fixed in the Unit #3 baghouse first.  Maintenance of the other shaker
systems in continuing.

Correcting deflation flowrate and the shaker systems will not eliminate bag
failures at Nucla.  It has been determined that ash abrasion on the fabric is
occurring in all baghouses, even the new Unit #4A baghouse.  It appears that the
problem is mainly due to the type of bag attachment used in these bag houses, a
snap ring method.  Some damage to all of the bags has very likely occurred and it
is just a matter of time before each of these bags will fail.

CONCLUSIONS AND RECOMMENDATIONS
The Nucla baghouse system has been in operation since mid-1987.  Shake/deflate
cleaning has provided stable operation within design parameters for two types of
coal under a variety of coal and sorbent feed rates.  Except for bag failures, the
system is performing well, operating within design values, and is able to provide
a low mass emission rate (0.007 Ib/MBtu (no bag failures)) at a fairly high
filtering air to-cloth ratio (2.4 to 2.9 acfm/ft2) with low to moderate tubesheet
(3.7 to 5.2 in. H20) and flange-to-flange (5.0 to 6.5 in. H20) pressure drops at
near full-load conditions.  The shake/deflate cleaning system is able to maintain
low bag weights (0.23 lb/ft2 residual dustcake areal density) even though the
                                      11-9

-------
inlet mass concentration is high (8.9 gr/dscf) compared to values typically
experienced at baghouses downstream from utility pulverized-coal boilers.

There have been 326 bag failures (out of 4,176 installed bags) since October 1988.
The majority of the bag failures have been due to ash abrasion on the lower
portion of the bag just above the snap ring attachment point.  This type of bag
attachment is exposing the fabric to abrasion by the ash-laden flue gases entering
the bag.  Bag failures were being exacerbated by the fact that overdeflation was
occurring (prior to correction) and that the bags continue to be shaken during
deflation, exaggerating stress on weakened areas of the fabric.

Several recommendations have been made to CUEA for improvement of the operation of
the baghouses at Nucla.  These include

1.   Reduction of the deflation flow rate such that the pressure drop during
     deflation is on the order of 0.25 to 0.5 in. H20 during full-load operation.

2.   Adjustment of the baghouse compartment cleaning sequence to have bag defla-
     tion precede bag shaking.  Bag deflation is necessary only to relax the bags
     so that the shaking energy can be transmitted along the entire length of the
     bag.  Deflation by itself is not intended to clean the bag.

3.   Completion of maintenance on the bag shaker mechanisms in the Unit #1 and
     Unit #2 baghouses.

4.   Trial testing of bags made from fabrics that could better withstand ash
     abrasion, or trial testing of bags manufactured with extended lower bag cuffs
     (two feet in length).  Bags with extended cuffs would place several layers of
     fabric in the area where ash abrasion occurs.  Although this approach would
     not eliminate ash abrasion, bag service life prior to failure would be
     extended.
                                      11-10

-------
ACKNOWLEDGMENTS

The following Southern Research Institute personnel participated  in the  special

four-day test at the CUEA Nucla AFBC Demonstration Plant:  Ms. Sherry Dawes,

Mr. Larry Felix, Mr. Terry Hammond, Mr. John Hester, Mr. Carl Landham, Mr. Charles

Lindsey, Mr. Randy Merritt, and Mr. Ron Young.  We appreciate the support  and

assistance of Mr. Bob Melvin of CUEA-Nucla and Mr. Steve Sitkovitz of Radian

Corporation during the on-site testing.  The data and analysis presented  in this

paper were funded by EPRI Projects RP3005-1 and RP2303-21.


REFERENCES

1.   K. M. Cushing, P. V. Bush, and T. R. Snyder.  Fabric Filter  Testing  at the
     TVA  Atmospheric Fluidized-Bed Combustion (AFBC) Pilot  Plant.  CS-5837.   Palo
     Alto, California.  Electric Power Research Institute, May 1988.

2.   W. E. Blunden.  Colorado-Ute Circulating AFBC Demonstration.  Volume  1:
     Pro.iect Origin.  CS-5831.  Palo Alto, California.   Electric  Power Research
     Institute, May 1988.

3.   P. V. Bush, T. R. Snyder, and W. B. Smith.   "Filtration Properties  of Fly Ash
     from Fluidized-Bed Combustion."  JAPCA 37(11):1292  (1987).

4.   P. V. Bush, T. R. Snyder, and R. L. Chang.   "Demonstration of Baghouse
     Performance from Coal and Ash Properties:  Part I."  JAPCA 39(2):228  (1989).
                                      11-11

-------
DUSE m
NUCLACFBC BOILER
BAGHOUSE 02 BAGHOUSE H3



— « 	 -« 	 -« 	
_,! _ LJ— , P-L.J 	 . BALANCE DAMPE
                                                                             Kr
                                                            BY PASS DAMPERS
                                  CUEA NUCLA GENERATING PLANT
Figure  1.   General  layout of the  four  baghouses  at the  CUEA Nucla 110-MW AFBC
(circulating bed)  Demonstration  Plant.   The Unit #4A  baghouse  is the  new
Research-Cottrel1  baghouse.
     Start
    Cleaning
     Cycl e
                           Close
                           Outlet
                           Damper
25  sec
Pause
Before
Opening
Deflate
Damper
                Pause
                Between
                Compartments
15 sec
Pause
Before
Opening
Outlet
Damper
1 ^ cor
Pause
Shake













5 sec
Timer

Start
Shakpr
                                                 Motor
                                                          Stop
                                                          Shaker
                                                          Motor
Step to
 next
 Comp.
                                                    Open
                                                   Outlet
                                                   Damper
                 Figure 2.   Nucla Baghouse Cleaning  Cycle Logic
                                         11-12

-------
                               Nucla Baghouse 4 Flow Measurements

                                       Calc.(Air Foil) vs Meas.(Pitot)
                200 -
                                         240         260
                                            (Thousands)
                                     Measured (Pitot) Flow Rate, ocfm
Figure  3.   Calculated Unit #4A  baghouse  air flow rate versus measured Unit
baghouse air  flow  rate.   Calculated air  flow  rate based on  air foil  measured
mass  flow rate.  Measured air flow rate  based  on pitot traverses.
                                   CUEA 110 MW AFBC (CIRCULATING BED) DEMONSTRATION
                                     PLANT (SHAKE/DEFLATE CLEANING)

                                  j TVA 160 MW AFBC (BUBBLING BED) DEMONSTRATION
                                     PLANT (REVERSE GAS CLEANING)
                                   TVA
                                                       ANT
                                     (REVERSE GAS CLEANING!

                                   TVA 20-MW AFBC (BUBBLING BED) PILOT PLANT
                                     (REVERSE GAS CLEANING WITH SONIC ASSISTANCE!

                                   FULL SCALE PULVERIZED COAL BAGHOUSES

                                    O REVERSE GAS CLEANING
                                    • REVERSE GAS CLEANING WITH SONIC ASSISTANCE
                                    V SHAKE/DEFLATE CLEANING
                                     FILTERING AIR TO CLOTH RATIO, ai
             Figure  4.  Comparison of  tubesheet pressure drop versus
             air-to-cloth  ratio  for several  baghouses on fluidized
             bed combustion boilers and  for  fabric  filters  downstream
             from pulverized-coal  utility boilers.
                                          11-13

-------
Figure 5.  Photograph of the slit failures in Bag B2 removed from the
CUEA Nucla Unit 2 Baghouse, Compartment R on January 20, 1989.   These
failures are 6 in. above the bottom of the bag.  Smooth and texturized
surface features are distinct.   Smooth areas are caused by fly ash
abrasion.  Slit failures are located within these areas of abrasion.

-------
                                   Table 1

       DESIGN INFORMATION FOR THE COLORADO UTE NUCLA STATION BAGHOUSES
Baghouse manufacturer
Number of compartments
  per baghouse
Bags per compartment
Bag size
Bag manufacturer and
  model  number
Bag fabric
Bag fabric finish
Bag cleaning method
Cloth area per bag
Cloth area per compartment
Cloth area per baghouse

Total cloth area
Design filtering air-to-
  cloth ratio, for all 30
  compartments and full-
  load flow of 414,000 acfm
Bag cleaning initiation
Bag cleaning set point

Compartment cleaning
  frequency
Compartment cleaning
  sequence
Deflation air-to-cloth
  ratio
Shake frequency
Shake amplitude
High pressure drop alarm
High pressure drop bypass
Bag tension
Compartment isolation
  available
Compartment vent system
High inlet temperature
  bypass
Low inlet temperature
  bypass
 Baqhouse #1. #2, 8. #3

Wheelabrator-Frye
6

112
8 in.  x 22 ft
Fabric Filters #504

3x1  twill,  warp out
10% Teflon B
Shake/deflate
44.31  ft2
4,963  ft2
29,778 ft2
                                                           Baqhouse #4A
Research-Cottrell
12

180
8 in. x 22 ft
Fabric Filters #504

3x1 twil1,  warp out
107. Teflon B
Shake/deflate
44.31 ft2
7,976 ft2
95,712 ft2
                  185,046 ft2
            2.24 acfm/ft2 (gross)
            2.50 acfm/ft2 (net)
            2.76 acfm/ft2 (net-net)
Pressure drop
5.0 in. H20 (slow mode)
6.0 in. H20 (fast mode)
25 s (slow mode)
  10 s (fast mode)
Null (25 s), Deflate
  (45 s),  Shake (5 s,
  15 s after deflation
  starts),  Null (15 s)
0.3 acfm/ft2

4 Hz
1 in.
7.0 in. H20
No bypass  available
60 Ib
Yes
Pressure drop
6.0 in. H20 (slow mode)
7.0 in. H20 (fast mode)
360 s  (slow mode)
  10 s (fast mode)
Null (25 s), Deflate
  (45 s),  Shake (5 s,
  15 s after deflation
  starts),  Null (15 s)
0.3 acfm/ft2
3 Hz
1 in.
8.0 in.
9.0 in.
60 Ib
Yes
H20
H-,0
No                       Yes
No bypass available      320°F
No bypass available
                         180°F
                                   Table  2

                   NUCLA DEMONSTRATION  PLANT  OPERATING  DATA
                                                    Average  Value
Description
Net load (MW)
Baghouse 4A inlet temp (DEG. F)
Baghouse inlet pressure (IN. WG)
Baghouse 4A outlet temp (DEG. F)
Baghouse 4A flue gas flow (KLB/HR)
Baghouse 1 pressure drop (IN. WC)
Baghouse 2 pressure drop (IN. WC)
Baghouse 3 pressure drop (IN. WC)
Baghouse 4A pressure drop (IN. WC)
Baghouse 4A compartment DP A (IN. WC)
Baghouse 4A compartment DP D (IN. WC)
Baghouse 4A compartment DP E (IN. WC)
Baghouse 4A compartment DP H (IN. WC)
Baghouse 4A compartment DP J (IN. WC)
Baghouse 4A compartment DP M (IN. WC)
ID fan inlet pressure (IN. WG)
ID fan inlet temp (DEG. F)
CEM opacity 6 min avg (%)
9/19/89
98.2
298.3
-14.1
288.5
552.6
4.9
4.8
4.8
6.0
4.4
4.5
4.5
4.6
4.9
5.6
-22.2
258.2
10.2
9/20/89
98.1
297.8
-14.0
279.4
526.2
4.5
4.3
4.4
5.6
4.1
4.2
4.2
4.3
4.5
4.8
-21.4
255.9
9.6
9/21/89
98.7
297.1
-14.3
279.1
525.1
4.5
4.4
4.4
5.6
4.1
4.2
4.3
4.4
4.6
4.8
-21.8
254.8
9.4
9/22/89
98.7
298.2
-14.4
281.6
534.4
4.6
4.5
4.5
5.7
4.2
4.2
4.8
4.3
4.6
4.8
-22.0
254.8
9.7
                                  11-15

-------
                                              Table 3



                  NUCLA  UNIT  #4A  BAGHOUSE  INLET AND OUTLET MASS CONCENTRATION DATA
Date
9/19/89
9/19/89
9/22/89
9/22/89


Mass Loading
qr/acf qr/scf
4.34 8.85
4.39 9.14
4.34
4.21
8.76
8.65
Inlet
Gas
acfm
274000
256700
244900
254700

Flow
scfm
134300
123300
121400
124000

Temp
deg-f
296
309
297
309


Mass Loading
qr/acf qr/scf
0.0017 0.0036
0.0017 0.0034
0.0020
0.0018
0.0040
0.0036
Outlet
Gas
acfm
264900
245000
242900
240600

Flow
dscfm
130000
119200
121100
118100

Temp
deg-f
284
291
279
289
Emiss
Ib/MBtu
0.0069
0.0067
0.0080
0.0071
Eff
%
99.959
99.963
99.954
99.958
Pen
%
0.041
0.037
0.046
0.042
Average    4.33    8.86   257575  125750   303    0.0018  0.0037  248350  122100   286    0.0072   99.959  0.041

-------
              ELECTROSTATIC PRECIPITATION OF PARTICLES PRODUCED
                   BY THREE UTILITY FLUIDIZED-BED COMBUSTORS
                                  E. C. Landham, Jr.
                                    M. G. Faulkner
                                     R. P. Young
                              Southern Research Institute
                                   P.  0. Box 55305
                                Birmingham, AL 35255
                                   Ralph F. Altman
                                  Ramsay L Chang
                            Electric Power Research Institute
                                   P 0. Box 10412
                                 Palo Alto, CA  94303
                                     ABSTRACT

Coal-fired,  atmospheric-pressure,  fluidized-bed  combustors  produce  effluents which  are
significantly different from those produced by conventional coal-fired boilers. These differences
have the potential to degrade the performance of downstream  electrostatic precipitators. The
characteristics of the effluent from two bubbling-bed and one circulating-bed design combustors
have been measured. The performance of the electrostatic precipitators installed on two of these
plants was determined and is  interpreted in terms of the effluent characteristics.  Precipitator
collection performance is predicted for all three sites with the EPA/SRI ESP model and these
results are compared with measured performance. Laboratory  measurements and techniques
for predicting dust resistivity are compared with data measured in situ.
        This paper was presented at the 8th Symposium on the Transfer and Utilization
        of Paniculate Control Technology; San Diego, California; March 1990.
                                          12-1

-------
              ELECTROSTATIC PRECIPITATION OF PARTICLES PRODUCED
                   BY THREE UTILITY FLUIDIZED-BED COMBUSTORS
INTRODUCTION

Atmospheric-pressure fluidized-bed combustors (AFBCs) have the potential to operate reliably
with low-grade fuels and to provide low S02 emissions without use of downstream scrubbers.
However, the particles produced by AFBCs can impact the operation of the paniculate control
device in ways which have not been well documented. This paper will discuss the results of an
Electric Power Research Institute study to determine the characteristics of the particles produced
by these processes and to evaluate their impact on electrostatic precipitators (ESPs).  A more
detailed discussion of fabric filter performance can be found  in a companion paper [1].

Detailed test programs have been conducted to date on three  utility AFBCs. The three plants are
R. M. Heskett Unit 2 of Montana-Dakota Utilities, Black Dog Unit 2 of Northern States Power, and
Nucla Station of Colorado Ute Electric Association. One coal or coal-sorbent combination was
evaluated  at each  site.  Field measurements were  made of  particle dust  loadings,  size
distributions,  electrical  resistivity,  gas composition, and  ESP operation and performance.
Comparisons were  conducted  between in situ resistivity and the results of laboratory and
predictive  techniques.  The accuracy of the  SRI/EPA  ESP  model in predicting actual  ESP
performance on these dusts was also assessed.
PLANT DESCRIPTIONS

Heskett Unit 2

Heskett Unit 2 went on-line with a retrofitted AFBC in 1987. The combustor is a B&W bubbling
bed design capable of producing 700,000 pounds of steam per hour at 1300 psig and 955° F (85
MW).  However, because of clinker formation at higher loads, the maximum load on Heskett 2 is
usually limited to the original nameplate value of 550,000 pounds of steam per hour (65 MW).
Fuel is added to the bed by overbed spreader-stokers.  Design bed velocity is 12 ft/sec and bed
temperature is 1500°F. Limited recycle is utilized, as only the material caught in the economizer
drop-out hopper is returned to the bed, while the particles caught in the mechanical collector
downstream of the economizer are discarded.

The coal burned during the test program was a lignite from the  Beulah deposit of the North
Dakota Fort Union region (Table 1). The coal typically has low sulfur, high moisture, and low ash.
Sand, sized to be between 4 and 16  mesh, is continuously dried and added to the bed as filler
material.  Use of a sorbent for SO2 control is not required.

The mechanical collector located between the economizer and the air heater was provided by
Joy-Western and has a design collection efficiency of 88%. Final particulate control is provided
by a Research-Cottrell cold-side electrostatic precipitator which was retrofitted to the plant in 1975.
The ESP has 5 electrical sections in the direction of gas flow with a design specific collection area
(SCA) of 360 ft2/1000 acfm. Three half-sections (out of a total of 20) were out of service during
the test program.  Collecting plate spacing is 9 inches with weighted wire discharge electrodes.
                                         12-2

-------
 ESP gas velocity was approximately 3.3 ft/sec during the test program.  Although maximum
 power input to this ESP would normally be limited by sparking, the plant manually limits the T-R
 set controls below the point where sparking would occur.

 Black Dog  Unit 2

 Black Dog  Unit 2 was originally a pulverized coal boiler which was modified for fluidized bed
 operation in 1986.  Unit 2 is a Foster-Wheeler bubbling bed design with total generation capacity
 of 130  MW.  Particles caught by a multiclone mechanical collector, located  between the
 economizer and air heater, are recycled to the bed in varying amounts.  Design air-based bed
 velocity is 10 ft/sec and the bed temperature is typically 1500°F. Pebble sized limestone was
 added to the bed for S02 control, but was added in  batches rather than continuously as a result
 of sorbent feeder malfunctions during the test. The rate at which batches of sorbent were added
 varied from once per hour to once per day during these tests. The coal burned during the test
 period was a 1% sulfur blend of Illinois bituminous and western subbituminous (Table 1).

 Particulate  collection  at Black Dog is by two ESP systems in series. Both sets of ESPs were
 designed for full load operation of 100 MW and a lower gas flow than now possible.  The primary
 ESPs were  installed in 1954 by Research-Cottrell, while the secondary ESP was retrofitted in 1974
 by Belco.   At 130 MW, the gas volume flow should be 500,000 acfm producing SCAs for the
 primary and secondary ESPs of 115 and 196 ft2/1000 acfm, respectively (total SCA = 311).
 During the  85 MW  tests discussed here, the total system SCA was 352 ft2/1000 acfm. Each of
 the ESP systems has three electrical fields in the direction of gas flow. Normally, five of the ESP
 fields (all but the primary  inlet)  are pulse energized. During the 85 MW tests, only four of the
 fields were pulsed due to equipment problems. Intermittent energization (IE) was used to power
 the remaining fields.

 The Black Dog AFBC is operated only two shifts per day requiring daily startup  and shutdown.
 High opacities have generally limited the operation of the unit to less than full load.   During this
 program, tests were conducted at boiler loads of 85, 75,  and 40 MW.  The ESP performance
 results discussed in this paper were obtained at 85 MW.

 Nucla Station

 Nucla Station was repowered with a 110 MW fluidized bed in 1987. The circulating fluidized bed
 is a Pyropower design capable of producing  925,000 pounds of steam per hour at 1510 psig and
 1005°F. The gas-based bed velocity was 17 ft/sec during the test program. A hot cyclone is
 installed between the radiant furnace and the convective heat transfer sections.  All of the particles
 collected in the cyclone are returned to the bed. Pulverized limestone is continuously added to
 the lower combustion chambers for S02 removal. The coal burned at Nucla Station during the
 test program was a low-sulfur, western subbituminous from  the Salt Creek mine  near  Grand
 Junction, Colorado (Table 1).

 Control of particulate emissions from Nucla is provided by four parallel fabric filters. Three of the
 baghouses  were originally installed on old 12 MW boilers which have been removed from service,
while the fourth was retrofitted at the same time as the AFBC. The performance evaluation of the
new baghouse #4 conducted during this test program is the subject of another paper.
                                          12-3

-------
PARTICLE CHARACTERISTICS

Total Mass Loading

The total particle mass loadings, gas volume flows, and gas temperatures measured at the inlet
and outlet of the particulate control devices installed on the three plants are shown in Table 2.
Heskett Station has the lowest inlet mass loading, which is consistent with the lack of sorbent use
in the bed. Chemical analyses indicate that most of the inert material (sand) added to the Heskett
bed is  discharged through the bed drains, and that the mass entering the  ESP is composed
primarily of coal ash. The very low recycle rate at Heskett may also reduce the amount of mass
that eventually penetrates the mechanical collector. The inlet mass loading is slightly lower than
the EPRI data base  [2] would predict for a pulverized coal fired boiler and the Heskett coal ash
content (2.5 gr/scf predicted vs 2.05 gr/scf measured).

At both Black Dog and Nucla, the measured inlet mass loadings are approximately a factor of 2
higher  than typical for a pulverized  coal (PC) boiler on these  coals. Nucla also has twice the
mass loading of Black Dog.  At both sites sorbent  is added to the bed, and material is recycled
from the mechanical collector or cyclone.  Sorbent is added to the Nucla bed continuously, while
limestone addition was conducted on a batch basis at Black Dog. Although the increase in mass
loading from Black Dog to Nucla is proportional to the ash content of the coal, many other factors
must be considered, including sorbent characteristics, sorbent addition procedures and rates, the
characteristics of the cyclones, bed  velocities, recycle rates, and bed drain removal rates.

Particle Size Distribution

The size distributions of the particles entering the control  devices were measured with modified
Brink cascade impactors and are compared in Figures  1 and 2. Figure 1 is a plot of cumulative
mass loading as a function of particle  size, while in Figure 2 the mass loading at a size interval
is converted to a percentage of the total mass.  Also shown on these figures is the average size
distribution of PC boilers  burning subbituminous coal from the EPRI data base. On Figure 1, the
lower limit of the plotted data base distribution is the -50% confidence interval for the Black Dog
ash content, while the upper limit is the +50% confidence interval  for the Nucla ash content.
Fewer  particles smaller than  1 ^m were measured at Black Dog than were observed at the other
two sites.  This result may be attributable to the bituminous fraction of the Black Dog coal blend.
Fewer fine particles are generally seen  on PC boilers burning bituminous coals rather than lower
grade coals [2].

The Heskett distribution has significantly higher content of particles smaller than 1 jim as a fraction
of total mass.  A  portion of this may  be due to the overall lower total mass, but part of the
difference is due to the unusually high level of sodium vaporization typical of North Dakota lignites
[3], Chemical analyses, discussed later, indicate that an unusually large fraction of the sodium
in the coal is vaporized in the bed and subsequently condensed to forrv. a submicron fume of
sodium sulfate.

The Nucla distribution shows higher mass across the entire range of particle sizes resolved  by
the impactors than do the other sites. The increased mass in the 1 ^m size range creates a more
severe  particulate control problem.  For a given  overall collection efficiency, an increase in
                                          12-4

-------
 particles in the 0.5 to 1.0 [im range will produce an increase in opacity at the outlet of either an
 ESP or a fabric filter.

 Dust Electrical Resistivity

 Resistivity of the dust produced by the three  AFBCs was measured both in situ  and in the
 laboratory. The in situ measurements were made with a point-plane probe located in the control
 device  inlet gas stream.  The  laboratory determinations  were acquired  according to IEEE
 Standard 548-1984,  descending temperature  technique, with modifications for ash-sorbent
 mixtures.  Previous work on ash-sorbent mixtures  has indicated that limiting the  maximum
 temperature  of the  sample  to  250° C  rather than  the standard 450° C prevents thermal
 decomposition of calcium hydroxide and produces more accurate results  [4]. The maximum
 temperature of the AFBC samples was  therefore limited to 250° C, below the decomposition
 temperature of calcium hydroxide.

 In addition to laboratory and in situ measurements, another useful tool for evaluating electrical
 characteristics of coal fly ash has been the predictive technique developed by Bickelhaupt [5].
 This technique uses an  empirical data base to  estimate the resistivity of an ash based on the
 chemical compositions of the dust and flue gas. However, the data base of dusts upon which
 these predictions were developed does not include the levels or forms of calcium present when
 sorbents are used for SO2 control.  One goal of this project is to evaluate and, if necessary,
 upgrade this model. The predicted resistivity discussed in this paper are based on the chemical
 compositions shown  in Table 3.

 The results of the in situ, laboratory, and predicted resistivity are compared for the Heskett AFBC
 in Figure 3.  Both the lab data and Bickelhaupt model indicate the dust resistivity drops by one
 and a half orders of magnitude from the inlet to outlet hoppers of the ESP. These data are quite
 unusual  and  are believed to be due to the high  degree of sodium volatilization mentioned
 previously. Sodium vapor should recondense on dust particles as a function of surface area,
 which  increases relative to mass with  decreasing  particle size.  Chemical  analysis of size
 segregated samples from  the cascade impactors confirmed that sodium was concentrated in
 smaller particles. Since ESPs collect large particles more readily than small, the smaller particles
 tend to concentrate toward the outlet fields. The total of NajO and S03, most of which probably
 exist as sodium sulfate, in the ESP hopper samples (Table 3) increases from 23% to 67% of the
 total ash from inlet  to outlet hopper.  (Note, however, that the particulate mass collected in the
 outlet hopper of the Heskett ESP was less than 1 % of the total inlet mass.)  The effect of these
 large  changes in  sodium concentration,  to  which resistivity is very sensitive,  are mainly
 responsible for the unusual resistivity results of the middle and outlet hoppers.

 Although the predictive model does indicate that resistivity should drop as a function of hopper
 location, the model does not correctly predict the temperature dependency nor the absolute value
 of resistivity at the temperature of interest (-300° F).  This is not surprising considering that the
 highest sodium ash in the  data base upon which the model is based contained less than 1 0%
The range of resistivity measured in situ at Heskett is indicated by the bars surrounding the solid
symbol in Figure 3. The Heskett dust was precipitated in the resistivity probe more readily than
any in our experience. Despite the relatively low mass loading, only 5 minutes was required to
                                          12-5

-------
collect a 1 mm deep layer.  This is contrasted with a 1 hour precipitation time generally required
for a typical PC generated fly ash with a 1011 ohm-cm resistivity.  The resulting dust  layer was
quite fluffy with high porosity. Collection of high porosity layers is expected with particles which
have highly cohesive surface properties, which is consistent with our experience on high sodium
ashes.  The in situ measurement avoids compaction of the dust layer,  while the laboratory
resistivity  measurement is  conducted  on  a highly  compacted  sample.  This difference in
compaction may be responsible for the disagreement in the resistivity values obtained by the two
techniques.  (Because of the size selective nature of electrostatic devices, the in situ probe data
should be compared with the inlet field lab data only.) The dust collected in the ESP is probably
somewhere in between the two techniques, with new dust collected in a fluffy outer  layer and
older  dust compacted somewhat by electrical and vibrational forces.

Resistivity data from Black  Dog are  shown in Figure 4.  All of the in situ data points are shown
because of the wide range of temperatures  encountered. Lab resistivity  data from inlet hopper
samples are shown for two days of the test program.  Measurements were also made on second
and third hoppers, but no trend in resistivity  was observed as a function of hopper location.  On
August 9, limestone was added to the combustor bed approximately once per hour throughout
testing, while on August 10, sorbent was not added after testing started.  The difference in
sorbent addition procedure is believed to  be responsible for the dramatic  difference  in lab
resistivity of the hopper samples. The lab data indicate why the in situ results, obtained over four
days of testing, are so scattered.  A rigid correlation  does not  exist between the in situ and lab
data from a given day, which is possibly due to  differences in coincidence of sample collection
and sorbent addition.

The resistivity model did a poor job of predicting absolute values and of explaining the differences
between  the two Black Dog hopper samples.  The model's failure to predict the 1  order of
magnitude difference in resistivity between the two days is indicative of the fact that no significant
changes were  detected in the dust components which have been found to be important to coal
ash resistivity.

Figure 5  compares the resistivity data from the Nucla AFBC.  Limestone addition at Nucla is
continuous, suggesting that these samples should be fairly uniform in composition. The  in situ
data are  the highest measured at any of the sites with an average value of 2x1013 ohm-cm at
307° F The moisture content of the  Nucla flue gas was also much lower than was measured at
Black Dog. Figure 6 indicates that the lower water vapor (8% at Nucla vs 12% at Black Dog) is
largely responsible for the higher resistivity at Nucla.  The lab data are about a factor of 3 lower
than the in situ data, although some overlap was observed. Chemical analyses indicated that the
dust collected in the in situ probe contained significant levels of unreacted  calcium sorbent, while
the baghouse hopper samples indicated little free sorbent. The desired calcium sulfate reaction
product remained approximately constant between the two sets of samples. The implication is
that the free sorbent  (CaOH) reacted with some component of the flue  gas (most likely COJ
during the fairly long  period of time it resided on the fabric filter bags.   The effect of such a
process on resistivity  is unknown at  this time.

Once  again the resistivity model was  not able to accurately predict the shape or magnitude of the
resistivity-temperature relationship at Nucla.  The model also predicted a difference between the
two hopper samples that is  counter  to the actual trend.
                                          12-6

-------
 The resistivity at Nucla is sufficiently high that some means of resistivity modification would
 probably be required for all but the very largest ESPs. Figure 6 shows the relationship between
 laboratory resistivity, temperature, and moisture level for the Nucla hopper samples. These data
 indicate that, if humidification were used to lower resistivity, a 6.4% increase (from 7.8% to 14.2%)
 in water content of the flue gas would lower the gas  temperature to 230° F and drop resistivity
 to the range of 4x1011 to 1 x1012 ohm-cm.  This would be a sufficient reduction in resistivity to allow
 a large modern  ESP  to perform  adequately.   However, the lab resistivity  technique  has
 consistently under-predicted the effect of water droplet humidification on resistivity of ash/sorbent
 mixtures, most recently at Edgewater [6].  Based on the Edgewater data, we  suspect  that
 humidification to 275° F would produce resistivity values in the 1011 ohm-cm range and allow
 adequate performance of a large modern ESP. However, this is speculative and will be further
 evaluated experimentally.

 The effect on lab resistivity of 5 ppm S03 added to the Nucla dust environment with 8% water
 present (no humidification) is also shown in Figure 6.  These data indicate that the Nucla dust is
 susceptible to conditioning with acid.  However, to produce the effect of the 5 ppm equilibrium
 laboratory concentration  in the field where active sorbent is present in the environment, may
 require large quantities of S03 to  be added to the duct. In pilot-scale tests, acid levels of up to
 100 ppm S03 were required to lower resistivity below 1012 ohm-cm [7].
 ESP PERFORMANCE

 Operating conditions and collection performance of the particle control devices were measured
 at all three sites. The inlet and outlet mass measurements were previously provided in Table 1.
 Several performance indicators are calculated from those data and are presented in Table 4. The
 collection efficiencies of the two ESPs (Heskett and Black Dog) were approximately the same at
 slightly above 99.4%. The higher inlet mass loading at Black Dog produced outlet emissions that
 were  approximately twice as high  as those measured at Heskett.   The particle penetration
 measured for the Nucla fabric filter #4 was more than an order of magnitude lower than for the
 two ESPs. However, when comparing these units  note that both ESPs had operating problems
 and that only the newest of the four Nucla baghouses was evaluated. The efficiency of the Black
 Dog ESP was affected by operation with one pulser off-line and by high levels of excess air which
 reduced SCA.  The Heskett  ESP was operated with reduced voltage and current because of
 controller problems, and 15% of the collecting area was off-line.

 Typical  electrical conditions measured on the Heskett ESP are shown in Figure  7. The small
 symbols connected with lines represent the voltage-current (V-l) characteristics of the fields, while
 the larger unconnected  symbols  are the automatic operating points of the ESP.  The Heskett
 transformer-rectifier (T-R)  set  controls  are  manually adjusted  to  avoid  all sparking  and
 consequently power input to the  ESP is  not limited by the dust resistivity.  Manual operation is
 necessary because of inadequate automatic controller operation. The manual operating points
 are in general agreement with typical values from the EPRI  data base  [2] for  the resistivity
 measured in situ at Heskett.  The V-l curves, which were run up to spark-limit, indicate that the
 outlet fields can run at much higher current density  consistent with lower resistivity in those fields.
The fourth field, which  does not follow this trend, was not limited  by normal sparking,  but
appeared to have a controller problem which prevented further increases in power.
                                          12-7

-------
During the 85 MW Black  Dog tests,  four of the ESP fields in the direction of gas flow were
operated in pulse energization mode. V-l curves could not be obtained on these  fields.  The
electrical conditions of the Black Dog ESP fields which were not operating in pulsed mode are
shown in Figure 8. These fields were operated in intermittent energization mode during testing,
which accounts for the very low operating points.  The low voltage and high current of the inlet
fields is due to barbed wire  discharge electrodes - the rest of  the fields have round wires.
Although back corona is often masked by the large numbers of particles close to the ESP inlet,
these inlet fields do indicate breakdown above 21  and 34 nA/cm2. The inlet and  outlet fields of
the secondary ESP, which correspond to the fourth and sixth fields in the direction of gas flow,
indicate back corona onset between 3 and 7 nA/cm2.  Back corona occurring below 10 nA/cm2
is consistent with a dust resistivity around 1x1012 ohm-cm, the upper range of values measured
in situ.
PERFORMANCE OF THE ESP MODEL

Revision 3 of the EPA/SRI ESP Model [8] was used to mathematically model ESP performance
for the three AFBC particle sources. The mechanical specifications of the actual ESPs at Heskett
and Black Dog were modelled while a fictitious ESP of typical design was created for projections
on the Nucla dust.  In all cases, the measured particle and flue gas characteristics were used as
model input.  The measured electrical operating conditions of the Heskett ESP were used in the
model, while the electrical conditions at Black Dog and Nucla were estimated from the correlation
between  resistivity, electrode spacing, and operating points  in the  EPRI data  base.  It was
necessary to use estimated electrical conditions for Black Dog because standard procedures for
modeling ESP performance during  pulsed or IE operation have not been  developed.

Two empirically-determined, non-ideal performance parameters are required for the model. These
parameters describe the fractional sneakage of dirty gas around each electrified section (s) and
the normalized standard deviation of the gas velocity distribution (og) in the ESP.  Increases in
either of these parameters tend to degrade ESP performance. Values of s  = 0.05 and og = 0.15
have been shown to provide the best agreement on modern ESPs in good condition, while s =
0.10  and  og = 0.25 more accurately represent older ESPs in questionable states of repair.

The results of the model computations for the three AFBC sites are shown as a function of SCA
in  Figures 9 (ESP efficiency)  and 10 (particle emissions).  The solid symbols represent the
measured performance of the Heskett and Black Dog ESPs.  The open symbols indicate the
modelled performance with the limits of the error bars corresponding to the two sets of non-ideal
conditions (s  =  0.05, og = 0.15;  and s = 0.10, og  = 0.25).  The Heskett ESP performance is
better than the model predicts with the  standard non-ideal conditions.   The measured
performance lies between the computed performance with ideal conditions (s = 0, og = 0) and
the better of the non-ideal conditions. Although not shown on the graphs, the ESP model also
projects that particle emissions would be reduced by a factor of 2 if the ESP were operated at
maximum power input.

Using the electrical conditions predicted from the data base, the model gives good agreement
with performance of the Black Dog ESP for s = 0.05 and og= 0.15. This is rather surprising when
considering the problems with this ESP system. The poor gas flow distribution and high velocities
in the primary ESPs should be sufficient to degrade performance considerably.
                                         12-8

-------
ESP performance on the Nucla dust was projected for a range of ESPs with SCAs from 200 to
600 ft2/1000 acfm and is represented by the bounded areas on Figures 9 and 10.  The upper and
lower limits of the bounded areas correspond to the two sets of non-ideal conditions. In addition
to the model runs using the very high measured resistivity, ESP performance was also projected
assuming  that  humidification was used  to  lower  resistivity to 1x1011 ohm-cm.   Without
humidification, the model indicates that an  ESP with SCA greater than 600 ft2/1000 acfm would
be required to achieve an emission rate of  0.1  lb/106 Btu. In our opinion, it is not clear that this
level of performance could be achieved. The accuracy of the electrical data base and ESP model
with dust resistivity values above 5x1012 ohm-cm is almost totally untested. Application of useful
power to the ESP may not be possible with resistivity values in the 1013 ohm-cm range.

When resistivity  of the Nucla dust was reduced to 1x1011 ohm-cm with humidification, the ESP
model predicts very good performance. An ESP larger than 300 ft2/1000 acfm should meet 0.1
lb/106 Btu, and SCAs greater than 450 ft2/1000 acfm should achieve 0.01 lb/108  Btu.
SUMMARY AND CONCLUSIONS

Detailed measurements have been made on 3 AFBCs - two bubbling beds and one circulating
design.  Particle and gas properties and their impact on ESP performance have been evaluated.
The results can be summarized as follows:

     •  The particle  mass  loading entering the control device was affected by the use of
        limestone for S02 control. When sand was  used as bed filler, the mass loading was
        consistent with the EPRI data base for PC boilers. On the units with limestone addition,
        the mass loading was a factor of 2 higher than the data base predicts.  For constant
        collection efficiency the emissions would also be twice as high with added sorbent.

     •  The number of fine particles generated  by these sources was surprisingly high.  The
        mass of particles smaller than 1 urn was as high or higher than one would expect to find
        on a typical PC boiler burning a subbituminous coal. The shape of the size distribution
        was affected  by limestone addition to the bed.

     •  The in situ resistivity ranged from 1x1012to 4x1013 ohm-cm when limestone addition was
        continuous or regular.  Variations in flue gas moisture content are largely responsible
        for the range of values.

     •  Reasonable agreement between  laboratory and in situ measured dust resistivity was
        observed with sorbent addition.

     •  These dusts  should  respond to resistivity  reduction  by humidification or  SO3
        conditioning.   However, high levels of S03  may have  to be added to produce the
        desired effect.

     •   The electrical operating conditions of the Heskett and Black Dog ESPs were consistent
        with the EPRI  data base for the resistivity values measured.
                                          12-9

-------
        When lime based  sorbents were present in the dust, changes in measured resistivity
        could not be correlated with variations in the chemical constituents of the dust known
        to affect coal ash resistivity. The fly ash resistivity model predictions were low by more
        than one order of magnitude and  generally ignored variations between samples.
        Additional work is required to identify sorbent contributed compounds and to determine
        how they affect the resistivity of ash-sorbent mixtures.

        The ESP model was able to provide a reasonable simulation of the  overall collection
        efficiency of the ESPs.  Agreement was  achieved with low values of the model's non-
        ideal conditions.

        Because of the  unusual nature of the coal and lack of sorbent use in the bed, the
        Heskett results are not applicable to  most applications.
REFERENCES

 1.    KM. Gushing, et al.  Fabric Filter Monitoring at the CUBA Nucla AFBC Demonstration
      Plant,  in Proceedings: Eighth  Symposium on the Transfer and Utilization of Particulate
      Control Technology. Electric Power Research Institute.  Palo Alto, CA.  March 1990.

 2.    J. L DuBard and R. S. Dahlin.  Precipitator Performance Estimation  Procedure.  Electric
      Power Research Institute.  Palo Alto, California.  EPRI CS-5040.  February 1987.

 3.    M. Neville and A. F Sarofim. The Fate of Sodium During Pulverized Coal Combustion.
      Fuel, March 1985, Vol 64, 384.

 4.    R.  P Young.  Measurement and Prediction of  the Resistivity of Ash/Sorbent Mixtures
      Produced by Sulfur Oxide Control Processes.  Draft Report to the  U. S.  Environmental
      Protection Agency, Research Triangle Park, NC.  Cooperative Agreement No. CR812282.
      1988.

 5.    R.  E. Bickelhaupt. A Technique  for Predicting Fly Ash Resistivity.   U. S.  Environmental
      Protection Agency. Research Triangle Park, North Carolina. EPA-600/7-79-204. July 1979.
 6.    R. F  Altman, et al. Electrostatic Precipitation of Particles Produced By Furnace Sorbent
      Injection at Edgewater.  in Proceedings: Eighth Symposium on the Transfer and Utilization
      of Particulate Control Technology. Electric Power Research Institute. Palo Alto, CA. March
      1990.

 7.    J. P. Gooch, J. L DuBard, and R. Beittel.  The Influence of Sorbent Injection on Precipitator
      Performance. Air Pollution Control Association 81st  Annual  Meeting. Dallas,  TX.  Paper
      #88-153.1. June 1988.

 8.    M.  G. Faulkner and J.  L DuBard.  A Mathematical  Model of Electrostatic Precipitation
      (Revision 3): Vol.  I.  U. S. Environmental Protection Agency.  Research  Triangle Park,
      North Carolina. EPA-600/7-84-069. July 1984.
                                          12-10

-------
    10 =
 £  io4
 >—i

 O
Wl



i
    10J
                 NUCLA
                                    SUBBITUH1NOUS COAL
                           -BUCK DOG
         10'1            10°            IO1            IO2


                  PARTICLE DIAMETER,  MICROMETERS



  Figure 1.  Cumulative Mass Particle Size Distribution.
   99.99



    99.9




 -    99


 ac
 LU


 S    90
 «c
 h*
 O

 z    70




 !    s°

 5    30
 _J
 «c


 5    10
    0.1
   0.01
                SUBBITUMINOUS COAL
           HESKETT
           NUCLA-
           BLACK DOS
        IO"1            10°            IO1            IO2


                 PARTICLE  DIAMETER, MICROMETERS



Rgure 2.  Cumulative  Percent Particle Size Distribution.
                             12-11

-------
   10'
   10'
   101
g  10'
   10"
   10'
   io6h
-I—.—I—I—I—'—I—•-

                   IN SITU
                                               INLET
         PREDICTED
                    _L
                          _L
                                             MIDDLE
                                               I   .	L
        3.0    2.8    2.6   2.4    2.2    2.0   1.8   1000/K
        60
              84
                    112
                          144    182    227   283   -C
        140    183    233   291    359    441   541   -F
                          TEMPERATURE

          Figure 3.  Heskett AFBC Dust  Resistivity.
    10'
    10'
        :  LAB 8/9
    10'
    10'
    10' -
             IN SITU
        :  us a/io
          PREDICTED
        •  8/9-8/10   '
        3.0   2.8    2.6    2.4   2.2    2.0    1.8   1000/K
        60    84     112    144   182    227    283   -C
        140   133    233    291   359    441    541   -F
                           TEMPERATURE

          Figure 4.  Black Dog AFBC Dust Resistivity.
                             12-12

-------
      10'
      101
   g  1012
      101
      10
        10'
                                      IN SITU
             PREDICTED
                  I
          3.0
          60
          140
2.3
84
183
2.6
112'
233
2.4    2.2    2.0
144    182    227
291    359    441
 TEMPERATURE
                   1.8   1000/K
                   283   -C
                   541   -F
               Rgure 5.   Nucla AFBC Dust  Resistivity.
      iolz
      10'
      10
        101
             7.8X WATER
            11.SX WATER
            14.« WATER
                               •  5 ppa 503
          3.0
          60
          140
2.3
34
133
2.6
112
233
2.4
144
291
      2.2    2.0
      182    227
      359    441
TEMPERATURE
1.8  1000/K
283  -C
541  -F
Figure 6.  Effect of Conditioning on the Resistivity of Nucla Dust.
                             12-13

-------
    60  -
    SO
    40
    30
I   20
    10
                                     44.8,72
        25
                                                   SO
                30       35      40       45
                       ESP  VOLTAGE,  icV
  Figure 7. Electrical Characteristics of the Heskett ESP.
     o -
                                        40
                                                   so
                   20         30
                       ESP VOLTAGE, KV
Figure 8.  Electrical Characteristics of the Black Dog ESP.
                           12-14

-------
    99.99
     99.9
   _
   E   99
       90
       70 -
                         HUCU WITH HIMID
              BLACK DOS
          ZOO
                                 NUCLA W/0 HUMID
                                                      600
                     300        400        500

                        SCA,  FT2/1000 ACFM

Figure 9.  Predicted and Measured ESP Collection Efficiency.
      10' r
      iou
     10"
                                       NUCLA tf/0 HUMID
               BLACK DOS
                   HESKETT
                            NUCLA KITH HUHIO
          200        300        400        500

                        SCA,  FT2/1000 ACFM
600
 Figure 10.  Predicted and Measured ESP Particle Emissions.
                             12-15

-------
TABLE 1. COAL COMPOSITION

% Moisture
% Ash
% Sulfur
% Volatile
% Fixed Carbon
HHV, Btu/lb
% Carbon
% Hydrogen
% Nitrogen
% Chlorine
% Oxygen
Heskett
36.22
8.20
1.22
26.01
29.57
6885
40.39
2.63
0.61
0.01
10.72
Black Dog
20.64
8.05
1.23
34.87
36.44
9603
55.09
3.63
0.92
0.05
10.39
Nucla
7.91
13.64
0.58
33.32
45.13
11088
62.71
4.19
1.32
0.01
7.91
TABLE 2. PARTICULATE MASS LOADINGS
PLANT
PARTICLE MASS
LOADING
gr/acf
gr/scf
GAS VOLUME
FLOW
acfm
scfm
TEMPERATURE
°F
CONTROL DEVICE INLET
Heskett
Black Dog
Nucla
1.16
2.69
4.32
2.05
4.29
8.85
343800
451500
257600
1 95800
283500
125800
296
280
303
CONTROL DEVICE OUTLET
Heskett
Black Dog
Nucla
0.0069
0.0163
0.0018
0.0112
0.0255
0.0037
341 300
432700
248400
205000
275400
122100
277
284
286
12-16

-------
TABLE 3. CHEMICAL COMPOSITION OF DUST SAMPLES

Li20
Na-jO
(C.O
MgO
CaO :
Fe203
AIA
SiO2
TiO2
P205
S03
LOI
Sol SO4
Heskett
Inlet
0.01
6.4
0.48
9.7
28.3
4.9
1.0
7.4
0.33
0.49
6.9
2.2
5.9
Middle
0.01
12.1
0.72
8.6
25.1
3.8
8.4
10.9
0.33
0.47
25.9
7.0
27.0
Outlet
0.03
25.9
1.2
4.8
13.7
2.2
4.4
4.5
0.25
0.27
41.1
4.1
46.3
Black Dog
8/9/88
0.01
1.2
1.3
3.7
26.2
8.2
15.2
31.8
0.92
0.64
9.0
11.0
10.3
8/10/88
0.01
1.3
1.2
3.9
23.6
8.0
16.3
34.8
0.92
0.72
8.1
10.5
8.8
Nucla
9/19/89
0.02
0.47
0.45
0.74
17.8
1.8
23.6
47.5
0.67
0.49
5.1
6.2
6.1
9/22/89
0.03
0.47
0.42
0.74
19.9
1.8
23.1
46.7
0.75
0.52
4.7
6.0
5.5
TABLE 4. CONTROL DEVICE COLLECTION PERFORMANCE

PLANT
Heskett ESP
Black Dog ESP
Nucla FF
EFFICIENCY
%
99.454
99.406
99.959
EMISSIONS
gr/acf
0.0069
0.0163
0.0018
OPACITY
%
7


SCA
ft2/kacfm
368
352

G)k"
cm/sec
38.8
37.6

* Effective Migration Velocity
12-17

-------
Choice Between ESP and Baghouse for Pakistan's First
               Coal-Fired Power Plant

                Ghulam Murtaza IIias
                 (no  paper provided)
                      13-1

-------
                   EFFECTS OF E-SOX TECHNOLOGY ON ESP PERFORMANCE
                                G. H. Marchant, Jr.
                                    J. P. Gooch
                                   M.  G.  Faulkner
                            Southern Research  Institute
                              2000 Ninth  Avenue, South
                             Birmingham,  Alabama   35205
                                    L. S. Hovis
                           Gas Cleaning Technology Branch
                             Pollution Control  Division
                   Air and Energy Engineering Research Laboratory
                       U. S.  Environmental Protection Agency
                   Research Triangle  Park, North  Carolina  27711
ABSTRACT

The E-SOX process is being evaluated at pilot scale at Ohio Edison's Burger Station.

Adequate S02  removal, while maintaining an  acceptable  particulate  emission  level

from the ESP, is the prime objective of this investigation.  This  paper describes

limited ESP performance testing under both  baseline and E-SOX conditions.   The ESP

data collected under E-SOX conditions,  which give the  required S02 removal, show

evidence of ESP behavior dominated by factors not represented in existing versions

of precipitator performance models.   Analyses of particle size fractions from

impactor stages revealed that the relative  calcium content of the  finer size

fractions increases from inlet to outlet.   From these  analyses and other consider-

ations, it appears that the factors  which dominate under the conditions tested are a

combination of instantaneous  reentrainment  of low resistivity ash/sorbent particles

and deagglomeration of  slurry residues  within the precipitator.   These observations

may be important to other sorbent injection processes  as well as E-SOX.  Further

work is underway to develop a better understanding of  these slurry residue proper-

ties so that  their effects can be addressed in additional E-SOX testing.


          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.
                                        14-1

-------
                   EFFECTS OF E-SOX TECHNOLOGY ON ESP PERFORMANCE
INTRODUCTION

The E-SOX process involves removal  of sulfur oxides prior to the inlet of an elec-
trostatic precipitator (ESP)  with an aqueous spray of an alkaline material.  The
entering fly ash and resultant particulate matter are then removed in the ESP.  A
research program to develop and demonstrate the process is in progress under the
sponsorship of the U.S.  EPA,  the Ohio Coal Development Office, and the Babcock &
Wilcox Company.


Process Description

Slaked lime slurry without the use of recycled material has been the source of
alkalinity for experiments performed to date.  Pebble lime was transferred
pneumatically from tank trucks to a storage bin.  The lime was then slaked and
placed in a slurry tank.   The slurry was then metered and injected into a spray
chamber through two B&W Mark 4 nozzles.  Dilution water was added with the slurry
prior to reaching the nozzle, depending on the calcium to sulfur ratio and approach
to saturation desired.  At the exit of the spray chamber and ahead of the ESP are
two rows of Droplet Impingement Devices (or DIDs) which are temperature-controlled
pipes to prevent entry of large wet particles into the ESP.  The flue gas and
uncollected particulate matter which exited the ESP were returned to the main
ductwork ahead of the main unit's ESP.  Figure 1 presents a schematic of the E-SOX
pilot facility.


Pilot ESP Description

The ESP installed at the E-SOX facility is EPA's pilot transportable ESP (TEP) which
was originally installed at Public Service Company of Colorado's Valmont station.
The pilot ESP was disassembled at Valmont and reassembled at the Burger station.
Personnel of B&W's Alliance Research Center supervised the reassembly of the TEP and
operated the pilot system during the test programs.  Figure 2 shows the E-SOX tran-
sition and ESP arrangement installed at the Burger station.

The ESP consists of four electrical fields in the direction of gas flow, and each
field is preceded by a cooled pipe precharger of Denver Research Institute design.
Each field consists of six gas passages, 9 in.* wide,  12 ft high, and 70 in. deep.
This results in a total collection area of 3360 ft2.   The discharge electrodes are
0.25 in. in diameter, 9 in. apart, and are held in a rigid frame.  The discharge
frames and collecting plates are rapped with a drop hammer type of rapping system.
All collecting plates in a section or all discharge wires  in  a section are rapped  at
once when a rap is called for during the rapping cycle.
'Readers more familiar with metric units may use conversion table at end of paper.
                                         14-2

-------
Ash collected in the precharger sections or collector fields is discharged from the
individual hoppers into a screw conveyor by a rotary valve.  The ash is then trans-
ported to a storage bin for disposal.  Ash from the transition section is also
removed by screw conveyors, but placed in a different storage bin.


E-SOX Test Program

The E-SOX pilot facility was operated and maintained by personnel from B&W's
Alliance Research Center.  The B&W test program included evaluations of:  spray
nozzles, SO, removal  efficiency at various stoichiometric ratios  and approaches to
saturation temperature, the Droplet Impingement Device, and equipment performance
during long term operation of the process.  Results from this work will be reported
elsewhere.

Southern Research Institute (SRI) was responsible for testing the ESP under
"baseline" and "E-SOX" conditions and for evaluating the performance of the
precipitation process.  These particulate characterization measurements and ESP
performance analyses are the subject of this paper.


PRECIPITATOR PERFORMANCE

The ESP, the fly ash, and fly ash/sorbent mixtures were characterized by measuring:

     •     Inlet and outlet mass concentration

     •     Inlet and outlet mass vs particle size with cascade impactors

     •     Real-time outlet mass concentration trends with an Environmental Systems
           Corporation PSA mass emissions monitor

     •     Secondary voltage-current relationships and operating points

     •     Inlet velocity traverses

     •     Inlet and outlet temperature traverses

     •     Laboratory and in situ resistivity

     •     Chemical analysis of bulk and size-fractionated samples

     •     Ash cohesivity and Bahco particle size


Baseline measurements were performed without the DID array, whereas the sorbent
injection tests necessarily were performed with the DID array present.  In addition
to preventing penetration of large, moist particles into the first field of the ESP,
the DID array was designed to minimize gas velocity non-uniformity due to flow
disturbances caused by the sorbent injection nozzles.

Emissions caused by rapping systems in pilot-scale ESPs are usually not representa-
tive of full-scale systems.  Therefore, the test program was conducted with rapping
systems de-energized during the time period that outlet measurements were underway.
Rappers were energized between tests to avoid excessive electrode buildups.  This
testing strategy allowed the overall and particle size dependent efficiencies to  be
compared with the "no rap" projections of the mathematical model, as will be dis-
cussed later.

                                        14-3

-------
Tables 1 and 2 present results from particulate mass concentration measurements
under baseline and slurry injection conditions, respectively.  Voltage-current
relationships for these test conditions are presented in Figures 3, 4, and 5, and
average electrical operating points are given in Table 3.  Figure 6 illustrates the
changes in the signal  from the outlet mass monitor as average temperature was
increased from 160 to  180°F at the inlet.   The decrease in "noise" due to non-
rapping emission spikes is apparent as temperature was increased.  The actual
decrease is larger than illustrated,  because the 180°F segment is recorded on a
scale with a maximum of 0.5, whereas  the 160°F segment was recorded with a maximum
scale reading of 2.0.

ESP performance changed from excellent under baseline conditions (0.012 or 0.008 lb/
MMBtu out) to unacceptable (0.113 to  1.55  Ib/MMBtu out) with slurry injection.  Note
that inlet data for slurry conditions were obtained downstream of the DID array in a
low velocity region, and therefore the efficiency data in Tables 1 and 2 are not
directly comparable.  Average process conditions on 10/22 and 10/25 were as follows:
Ca/S ratio = 1.44; S02 removal  =  48.5%;  and the estimated total  mass loading of
dried, partially sulfated sorbent and fly  ash was 10.2 gr/scf.  Using this estimate
as a basis, the mass train traverse in the low velocity region recovered about 38%
of the total mass.

"Omega K" values in Tables 1 and  2 represent an ESP performance parameter that
provides a semi-quantitative means of comparing performance under various con-
ditions.2  As points of reference,  omega k values of full  scale  ESPs collecting ash
downstream of spray dryers have been  reported to range from 27 to 62 cm/sec.  The
recent paper by Durham, et al .3 provided data on the Shawnee  TVA ESP Spray Dryer
pilot plant which indicated omega k values ranging from 32 to 53 cm/sec.  Thus, the
highest efficiency "no-rap" data  from the  E-SOX system at 180°F indicate a per-
formance parameter in  the lower portion of the range reported for spray dryer
applications.

The decrease in outlet emissions  measured  by the mass train with increasing temper-
ature on 10/25/89 is confirmed by the P5A  trace shown in Figure 6.  This trend
toward higher emissions at lower  operating temperatures was observed at earlier
times in the test program, and proved to be a reproducible phenomenon.  The large
spikes appear to represent non-rapping reentrainment occurring on a massive scale at
the lower temperatures.

A related observation  concerns the appearance of rapping spikes on the P5A output
which coincides with rapping of the DID array.  This rapping process is expected to
produce relatively large particles which would be charged and be driven to the
collecting electrodes  quickly with the observed voltages and currents.  However,
large rapping spikes were observable  at the outlet with a time lag corresponding to
the gas transit time between the  DID  array and the P5A sampling point.  This obser-
vation indicates that  a large fraction of  the sorbent/ash mixture is instantaneously
and repeatedly reentrained by electrical forces as it travels through the electrical
fields.

The electrical operating points in Table 3, and the voltage-current curves in
Figures 3 through 5, reveal no anomalies which would explain the extremely high
outlet emissions with  sorbent injection.  As will be shown in a subsequent dis-
cussion of observed vs predicted  performance, the measured voltages and currents
indicated that very high electrical migration velocities and collection efficiencies
would be predicted under both baseline and slurry conditions.  It is also of inter-
est to note that a significant change in voltages and currents did not occur as
temperature was increased from 162 to 181°F, although mass emissions decreased by  a
factor of 13.7.
                                        14-4

-------
An extensive trouble shooting effort was performed during the test program in an
attempt to locate possible mechanical problems that might be responsible for the
excessive particulate emissions with slurry injection.


Specifically:

     •    Hopper fluidizing air was turned off and on;

     •    The ash removal screw conveying system was turned off and on;

     •    Nozzles and the DID array were periodically cleaned both on line
          and while the system was down for short-term repairs; and

     •    Voltages were held to values below those which would cause
          excessive sparking during the test periods.


The above items related to mechanical issues did not result in a reduction in outlet
emissions; excessive sparking did, however, increase outlet emissions.  Also, the
No. 4 precharger was energized and de-energized, with no apparent effect on the
outlet mass monitor.

Overflowing hoppers are one source of emissions which could not be directly ruled
out by observation.  However, the ash removal  system was monitored to ensure that it
was operating.  Furthermore, the reproducible change in outlet emissions with tem-
perature is not explainable by hopper overflow since the lower emission data were
obtained later during the same test day.

In view of potential flow disturbances due to the presence of sorbent injection
nozzles and the DID array, temperature traverses were conducted during the test
program.  A velocity traverse with air flow was also performed after the test
program with the perforated plate downstream from the DID array in an uncleaned
condition.  Tables 4 and 5 illustrate the inlet temperature distribution on October
25 when the average inlet temperatures were 160 and 180°F, respectively.  Table 6
contains the inlet velocity traverse.  Note that high velocity and low temperature
areas coincide near the bottom of the ESP.  This combination of conditions, where
reentrainment by electrical forces is greatest in the region of highest velocity,
would be expected to exacerbate the ESP performance problems.  Also, during the test
program, an average temperature of 155°F was obtained from a traverse at the outlet
of the fourth field of the ESP, while the average inlet temperature was 173°F.  As
expected, the lowest outlet temperatures occurred near the bottom of the ESP.

The test program data clearly indicated that operation in the 180°F region improved
ESP performance.  However, operation at temperatures far above the adiabatic satu-
ration point is not an acceptable solution for excessive particulate emissions
because of the adverse effects on S02 removal.


PARTICULATE CHARACTERIZATION

Figure 7 contains cumulative inlet size distributions obtained by impactors at the
E-SOX pilot facility, and Figure 8 illustrates baseline and E-SOX size-dependent
efficiencies obtained from impactor data.  Also shown on Figure 8 are ESP model
projections for baseline and E-SOX conditions.  These projections include the
effects of 10% sneakage and reentrainment with a gas velocity standard deviation  of
0.25.   The model projections will be discussed later.  Total mass loadings obtained
with the impactor traverses are presented in Table 7.


                                        14-5

-------
If it is assumed that each slurry droplet produces one agglomerated particle upon
drying,  a size distribution can be estimated for dried and partially sulfated
sorbent.  Figure 9 contains an estimated size distribution of the dried sorbent
downstream of the DID array that was supplied by B&W.4  A Bahco-derived size dis-
tribution of ash/sorbent mixture obtained from the ESP hopper is also provided on
Figure 9.

The impactor data in Figure 7 indicate no significant difference in total inlet
loading  between the E-SOX and baseline conditions in the size range resolved by the
impactors (below about 8 ^m diameter).  An examination of the dried agglomerate size
distribution in Figure 9 reveals that only 15% of the slurry residue would be
expected to consist of particles 8 /j.m in diameter and smaller, which is consistent
with the lack of increase that was observed in this size range by inlet impactors.
The low  total mass loading obtained with the impactors under E-SOX conditions
results  from the fallout and impaction which occur in the spray chamber and on the
DID array, and from the difficulties in obtaining a representative total mass sample
in the low velocity region behind the DID array.

A comparison of the baseline and E-SOX fractional efficiency curves in Figure 8
indicates a large drop in efficiency under E-SOX conditions across the entire size
range.  Total average mass loading obtained with the outlet impactors on 10/20 and
10/23 at 170°F (Table 7) is similar to that obtained with the mass trains at 180°F
on 10/25 (0.0497 vs. 0.0464 gr/scf).  It is of interest to note that the penetration
ratio of E-SOX to baseline conditions at 3 urn diameter is similar to the results of
Durham3  for the Shawnee spray dryer ESP (13 for  E-SOX and 10 for Shawnee).

A comparison of the ESP model projections with baseline data in Figure 8 shows good
agreement between model projections and measured results.  In contrast, the E-SOX
measured data exhibit a large decrease in efficiency instead of the increased
performance projected by the ESP model.  This indicates that ESP performance with
sorbent  is dominated by factors which are not represented in the existing model.
All of the model projections are "no rap" cases, since the pilot ESP rapping system
was not  energized during sampling.

Penetration vs particle size curves such as those in Figure 8 are calculated by
taking the ratio of outlet to inlet particle mass in a given size interval.
However, the dried slurry residue is an agglomeration of smaller particles which
originated from the lime slaking process, and the potential exists for deagglomer-
ation to occur in the ESP.  It has been hypothesized that slurry agglomerates could
be broken apart due to internal forces arising from the relatively high values of
charge acquired by particles in the interelectrode region.  These forces must
overcome the cohesive forces between the individual  particles.  The ID fan upstream
of the outlet sampling ports would not be expected to provide sufficient shear
forces to deagglomerate particles in the size range of interest.  If deagglomeration
or decrepitation did occur, the fractional efficiency curves would represent the net
of the collection and decrepitation processes.

Evidence that a significant degree of slurry residue decrepitation did occur is
presented in Figures 10 and 11.  These figures contain plots of the signal ratio of
calcium to iron and calcium to silicon from an energy dispersive X-ray (EDX)
analysis of inlet and outlet impactor substrate samples.  There is a large change in
the relative amounts of calcium to iron and silica from the inlet to the outlet.
This change is consistent with the hypothesis that slurry residue agglomerates were
broken apart, so that the relatively fine, calcium-rich particles dominate the
smaller  size fractions in the outlet samples.

Photomicrographs of outlet impactor stages 4 and 6 are presented in Figure 12 for
baseline and E-SOX conditions.  Also illustrated is a large agglomerate captured  in


                                        14-6

-------
the cyclone stage of the inlet impactors.  The outlet stages illustrate that the
impactors were classifying the sampled particles, and no evidence of gross reen-
trainment of larger particles from the upper stages was observed.  If reentrainment
occurred in the impactors to a significant degree, the change in composition with
impactor stage could not be attributed to compositional differences as a function of
particle size.

Further evidence that deagglomeration of slurry residue can occur is given in Figure
9, in which it can be observed that a Bahco size distribution of a sample obtained
from the ESP hoppers contains more fine particles on a relative basis than the
estimated slurry residue distribution entering the ESP.  Figure 13, which provides
composition of the Bahco size fractions as a function of particle size, illustrates
that the finer size fractions are dominated by calcium-rich material.

Chemical analyses of samples collected from several points in the system are con-
tained in Table 8, and coal compositions of representative samples are presented in
Table 9.  It is of interest to note that the outlet sample composition is very
similar to that obtained at the ESP inlet.  This observation provides further
evidence that the slurry droplet residues were not retained in the ESP as predicted.
Since the agglomerated slurry residue has a small fraction of fine particles,
theoretical collection efficiency vs particle size relationships predict that the
ESP would selectively collect the slurry residue so that the outlet particle mass
would contain a significantly smaller fraction of calcium compounds.

Reliable values for the electrical resistivity of sorbent/ash mixtures are difficult
to obtain with standard methods.  The in situ point-plane resistivity probe has been
reported to selectively reentrain low resistivity sorbent particles,  and this
process is likely to have biased data obtained with the probe during this test
series.  In addition, samples collected at various points in the system at different
times during the slurry injection test program exhibited significantly different
resistivity vs temperature curves, as Figure 14 illustrates.  The laboratory data
were obtained using a modified procedure that was adapted for samples containing
calcium sorbents.5

An examination of the in situ and laboratory data for 10/24 and 10/25 (Figure 14)
indicates resistivity values in the 1010  to  10   ohm-cm range at 160-180°F,  which
should not result in low resistivity reentrainment.  However, the samples of 10/3
indicate resistivity values below 1010 ohm-cm  at  170°F,  and  exhibit  a  very  steep
slope which would result in resistivity values of less than 108 ohm-cm at 140°F.   It
should be noted that the laboratory atmosphere is only a static simulation of the
dynamic environment which exists in the ESP.  It could be argued that the presence
of sulfur oxides and surface moisture in the actual environment is likely to produce
a lower real-time resistivity value with a representative sample than would be
obtained in the laboratory air-water vapor environment with samples obtained from
system hoppers.

Baseline resistivity data are also shown on Figure 14.  The probe provided data  in
good agreement with the laboratory data in equilibrium with 7 ppm S03.  This value
of SO,  was  estimated  to  result  from the  sulfur content of the coal.   The observed
fact that neither the probe nor the ESP experienced any difficulty in collecting an
ash with a resistivity of 109 ohm-cm suggests  that low resistivity may not be the
only property of a dust responsible for excessive reentrainment.

Ash cohesivity is also expected to be relevant in efforts to quantify factors
responsible for excessive reentrainment.   Measurements of cohesivity of fly ash  and
ash/sorbent mixtures were performed on fly ash alone and on the ash/sorbent mixtures
from the ESP with slurry injection.  These measurements produced the surprising
result  that the E-SOX solids at 145°F exhibited a cohesivity in the low end of the


                                         14-7

-------
range (40.3°,  angle of internal  friction6)  measured  for a large number of fly ash
samples.   Furthermore, the angle of internal  friction of the fly ash at 285°F was
45.5°,  which is significantly higher than the E-SOX solids at 145°F-  Low cohesivity
would be  expected to aggravate a reentrainment tendency resulting from low dust
resi stivity.


DISCUSSION

Table 10  contains a summary of measured and model  predictions of outlet mass loading
and efficiencies for the E-SOX pilot facility.  Since all measurements were con-
ducted  with rappers off, the model  estimates  are all "no rap" values.  The modeling
factors represent the fraction of sneakage/reentrainment which is assumed to occur
over four stages, and the sigma g value is  the normalized standard deviation of the
gas velocity distribution.  The value of 0.25 is assumed for the baseline test; the
value of  0.36 is based on the measurements  presented in Table 6.

A comparison of measured and predicted results shows that the model  failed to
predict the performance trends as well as the absolute value of outlet emissions.
Performance improvements were predicted because of increased electrical  migration
velocities at the lower temperatures with slurry injection.   However, outlet
emissions increased with slurry injection from predicted values by factors ranging
from 28 to 390.  Model output could be forced to match the measured  results by
assigning reentrainment values per stage ranging from 40 to 85%.

These results are qualitatively similar to  those of Durham,3 in  which model  pre-
dictions  of the Shawnee spray dryer ESP were  off by a factor of 80,  and a reen-
trainment factor of 60% was required to match measured and modeled penetrations
under spray dryer conditions.  However, there are  significant quantitative
differences in that omega k values  for the  E-SOX unit ranged from a  low of 4.7
cm/sec  to a high of 32.9 cm/sec.  In contrast, the Shawnee data indicated omega k
values  ranging from a low of 32.3 cm/sec with a two  field configuration to 51 to 53
cm/sec  with three or four fields.

The extreme temperature sensitivity of the  E-SOX ESP performance is  believed to
result  from the combined high-velocity/low-temperature regions near  the bottom of
the precipitator.  These conditions would magnify  the process of electrical
reentrainment which also appears to be occurring to  some degree at the higher
temperatures.


CONCLUSIONS

1.   Analysis of particle size fractions collected on impactor stages at the inlet
     and  outlet of the E-SOX ESP showed a large increase in the relative calcium
     content of the finer size fractions across the  ESP.

2.   Massive reentrainment of ash/sorbent mixtures could be induced  without elec-
     trode rappers in service by lowering the operating temperature  of the ESP
     inlet.  The reentrainment could be reduced by elevating the average inlet
     operating temperature 20°F with no accompanying change in secondary voltages
     and  currents.

3.   ESP  performance for the E-SOX  process  is dominated by two factors not repre-
     sented in the existing EPA-SRI versions  of the mathematical model of ESP
     performance.  These factors are instantaneous reentrainment of low resistivity
     ash/sorbent particles and deagglomeration of  slurry residue within the ESP.
                                        14-8

-------
     Significant improvement of the E-SOX ESP performance is expected to result from
     correction of velocity and temperature non-uniformities downstream from the DID
     array.

     Additional work is required to develop a quantitative understanding of the
     chemical  and physical  properties of slurry residues which result in poor ESP
     performance.  Slurry additives to effect desirable changes in such properties
     and ESP electrode design modifications are both remedial  approaches which
     should  be considered for E-SOX and other low-temperature sorbent injection
     processes.
REFERENCES

1.
K. Redinger, et al.   "E-SOX 5 MW Pilot Demonstration Results," to be presented
at the 1990 S02 Control  Symposium,  May 1990.

S. Maartmann.  "Experience with Cold Side Precipitators on Low Sulfur Coals."
In:  Symposium on the Transfer and Utilization of Particulate Control
Technology, Volume I, EPA-600/7-79-044a (NTIS PB295226), February 1979.

M. D. Durham, D. E.  Rugg, R. G. Rhudy, and E. J. Puschauel.   "Low-Resistivity
Related ESP Performance Problems in Dry Scrubbing Applications."  J. Air Waste
Manage. Assoc. 40:112(1990).

Correspondence from Kevin Redinger of B&W Alliance Research  Center to G. H.
Marchant, Jr. of Southern Research Institute, January 23, 1990.

R. P. Young, J. L. DuBard, and L. S. Hovis.  "Resistivity of Fly Ash/Sorbent
Mixtures."  In Proceedings:  Seventh Symposium on the Transfer and Utilization
of Particulate Control  Technology, Volume I,  EPA-600/9-89-046a (NTIS PB89-
194039),  May 1989.

0. Molerns.  "Theory of Yield of Cohesive Powders."  Powder  Technology 12:259.
1975.
        TO CONVERT FROM

            in.
            ft
            ft3
            Ib
            gr   ^
            lb/ft3
            gr/ft3
            ftymin
            °F
            ft2/kacfm
            1 b/MMBtu
                           UNIT  CONVERSION  TABLE

                                    TO

                                cm
                                m
                                m3
                                g
            ft/sec
                                m /sec
                                °C
                                m2/(m3/sec)
                                ng/J
                                nA/cm
                                m/sec
MULTIPLY BY

 2.54
 0.3048
 0.02832
 453.6
 0.06480
 1.602 x  104
 2.288
 0.000472
 5/9  (°F-32)
 0.19685
 430
 1.08
 0.3048
                                        14-9

-------
                                                                      E-SOX
                                                            PILOT  PROGRAM
                                                            OHIO EDISON COMPANY
                                                               R.E. BURGER  PLANT
                            Figure / E-SOX Pilot Facility Schematic
                                 Figure 2  E-SOX Transition and TEP Elevation
O INLET FIELD
• SECOND FIELD
A THIRD FIELD
A OUTLET FIELD
  ZO        30
O INLET FIELD
• SECOND FIELD
A THIRD FIELD
A OUTLET FIELD
                                                         10        20
                                                                                              50        60
       SECONDARY VOLTAGE kV
                                                                        SECONDARY VOLTAGE, kV
                                                  Figure 4 EPA Piloi-ESP Voltage-Current Curves E-SOX Slurry Test, f 0/25/89, 160°F
                                             14-10

-------
                     O  INLET FIELD
                     •  SECOND FIELD
                     A  THIRD FIELD
                     A  OUTLET FIELD
                                30
                                                         50
                                                                     60
                           SECONDARY VOLTAGE, kV

Figure 5  EPA Pilot-ESP Voltage-Current Curves E-SOX Slurry Test. 10/25/89. 180 °F
                                                                                           ,   j _J»|	,	]	1	!	 1 ST MASS TRAIN TEST 	,—
                                                                                                                  MAES TRAIN = 1 1 g/ acm OR 1 55 lt>/106BTU
                                                                                                                                   10/25/69

                                                                                                                                     1509
                                  PSA FULL SCALE - 0 5g/st
                                            1530 I    li
                                                                                                                           - 2ND MASS TRAIN TEST -
  . MASS TRAIN •* 0.08 g/ncm OR 0 11 lb/106BTU



Figure 6  Strip Charr Recording of PSA Output
Figure 7    Inlet Cumulative Mass vs Panicle Site for Baseline and Post-DID E-SOX
           Conditions
                                                                                                                            PARTICLE SIZE.fj.rn

                                                                                           Figure 8  Measured and Modeled Penetration vs Particle Size, E-SOX Demonstration Tes
                                                                              14-11

-------
                                         O PREDICTED POST DID
                                             SLURRY RESIDUE
                                         O BAHCO DATA ESP
                                             HOPPER SAMPLE
                            ARTICLE DIAMETER, urn
           Figure 9  Bahco and Predicted Slurry Residue Size Distribution
I,
                                                                                                                   J	L
                                                                                                       0 2           0.5       1.0       2.0
                                                                                                                                                   5.0       100
                                                                                          Figure TO  EDX Derived Calcium la Silicon Ratios vs Particle Size lor E-SOX Impactc
                                                                                                   Substrate Samples
                                                                                  Figure 12 Photomicrographs of Impartor Catches - Baseline and £ SOX Candida
01        02
                                                     50       100
   e t!  EOX Derived Calaum to Iron Ratios vs Candle Size for E SOX t
        Substrate Samples
                                                                            14-12

-------
                                                                                                                    LABORATORY
                                                                                                                        ESP, 6/26:67%H20, 7 p
                                                                                                                      O ESP, 10/3. 10.9% H20
                                                                                                                      D ESP, 10/24  9B%H20
                                                                                                                      O ESP, 10/26  9 6% H2O

                                                                                                                    IN SITU

                                                                                                                      A BASELINE
                                                                                                                      A E-SOX CONDITION
                            PARTICLE SIZE,
                                                                                                                    TEMPERATURE
Figure 13. Analysis of Bahco Size Fractions from ESP Hoppers with Slurry In/ectii
                                                                                                       iry and In Situ Res
                                                                                                                                    wnts. E-SOX Demonstrat/o
                             PILOT ESP RESULTS
                        MASS MEASUREMENTS, BASELINE
                  PILOT ESP RESULTS
         MASS MEASUREMENTS, E-SOX CONDITIONS
                                    6/24/89
SYSTEM INLET
Temp., -F
gr/scf
1 b/HHBtu
OSCFH
OUTLET'
Temp . , ' F
gr/scf
1 b/HHBtu
EFFICIENCY*,?.
SCA, ftVkacfm
OMEGA K, cm/sec
'No Rapping
^Based on 1 b/HHBtu
308
4.199
9.258
8553
276
0.0045
0.012
99.87
231
97.1
315
4.018
8.296
9017
277
0.0033
0.008
99.90
215
112.7
POST DID INLET
Temp . , 'F
gr/scf
1 b/MHBtu
DSCFM
OUTLET*
# of Runs
Start Time
End Time
Temp., -F
gr/scf
1 b/HHBtu
EFFICIENCY', X
SCA, ftz/kacfm
OHEGA K, cm/sec

170
3.891
7.867
9267

2
1116
1527
175
0.1288
0.313
96.02
257
20.5

160
3.891'
7.867
8493

1
1002
1110
162
0.6352
1.552
80.27
287
4.7

180
3.891'
7.867
8493

1
1430
1546
184
0.0464
0.113
98.56
278
32.9
                                                                                               'Inlet  Data  of 10/22/89 Used
                                                                                               ^No  Rapping
                                                                                               'Based  on  1b/HHBtu
                                 Table  3

                 AVERAGE  ELECTRICAL OPERATING CONDITIONS
          Baseline,  6/25/89,  281'F

          E-SOX  Slurry,  10/25/89,  180'F

          E-SOX  Slurry,  10/25/89,  160'F
       TEMPERATURE DISTRIBUTION AT  ESP  INLET
    AVERAGE INLET GRID TEMPERATURE  EQUALS 160'F


             Position  Position   Position  Position
               One       Two      Three      Four
               •F        -F         -F         -F

Top Row        172       175        176        177

Third Row      168       165        164        164

Second Row     157       153        156        150

Bottom Row     148       147        147        146
                                                                                                 GRID IS VIEWED  IN  DIRECTION  OF GAS FLOU
                                                                        14-13

-------
         TEMPERATURE  DISTRIBUTION AT ESP INLET
      AVERAGE  INLET GRID TEMPERATURE EQUALS 180'F
                                                                      GAS  FLOU  MEASUREMENTS AT PILOT ESP  INLET
                                                                           VELOCITIES IN FEET PER MINUTE
               Position   Position  Position  Position


Top Row
Third Row
Second Row
Bottom Row
One
•F
195
187
176
173
Two
•F
186
182
171
166
Three
•f
190
184
17Z
165
Four
•F
196
187
175
172
      GRID IS VIEWED IN DIRECTION OF SAS FLOH
                                                                                                  Position  Position   Position   Position
                                                                                                    One       Two       Three       Four
                                                                                                   ft/min    ft/min     ft/min     ft/min
                                                                                     Top Row

                                                                                     Third Row

                                                                                     Second Row

                                                                                     Bottom Row
                                                                              388

                                                                              385

                                                                              282

                                                                              655
                                                                          275

                                                                          235

                                                                          205

                                                                          520
335

245

230

455
400

325

465

640
                                                                    GRID  IS  VIEWED  IN DIRECTION OF GAS FLOH

                                                                     1)   Perforated  Plate Uncleaned From
                                                                         Operability Test
                                                                     2)   Mass Gas Flow 44,000 Ib/hr, Temp. 91
                                                                     3)   Grid Average is 378 ft/nln.
                       Table 7

          PILOT  ESP  IHPACTOR DATA, BASELINE
Date
Mass


Avg.
Avg.
Avg.

Date
Mass


Avg
Avg.
Avg.
Loading
gr/acf
gr/scf
Temp., -F
HMD, Jim
SCA, ft2/kacf
PILOT ESP

Loading
gr/acf
gr/scf
Temp . , ' F
HMD, pi
SCA, ftVkacf
Inlet
6/25 t 6/27
1.7834
3.0160
306
19.9
m 208
IMPACTOR DATA,
System Inlet
10/20

2.163
3.405
264
21
m 272
Outlet
6/25
0.0034
0.0053
270
5.1







E-SOX CONDITION
ESP Inlet
10/23

1.114
1.567
170
17

Outlet
10/20 i 10/23

0.0372
0.0497
170
4.J

                                                                                                           Table  8

                                                                                              BURGER PILOT SYSTEM E-SOX SOLIDS

«
Li,0
Na,0
M
MgO
CaO
Fe,0,
Al |2
SiO?
TiO,
P,0,
S°3
a. Ash
b. Ash
Mass
Train
Inlet"
0.03
0.3
1.9
1.1
3.0
22.8
22.4
47.0
1.3
0.5
0.6
only
Mass Train
Trans.
0.02
0.2
1.0
1.4
30.6
13.9
11.9
26.0
0.6
0.1
13.6

_Ei£_
0.02
0.2
0.9
1.5
40.2
7.7
10.1
20.9
0.6
0.1
16.2

DID'
0.01
0.2
0.8
1.5
39.2
6.6
9.0
17.9
0.5
0.2
21.0

Outlet1*
0.01
0.1
0.7
1.3
38.3
6.4
8.5
17.3
0.4
0.2
24.7

plus sorbent
COAL COMPOSITIONS  FROM  BASELINE AND E-SOX TEST SERIES
                  ANALYSES BY CTSE
                                                                                     Table 10

                                                                        MODELING RESULTS  -  BURGER PILOT ESP
 H.O
 Cl
 s
 Ash

 Volatile
 Fix. C
 Btu/lb
Baseline

  8.25
 64.83
  4.25
  1.24
  0.00
  2.64
 12.18

 33.50
 46.08
 11627
E-SOX Condition

     6.67
    67.89
     4.32
     1.33
     0.06
     2.67
    11.02

    34.78
    47.54
    12120
Mass Loading
Baseline 0.008
E-SOX (anticipated)
E-SOX (180-F) 0.113
E-SOX (170'F) 0.313
E-SOX (160'F) 1.55
Efficiency
99.92

98.56
96.02
80.27
Mass Loading
1 b/MHBtu
0.009
0.004
0.100
0.274
1.8
Efficiency Modeling
99.91 0.10/0.25
99.94
98.7
96.5
76.9
0.10/0.25
0.40/0.36
0.55/0.36
0.85/0.36
                                                                      "Combined sneakage & reentralnment/ff,
                                                               14-14

-------
 IDENTIFICATION OF LOW-RESISTIVITY REENTRAINMENT IN ESPS OPERATING IN DRY
                            SCRUBBING APPLICATIONS


                                Michael D. Durham
                              ADA Technologies, Inc.
                          304 Inverness Way So. Suite 110
                               Englewood, CO 80112
                                 Richard G. Rhudy
                          Electric Power Research Institute
                                Palo Alto, CA 94303
              Thomas A. Burnett, Jose DeGuzman, and Gerald A. Hollinden
                             Tennessee Valley Authority
                              Chattanooga, TN 37401
                      Robert A. Barton and Charles W. Dawson
                                  Ontario Hydro
                         Toronto, Ontario, Canada M8Z 5S4
ABSTRACT

The evaluation of the performance of ESPs operating downstream of spray dryers on highl-
and medium-sulfur coal is described.  Tests were conducted at the TVA 10 MW Spray
Dryer/ESP  Pilot  Plant  and the EPRI High Sulfur Test Center.   Analyses of particle
characteristics, spray dryer operating parameters, and ESP operating variables were used to
identify the occurrence of unexpectedly high particle reentrainment due to the low resistivity
(108 ohm-cm and lower) of the sorbent/flyash mixtures.  This reentrainment has a significant
impact on collection  efficiency and  these results should also be relevant to other  dry
scrubbing processes  in a humidified environment.  A  theoretical analysis and laboratory
experiments on reentrainment are described. Data are presented which demonstrate that
the chloride content of the coal is an  important parameter effecting the performance of the
ESP.   Implications of these results  relative to ESP upgrades are presented.  Special
techniques for measuring particle resistivity at spray dryer conditions are described.
                                       15-1

-------
                                  INTRODUCTION
Dry scrubbing processes, such as spray drying, duct sorbent injection, and furnace sorbent
injection with humidification, offer potential cost-effective  retrofit  flue gas  desulfurization
(FGD) technologies for existing  coal-fired  electric generating stations faced  with proposed
acid rain legislation. However, the favorable economics for these processes require that the
existing particulate  control  equipment  be capable of collecting both  the  flyash and the
injected sorbent.   Since most  of the  plants that would  be  impacted  by the proposed
legislation have electrostatic precipitators  (ESPs),  the performance of ESPs on humidified
sorbent/flyash mixtures is critical to the feasibility of dry scrubbing.


The performance of ESPs downstream of  spray dryers in high- and medium-sulfur coal flue
gas streams is currently  being investigated at both the Tennessee Valley Authority (TVA)  10
MW Spray Dryer/ESP Pilot  Plant at the Shawnee Steam Plant (TVA, 1988) and the Electric
Power Research Institute (EPRI) High Sulfur Test Center  (HSTC) at the New York State
Electric and Gas Company's Somerset station (Blythe et al., 1988).  The TVA program is co-
sponsored by  TVA, EPRI, and Ontario Hydro,  while the HSTC  is  supported by EPRI, the
Empire State Electric Energy Research Corporation, New York State Energy Research and
Development Authority, Consolidation Coal Company, and the U.S. Department of Energy.


There is  a need to define  the operating  characteristics of spray dryers and  downstream
particulate control equipment especially ESPs operating on flue gas from medium- and high-
sulfur gas streams (> 2.5% S).  Most of the available operating  experience on spray dryer
systems is from existing  full-scale systems operating on low-sulfur flue gas. In addition, only
three  of the fifteen operating spray dryer  units operating in the U.S.  have ESPs, and all of
these ESPs have very large specific collection areas,  650 to 725 ft2/kacfm  (Bradburn and
Mauritzson, 1988;  Donnelly and Quach, 1988; Doyle  et al.,  1986).  The potential medium-
and high-sulfur applications will involve retrofits to older ESPs with SCA more in the range of
180to450ft2/kacfm.


This paper evaluates the particulate control performance of ESPs operating downstream of
spray dryers for medium- and high-sulfur  coal applications. The measurement  and analysis
of particle characteristics (loading, size distribution, electrical resistivity, and  chloride level),
spray  dryer operating  parameters (inlet temperature,  reagent ratio,  and approach-to-
saturation temperature)  and  ESP operating variables (electrical  characteristics,  specific
collection area, gas velocity, and rapping frequency and intensity) is described.  The results
of  this  analysis  provides  a  strong indication   of  the  occurrence  of  severe   particle
reentrainment due  to the low resistivity of the sorbent/flyash mixtures at low approach-to-
saturation temperatures.  The  reentrainment has a  significant impact  on the collection
efficiency of  ESPs which  could represent  a fundamental limitation  on  their  ability  to
adequately perform in this environment. Although,  this program has been focused on spray
dryer  applications,  because  of the  similarities  of  the  gas  and  particle  characteristics
produced from spray drying  and  other  dry scrubbing processes, the  results also have

                                         15-2

-------
implications to duct slurry injection, dry sorbent injection with humidification, and processes
involving furnace sorbent injection with downstream humidification  such as the Tampella
LIFAC process.

                       IMPACT OF DRY SCRUBBING ON ESPS
The application of a humidified dry scrubbing system upstream of an ESP will provide a
multitude of changes that will impact the operation and performance of the ESP  Dealing
with these impacts could be the most serious problem facing the application of dry sorbent
injection technologies.  The primary effects of these processes on a downstream ESP are
due to the addition of large amounts of water and reacted and unreacted reagent to the flue
gas which results in changes to the mass loading, particle resistivity, particle size distribution,
and gas temperature.


EFFECT OF INCREASED MASS LOADING


The injection of sorbent into the gas stream will increase the particulate loading by 5 to 15
gr/acf, which could increase the loading to the particulate control device by a factor of from
3 to 12. The amount  of the increase will depend on the amount of sorbent required to
achieve the  desired level of SO2 removal and the amount of recycle.  The fine particles of
sorbent material become suspended in the flue gas and must be removed by the ESP, along
with the fly ash, in order to meet the particulate emission requirements.


This increase in inlet loading will result in an increase in emissions at the ESP outlet, even if
the collection  efficiency remains unaffected by any changes in the flue  gas.  Therefore,
neglecting any other factors that may be detrimental to the performance of the ESP, the ESP
must perform at a higher efficiency with  dry scrubbing  in order to maintain  the  same
emission levels.


EFFECT OF PARTICLE SIZE DISTRIBUTION


The particle  size of the sorbent is important because the ESP collection efficiency is related
to particle size. The size distribution produced  by a  spray dryer is quite large.  Figure 1
shows a comparison  of the inlet particle size distributions for baseline  and  spray dryer
conditions obtained at the 10 MW TVA test facility.  The data are plotted as  a differential
distribution such that the area under each curve is proportional to the mass associated with
each size  fraction.  As can be seen the spray dryer material is primarily  associated with
particles greater than 5 um physical diameter.  This material should be much easier to collect
than the flyash and should not lead to an increase in corona quenching.


EFFECT ON ELECTRICAL OPERATION


The humidification of the flue gas stream that occurs as part of the dry scrubbing process
can provide  an improvement in electrical characteristics of the ESP  Since the spark-over
voltage, which sets the maximum ESP operating voltage, is proportional to the  gas density,
the decreased temperature can result in an increase in operating voltages.  The effects of a
spray dryer  on electrical operation of an  ESP  are demonstrated by the voltage-current

                                       15-3

-------
characteristics plotted in Figure 2 (Durham et al., 1988a). The characteristics for the first and
fourth fields are shown for both fly ash only and spray dryer conditions. A comparison of the
first fields shows the effects of both the increased mass loading and the lower temperature.
Since space  charge effects due to particle concentrations were not found to penetrate
beyond the first field, the changes in the electrical characteristics in the final section would be
solely due to changes in the gas temperature and moisture.  These curves demonstrate that
improved  electrical  operation  was  achieved  in  both the  upstream  section  and  the
downstream section.  The electric field strength increased by almost 30% in electrical section
1 and 20% in section 4.  It should be noted that if the operating points were not limited by the
200 mA power supply, the improvement might have been even greater.  Although the mass
loading at the ESP inlet increased by nearly a factor of 10, there was limited quenching of the
corona current  because most of the increased mass is in the larger size  intervals.  Similar
improvements in electrical conditions were measured at the EPRI Arapahoe test facility under
both humidified dry-injection conditions and slurry injection of sorbent (Blythe et al, 1986).


RESISTIVITY OF SORBENT FLYASH MIXTURES


Another important parameter that determines the performance of an ESP is the electrical
resistivity of the  collected  particles.   The  resistivity determines  the rate at which ionic
conduction occurs in the dust layer. The range for optimum performance of an ESP is 10=9
to 1010 ohm-cm.  When the resistivity is  below 109,  the  electrostatic force holding  the
particles onto the collector plates is reduced and the particles are easily reentrained  (Katz,
1981). In a laboratory experiment, Spencer (1976)  found that at  low dust resistivity levels,
the corona wind alone was sufficient to reentrain the collected dust.


Dry scrubbing  processes require that the gas  stream  be humidified to low approach-to-
saturation temperatures in order to achieve high SO2 removal efficiencies.  Humidification
also has an effect on  the  particle resistivity.  Although the injected material has a high
resistivity (1013 to 1014 ohm-cm)  at  300  op, the  surface  conditioning  provided by  the
increased moisture and flue gas cooling is sufficient to reduce the resistivity to levels much
below the optimum level of 109-1010 ohm-cm.  Operating at these low-temperature, high-
humidity  conditions is significantly different  from all other ESP applications  and will have
implications relative to the operation of the ESP and the measurement of particle resistivity.


Table 1 provides a summary of laboratory and in-situ measurements of particle resistivity for
dry scrubbing processes.   The in-situ  resistivity measurements shown in this table were
made using the in-situ point-plane probe developed by the Southern Research Institute (SRI)
(Smith et al., 1977). Two interesting observations can be made from the data in Table 1. A
comparison of the laboratory resistivity measurements demonstrates that at gas conditions
representative of 20 to 40 °F approach-to-saturation temperatures (i.e. 145-165 °F with 15%
moisture), the  resistivity values of  the  humidified sorbent material are similar for all  dry
scrubbing processes. This indicates that the particle resistivity is primarily a function of the
particle and gas stream conditions and less a function of how or where  the sorbent was
injected and humidified.


The second point to be  noted in Table 1 is that there is an obvious discrepancy between the
measurements  made in the laboratory  and  those made using the in-situ probe.  In most
cases, the  in-situ  measurements are nearly  two orders of magnitude greater than  the
laboratory measurements.   This discrepancy  between laboratory  and  in-situ resistivity


                                         15-4

-------
measurements is most likely due to reentrainment from the point-plane probe.  When the
particle resistivity is very low, an ESP experiences excessive amounts of reentrainment even
when the gas velocity is only 5 ft/s.  Since reentrainment  increases at higher velocities,
significantly  greater reentrainment  would  be  expected while attempting  to  precipitate a
sample using the in-situ probe in a duct with a gas velocity on the order of 60 ft/s.   The
higher velocity also increases  the potential for biasing the particle size distribution of the
sample either aerodynamically, by distorting the particle size penetrating the shroud of the
probe, or electrostatically, because larger particles are more likely to precipitate.  For  very
low-resistivity dusts, it is not possible to precipitate a representative paniculate sample at
high velocity. The difference between the laboratory and in-situ measurements is most likely
explained by the selective reentrainment from the in-situ probe of the very  low-resistivity
material such that the remaining sample contains only higher resistivity particles.

                                      Table 1.

        PARTICLE RESISTIVITY MEASURED IN DRY SCRUBBING APPLICATIONS
   Program
Temp
 (oF)
 Laboratory
 (ohms-cm)
   In-Situ
 (ohms-cm)
Ref
   HALT @ Toronto
   TVA10MW
   LIMB
   ESOX
   EPRI @ HSTC
   Joy Riverside
   TVA 1 MW
   Flakt GRDA
   G.E. IDS
 145

 145

 160

 160

 145

 163

 155

 170

 190
   108-109

    1x108
     108

     1Q8

   107-1Q8

   107-109


   107-108

6x109-6x1011
  1010-1011

   1x1010
  1010.1Q11


     109


9x109-3x1011
 a
 b
 c
 d
 e
 f

 g
 h
 i
a. Goochetal., (1988)
b. Durham etal., (1988b)
c. Faulkner et al, (1988a)
d. Young etal., (1988)
e. Blytheetal., (1988)
                f.    Chen etal., (1984)
                g.    Research Cottrell and TVA (1986)
                h.    Bradburn and Mauritzson (1988)
                i.    Drummond (1988)
Measurement of Sorbent Resistivity Using an Extractive System
A series of resistivity measurements were made at the TVA Spray Dryer/ESP 10 MW Pilot
Plant using an extractive resistivity apparatus to demonstrate that this discrepancy between
laboratory  and in-situ  resistivity  measurements is  due  to problems  in  precipitating a
representative sample. The extractive resistivity apparatus incorporates a modification to the
conventional  system  such  that the  point-plane  sample  precipitation  and  resistivity
                                        15-5

-------
measurement are performed external to the duct. A probe is used to isokinetically extract a
sample of dust from the gas  stream  to a temperature controlled chamber where a point-
plane precipitator deposits the dust onto a disc.   The gas velocity in  the  chamber is
maintained at a level similar to that in an actual ESP ( 3 ft/s)  and therefore, the potential for
reentrainment is reduced.  The thickness of the dust layer is measured by a micrometer
external to the precipitation chamber.  Unlike the SRI probe  in which the  measurement is
made blind to the operator, the extractive device has a window to observe the lowering of the
upper disc onto the dust layer.  The chamber can be isolated from the gas flow to prevent
disturbing the dust layer.  With the extractive system the flue gas temperature during the
resistivity measurement can be varied independently of the duct temperature. Therefore, the
resistivity data is not limited to a single temperature, as is the case with the SRI probe, but
can  be  measured at a range of  temperatures without having  to change  the process
conditions.


Normally, after the sample  is precipitated and the thickness is  measured,  increasing voltage
is then applied  to the dust layer and the  resulting current  is  recorded until the dust  layer
breaks down electrically  and  sparkover  occurs (ASME,  1965).   However,  for the low-
resistivity material it was necessary to modify this procedure because the combination of a
high sparkover voltage and resulting high current density  is sufficient to  cause localized
heating of the sample which changes its resistivity.  For example,  at  108 ohm-cm, a voltage
drop of 1.6  kV across a 1 mm thick  dust layer would  produce a  current density of 185
uA/cm2  This produces 0.5 W/cm2 of energy which must  be dissipated in the dust layer,
and  results in an increase in  the sample  temperature of approximately 2 °F per second.
Since the resistivity doubles every 6 °F, this is obviously a significant concern.


To overcome this  problem, the high voltage was only  applied long  enough to read the
voltage and current at a single point and quickly turned off.  In order to select a voltage to
make the measurements, the resistivity of a sample of spray  dryer material was measured at
voltages representative of field strengths of 2, 4,  and 8 kV/cm. The results indicated that for
samples  with low levels of resistivity,  if the power was only momentarily energized, the
measured resistivity was independent of field strength. Therefore,  a voltage representative
of an average electric field of 4 kV/cm was selected for testing. This value is approximately
equal to the level experienced in an operating ESP but significantly  below  the electric field
necessary for breakdown.  This procedure  prevented  electrical heating  of sample and
provided a technique which produced consistent results.   The heating  of the dust layer,
which is proportional to the square of the current density, would not be  expected on an ESP
collector  plate because the current density in an operating  ESP  is three or four orders of
magnitude lower than that obtained during the resistivity measurement.


Figure 3  shows a  comparison of the   results from  the extractive device  along  with
measurements made with the SRI probe and laboratory measurements. The measurements
using the SRI probe were made with the probe shroud attached. The measurements made
with the extractive system and the laboratory measurements  show excellent agreement over
a range  of temperatures.   At temperatures characteristic of 20  to  40 oF approach-to-
saturation temperatures, the resistivity values are in the range of 107 ohm-cm.  For the same
range, the measurements made using the in-situ probe are two orders of magnitude greater.
This  data indicates that meaningful in-situ  resistivity measurements can be made using the
procedure described.
                                        15-6

-------
 ESP PERFORMANCE AT HIGH-SULFUR SPRAY DRYER CONDITIONS


 The performance of an ESP at spray dryer conditions was measured at both the TVA Spray
 Dryer/ESP Pilot Plant and the EPRI High Sulfur Test Center. Details on the two pilot units are
 presented in Table 2.

                                     Table 2.

   COMPARISON OF DESIGN CHARACTERISTICS OF THE TVA AND HSTC PILOT ESPS
            ESP Parameter               TVA             HSTC


              Pilot Size                 10MW            4 MW
       Nominal Flow Rate (acfm)          35,000            13,000
           Plate Height (ft.)                23                12
           Plate Length (ft.)               9.2                12
          Plate Spacing (in.)               10                12
         Gas Flow Passages              8                 5
       Plate Area per Section (ft2)          3,382             1,440
         Number of Sections              4                 5
         Total Plate Area (ft2)            13,528            7,200
           Power Supplies          55 kV; 200 mA     65 kV; 200 mA
              Rappers           Tumbling Hammers  Electromagnetic
          Corona Wire Dia.              0.1 in.            0.19 in.
The test program at TVA involved two series of tests which were run while a 4% sulfur coal
was being burned.  Each test series was set up  to evaluate the effect of specific collection
area on ESP performance at baseline and spray dryer conditions.  The effect of SCA was
determined by measuring the  collection efficiency at  a constant gas flow rate  but with a
different number of ESP sections energized. The effect of velocity was determined  by
increasing the gas flow rate and energizing an additional ESP section such that the SCA was
constant for both velocities.


At baseline conditions the ESP was operated at a constant temperature of 320 OF for the
entire series.  During the spray dryer test program, the spray dryer  was operated under
constant conditions. The inlet temperature to the  spray dryer was maintained at 320 oF The
reagent ratio was set at a value of 1.3 moles  of calcium in the fresh lime slurry per mole of
SO2 in the flue gas.  Maximum recycle was used to control the solids content of the slurry to
the wheel. The slurry feed rate was set to provide an approach-to-saturation temperature of
20 oF which corresponds to an ESP inlet  temperature of 145 °F.


                                       15-7

-------
Table 3 shows a comparison of the ESP performance data as a function of SCA.  These data
represent an average of from five to seven sets of measurements at the inlet and outlet of the
ESP for each condition.   For each ESP configuration, the  measurements were tightly
grouped indicating consistent performance of the ESP.  For Example,  the seven efficiency
measurements made at the 4 field  ESP condition shown in Table 3 were 99.79, 99.83, 99.83,
99.84, 99.84, 99.85, and 99.87.

                                     Table 3.

      COMPARISON OF ESP PERFORMANCE FOR BASELINE AND SPRAY DRYER
       CONDITIONS AS A FUNCTION OF SCA AT THE TVA 10 MW PILOT PLANT
Test


Baseline
SD
Baseline
SD
Baseline
SD
Flow
Rate
acfm
34,614
32,832
34,961
32,623
35,124
31,856
Vel.
ft/s

3.8
3.6
3.8
3.6
3.8
3.5
Fields


4
4
3
3
2
2
SCA
ft2/kacfm

390
412
290
311
192
212
Inlet
Cone.
gr/acf
1.151
9.223
0.942
8.987
0.646
8.58
Outlet
Cone.
gr/acf
0.00198
0.0155
0.00240
0.0304
0.00942
0.219
Effic.
%

99.82
99.84
99.73
99.66
98.48
97.46
The spray dryer test conditions were designed to duplicate the baseline tests except that the
reduced gas flow rate due to the cooling of the gas resulted in lower velocities and higher
values of SCA.  As expected,  the  primary differences in the gas stream characteristics
between the baseline and spray dryer conditions is the increase in mass loading by a factor
of 9.  The  results show that relative to collection efficiency,  there was very little change
between baseline and spray dryer conditions for the 3 and 4 field ESP cases and a reduced
efficiency at spray dryer conditions for the 2 field case.  However, the increased inlet loading
resulted in a large increase in the outlet emissions for all cases.


A series of  tests was then run  at the High Sulfur Test Center.  Data comparing baseline
conditions with three sets of measurements at spray dryer conditions for the two sites is
presented in Table 4.  At the baseline conditions and for all three values of SCA at the spray
dryer  conditions, the  collection efficiencies  at the HSTC were higher than those  at TVA.
During the spray dryer tests, the SCA and electrical characteristics at HSTC were set to be
equivalent to the TVA test conditions.  However, the  TVA pilot unit extracts a flue gas
slipstream downstream  of a multiclone system which removes many of the larger easy to
collect particles. Therefore, the higher efficiency at the HSTC appears to be largely due to a
difference in particle size distribution  at the two sites. In  addition, the shorter plates at HSTC
may lead to increased collection and a reduced  amount of reentrainment because of the
more favorable aspect ratio.
                                        15-8

-------
                                      Table 4.

     COMPARISON OF ESP PERFORMANCE AT THE TVA AND HSTC PILOT PLANTS
Test
COND.
HSTC Baseline
TVA Baseline
HSTC SD
TVASD
HSTC SD
TVASD
HSTC SD
TVASD
HSTC Baseline
HSTC SD
Fields
4
4
4
4
3
3
2
2
4
4
SCA
ft2/kacfm
348
390
409
412
312
311
211
212
348
409
Vel.
ft/s
4.6
3.8
3.9
3.6
3.85
3.6
3.8
3.5
4.6
3.9
Inlet
Cone.
gr/acf
1.19
1.15
5.70
9.2
6.34
8.99
8.77
8.58
1.19
5.70
Outlet
Cone.
gr/acf
0.0009
0.002
0.0045
0.015
0.015
0.03
0.074
0.219
0.0009
0.0045
Effic.
%
99.92
99.82
99.92
99.84
99.76
99.66
99.16
97.46
99.92
99.92
In spite of differences in the values of collection efficiency at the two pilot plants, the relative
effect of the spray dryer conditions on the performance of the ESP was found to be the same
at both sites.  As can be seen by the last grouping of data in Table 4, the collection efficiency
measured at spray dryer conditions was identical to that obtained at baseline conditions.


The fact that the collection efficiency did not improve at spray dryer conditions at either TVA
or HSTC is a surprising phenomenon. The increased loading was primarily in the larger size
ranges, and therefore, should have been easier to collect. At spray dryer conditions the ESP
operated at field strengths 20 to 30% higher  than the base case. In addition the flue gas
viscosity was reduced 20% due to the reduced temperature.  The combination  of these
factors should have provided a significant increase in overall ESP collection efficiency.


Particle size distributions from the TVA tests were analyzed to determine why the efficiency
did not improve at the spray dryer conditions.  Figure 4 is a plot of the outlet particle size
distribution for both baseline and spray dryer conditions.  Most of the increased emissions at
spray dryer conditions occur in a size range between 2 and 10 micrometers. Penetration in
this  size  range  is  usually associated  with  non-ideal  effects  such  as sneakage  or
reentrainment. Since, the configuration of the ESP was  identical under baseline and spray
dryer cases, and the velocity  was slightly lower in the spray dryer case, it is unlikely that
sneakage  would be  increased at spray dryer  conditions. Therefore, the problem was most
likely some characteristic of the sorbent/flyash mixture that made it more easily re-entrained.

                                        15-9

-------
Another interesting phenomenon was detected from the particle size measurements.  From
Figure 4 it can be seen that at spray dryer conditions there is a significant amount of the
outlet emissions associated with 1-5 micrometer particles. The inlet size distribution in Figure
1 shows that the spray dryer material is primarily associated with particles greater than 5 um.
Therefore, one would expect that these 1-5 um particles, which represent over 50% of the
outlet emissions, would be flyash particles. However, analysis of the inlet and outlet samples
by scanning electron microscopy determined that the ratios of flyash to sorbent in both the
inlet and outlet were approximately the same.  This implies that the sorbent material is
breaking up in the ESP and generating smaller particles.


It should be noted that the tests run at both TVA and the HSTC were for ESP velocities of 4
to 5 ft/s which are typical of the newer and larger SCA ESPs. However, many of the plants
that will be faced with  retrofit SCL control have older ESPs which not only have smaller
collection areas but were designecffor gas velocities of 6 to 7 ft/s.  Since gas velocity has a
strong  impact  on reentrainment,  these older  ESPs  may experience  even  greater
reentrainment at dry scrubbing conditions.


Modeling of ESP Performance


Version III of the SRI computer ESP model (Faulkner and DuBard, 1984) was used to predict
the performance  of the  TVA pilot ESP.  Data  on the  design of the ESP, flue  gas
characteristics, ESP electrical operating  conditions,  and measured  particle size distributions
were  input  into the model.  The  model was then  run for  several experimental conditions
which reflected changes in specific collection area  and velocity. At each condition several
values for the non-ideal parameters of sneakage and gas flow  distribution were evaluated.
The results  of the model predictions were then  compared to the  measured  results  to
determine the optimal values for parameters that  control the non-ideal effects.  Based upon
the results of this comparison, values of 5% for sneakage and 15% for the standard deviation
of the gas flow distribution were selected. These are also the values recommended by the
EPRI  study (DuBard and  Dahlin,  1987).  Using these values for the  non-ideal effects,  the
model predictions compared favorably with the 4 field ESP  performance.  For the two test
conditions, the model predicted 99.92% vs. 99.82%  measured at one velocity and predicted
99.62% vs. 99.69% measured at a higher velocity.


The model was then used to predict the performance of the TVA pilot ESP under spray dryer
conditions.    The  measured operating conditions were input and  the  model was  run
predicting an efficiency of 99.998% (0.002% penetration).   The model predicted that  the
improvements in  the ESP operating  conditions  were more than  sufficient to  handle  the
increased loading produced by the spray dryer. However, the measured penetration for this
test was 0.16%, which means that the model prediction was off by  a factor of 80. This was
interpreted  not as  a  deficiency  in  the model  but  a statement  of  the  severity  of  the
reentrainment problem.


The ESP model was then set up to simulate a  condition of continuous reentrainment to try
and develop a quantitative feel for the magnitude of the reentrainment. The model was  run
with various levels of reentrainment until the predicted performance matched the measured
performance.  It was determined that a reentrainment factor of 60% was required for  the
model to correctly predict the ESP efficiency at spray dryer  conditions. This means that at
any point in the ESP, 60%  of the particles collected at that point are reentrained back into the


                                       15-10

-------
 gas stream where they begin to be recharged and recollected later downstream. Although
 the model handles reentrainment in  a  somewhat simplistic manner, this does provide an
 indication of how reentrainment dominates the performance of the ESP.


 RAPPING CHARACTERISTICS OF SORBENT/FLYASH PARTICLES


 Reentrainment could be due to either particles dislodged during rapping or the scouring of
 the particles from the collector plates by the gas flow. Tests conducted by EPRI on full scale
 boilers (Gooch and Merchant,  1978) have shown that rapping puffs have a mass  median
 diameter of 6 micrometers, which is similar to the outlet size distribution shown in Figure 4.
 Therefore, the increased emissions could be due to improper rapping.


 Effect of Rapping Schedule


 A test program was conducted at the TVA Pilot Plant while operating with medium sulfur coal
 to determine  if it was possible to reduce emissions from the ESP by extending the period
 between  raps (Tennessee Valley Authority,  1990a).  The extended rapping schedule was
 determined by calculating the time required in each section to collect an average dust layer
 of 1/16th of an inch before rapping.  The rapping schedules for both baseline and extended
 rapping are presented in Table 5 which shows that the frequency of the extended  rapping
 schedule was increased by a factor of six to eight over the baseline case.

                                     Table 5.

                              RAPPING SCHEDULES
Rapping
Program
BASELINE
EXTENDED CYCLE
Section
1
5 min.
27 min.
Section
2
25 min.
180 min.
Section
3
2hrs.
18hrs.
Section
4
10hrs.
72 hrs.
The collection efficiency of the ESP was first measured at spray dryer conditions with the
baseline rapping schedule.  The average  efficiency under these  conditions was 99.87%.
Following  the baseline tests,  all sections  of the  ESP were  repeatedly rapped.  Mass
measurements were  then  made  to determine pseudo-clean plate  performance.   It was
believed that this case was representative of the no-rap ESP performance and therefore, the
maximum efficiency that could be obtained by optimizing the rapping. These tests averaged
99.90% removal. The ESP was  then operated for 72 hours (i.e. the length of the longest
rapping  cycle)  before the next series of measurements were made.  With the extended
rapping schedule the efficiency was 99.89%.


The performance of the ESP for the three different rapping schedules is summarized in Table
6.   It is concluded from this data that rapping  schedule has a minimal  effect on the
performance of the  ESP.  It is therefore unlikely that the ESP performance can be improved
by optimizing the rapping frequency.

                                      15-11

-------
                                     Table 6.

             EFFECT OF RAPPING FREQUENCY ON ESP PERFORMANCE
Test Descriptiom


Baseline Rapping
Clean Plate
Extended Rapping
Inlet
Cone.
gr/acf
8.5
8.4
7.0
Outlet
Cone.
gr/acf
0.011
0.008
0.007
Effic.
%

99.87
99.90
99.89
Effect of Rapping Intensity


A test program was conducted at the HSTC operating at high sulfur conditions to evaluate
the effect of rapping intensity on ESP performance at spray dryer conditions. Tests were run
with only two fields of the ESP energized.  This provided for an arrangement where the
effects of changes to the rapping system  could be determined in the minimum amount of
time.


The rapping intensity was first set at the default values prescribed by the equipment vendor
of 10  ft-lbs.  At these conditions there were opacity spikes associated with rapping the first
section plates that exceeded 60%. The rapping intensity was then reduced to 3 ft-lbs and the
rapping spikes were greatly reduced with the largest spikes on the order of 30%. The ESP
was allowed to operate with this setting overnight. One of the potential detrimental effects of
reducing the rapping intensity is that the  electrodes  don't get cleaned  sufficiently and the
electrical characteristics deteriorate.  However, after approximately 20 hours of operating in
this manner, there was no change in the electrical characteristics.  The rapping intensity was
then further reduced to 2 ft-lbs which further reduced the rapping spikes to less than 20%.


A set of mass emission  measurements  were made to determine if the  optimization  of the
rapping system could be quantified with reduced outlet loadings.  Comparing  the results
before and after optimization of the rappers demonstrated that although the opacity spikes
were  reduced, the mass emissions were not  effected.  This indicates that  reentrainment
occurring during spray dryer  conditions is probably  not due to  rapping  emissions  but to
scouring.  Attempts have been made to reduce rapping reentrainment by optimizing both the
frequency and the intensity of rapping and have found little effect.  The fact that it is possible
to effectively clean the plates with a greatly reduced rapping intensity does indicate further
that the material is rather loosely held onto the plates.
                                        15-12

-------
NON-RAPPING REENTRAINMENT


Precipitated particles can be reentrained from the collector plates and remixed in the gas
stream. This can be the result of scouring by the mean gas flow and occurs when the forces
holding the material onto the plate are reduced. The collected dust is held onto the plate by
both electrical and mechanical forces.  The electrical forces  are due to the current flowing
through the dust to the collector plates.  The mechanical forces are due to surface cohesion
of the  particles.   Since the electrical force  is proportional to  the product of current density
and particle resistivity, this force is greatly diminished at low-resistivity conditions. In fact, it is
even possible for the electrical forces to reverse and  repel the collected particles (Penny,
1962, 1975).


A theoretical  examination of particle reentrainment was undertaken  in order to  evaluate the
roles that the electric field, current density, and particle resistivity play in the  electrostatic
aspects  of reentrainment.  However, the extent of reentrainment depends  upon  many
factors, such as particle cohesion and gas  stream turbulence, in addition to the  electrostatic
forces.   Therefore,  conditions  are discussed  in  terms  of  whether  the  electrostatic
reentrainment forces are increased or decreased.


Figure 6a shows the boundary conditions for a layer of particles. An electric field, E   exists
at the surface of the layer due to the corona wire voltage which also creates the ion density,
N|, at the surface.  The resulting current  density, J   is in the positive x direction although the
negative ions move toward the plate.

The ion current density is

             Jg  = Nieb,Eg        (A/m2)                                             (1)

where  N| is the ion density, e is the charge per ion and bj is the ion mobility.


The particle layer is assumed to be homogeneous and at  steady  state conditions.  It is
realized that large fields and forces  exist at the particle to particle contact points.  However,
several useful relationships  can be developed  using the assumption  of homogeneity and
current continuity. Therefore,

             Jg  = J, = Jp        (A/nr.2)                                             (2)

where  J7 is the current density through the layer and J  is the  current density to the collector
plate which is the measured current divided by the plate area. The electric field within the
layer, E^ is related to the current density

             E/ = J7 e{           (V/m)                                               (3)

where  P/ is the layer resistivity in  ohm-m.  Figure  6b shows the  step  change in electric
displacement at the surface of the layer.  A surface charge density, a ,  exists on the surface.
where eQ is the dielectric constant of free space and e/ is the relative dielectric constant of
the layer.  Equation (4) shows the surface charge density can be either positive or negative

                                         15-13

-------
depending upon the electrical and layer conditions.

An electric force per unit area fx acts on the surface charge

            fx = a(Eg+E,)/2                (N/m2)                                (5)


The force fx tends to pull the particles off the surface of the layer back into the gas if a is
positive.  If a is negative, the force tends to hold the layer against the plate.


Figure 7 is a plot  of the calculated electrostatic force on the  dust layer as a function of
resistivity  and  the  electric field strength  at the plate.  The  calculations were made for a
constant current density of 60 nA/cm2, which was equivalent to the average current density
at the TVA pilot ESP   This family  of  curves demonstrates the  general trends of the
relationships defined by Equations 1-5.


When the resistivity is greater than 1010 ohm-cm, the electrostatic force rapidly increases in
the negative x  direction toward the plate (i.e.  increased clamping force) as the resistivity
increases. At these resistivity levels, the force is independent of the electric field at the plate
and only a function of the product of the current density and  particle resistivity.  For high-
resistivity  applications (i.e.  > 1011 ohm-cm) the holding force  is so great that "power off"
rapping is often required to remove the dust from the plates.


At particle resistivity levels below 6 x 109  ohm-cm, the electrostatic forces reverse and tend
to pull the particles off the plates.  As can be  seen from Figure 7, for very low-resistivity
levels, the repulsion force becomes relatively independent of resistivity, and correspondingly
current density, and only proportional to the square of the electric field strength.  Therefore,
although  high electric  fields  increase the primary precipitation rate of particles,  for low-
resistivity  applications the high electric fields also increase the reentrainment of the collected
particles.


A series  of experiments were run in the  laboratory  to determine if visual evidence of
reentrainment could be detected using the ADA Field  Resistivity Device.  Samples  of spray
dryer/flyash recycle material were placed  on  bottom disc such that the thickness was
approximately  1/16th inch.   The chamber was conditioned with a flow of 145 °F air which
was  humidified  to 14% H2O by  volume to simulate  spray  dryer  gas  stream conditions.
Resistivity measurements were made to verify the conditions and levels on the order of 108
ohm-cm were measured.


The instrument was then operated in the  precipitation mode  with comparable voltages and
currents to the pilot ESP  A video  camera was used to observe and document the impact on
the dust.  The video tape showed that indeed the dust reversed its charge and was repelled
by the plate. The repulsion force was high enough to overcome gravity as the discs were set
up in  a horizontal  orientation.  During these tests it was possible to change the electrical
characteristics of the discharge electrode  by lowering the top disc over the corona electrode.
The results qualitatively indicated that the magnitude of the particle repulsion, as evidenced
by the reentrainment of dust from the lower disc, was somewhat independent of current
density, but a very strong function of the applied electric field.   This effect is predicted by
Equation 5.
                                         15-14

-------
 Equations 4 and 5 indicate that the reentrainment of the collected particles is proportional to
 the square of the electric field and  therefore reduced reentrainment should result from a
 decrease in operating voltage.  However, the decreased voltage would be detrimental to the
 primary particle collection mechanisms which are also proportional to square of the electric
 field.


 In order to investigate the effect of  reducing voltage on ESP performance at  spray  dryer
 conditions, tests  were run at the TVA pilot plant at two different voltage levels.   ESP
 performance was measured  with all four sections operating  at the limits  of the power
 supplies of 200 mA per section for the first test and at a 50 mA limit on  all sections for the
 second test.


 The computer ESP model was run to determine the effect that this change would have on the
 ESP performance.   The model  predicted  that if reentrainment were not effected,  then
 theoretically the decrease in electrical operating conditions associated with  a  50 mA limit
 should produce a factor of 10 increase in penetration.  However, the experimental results
 indicated that  the  reduction  in  electrical  operating  conditions  had  no effect on  the
 performance of the ESP as both the efficiency and the opacity remained  relatively constant.
 This can only be  explained if the  reduced operating voltage resulted in lower reentrainment
 from the plate.  The  model was then run  to determine how much the reentrainment would
 have to be decreased in order for the modelled results to equal the measured results.   The
 model predicted that the reentrainment decreased from 71% to 43% at the lower voltage.


 EFFECT OF CHLORIDE LEVEL ON ESP PERFORMANCE


 Laboratory and pilot-scale research has demonstrated  that the chloride content in  the
 sorbent effects the absorption rate of SO2 (Brown et al., 1988;  Karlsson et al.,  1983). This is
 due to the formation of calcium chloride,  a deliquescent salt which increases the affinity of
 the sorbent to hold moisture. The increased hydrophilic nature of the sorbent improves SO2
 absorption by extending the  time for drying of the  droplet  and this increases the time
 available for reaction during the reactive wet phase.


 In addition to enhancing SO2 removal, the chloride content has been found  to also be an
 important parameter effecting the performance of the ESP  (Tennessee Valley  Authority,
 1990b). As the chloride level increases, the  moisture content of the sorbent collected in the
 ESP also increases.  The increased moisture increases the surface conductivity of  the
 collected particles which lowers the particle resistivity.


 To investigate the effect of chloride content on the resistivity of spray dryer material, the ADA
 resistivity instrument was used to make  in-situ  resistivity measurements at  the  TVA pilot
 plant.  The first set  of measurements  were made during the  spray dryer testing with a
 medium-sulfur (2.8-3% S) low-chloride (0.1% Cl) coal.  The extractive instrument  provides
 independent control  of the temperature of  the resistivity cell so it  was  possible to  make
 measurements  as a function of temperature  without  having to change the process
 conditions. Particle resistivity is plotted in  Figure 8  as a function of temperature. Each data
 point  represents  the precipitation and  measurement of a  new sample at  a constant
temperature.
                                       15-15

-------
A second series of tests were made while calcium chloride was added to the recycle slurry
feed to spike the chloride level to an equivalent coal level of 0.3%. This data is plotted in
Figure 8 along with the measurements made at the unspiked conditions. As can be seen,
the addition  of the chloride reduces the resistivity by an order of magnitude or more at the
low approach temperatures.  Also plotted in Figure 8 are the extractive  in-situ measurements
made earlier on the high-chloride (0.3% Cl) medium-sulfur (2.6-2.7% S) coal.  The resistivity
measurements are nearly identical for the both high-chloride conditions, and it appears that
the effect of the chloride is independent of whether the chloride is inherent in the coal or
added as calcium chloride in the recycle slurry.


In addition to its effect on resistivity, the increased moisture in the collected solids associated
with the chloride levels also  appears to increase the particle cohesion.  Particle cohesion
plays an important role in reentrainment.  If cohesion is high, the particles will tend to stick
together on  the collector plate making reentrainment more difficult.  Cohesion will also lead
to increased agglomeration of particles. This will improve the recollection  of particles and
increase their likelihood of falling  into the hopper. Therefore,  any mechanism capable of
increasing the cohesive characteristics of particles should improve the collection efficiency of
the ESP.


 1. High Chloride Levels


During the testing with medium-sulfur coal, chloride addition was found to  be effective for
improving the performance of the ESP. Table 7 demonstrates how ESP performance can be
improved by increasing the chloride level. For all of these tests, the flow characteristics and
ESP operating parameters were help approximately constant.  The only electrostatic effect of
the increase in chloride level  would be a decrease in particle resistivity  which would increase
reentrainment and decrease efficiency.  However, it can be seen  that  for these spray dryer
conditions, the efficiency improved at the higher chloride levels.  An increase  in particle
cohesion could help explain this phenomenon.

                                      Table 7.

      IMPROVED ESP COLLECTION EFFICIENCY AT ELEVATED CHLORIDE LEVELS.
Test


Approach
Temp
oF
SD
Inlet Temp
oF
Sorbent
Ratio

Coal
Chloride
(%)
Efficiency

(%)
Penet.

(%)
4-K-08
4-K-16
4-M-03
4-M-10
18
18
18
18
320
320
320
320
1.0
1.0
1.0
1.0
0.06
0.06
0.3
0.5
99.41
99.45
99.82
99.96
0.59
0.55
0.18
0.04
                                        15-16

-------
ESP UPGRADES FOR LOW-RESISTIVITY REENTRAINMENT PROBLEMS


Unfortunately there are a limited number of options for low-resistivity reentrainment because
most upgrades are designed for solving problems related to high resistivity. The resistivity is
already low so that sulfur trioxide conditioning would serve no purpose.  Since the pre-NSPS
ESPs are fabricated from  carbon steel, running the  ESP wet  would  have devastating
corrosion problems.  For low-resistivity applications, intermittent energization and increased
plate spacing would be primarily used to reduce  costs with limited ability for  improving
performance.


The conventional technique for increasing ESP performance is to increase the specific
collection area (SCA) of the ESP.  However, tests conducted at TVA have shown that only
marginal performance increases are  obtained  by going from a three field (300 ft ^acfm) to
a four field (400 ft j^kacfm) ESP.   Further increase in SCA would not only be costly and
require additional space that might not  be available in a retrofit situation,  it would probably
provide only diminishing returns.


In order  to directly treat the reentrainment problem, it is necessary to either increase the
particle cohesion, modify the resistivity, or decrease the electrostatic reentrainment forces.
The cohesive characteristics of the  dust  might  be  improved by using  additives such as
ammonia.  However, the fate of the ammonium compounds in the presence  of unreacted
calcium hydroxide is unknown.


Modification of the resistivity of the dry scrubbing material could be performed  by increasing
the temperature of the ESP.  Since  the  low approach-to-saturation temperatures are
necessary to obtain the required levels of SO2 removal,  increasing the temperature might be
unacceptable. However, some form of reheat system might be considered in which the gas
is heated downstream of the dry scrubbing system prior to treatment in the ESP.  Although
this would be detrimental to SO2 removal in the ESP, it might improve  the efficiency of the
ESP.


It might also  be possible to reduce reentrainment and realize some improvement in  ESP
performance by modifying the electrical characteristics.  Equations  4 and  5 indicate that for
low-resistivity dusts (< 109 ohm-cm),  reentrainment will be proportional  to the  square of the
electric field at the plate. The magnitude  of the  electric field at the plate is due to both the
applied electric field (i.e. the operating voltage divided by the wire-to-plate spacing) and the
electric field produced by the charged particles  and ions in  the interelectrode space.  The
contribution of the field due to  space charge will depend upon the particle loading and the
current density.  It might be possible to reduce this  electric field by modifying  the design of
the emitting electrode.


However, from a standpoint of fundamental particle collection, it is important to maximize the
operating voltage and current density as particle  collection is a function  of the  square of the
electric field strength.  Since the requirements for reduced  reentrainment conflict with the
requirements for particle collection,  it is  difficult to determine if an electrical modification
would produce an improvement in performance.
                                        15-17

-------
CONCLUSIONS


Pilot-scale test programs designed to evaluate the impact of medium and high-sulfur spray
drying  on  ESPs,  have  identified  the  occurrence  of  low-resistivity  related  particle
reentrainment which appears to provide a fundamental limitation to the performance of ESPs
in this application.  The data indicates that  for spray dryer conditions, it is possible to
maintain emission levels below the 0.1 Ib/MMBtu limits with SCAs on the order of 400-450
ft2/kacfm.  For older ESPs, which are characteristically smaller and designed for higher ESP
gas velocities, ESP enhancements will be required to provide acceptable performance.


In addition to spray  dryer applications, published  data indicate that other dry scrubbing
systems, such  as  duct sorbent injection and furnace  sorbent  injection with  downstream
humidification, also produce low-resistivity particles and therefore would also be expected to
experience severe reentrainment. Reentrainment in these applications has not been detailed
in the past because its presence would be disguised somewhat in test programs with very
large SCA ESPs, higher approach-to-saturation temperatures,  in pilot ESPs with small plate
heights, and for tests run at lower than typical ESP gas velocities.


Other specific conclusions about reentrainment include:

      e  In-situ resistivity measuring systems which require precipitating a sample at
             high velocities (30-60 ft/s) cannot obtain a representative sample of low-
             resistivity material.

      0  Comparing  the outlet emission levels under  baseline and dry scrubbing
             conditions, rather than collection efficiency, is a more adequate means
             to determine the performance of an ESP in dry scrubbing applications.

      •  Optimizing ESP rapping intensity and frequency for spray dryer applications
             reduces  the  opacity spikes but  provides  little improvement in  overall
             mass emission reduction.

      e  The chloride  content of  the coal,  and resulting  chloride content of the
             sorbent,  effect  both  the SO2 removal  in the system  and the particle
             resistivity and cohesion of the sorbent which subsequently effects ESP
             performance.

      0  There are limited ESP upgrade options for low-resistivity applications.  A
             combination of additives to increase particle cohesion and discharge
             electrodes designed  for   low-voltage   high-current operation  might
             provide at least a partial solution to the problem.


The problem  with re-entrainment will be  investigated during future testing at both the TVA
and the HSTC pilot plants.  Understanding this problem will provide  a major step toward a
solution that  would result in  a potentially cost effective  alternative for retrofit flue gas
desulfurization.
                                        15-18

-------
                                  REFERENCES

American Society of Mechanical Engineers.  "Determining the Properties of Fine Particulate
      Matter," ASME.  PTC 28. The American Society of Mechanical  Engineers, 345 East
      47th Street, New York, NY 10017, 1965.

Blythe, G., L Lepovitz and R. Rhudy. "Results from EPRI HSTC High Sulfur Spray Dryer Pilot
      Tests," Presented at EPA  and EPRI  First Combined  FGD and Dry SO2 Control
      Symposium. St.  Louis, Missouri, October 25 - 28, 1988.

Blythe, G., V.  Bland,  C. Martin, M.  McElroy and  R. Rhudy, "Pilot-Scale Studies of SO2
      Removal by the Addition of Calcium-Based  Sorbents  Upstream of a  Particulate
      Control Device," Presented at Tenth Symposium on Flue Gas Desulfurization, Atlanta,
      Georgia. November, 1986.

Bradburn, K. and C. Mauritzson.  "Full Scale Precipitator Experience Following a Lime Slurry
      Flue Gas Desulfurization System," Paper 5B3, presented at the EPA\EPRI Seventh
      Symposium on Transfer and Utilization of Particulate Control Technology. Nashville,
      Tennessee, March 22-25, 1988.

Brown, C.A.,  G.M. Blythe,  LR. Humphries, R.F. Robards,  R.A. Runyan and R.G. Rhudy.
      "Results  from the TVA 10-MW  Spray  Dryer/ESP  Evaluation," EPA/EPRI  First
      Combined FGD  and Dry SO2 Control Symposium. St. Louis, Missouri, October, 1988.

Chen,  Y.J.,  H.V.  Krigmont,  R.T.  Triscori  and H.W.  Spencer  (1984).   "Electrostatic
      Precipitators  in  Dry FGD  Applications," Second  International Conference  on
      Electrostatic Precipitation. Kyoto, Japan, November.

Donnelly, J.R. and M.T. Quach. "Update of JOY/Niro U.S. Utility Spray Dryer FGD Systems,"
      EPA/EPRI First Combined FGD and Dry SO2 Control Symposium, St. Louis, Missouri,
      October, 1988.

Doyle, J. B., B. J. Jankura and  R. C. Vetterick. "Comparison of Dry Scrubbing Operation of
      Laramie River and Craig  Stations,"   Paper  RDTPA 86-29, no.  9b, Presented  at
      Symposium on Flue Gas Desulfurization. Atlanta, Georgia, November 16-21, 1986.

Drummond, C.J.,Editor; M.Babu, A.Demian, D.S. Henzel,  D.A. Kanary, D. Kerivan, K. Lee,
      K.R. Murphy, J.T. Newman, E.A. Samuel, R.M. Statnick and M.R. Stouffer.  " Duct
      Injection Technologies For SO2 Control,"  Paper 10-2, presented at First  Combined
      FGD and Dry SO2 Control Symposium. St. Louis, MO.  October 25-28, 1988.

DuBard, J.L  and R.S.  Dahlin. "Precipitator  Performance Estimation Procedure,"  Research
      Project 629-5, Final Report, EPRI CS-5040, February, 1987.

Durham, M.D., C.M. Huang, J. DeGuzman,  B.F. Kee and  R.G. Rhudy.  "Analysis of the
      Performance of an ESP Operating Downstream of a Spray Dryer on High-Sulfur-Coal
      Flue Gas,"   Paper 5B1, presented  at the EPA/EPRI Seventh Symposium on  the
      Transfer and Utilization of Particulate Control Technology. Nashville, Tennessee,
      March 22-25, 1988a.

Durham, M.D., C.M. Huang, J. DeGuzman, B.F. Kee and R.G. Rhudy.  "The  Effect of A
      Retrofit Application of a Spray Dryer on the Performance of an ESP," Paper 88153.2,
      Air Polut. Control Assn. 81st Annual Meeting, Dallas, Texas, June 20-24, 1988b.


                                      15-19

-------
Faulkner, M.G.  and J.L.  DuBard.  "A  Mathematical  Model of Electrostatic Precipitation
      (Revision 3)," EPA-600/7-84-069a,b,c. (Volume I, Modeling and Programming:  NTIS
      PB84-212-679; Volume II,  User's Manual:  NTIS PB84-212-687;  FORTRAN Source
      Code Tape: NTIS PB84-232-990). August, 1984.

Gooch, J.P, G.H. Marchant, Jr., R. Beittel and J.L. DuBard. "The Effect of Hydrate Injection
      of ESP Performance," Paper 88-153.1, presented at the 81 st Annual Meeting of the Air
      Pollution Control Association. June 20-24,1988.

Gooch, J.P and G.H. Marchant (1978).  "Electrostatic Precipitator Rapping Reentrainment
      and Computer Model Studies," EPRI Report No FP-792.3, EPRI Contract RP 413-1,
      June.

Hendriks, R.V. and P.S.  Nolan.  "EPA's LIMB Development and Demonstration Program,"
      Journal of the Air Pollution Control Association, 36: (4),432.  1986.

Karlsson, H.T.,  J. Klingspor, M. Linne,  !. Bjerie.  "Activated Wet-Dry Scrubbing of SO2  "
      Journal of the Air Pollution Control Association, Volume 33, no.1, January, 1983. pp.
      23-28.

Katz, J.  The Art of Electrostatic Precipitation  Precipitator Technology, Inc., S&S Printing
      Company, Inc. Pittsburgh, Pennsylvania. 2nd Printing January 1981.

Penney, G.W.  "Adhesive Behavior of Dust in Electrostatic Precipitators,"  Symposium on
      Electrostatic Precipitators for the Control of Fine Particles. EPA-650/2-75-016: 65.
      1975.

Penney, G.W.  "Role of Adhesion in Electrostatic Precipitation,"  Archives of Environmental
      Health, Volume 4, p. 301, March, 1962.

Research-Cottrell, Tennessee Valley Authority.    "High-Sulfur Spray Dryer Shawnee Test
      Program, Volume VI: R-C/TVA Spray Dryer/ESP Evaluation," TVA/OP/EDT-86/15.
      1986.

Smith,  W.B., K.M.  Gushing  and J.D.  McCain.   "Procedures  Manual for Electrostatic
      Precipitator Evaluation," EPA-600/7-77-059, NTIS PB269698, June, 1977.

Spencer, H.W.,  III.  "Rapping  Reentrainment  in a  Nearly  Full Scale Pilot Electrostatic
      Precipitator," EPA-600/2-76-140: 76. 1976.

Tennessee Valley Authority.  "10-MW Spray Dryer/ESP Medium-Sulfur Coal Testing (Phase
      IV):  Basic Evaluations," Final Report, TVA/R&D, January 1990a.

Tennessee Valley Authority.  "10-MW Spray Dryer/ESP Medium-Sulfur Coal Testing (Phase
      IV):  Coal Chloride Evaluation," Draft Report, January 1990b.

Tennessee Valley Authority. "10-MW Spray Dryer/ESP Pilot Plant Test Program High Sulfur
      Coal Test Phase (Phase III)," Final Report TVA/ED&T-88/35, July 1988.

Young, R.P.,  J.L DuBard and L.S. Hovis.  "Resistivity of Fly Ash/Sorbent Mixtures," Paper
      1A1, Presented at the EPA\EPRI  Seventh Symposium on Transfer and Utilization of
      Paniculate Control Technology. Nashville, Tennessee, March 22-25. 1988.
                                       15-20

-------
         —Tost 3-C-01 with A fields
          Spray Dryer Conditions
              .000   1.00  Z.OO   3.00  10.0  20.0
             PARTICLE PHYSICAL DIAMETER (ulcronioterB)
                                                    220.


                                                    200. -


                                                    1BO. -


                                                    160. -


                                                    140. -


                                                    120. -


                                                    100. -


                                                    BO.O -


                                                    60.0-


                                                    40.0-


                                                    20.0-
                                                              AVERAGE ELECTRIC FIELD STRENGTH (kV/cm)

                                                            0.5   1.0   1.5   2.0   25   3.0   35
Teat a-A-01

Baseline Conditions

0  Section 1

*  Section A



Test 3-C-01

Spray Dryer Condition

*  Section 1

x  Section A
                                                                       VOLTAGE (kVl
 Figure  1.   Comparison  of inlet  particle    Figure 2.   Comparison of voltage  and
 size distributions  for baseline  and         current characteristics  at baseline and
 spray dryer conditions at the TVA 10  MW    spray dryer conditions at the  TVA 10 MW
 Pilot Plant.                                     Pilot Plant.
  10 "
I  10-
2f
is
s
  10 ';
               00 O
                      ft  t
                        fciiUfift Loborotory Descending Temp.
                        QDDDO Laboratory Ascending T«mp.
                        ooooo |n_situ with SRI Probe
                        A.AAAA In-Situ with ADA Extractive Syaterr
   ) 'I	i i i I i i i i i i
   1 20.0      1 40.0
                    160.0      180.0
                    Temperature (*F)
                                                         100   .EOO   .500  1.00  2.00   S.OO  10.0  20.0   50.0  1 0.


                                                                 PARTICLE PHYSICAL DIAMETER (mlcrometera)
Figure 3.   Comparison  of laboratory  and     Figure  4.   Comparison  of  outlet particle
in-situ  resistivity measurements  made at    size distributions for baseline and
the  TVA  shawnee spray  dryer  /ESP Pilot
Plant with medium-Sulfur,  high-chloride
coal.
                                                    spray dryer  conditions  at  the TVA  10
                                                    Pilot Plant.
                                               15-21

-------
PLATE
     )
                 CT
LAYER
H3
/ ( ^l
\ \^}
/ (~~\
_S_
/


©
pn
©
©
>
©
>
©
GAS
/f~~\
\ v^
e© Nl
/ f" N
\ V,^^
_^
J9 x

                                               €E
                                                    LAYER
                                                                            GAS
 Figure 5a.  Boundary conditions  for a
 layer of collected particles.
Figure  5b.   Step  change in  electric
displacement at layer-gas boundary.
               10 '      10 '     10 '
                PARTICLE RESISTMTY (ohm-cm)
   Figure 6.   electrostatic  force acting
   on a dust  layer at a  current density
   of 60 nA/cm  as a function of particle
   resistivity and electric  field at  the
   Plate.
                                                 lO'+TT-
                                                  120.0
                                                            0  ,-'i   ooooo Low Cl, Medium S; 6/89
                                                            iff     04.0t5 High Cl. Medium S; 10/88
                                                                  wt*»Hl9h Cl (Splk.d), M.d S; 8/89
                  180.0     180.0
                  Temperatura (*F)
 Figure 7.  Comparison  of in situ
 resistivty mesurements  as a function
 of  chloride  level.
                                           15-22

-------
             ELECTROSTATIC PRECIPITATION OF PARTICLES PRODUCED
                 BY FURNACE SORBENT INJECTION AT EDGEWATER
                                  Ralph F. Altman
                           Electric Power Research Institute
                                  P.O. Box 10412
                                Palo Alto, CA  94303
                                 E. C. Landham, Jr.
                                  E. B. Dismukes
                                  M. G. Faulkner
                                    R. P  Young
                             Southern Research Institute
                                  P  O. Box 55305
                               Birmingham, AL  35255
                                   Louis S.  Hovis
                        U.S. Environmental Protection Agency
                   Air and Energy Engineering Research Laboratory
                          Research Triangle Park, NC 27711
                                    ABSTRACT

LIMB (Limestone Injection Multistage Burners) is a system which combines calcium-based
sorbent  injection at  furnace temperatures with  low-NOx  burners to reduce SO2 and NOX
emissions from coal-fired utility boilers.  The particles generated by this process can have a
severe detrimental impact on the performance of a downstream electrostatic precipitator (ESP).
A demonstration of LIMB was conducted at the Edgewater  Station of Ohio Edison.  This paper
describes the performance of the Edgewater  ESP during LIMB  operation with and without
humidification for resistivity reduction.  The results of the test program are compared to the
predictions of the EPA/SRI ESP Model, and the accuracy of laboratory resistivity measurement
on this dust is assessed.
      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.
                                       16-1

-------
             ELECTROSTATIC PRECIPITATION OF PARTICLES PRODUCED
                 BY FURNACE SORBENT INJECTION AT EDGEWATER
INTRODUCTION

LIMB (Limestone Injection Multistage Burners) is a retrofit control technology for SOX and NOX
emissions that may be applicable to a number of existing wall-fired and tangentially fired boilers.
LIMB technology  involves the injection of a calcium-based sorbent directly into  a furnace
equipped with low-NOx burners.

Candidate boilers for a retrofit application of LIMB technology are mostly equipped with small
ESPs (<300 ft2/"!000 acfm at 9 in. plate spacing) that were designed to collect low-resistivity fly
ash from high-sulfur coal combustion. A retrofit application of LIMB technology will result in
changes in the concentration and properties of suspended particulate matter that will probably
present a difficult challenge to these ESPs.

To  investigate the impact of LIMB on an  existing boiler, the Air and  Energy Engineering
Research Laboratory of EPA supported a demonstration of this technology at Edgewater Station
Unit 4 of Ohio Edison Company, where preliminary tests took place during September 1987.
LIMB operation during these tests was limited to short time periods because the performance
of the ESP degraded rapidly and the stack opacity began to approach the legal limit for the unit.
To overcome this problem, a flue gas humidification system was installed on Unit 4 between the
air heater and the ESP to reduce dust resistivity.

Humidification produced a significant improvement in  ESP performance.  EPRI and EPA co-
funded a comprehensive  1-week performance test to study the  operation of the ESP with
humidification. A detailed presentation of the background for this work is contained in Reference
1. This paper provides a brief summary of observations concerning ESP performance during
early periods of LIMB testing with no humidification and a more comprehensive report of results
from the humidification tests.
Plant and Sorbent System Description

Unit 4 of Edgewater Station is a Babcock & Wilcox wall-fired boiler rated at 105 MWe. The fuel
burned during the LIMB test program was an Ohio bituminous coal with a sulfur content of
around 3%, although during the ESP performance test discussed here the coal sulfur averaged
2.6%.  During  all of the tests described  in this paper, commercial calcitic hydrated  lime was
injected into the boiler close to the nose of the furnace. The gas temperature at this elevation
was approximately 2300°F  To improve ESP performance, a flue gas humidifier was added
between the air preheater and the ESP  The humidifier is a horizontal chamber 14.6 x 14.6 ft
in cross section and 56.5 ft long.  It provides a residence time of roughly 2 seconds if it is


                                        16-2

-------
treating all of the flue gas generated at full boiler load.  Water is injected in the humidification
chamber  through an array  of 110  air/water  spray  atomizers supported in 22  lances.
Compressed air is used to produce a fine droplet formation in the spray pattern.

The ESP  at  Edgewater  Unit 4 was constructed by Lodge-Cottrell with  a nominal specific
collection  area (SCA) of 612 ft2/1000 acfm at full load (105 MW). There are six separately
energized electrical fields in the direction of gas flow,  with each  field consisting of two bus
sections.  Separate transformer-rectifier (T-R) sets supply power to the bus sections  in the first
three fields, while a single T-R set supplies the two bus sections in the last three fields. The ESP
has 44 lanes that are  12 in. wide.  It  was designed to  process  518,000 acfm  at  a gas
temperature of 280 °F with a gas velocity of 4.9 ft/s.  The exhaust gas from the ESP  goes to a
single stack that has an exit diameter of 10.5 ft.
 Description of Test Program

 Three sets of data were obtained on the operation of the Edgewater ESP during the LIMB test
 program:

    1.      Baseline without LIMB,
    2.      LIMB without humidification, and
    3.      LIMB with humidification.

 The baseline and  LIMB without humidification tests included collection of limited data only.  For
 these tests dust resistivity,  ESP electrical characteristics, and stack opacity were obtained.  For
 the LIMB  test with  humidification,  however, extensive test data were  obtained and will be
 described. Five days of testing were conducted on the Edgewater Unit 4 ESP during May 1989.
 The test program consisted of measuring the particulate mass loadings at the ESP inlet  and
 outlet, determining the  inlet and outlet particle size distributions, measuring the inlet and outlet
 gas flows, determining  the ash resistivity in situ, and measuring the ESP electrical conditions.
 In addition to  the field  data, the study supported a sizable laboratory  study which included
 chemical analyses, determinations of resistivity, determinations of BET surface area, and Bahco
 particle-size analyses for selected ash samples.
ESP PERFORMANCE RESULTS

The fuel burned in Edgewater Unit 4 during the ESP tests was a bituminous coal with a sulfur
content of approximately 2.6%.  In the absence  of sorbent addition, this fuel should generate
sufficient SO3 to produce a low fly ash resistivity  that would not limit the electrical operation or
collection performance of the ESP.  Measurements of in situ resistivity with a point-plane probe
indicated that the ash had a baseline value of  3x1010 ohm-cm  at 305°F [1].  During these
baseline tests, the T-R sets operated at current limit and stack opacity was indicated to be below
2%.

Without any form  of resistivity conditioning, hydrated lime addition to the furnace produced a
severe impact on  ESP performance. As  a consequence, LIMB operation could be conducted
only a few hours at a time before  opacity approached the 20% limit,  and only limited data for
this method of operation were obtained.  However, sufficient time was available during these
                                         16-3

-------
short term tests to  evaluate the characteristics of the particles produced and to judge the
severity of the impact on ESP operation.

For LIMB operation  without humidification, the in situ measured dust resistivity values were in
the range from 1x1012 to 5x1012 ohm-cm at 300 to  350 °F  The resistivity measurements were
made using the same probe that was used during the baseline tests. Laboratory measurements
suggest an even higher resistivity, as  high as 1x1014 ohm-cm [1].

The effect of LIMB without humidification on ESP electrical operating conditions can be seen in
the voltage-current (V-l) curves of Figures  1 and 2.  The curves labeled "without  LIMB" were
measured just before the start of sorbent addition,  while the "with LIMB" curves were obtained
after 3  hours of sorbent injection just before termination of the test due to high opacity. Severe
back corona is apparent in the V-l curves obtained with LIMB, and the decrease in useful power
with time is responsible for the loss of performance.  Degradation of the electrical conditions
initially occurred in the inlet field and gradually worked toward the outlet.  In general, there was
very little sparking and the  controllers maintained current limit while voltage was lost to back
corona.

The addition of the humidification system to Edgewater Unit 4 allowed continuous operation with
LIMB  without  risk of opacity  violations. Accordingly, 5 days  of  detailed  ESP testing were
conducted  during May 1989.   During this  period  the unit was operating at  75 MWe with a
sorbent injection rate that produced a calcium to sulfur ratio of 2.0.  The ESP inlet temperature
was held constant at 275 °F  during most of the tests, but was reduced to 165°F on 1 day. Most
of the  tests were conducted with only the last five fields of the  ESP energized to reduce the
effective size of this large ESP   During one of the test days, the number of energized fields was
reduced to three.

The data related to the total mass collection  efficiency of the ESP on each  of the four test
periods are summarized in Table 1. In general, the performance of the ESP during LIMB with
humidification  was very good.  ESP particle collection efficiency  was in excess of 99.9% on all
days.  Since the SCA varied  for each  test day  due to changes in fields on line and gas
temperature,  effective migration velocity (uk) is a more effective  measure of ESP performance
due to its insensitivity to ESP size.  A consistent increase in wk with time can be observed during
the test program which appears to  be somewhat independent  of the test  conditions.  A
corresponding trend in ESP  electrical conditions was observed which indicates that some aspect
of ESP performance was not stable.

This trend  makes it difficult to  establish the exact effect of the 165°F ESP  temperature on
performance.  However, the increase between day 1  and day 2 was much sharper than the
general increasing trend, suggesting  a positive effect due to higher levels of humidification.
Since the high humidification level produced the first observed increase in ukl it is possible that
this test condition (low temperature, high moisture)  had an irreversible effect on ESP operation.
If this is the case, the mechanism is not clear.

The highest uk of the week was measured during  the three field tests of 5/26. It cannot be
determined to what extent this increase is  a continuation of the generally improving trend or
specific to this test condition. However, even with only three fields on line, the SCA was still very
high at 439 ft2/1000 acfm.  Stabilization of the dust layers after  a major reconfiguration of the
fields is likely to take a significant period of time.  The ESP  had been operating in three-field
configuration for only 15 hours prior to the  test.  Consequently,  the unexpectedly  good three-


                                         16-4

-------
field performance could  have been partially caused by  lower rapping emissions  due to
insufficient stabilization time.   Since other studies have indicated that rapping emissions can
constitute up to 80% of total emissions for large ESPs [2],  incorrect treatment of rapping issues
has the potential to produce large errors.  There is the possibility that the dust burdens in the
three fields had not adjusted to  their  new steady-state  level.   Consequently, the collection
efficiency measured during three-field operation may not reflect stable operation and should be
interpreted carefully.

The average inlet mass loading during  the LIMB test program was approximately 6.6 gr/dscf.
This is a factor of 2 higher than the inlet  mass  loading  predicted  by the EPRI data base on
pulverized-coal-fired  boilers operating without sorbent addition [3].  For bituminous coal with
10.49% ash, the  data base predicts an inlet mass  burden of 3.15 ±  0.78  gr/dscf  without
sorbent.

The inlet and outlet particle size distributions were measured with cascade impactors during a
test day when five ESP fields were in service. The cumulative mass size distributions at the inlet
and outlet are shown in Figures 3  and 4. The plotted  curve for the inlet represents the average
of 10 impactor  runs; the curve for the  outlet is  the average of 2 runs.  Assuming log-normal
distribution, the inlet and outlet particle size distributions can be described as:
Measurement Location
ESP Inlet
ESP Outlet
Mean Size, [im
10.7
1.7
Standard Deviation
3.1
2.0
Data on the ESP fractional efficiency are presented in Figure 5.  The data indicate the minimum
collection efficiency or the  maximum penetration  occurred  near a particle size of 1 /un.  For
comparison, the particle size dependent collection efficiency of the fabric filter on the circulating
fluidized bed combustor at Colorado Ute's Nucla Station is also plotted in Figure 5. These two
curves illustrate that under favorable operating conditions both ESPs and fabric filters  can
produce  high collection efficiency for the entire  range of particle sizes produced by these two
S02 control processes.

The dust resistivity  measured in situ during the humidification tests averaged approximately
2.3x1011 ohm-cm. This value of resistivity is consistent with  highly efficient ESP operation  with
SCAs as high as that which prevailed during the testing at Edgewater.  However, this level of
dust resistivity will limit ESP electrical operation to some extent and could produce unacceptable
performance in small ESPs (SCA  < 300 ft2/1000 acfm).

Humidification during sorbent  addition significantly improved  ESP electrical  operation consistent
with the observed reduction in dust resistivity.  The ESP V-l curves of May 25 during five-field
operation (Figure 6)  are representative of 275 °F operation. The slight indications of back corona
in the curves suggest that the 275 °F temperature is marginal for resistivity  reduction. There is,
however, a very high degree  of suppression of  the back corona previously evident in  Figures
1  and  2  for LIMB without  humidification.  Figure 7 indicates that further lowering the  gas
temperature to  165°F eliminated the back corona and increased useful power to the ESP.
                                          16-5

-------
RESULTS OF LABORATORY STUDIES

The results of laboratory resistivity measurements made on four dust samples obtained on May
24 and 25 are shown in Figure 8.  The resistivity data were obtained according to the procedure
described in IEEE Standard 548-1984 [4], except that the maximum temperature to which the
samples were exposed was only 480 °F  This limit on temperature was necessary with the
ash/sorbent samples to avoid thermally decomposing Ca(OH)2 and converting it to CaO, thus
substantially changing the sample resistivity.  The resistivity data were determined at a constant
water vapor concentration of 9% by volume.  This water vapor concentration was approximately
equal to that in  the ESP at Edgewater during the  high temperature tests. Average flue gas
moisture  levels measured during  LIMB were 8.9% during  the 275°F tests and 10.8% for the
165 °F tests.

The results  obtained from the laboratory resistivity measurements are self-consistent; the only
significant deviation occurred with the data for the mass train sample. At the temperature of flue
gas at the inlet  of the Edgewater ESP (approximately 275°F in three  of the tests previously
described),  the laboratory data range from 5x1012 to 2x1013 ohm-cm. Even the lower end of this
range is  still  approximately an order of magnitude  higher than the range of data from the
measurements of resistivity in situ.  The discrepancy between values of resistivity measured in
the laboratory and in situ was encountered earlier in experiments involving the humidification of
LIMB ash in the Southern Research Institute (SRI) pilot-scale coal combustor.  The reason for
the discrepancy is not known, and further work is needed  to try to resolve this problem.
ESP MODELING RESULTS

Revision 3 of the EPA/SRI ESP performance model  [5] was used to simulate the performance
of the Edgewater Unit 4 ESP  The ESP model performs a detailed numerical simulation of the
charging and collecting of dust particles along one gas passage of an ESP and makes empirical
corrections for  non-ideal gas flow and  rapping reentrainment.   The  performance of the
Edgewater ESP with humidified LIMB ash was modeled using data obtained during tests on May
24 and 25 (five fields on-line) and May 26 (three fields on-line). The data were modeled for two
non-ideal gas conditions. These are expressed as the amount of gas sneakage (fraction of gas
bypassing the electrified portion of the ESP and therefore not subject to particle collection) and
the value of ag (a measure of gas velocity non-uniformity, which degrades ESP performance).
The values used were:

    1.     sneakage = 0.05 and  ag = 0.15 (typical  of a modern ESP in good  repair), and
    2.     sneakage = 0.10 and  ag = 0.25 (typical  of older ESPs).

The ESP electrical conditions used in the model were obtained from V-l curves and chosen to
exclude back corona current.

The results of the modeling are compared to the measured data in Table 2.  For five fields in
operation, the model accurately predicts actual collection performance with sneakage = 0.10
and ffg =  0.25.  These  non-ideal parameters are  somewhat higher  than  expected for the
Edgewater ESP  which is relatively new and in good  condition.

For three fields (the lower SCA) computed performance is provided in the table both with and
without rapping.  Since the time for readjustment of the  dust layers in the  outlet field (upon


                                        16-6

-------
which rapping emissions  are dependent) was short, rapping  emissions may be lower than
expected.  The actual performance was better than that predicted by the model when rapping
emissions  were included in the calculation.  Ignoring rapping emissions, the actual collection
performance falls between the predictions for  non-ideal conditions of s  = 0, ag = 0; and s =
0.05, ag  =  0.15.

CONCLUSIONS

Minimal  humidification of the  flue gas at Edgewater Unit  4  made it  possible to maintain
satisfactory ESP performance.  Without humidification, the ESP performed  so  poorly that the
injection of sorbent had to be suspended despite its very high SCA. With humidification, on the
other hand, the opacity could be maintained at 5% or less, and the efficiency of dust removal
was in the range above 99.9% on the overall mass basis.

Deliberate  efforts were made to determine the effects on ESP performance  of two changes in
operating conditions: 1)  further lowering of the temperature to 165°F, and 2)  lowering of the
SCA from  about 700 to about 440 ft2/1000 acfm.  Lowering the temperature did produce an
improvement  in the  electrical conditions in the ESP   An increase in effective  ESP migration
velocity (uk) was also noted which appears to be due to additional humidification. Contrary to
expectations,  no linear degradation from the SCA  reduction  was evident, but lower SCAs
generally gave higher penetrations.

There was significant improvement  in ESP  performance  on  LIMB dust  due to  flue gas
humidification. However,  caution  is advised in applying the results of this test to the general
population of  ESPs which are candidates for LIMB technology.  Most of the measurements of
ESP performance at Edgewater were made at an SCA of around  700 ft2/"!000 acfm. Candidates
for LIMB retrofits typically have small ESPs requiring low resistivity (2 x 1010 ohm-cm) to perform
adequately. At Edgewater, resistivity was lowered to only 2.3 x 1011  ohm-cm using low levels
of humidification. The general adequacy of humidification for overcoming the deleterious effects
of LIMB on ESP performance would have to be evaluated in future tests in which greater levels
of humidification are  more fully tested.
UNIT CONVERSION TABLE
To Convert From:
ft/sec
ft
ft2/kacfm
gr/ft3
in.
lb/106 Btu
°F
To:
m/sec
m
m2/(m3/sec)
g/m3
cm
ng/J
°C
Multiply by:
0.3048
0.3048
0.19685
2.288
2.540
430
5/9*(°F-32)
                                        16-7

-------
REFERENCES

1.  J. P Gooch, J. L. DuBard.  Evaluations of Electrostatic Precipitator Performance at
   Edgewater Unit 4 LIMB Demonstration.  U.S. Environmental Protection Agency.
   Research Triangle Park, NC. EPA-600/7-88-020, (NTIS PB-89-109177), September 1988.

2.  J. L. DuBard and R. S.  Dahlin.  Precipitator Performance Estimation Procedure. Electric
   Power Research Institute.  Palo Alto, CA.  EPRI CS-5040. February 1987.

3.  J. P. Gooch and G.  H. Merchant.  Electrostatic Precipitator Rapping Reentrainment and
   Computer Model Studies.  Electric Power Research Institute.  Palo Alto, CA.  EPRI
   FP-792, vol 3.  August 1978.

4.  IEEE Standard Criteria and Guidelines for the Laboratory Measurement and Reporting of
   Fly Ash Resistivity.  IEEE Standard 548-1984.  Institute of Electrical and Electronic
   Engineers,  Inc.  New York, New York.  1984.

5.  M.  G. Faulkner and  J. L. DuBard.  A Mathematical Model of Electrostatic  Precipitation
   (Revision 3):  Vol. I   U.S. Environmental Protection Agency.  Research Triangle Park,
   NC. EPA-600/7-84-069a, (NTIS PB-84-212679), June 1984.

-------
TABLE 1. ESP PERFORMANCE DATA
Date
5/22/89
5/23/89
5/24/89
5/26/89
SCA,
ft2/kacfm
689
853
715
439
ESP
Temperature,
°F
275
165
273
277
Particle
Emissions,
lb/106 Btu
0.0136
0.0018
0.0043
0.0106
Collection
Efficiency,
%
99.922
99.987
99.979
99.933
Particle
Penetration,
%
0.078
0.013
0.021
0.067
«k,*
cm/sec
38
48
51
62
Stack
Opacity,
%
6.2
3.4
4.6
6.0
"Effective migration velocity.
TABLE 2. ESP MODEL RESULTS WITH MEASURED ELECTRICAL CONDITIONS

Non-Ideal
Parameters,
s' <
Penetration,
%
Efficiency,
%
Emissions,
lb/106 Btu
Opacity,
%
5 ESP FIELDS ON LINE
Measured
Calculated
(Rap+No Rap)
NA
0.00,0.00
0.05,0.15
0.10,0.25
0.021
0.002
0.006
0.020
99.979
99.998
99.994
99.980
0.0043
0.0004
0.0012
0.0041
4.6
0.1
0.2
0.7
3 ESP FIELDS ON LINE
Measured
Calculated
(Rap + No Rap)
Calculated
(No Rap Only)
NA
0.00,0.00
0.05,0.15
0.10,0.25
0.00,0.00
0.05,0.15
0.10,0.25
0.067
0.122
0.267
0.602
0.058
0.155
0.409
99.933
99.878
99.733
99.398
99.942
99.845
99.591
0.0106
0.0192
0.0420
0.0947
0.0091
0.0244
0.0643
6.0
3.9
7.2
13.0
3.1
5.9
11.0
*s = Fractional gas sneakage, og = Normalized velocity profile standard deviation.
16-9

-------
   io
   10'
   10'
   IO1
        _L
          "1
                NUCLA
                                  SUBBITUMINOUS COAL
                          -BLACK DOG
                       li.
                                                   ' ' '
                                        1
        io"             10             io             io
                 PARTICLE DIAMETER, MICROMETERS

  Figure 1.  Cumulative Mass Particle Size Distribution.
  99.99



   99.9




 p   99


 J

 r   90
 c
 H


 ::   70
 c

 :   50


 \   30
 J



 \   10
 T
 C



      1



    0.1



   0.01
BLACK DOG
                                      li.
        10"
                        ^0
                                        ,1
                                      IO1            10'

                 PARTICLE DIAMETER,  MICROMETERS
Figure 2.  Cumulative Percent Particle Size Distribution.
                            16-10

-------
   10
     12
                                  I       1
   10
     11
                                 IN SITU
   101
o

>-
                                               INLET
   10
   10C
   10'
   10C
         PREDICTED
        I
               I
                     I
                            I
                                  I
                                         I
       3.0    2.8   2.6    2.4   2.2    2.0    1.8   1000/K
       60     84    112    144   182    227    283   -C
       140    183   233    291   359    441    541   -F
                           TEMPERATURE

          Figure 3.  Heskett AFBC Dust Resistivity.
    10
      13
             IN SITU
    10
      12
        '  LAB 8/9
   10
      11
          LAB 8/10
LLJ
CC
   10
      10
   10"  -
          PREDICTED
          8/9-8/10


             	I
                     J_
                                         I
        3.0    2.8    2.6    2.4    2.2    2.0    1.8   1000/K
        60     84     112    144    182    227    283   °C
        140    183    233    291    359    441    541   -F
                           TEMPERATURE

         Figure 4.  Black Dog AFBC Dust Resistivity.
                          16-11

-------
     10
     10
       13
     10
       12
     10
        11
     10
        10
                                     IN SITU
          J_
                _1_
                       _L
                             _L
                                     1
                                           1
                                                  1
                                                        1
          3.0
          60
          140
  2  8    2.6    2.4    2.2   2.0    1.8  1000/K
  84    112    144    182   227    283  °C
  183    233    291    359   441    541  -F
               TEMPERATURE

Figure 5.  Nucla AFBC Dust Resistivity.
      io13-
      10
        12
      10
        11
      10
        10
             7.87. WATER
            11.5% WATER
            14.2% WATER
          _L
                 J_
                       _L
                               «  5 ppm S03
                                    _L
                                          _L
                                                 _L
                                                        _L
          3.0
          60
          140
  2.8
  84
  183
2.6
112
233
2.4
144
291
2.2
182
359
2.0
227
441
1.8   1000/K
283   -C
541   -F
                              TEMPERATURE
Figure 6.  Effect of Conditioning on the  Resistivity of Nucla Dust.
                             16-12

-------
    60
    50
    40
S   30
LU
O
§   20
o
    10
     0
                                     44.8,72
        25
                                                   50
              30       35       40       45

                     ISP VOLTAGE,  KV

Figure 7  Electrical Characteristics of the Heskett ESP
i
o
LU
0
    50
    40
    30
LU

    20
    10
        10
                20
40
                                                   50
                              30

                       ESP VOLTAGE, KV

Figure 8.  Electrical Characteristics of the Black Dog ESP
                         16-13

-------
    99.99
     99.9
       99
       90
       70 -
                         NUCLA WITH HUMID
              BLACK DOG
                                 NUCLA W/0 HUMID
          200        300        400        500
                        SCA,  FT2/1000 ACFM
                                   600
Figure 9.  Predicted and  Measured ESP Collection Efficiency.
      101 -
   Si io-2
     10
       -3
                                       NUCLA W/0 HUMID
               BLACK DOG
HESKETT
                      L
                            NUCLA WITH HUMID
                                 _L
                                            _L
          200        300         400         500
                        SCA, FT2/1000 ACFM
                                   600
 Figure 10.  Predicted and Measured ESP Particle Emissions.
                            16-14

-------
TABLE 1. COAL COMPOSITION

% Moisture
% Ash
% Sulfur
% Volatile
% Fixed Carbon
HHV, Btu/lb
% Carbon
% Hydrogen
% Nitrogen
% Chlorine
% Oxygen
Heskett
36.22
8.20
1.22
26.01
29.57
6885
40.39
2.63
0.61
0.01
10.72
Black Dog
20.64
8.05
1.23
34.87
36.44
9603
55.09
3.63
0.92
0.05
10.39
Nucla
7.91
13.64
0.58
33.32
45.13
11088
62.71
4.19
1.32
0.01
7.91
TABLE 2. PARTICULATE MASS LOADINGS

PLANT
PARTICLE MASS LOADING
gr/acf
gr/scf
GAS VOLUME FLOW
acfm
scfm
TEMPERATURE
°F
CONTROL DEVICE INLET
Heskett
Black Dog
Nucla
1.16
2.69
4.32
2.05
4.29
8.85
343800
451500
257600
195800
283500
125800
296
280
303
CONTROL DEVICE OUTLET
Heskett
Black Dog
Nucla
0.0069
0.0163
0.0018
0.0112
0.0255
0.0037
341300
432700
248400
205000
275400
122100
277
284
286
16-15

-------
TABLE 3. CHEMICAL COMPOSITION OF DUST SAMPLES

Li2O
Na2O
K2O
MgO
CaO
Fe203
AI203
SiO2
TiO2
P205
S03
LOI
Sol SO4
Heskett
Inlet
0.01
6.4
0.48
9.7
28.3
4.9
11.0
17.4
0.33
0.49
16.9
12.2
15.9
Middle
0.01
12.1
0.72
8.6
25.1
3.8
8.4
10.9
0.33
0.47
25.9
7.0
27.0
Outlet
0.03
25.9
1.2
4.8
13.7
2.2
4.4
4.5
0.25
0.27
41.1
4.1
46.3
Black Dog
8/9/88
0.01
1.2
1.3
3.7
26.2
8.2
15.2
31.8
0.92
0.64
9.0
11.0
10.3
8/10/88
0.01
1.3
1.2
3.9
23.6
8.0
16.3
34.8
0.92
0.72
8.1
10.5
8.8
Nucla
9/19/89
0.02
0.47
0.45
0.74
17.8
1.8
23.6
47.5
0.67
0.49
5.1
6.2
6.1
9/22/89
0.03
0.47
0.42
0.74
19.9
1.8
23.1
46.7
0.75
0.52
4.7
6.0
5.5
TABLE 4. CONTROL DEVICE COLLECTION PERFORMANCE

PLANT
Heskett ESP
Black Dog ESP
Nucla FF
EFFICIENCY
%
99.454
99.406
99.959
EMISSIONS
gr/acf
0.0069
0.0163
0.0018
OPACITY
%
7


SCA
ft2/kacfm
368
352

S*
cm/sec
38.8
37.6

* Effective Migration Velocity
16-16

-------
   PROPOSED DEMONSTRATION OF HYPAS ON DUKE POWER'S MARSHALL STATION UNIT 2:
        AN INTEGRATED APPROACH TO PARTICULATE UPGRADES AND S02 CONTROL

                                 Kris W. Knudsen
                               Duke Power Company
                              422 S. Church Street
                        Charlotte, North Carolina  28242'

                                 Robert C. Carr
                           Electric Power Technologies
                         695 Oak Grove Avenue, Suite 2C
                          Menlo Park, California  94025

                                Richard G. Rhudy
                        Electric Power Research Institute
                              3412 Hillview Avenue
                          Palo Alto, California  94303
The HYPAS project will incorporate the first full-scale utility application of a
pulse-jet baghouse in the U.S. on the 400 MW Marshall Unit 2.  The HYPAS process
will be demonstrated on one 200 MW duct, incorporating humidification and dry lime
injection ahead of the pulse-jet baghouse,  with the goal  of achieving a 50 %
minimum reduction in S02 while reducing particulate emissions to less than NSPS.
Parallel  testing will be performed on the other 200 MW duct to provide baseline
pulse-jet performance without S02 control.

HYPAS provides an important strategy for utilities which  are required to upgrade
particulate control  and which expect the need for future  S02 reductions to meet
acid rain requirements.  The process is well suited to units with limited space for
retrofit  of a large  precipitator, a low-ratio baghouse, or an S02 scrubber.  The
pulse-jet baghouse is economically competitive with replacement ESP's and favorable
to low-ratio baghouses, and provides the bonus of simple  retrofit for future S02
control.   A precipitator/pulse-jet series arrangement for particulate collection
also may  be favorable from a maintenance aspect by extending bag life through
reduced particulate  loading and less frequent cleaning.

                                        17-1

-------
   PROPOSED DEMONSTRATION OF HYPAS ON DUKE POWER'S MARSHALL STATION UNIT 2:

        AN INTEGRATED APPROACH TO PARTICULATE UPGRADES AND S02 CONTROL


INTRODUCTION

The 1990's are certain to present increasing challenges to utilities with regard to

environmental  control on existing fossil-fired boilers.  Some of the key issues

which must be  addressed include:

     •    Aging Particulate Control  Systems:  A large class of retrofit
          precipitators installed in response to the Clean Air Act of 1970 are
          approaching 20 years of service.  Many older precipitators also remain in
          service,  either alone or in conjunction with retrofit collectors.  While
          most are  structurally sound and can continue to provide useful service,
          mechanical  degradation  has led in many cases to deteriorating
          performance.

     t    Inadequate  Original  Designs:   In the rush to meet compliance deadlines of
          the  early 1970's, neither utilities nor vendors were able to develop the
          expertise to adequately specify precipitator designs which would assure
          the  high  efficiency performance needed to meet continuous compliance.
          This has  led to a continuing  struggle to develop enhancements to make an
          inadequate  design work, but potential changes in operation, such as use
          of alternate fuels for  acid rain requirements, will increase the
          probability of unacceptable performance.

     •    Stricter  Enforcement Policy:   Because the risks of a potential
          non-compliance action are becoming less acceptable (civil penalties;
          forced load reductions; adverse public reaction; etc.), many companies
          with marginally complying units will  seek upgrades to provide a greater
          margin of safety.  Continuous monitoring requirements and tightening coal
          markets may affect the  ability to meet S02 standards using present coal
          specifications.  Consideration of one-hour ambient standards and stricter
          EPA  policy  on plant renovations may also result in tighter emission
          standards and/or require additional controls.

     •    New  Regulations:  The current debate in Congress will ultimately lead to
          extensive control of S02 on existing units.  Emissions of hazardous
          pollutants  is also under review and may result in further control of
          particulate or gaseous  emissions.

While there is an array of options for  control  of either particulates or S02,

retrofit control is limited by a  number of constraints.  These include: limited

space available for the additional equipment needed for high efficiency control;

the desire to  maintain existing waste handling operations including wet or dry

disposal of ash and/or sale of ash;  and increased complexity of operation requiring

additional  investments at older plants  in staff, facilities, and training.

                                       17-2

-------
Further, in considering control system upgrades to meet present and future
requirements, it is important to consider the integrated impact of one solution on
other possible needs.  For example, a precipitator upgrade or replacement may not
give the flexibility to install a dry S02 process without leading to increased
particulate emissions.  A high efficiency precipitator or low-ratio baghouse may
take so much of a station's limited space that any additional equipment for S02
control is not possible.
Duke Power has been faced with all of the above concerns in reviewing options for
improved environmental control at Marshall Steam Station.  Particularly, the
station has a need to improve the particulate collection on Unit 2 and recognizes
the probability of requirements for control of S02.  Study of available options has
led to the conclusion to pursue a full-scale demonstration of the HYPAS process
being developed by EPRI.  Duke Power believes that pulse-jet baghouse technology,
which is at the heart of HYPAS, has been demonstrated as a viable alternative for
particulate control and will preserve the option to use dry S02 processes.

MARSHALL UNIT 2 PARTICULATE / S02 EMISSIONS CONTROL STUDY
Site Description
Marshall Unit 2 is rated at 400 MW (gross) and began operation in 1966.  The boiler
is a Combustion Engineering split furnace, tangential-fired boiler rated at 2.4
million Ib/hr steam at 2450 psi / 1050 °F superheat steam conditions and 1000 °F
reheat temperature.  At full load conditions, the flue gas leaving the air
preheaters is typically 1.4 million ACFM at about 290 °F.
Particulate emissions are controlled by a series arrangement of a mechanical
collector installed as original equipment with a cold-side ESP installed in 1971.
The precipitator is a Buell (G.E.) design, two chambers across and four fields
deep,  with eight TR's which can be isolated into 16 sections.  The design SCA is
175 ft2/MACFM,  but at actual full load conditions the SCA is approximately
150 ft2/MACFM.   S0_ conditioning is used to enhance the resistivity.  Particulate
                  O
emission limits under the State SIP include an annual  stack test limit of
0.20 Ib/mBtu;  a six-minute average opacity limit of 40 %; and a rolling annual
average opacity limit of 20 %.  The annual average opacity limit includes only data
collected while the unit is on-line.
There  are no systems installed for control of S02 or NOx.  The S02 limit of 2.3
Ib/mBtu is achieved by purchasing low sulfur (1 %) Central  Appalachian coal which
gives  average  emissions of about 1.5 Ib/mBtu.  The State NOx limit of 1.8 Ib/mBtu
does not require any control based on AP-42 emission factors for tangential-fired
boilers.   NOx  emissions have not been tested.
                                        17-3

-------
 Original wet pond fly  ash  disposal was  replaced  during  the  mid-1980's with dry
 disposal in an  unlined  landfill  (clay soil).   Bottom ash  continues to go to the wet
 settling pond.  Facilities are set up for  sale of fly ash for  cement manufacture as
 permitted by ash characteristics.  Landfill  space is not  foreseen  to be a problem,
 but  opportunities are  being sought to maximize ash utilization to  assure that this
 will not be a constraint on future operations.

 Description of  Problem
 Particulate Emissions.  Marshall Unit 2 has  long  presented  a challenge in complying
 with particulate limits.  The retrofit precipitator  design  did not adequately
 account for the difficulty in collecting the fine ash leaving  the  mechanical
 collector nor was there enough conservatism  to allow for  variability in actual
 operating conditions (higher flue gas flow).  Prior  to  installing  the  SCL
                                                                         O
 conditioning system in  1982, the maximum capacity of the  unit  was  frequently
 limited by the  40 % six-minute opacity limit.  SCL conditioning provided an
                                                  •u
 adequate margin below the 40 % limit.
 However, in 1986, the North Carolina environmental regulatory  agency established
 annual average  opacity  limits to help assure continuous good operation  and
 maintenance of  precipitators.   The limits were further  set  at  a level  which would
 approximate an  emission rate of 0.10 Ib/mBtu based on mass  versus  opacity curves
 generated from  annual  stack tests.   (Note that many  units have  limits  below 20 %,
 including several  with  limits below 10 %.)
 Since 1987,  when the regulation became effective, Unit 2  has consistently been near
 the annual  average opacity limit although Duke Power's fossil  operations  have been
 at an all  time  low because of recent completion of baseload nuclear  capacity.
 Forecasted growth in demand is to be met by increased utilization  of fossil units,
 including Unit 2.   Figure 1 presents daily average opacity  data versus  the daily
 average utilization of Unit 2  for the period from February  1989 to March  1990.  This
 data indicates that based on historic operation,   the unit will  be  in violation if
 the output factor on the unit  exceeds about 75 %.
 Figure 2 presents  projected operations on Unit 2.  Beginning in 1992,  the output
 factor is  expected to  exceed 80 %,  and will reach 90 % by 1994.  The increasing
 capacity factor  also indicates that there will  be less opportunity for  outages to
 install  new  equipment  or to make  repairs.   This analysis would indicate  that
decisions  are  needed now to assure  full  compliance on the unit without  any load
restrictions.   (Note that performance on  Unit 2 has shown a  significant  improvement
                                       17-4

-------
since January 1990 after minor improvements were made to flow distribution  in the
precipitator inlet.  It is too early to draw long term conclusions from this
1imited data.)

SO, Control / Acid Rain Legislation.  The Clean Air Act proposal presently  before
Congress would require an emissions cap based on generation in the years 1985 -
1987.  Because increases in nuclear baseload capacity temporarily reduced Duke's
fossil generation during this baseline period, the company will be forced to
significantly control its emissions to accommodate increased utilization of existing
units from now to the compliance year of 2000 or 2001 (Figure 3).  Coal switching
is not an option of itself for our units under the present proposal.  To minimize
the capital cost, control  systems will be considered first on larger units with
expected high capacity factors, to offset the need for control on smaller units.
Marshall Unit 2 with a capacity of 400 MW and projected capacity factors greater
than 70 % is a prime candidate for control.

Hazardous Pollutants.  Likewise, Clean Air Act proposals are considering further
regulation of particulate and gaseous emissions which may be considered hazardous.
Recent regulations passed by North Carolina also require consideration of impacts
of hazardous pollutants.  Although utility boilers already typically remove much
greater than 95 % of the particulate, permit review of any major particulate
control upgrades is likely to consider the cost/benefits of very high efficiency
col lection.

Constraints On Solutions.   The most serious constraint limiting control options at
the Marshall  station is a lack of adequate space to install  large pieces of new
equipment.   Figure 4 shows the layout of Unit 2 from the boiler to the stack.  The
station is built up against the cooling water intake canal,  with only a necessary
service road between the stacks and the intake structure.   Lateral space is also
limited on one side by the bank of a hill  from which the site was excavated and on
the other side by support buildings, coal  handling, and the bottom ash settling
pond.   Because of this constraint, Marshall could not install a low-ratio baghouse
or a wet or spray dry scrubber system without incurring very high costs for a very
difficult retrofit.
The inadequate size of the existing precipitator will most likely rule out  the use
of dry injection processes for S02 control  where the sorbent and ash are collected
together in an existing precipitator.  Higher particulate loadings and recent
reports of very low resistivity from such processes will  only add to existing
particulate compliance problems (Ref. 1).   Further, a process which generates a

                                       17-5

-------
mixed fly ash/sorbent waste is not desired at Marshall because it may limit the
opportunity for diversified sale or reuse of the ash and/or spent sorbent.

DESCRIPTION OF HYPAS
HYPAS is a calcium-based, dry injection process for simultaneous control of S02 and
particulate emissions from fossil-fired power plants.  As depicted in Figure 5, the
HYPAS process is comprised of four basic steps which occur sequentially in the flue
gas stream: (1) fly ash removal  by an existing electrostatic precipitator (ESP);
(2) evaporative cooling by water injection (humidification) to within 10   20 ° F
of the flue gas saturation temperature; (3) dry injection of hydrated lime for
reaction with S02 in the flue gas; and (4) collection of sorbent by-products and
the remaining fly ash in a retrofit pulse-jet fabric filter.  A pilot-scale
evaluation of HYPAS is currently underway at EPRI's High Sulfur Test Center (HSTC)
at the Somerset Station of New York State Electric and Gas.

Particulate Control
HYPAS is one of the few S02 control processes specifically designed to enhance the
overall  particulate removal from a plant.  In fact, if S02 control  is not needed
immediately, the major component of HYPAS (the baghouse) can be configured as a
stand-alone system for meeting current particulate needs while providing the
flexibility to control S02 by 50 to 70 % in the future.
The pulse-jet baghouse installed on the HSTC pilot has been subjected to the full
range of operations including use as a primary ash collector,  as a secondary ash
collector, and as the collection device for the sorbent injection process.
Approach to saturation temperature has been maintained as low as 10 °F.   Upset
conditions have resulted in wetting of bags.   Throughout these operations, the
baghouse has performed extremely well, with typical outlet emissions of
0.002 Ib/mBtu or less and tube sheet pressure drop maintained below 5 in H20.
Isolated bag failures have occurred as a result of bag-to-cage designs,  but design
corrections seem to have eliminated this concern.  (Ref.  2, 3,  4)

SO, Control
S02 control in HYPAS is based on the increased reaction of hydrated lime and S02 as
temperature approaches saturation.   At 10 to 20 °F above saturation, 50 to 70 %
removal can be expected at a reasonable calcium to sulfur ratio.  Because the
by-product of the reaction is relatively pure (free of ash), recycle may be used to
increase utilization of sorbent or  to push towards the higher removal efficiencies.
                                      17-6

-------
Testing at the HSTC has confirmed the ability of HYPAS to achieve significant
emission reductions despite some problems associated with pilot-scale operations.
Figure 6 shows actual test results at varying S02 concentrations and operating
conditions. (Ref. 5)  Greater than 70 % reduction has been achieved at 700 ppm S02
(approximately 1 % sulfur coal).  A reduction of 50 % appears to be easily
achievable at moderate S02 concentrations (1.5 to 2 % su'ifur coal).  More recent
tests have indicated that grinding the spent sorbent prior to recycle may
reactivate the lime and achieve higher reductions (nearly 70 % reduction
with a 1.5 % sulfur coal).  Full-scale design improvements are expected to enhance
the ability of the process to achieve desired removal.  For example, wall
impingement in a small diameter duct affects the reliability of the humidification
process.  Also, full-scale designs will incorporate flow models to enhance
residence time in the duct and in the baghouse to promote better removal.
While the HYPAS process is not expected to consistently achieve the high removal
efficiency to meet New Source Performance Standard S02 requirements, moderately
high removal  in a totally dry state is a significant accomplishment for retrofit
applications where higher removal may not be justified by costs, age of the unit,
or physical constraints of the site.  HYPAS can be coupled with other choices
available to meet S02 standards.  For example, switching from a 3 % sulfur coal to
a 1.5 % sulfur coal along with a HYPAS removal efficiency of 50 % will  reduce
overall emissions by 75 %.

DUKE POWER'S APPROACH TO THE SOLUTION
Review of retrofit options for particulate control  and S02 control  proceeded along
separate paths within Duke Power until 1987, when EPRI and Electric Power
Technologies presented the concept of HYPAS to integrate these needs.  HYPAS
appeared to be well suited to the company's overall  retrofit needs because a number
of units have marginally designed precipitators and because most of the plants have
limited space available for retrofit S02 control.  Duke Power has supported EPRI's
development of the process by providing engineering support for an economic
analysis of commercial-scale application of HYPAS to utility boilers. (Ref. 6)
This analysis showed that HYPAS could provide moderate S02 control  at less than
half the cost of wet scrubbers.
In January 1989,  Duke Power initiated a study of options to improve particulate
control  on Unit 2 when review of the data predicted future non-compliance. The full
range of options was considered, from simple precipitator upgrades to a new
precipitator  to a new low or high ratio baghouse.  The incremental  cost of adding
                                      17-7

-------
HYPAS to the high ratio (pulse-jet) baghouse was also considered in light of
concerns over possible acid rain legislation.
One purpose of the study was to determine whether to submit a DOE Clean Coal
Technology proposal  for demonstration of HYPAS.  Factors to consider included
whether the pulse-jet baghouse was cost-competitive with other options; whether the
baghouse was a proven, reliable device for particulate control; and whether a
demonstration of the unproven HYPAS process could jeopardize station operations.
To address these concerns,  a detailed evaluation of pulse-jet technology was made
including site visits to U.S. and Australian facilities and consideration was given
to on-going test results from pilot-scale HYPAS testing.  This evaluation indicated
that pulse-jet baghouses were a commercially viable option for particulate control.
Test results also indicate  that the baghouse could respond to any HYPAS process
upsets caused by problems in scale-up, with little impact on boiler operation.
The results of the engineering study are presented in Table 1.  The study showed
that the pulse-jet/HYPAS option would provide the best overall environmental
performance at a comparable capital cost to a new precipitator, and in fact at a
significantly lower overall life of project cost (including present worth cost of
capital, O&M, and outage costs).  Life of project costs are included only for the
particulate control  portion of the pulse-jet/HYPAS option because other options do
not include cost of any additional S02 control.  Precipitator upgrades were
projected to give only marginal improvement in performance (sufficient to assure
compliance with present regulations), but at a much lower cost. (Ref. 7)
The key to the favorable economics of the pulse-jet/HYPAS option was the small area
required for the pulse-jet  as compared to a new precipitator.  A pulse-jet with an
A/C cloth ratio of 4.0 (3.5 at cooler HYPAS temperature) will fit into the space of
an ESP with an SCA of about 275 ftVMACFM, whereas a comparable efficiency new ESP
would require an SCA of over 400 ftVMACFM.  The smaller size of the baghouse
allowed for construction over the existing ESP while the unit remained on-line.
The new ESP would require demolition of existing equipment and an extensive outage.
Based on these findings, Duke Power decided in August 1989 to proceed with a
proposal to DOE.  The cost  of the project would be $ 47 million, including the
$ 27 million capital cost and the operating, testing, and management costs of the
project.  The project would be eligible for 50 % cost sharing from DOE, and other
supporters including pulse-jet vendors, interested utilities, and EPRI would
further help to offset Duke Power's cost.  The project would involve the first
full-scale (400 MW)  installation of a pulse-jet baghouse on a utility boiler  in the
                                       17-8

-------
U.S., and would demonstrate the commercial viability of the HYPAS process by
parallel testing between the pulse-jet alone on one half of the unit and use of the
HYPAS process (including humidification and dry injection) on the other half.
Modifications to Unit 2 for the HYPAS project are illustrated in Figure 7.
In December 1989, DOE announced awards for Clean Coal Round 3 projects.  HYPAS was
not selected.  Duke Power was briefed by DOE in January 1990, identifying why the
project was not selected.  While HYPAS received high marks for technical merit, DOE
felt the proposal was weak in its commitment to commercial development following
successful demonstration and that project funding from outside sources needed to be
more clearly established.  Duke anticipates that these weaknesses can be addressed
by additional negotiations with vendors, EPRI, and other interested organizations.
Round 4 projects are to be solicited by DOE in June 1990, and Duke Power is
expecting to resubmit the HYPAS project pending successful negotiations.  A key to
a new submittal  will be developing a licensing package for marketing HYPAS.   The
licensing rights are owned by EPRI.  At least two pulse-jet baghouse vendors have
expressed interest in commercializing HYPAS.
Because there is no assurance that DOE will approve HYPAS in Round 4, Duke will
also plan to initiate first phase design of particulate-only control options.  This
design phase will provide detailed cost and engineering analysis sufficient to
issue purchase specifications for the selected option.  Phase I design will  be
limited to a choice between limited precipitator upgrades and installation of a
pulse-jet baghouse in series with the existing precipitator (suitable for future
retrofit to HYPAS).  A replacement precipitator will not be considered.
Precipitator upgrade is recognized as a limited option which may not be sufficient
for future needs, but may be justified if the costs are reasonably low and the
extent of future requirements is not clearly defined.
To support a possible decision to select a pulse-jet baghouse, Duke is also
considering installation of a pilot pulse-jet at Marshal", depending on
availability from vendors.   This will provide operating experience with the
technology and may help the selection of components including bag material,  cage
design,  etc.
                                       17-9

-------
SUMMARY
HYPAS is an important option to consider for older plants which face upgrade of
participate control  systems.  The advantages of HYPAS include:
     •    S02 control combined with improved particulate emissions and, most
          likely,  reductions in hazardous pollutants.
     •    Segregated waste/by-product handling.
     •    Dry processing of all materials,  as well as simplified process control.
     •    The pulse-jet baghouse minimizes  space and wil"1  have a minimum impact on
          station  operations (low pressure  drop and minimum maintenance).
The pulse-jet baghouse appears to be a good choice if particulate control  is the
only immediate requirement, and can still provide the option to retrofit HYPAS if
installed in series  with an existing precipitator.  The series arrangement may also
provide the benefit  of extending bag life (in the particulate collection mode)
because less frequent cleaning cycles will  be required.   Duke's analysis of
pulse-jet baghouses  shows the technology to be cost-competitive to high efficiency
precipitators.
If a pulse-jet is  to be constructed with the option to retrofit HYPAS,  the design
specifications should consider ideas to maximize residence time for sorbent to
react with lime and  to minimize potential for corrosion  at close approach  to
saturation temperature.   Further developments from EPRI's  pilot project will
provide the basis  for optimizing design.
                                     17-10

-------
                                   REFERENCES

1.    M.  D.  Durham,  R.  G.  Rhudy,  E.  J.  Puschaver.  "Low-Resistivity Related ESP
     Performance Problems in Dry Scrubbing Applications."  Journal of the Air and
     Waste  Management  Association,  January 1990, pp.  112   124.

2.    HYPAS  Pilot Plant Evaluation.   Interim Progress  Report No.  1, EPRI RP2934-1,
     Electric Power Technologies,  November 1988, pp.  3 - 12.

3.    HYPAS  Pilot Plant Evaluation.   Interim Progress  Report No.  8, EPRI RP2934-1,
     Electric Power Technologies,  October 1989,  pp.  7 - 13.

4.    HYPAS  Pilot Plant Evaluation.   Interim Progress  Report No.  9, EPRI RP2934-1,
     Electric Power Technologies,  November 1989, pp.  13 - 14.

5.    HYPAS  Pilot Plant Evaluation.   Interim Progress  Report No.  10,  EPRI  RP2934-1,
     Electric Power Technologies,  December 1989.

6.    D.  V.  Giovanni, R.  G.  Rhudy,  J.  J.  McCarthy,  R.  C. Giles.   "Technical  and
     Economic Evaluation  of HYPAS."  Paper No.  8-3 Presented  at  the  llth  EPRI/EPA
     Symposium on Flue Gas  Desulfurization, Kansas City, Mo.,  October 28,  1988.

7.    K.  S.  Johnson.  Duke Power  Company  Marshall Steam Station  Particulate Control
     Study.   Internal  Report,  August 1989, pp.  11  -  12.
                                      17-11

-------
a.
o

u
o
                 ANNUAL AVERAGE OPACITY LIMIT
20 -
         15 -
         10 -
         5 -
                          20
                                     40          60



                               DAILY OUTPUT FACTOR (%)
                                                                       100
      Figure  1.   Daily Average Output  Factor Versus Opacity.

                  Marshall  Unit 2, February 1,  1989 - March  5,  1990.
 CO
 o:
 o
100





 90




 80




 70





 60





 50





 40





 30





 20




 10




  0
                                         CAPACITY FACTOR
              1989   1990   1991    1992   1993    1994   1993   1996   1997   1998
      Figure 2.  Projected Operating Factors For Marshall Unit  2.
                                    17-12

-------
u
a
x
o
M
       300
       4OO
       300
       200
       100
                                                         NO NEW REGULATION
                                                          •85-'87 BASEUNE
             1980
                       1985
                                  1990
                                            1995
                                                      20OO
                                                                2005
     Figure 3.  Projected Duke Power  Systemwide S02 Emissions,

                Considering Proposed  Acid Rain Legislation.
          POWERHOUSE
                               PRECIPITATOR



                       MECH COLLECTOR
                                                         INTAKE STRUCTURE
                                               YARD EL 77T-6'
                                      CCW SUPPLY
    Figure 4.  Layout  of Marshall Unit  2 from Boiler  to  Stack,

                Showing Relation to  Existing Water  Intake.
                                  17-13

-------
I LUE GHS
IHOH BOILER
  TO FLY FISH
  DISPOSflL
                                                       1.0. FRN
                                                 DRY
                                                 SORBEfU
                                                 INJECTION
                                          PUMP
 Figure  5.  Conceptual  Arrangement Showing Components
             of  a HYPAS  Retrofit.
                          3.5 A/C. TOP 4: BOTTOM PJ INLET
    80
    70 -
    65 -
    55 -
    45 -
     35
       200
                             O Co/S=2, 20 DEC APP. ONCE-THROUGH
                             t Ca/S=2. 10 DEC APP, ONCE-THROUGH
                             O Co/S = 2, 20 DEC APP. 1:1  RECYCLE
                             A Ca/S = 2, 10 DEC APP, 1:1  RECYCLE
                             X Co/S=1, 20 DEC APP, ONCE-THROUGH
                             V Ca/S=1, 10 DEC APP, ONCE-THROUGH
                —T	
                 600
                          1000
                                   1400
                                            1800
                                                      2200
                                                               2600
                       SYSTEM INLET S02 CONCENTRATION (PPM) .
 Figure  6.
S02 Removal  Efficiency Versus Inlet  S02 Concentration From
Pilot-Scale  Testing  at the  EPRI High Sulfur  Test  Center.
(Source:   Reference   5,  Interim Progress  Report No.  10)
                                17-14

-------
         <1
     Boiler
              \/
                           Air     Mech.   V V
                          Heater   Collector
                                 (Gutted)
                                                                     _ Stack
                                                                      Liner
Figure 7.   Proposed  Modifications to Marshall Unit  2  for
            Full-Scale  Demonstration of HYPAS Retrofit.
                                 17-15

-------
                           Table 1
              Marshall  Unit 2 Engineering Study
TECHNOLOGY
INSTALLED  SYSTEM     PARTICULATE    AVERAGE
  COST      COST*     PERFORMANCE    OPACITY
(mi 11 ions)(mi 11 ions)    (Ib/mBtu)        (%)
ESP Upgrade

New ESP
  $2.9      $6.9
  17.7
41.4
             < 0.10
< 0.05
               20
< 10
New PJFF
  22.2
31.4
< 0.03
<  5
HYPAS
  27.7
               0.03
             <  5
*  30 year life of project,  1989 dollars
                             17-16

-------
     INFLUENCE OF  A SOCK BETWEEN SUPPORTING CAGE AND  BAG
                      ON FILTER PERFORMANCE

                             E.  Schmidt
                             F.  Loffler
                     Universitat Karlsruhe (TH)
       Institut  fur  Mechanische  Verfahrenstechnik und Mechanik
            Postfach 6980,  D-7500 Karlsruhe, West Germany
ABSTRACT

In spite of the high collection efficiency which pulse-jet filters
normally achieve,  intensive research is still in progress with the
aim of improving their operational characteristics.  This presenta-
tion introduces the results of investigations into the effect  of a
protective sock (Teflon®,  with  1  cm mesh) which  is  fitted over  the
supporting cage under the filter bag.  Special emphasis is given to
the filter medium's motion during the  pulse-jet cleaning phase (de-
termined with the aid of high-speed cinematography and acceleration
transducers),  the particle penetration during,  and shortly after the
bag's regeneration (light scattering size analysis)  and the  bag's
residual dust mass.
                                 18-1

-------
     INFLUENCE OF  A SOCK  BETWEEN SUPPORTING  CAGE AND BAG
                      ON FILTER PERFORMANCE
INTRODUCTION

Whilst pulse-jet bag filters have attained an indisputable market
relevance,  such systems still incorporate a few undesirable weak-
nesses which offer scope for improvement. One such item is a protec-
tive Teflon®  sock, developed by Du Pont,  which  is  pulled  over  the
supporting cage, underneath the filter bag. Its purpose is to in-
crease the bag's service life by damping its movement during the
cleaning cycle and preventing it from striking the rigid cage under-
neath. As a result, one could additionally cut financial investments
by being able to operate with lighter filter media and with support-
ing cages fabricated from less abrasion-resistant materials.

This publication is aimed at introducing the reader to the results
of investigations into the influence of such protective socks on the
operational characteristics of a pilot filter unit, in view of the
particle collection,  the pressure drop and regeneration proficiency.
EXPERIMENTAL  PROCEDURE

The investigations were conducted on the pilot-scale pulse-jet fil-
ter apparatus,  schematically illustrated in Fig. 1  1|. The test
dust,  consisting of limestone particles with a mass distribution me-
dian of 4 .5 |im,  was injected to the inlet duct by a dust disperser.
The particle-laden gas (concentration CQ = 5 g/m3)  entered the test
filter at the bottom of the housing and was sucked through the sin-
gle bag (Tefaire®  needled  felt, 2.4 m  long,  0.115  m in diameter)  by
a rotary compressor. The filtered gas left the apparatus at the top
                                 18-2

-------
via a gauge chamber. All experiments were conducted at a  constant
filtration rate of  120 m3/(m2-h) .  The pressure drop Ap across the
bag was continually metered. After attaining a specific level of
Apmax = 1500  Pa,  the bag was cleaned by a compressed air pulse.

In order to determine the cleaning conditions inside the  filter bag,
fast differential pressure transducers  (response time < 0.1 ms) were
mounted at four locations along the bag  (M-j_ - M4) .  The medium's ac-
celeration during the cleaning was measured with two miniature ac-
celerometers  (M-]_, M2) •  The in-situ measurement of the bag's dust
loading before and  after regeneration was accomplished by a special-
ly developed method. This involves a unit which basically allows a
radioactive source  (Am241-source with Zr-target)  to be traversed
along the inside of the bag, together with three externally mounted
scintillation counters. The mass of dust which was retained in and
upon the filter medium could then be determined by means  of the X-
ray absorption.

In order to investigate the influence exerted by such a modification
on the operational behaviour, 60 filtration cycles were conducted
with and without the sock, whereby the chief characteristic parame-
ters were recorded. The results derived are discussed in  the follow-
ing sections .
RESULTS

Cleaning  efficiency

The cleaning conditions were investigated by measuring the differen-
tial pressure established across the bag during each cleaning cycle
at four different bag heights. Whilst the complete range of measure-
ments needs not to be discussed, attention is to be focussed to the
maximum excess pressures attained during each cleaning pulse for the
initial and final cycles as illustrated in Fig. 2. As earlier publi-
cations have already verified, these values do indeed mirror the
existing bag regeneration conditions |2|.
                                 18-3

-------
One can observe that with the exception of the uppermost bag loca-
tion,  the pressure maxima of the last cleaning pulses  (58-60) are
clearly higher than those of the initial cycles  (1-6), both with and
without the sock. This is caused by the substantially  larger quanti-
ty of residual dust which accumulates on, and in the bag. From these
pressure measurements, one can deduce that as far as the regenera-
tion is concerned, the use of the supporting sock does not substan-
tially influence the usual behaviour, so that normal cleaning re-
sults can be anticipated.

The regeneration effectivity can either be characterized by the fil-
ter bag's pressure drop at the end of the cleaning cycle or by the
post-cleaning areal dust mass still remaining on or in the medium.
Fig. 3 illustrates the residual bag pressure drop, measured after
each cleaning phase. As expected, no significant difference exists
between the conventional and the sock-equipped operation (the dif-
ferences between the 10th and 20th cycle  was  caused by  a  fault  in
the volume flow regulation when operating without the  sock) . It
should be noted, however, that in both cases, the pressure drop has
not reached a steady-state condition.

Fig. 4 illustrates the so-called residual areal dust mass after the
respective 1st and GO1-*1  cleaning  cycle. In accordance with  the  high-
er pressure drop, these profiles demonstrate that the  residual areal
dust mass which remained after the final cleaning cycle is signifi-
cantly larger than after the first.  Nevertheless, the  average value
is approximately the same, both with and without the sock.  When
using the sock, however, the bag's residual dust would appear to be
more uniformly distributed.  This tendency becomes more pronounced
with increasing number of cleaning cycles.

It may therefore be stated that the regeneration of the filter bag
is not significantly modified when the cage is covered with a sup-
porting sock.
                                 18-4

-------
Particle  penetration

The pilot unit's collection efficiency was assessed with the aid of
a  light  scattering size analyzer. The particle penetration through
the filter bag during and shortly after the cleaning cycle is repre-
sented in Fig. 5, which compares the counted number of particles per
unit period as temporal functions for the two different systems. For
this purpose, the values measured during the cycles 1-6 and 58-60
were respectively averaged, and plotted using a logarithmic ordi-
nate.

The fact that the use of the sock reduces the tendency for the dust
to penetrate now becomes evident. All experiments possessed the com-
mon feature that the pressure pulses induced high penetration rates
during the first 10 seconds, which decreased, however, quite rapidly
as the subsequent filtration cycle commenced. Initially, the clean
gas particle concentrations were smaller during the first 6 cycles
than in the last three, this situation being reversed after approxi-
mately 20 seconds. The inversion is clearly due to the larger resid-
ual areal dust mass and the associated faster transition from the
bridging phase to the cake filtration phase.

This residual dust mass is also responsible for the high penetration
rates which occur at the start of each new filtration cycle.  This
becomes evident upon regarding the basic motion of the filter cloth
during the on-line cleaning pulse (see fig. 6).  Whilst the filtra-
tion cycle is still in progress (and hence directly before intro-
ducing the cleaning pulse),  the filter fabric is sucked around the
supporting cage.  Upon injecting the compressed-air,  the bag rapidly
inflates, causing the medium to accelerate away from the cage until
its full volume is attained. A filter element which happens to be
situated in-between the cage wires is therefore initially slowly ac-
celerated,  but abruptly checked when the bag can expand no further.
At this point,  the filter cake is usually thrown off the medium.

After a short residence time (which depends upon the injection peri-
od)  this filter element is  then accelerated back to the cage by the
uninterrupted filtration gas stream.  Upon striking the cage,  the
                                 18-5

-------
filter fabric is obviously again rapidly decelerated, causing parti-
cles which still remain on (or in)  the medium to be ejected into the
clean gas stream to an extent which depends on the residual areal
dust mass ("carpet beating effect"),  hence inducing the comparative-
ly high penetration rates which follow each cleaning cycle.

The mass-specific penetration functions plotted in Fig. 7 again
clearly demonstrate that the use of a protective sock can reduce the
cleaning-induced particle penetration by a factor of three. These
measurements have verified moreover,  that up to 80 % of the total
dust emitted by a pulse-jet bag filter unit emerges as a result of
the bag regeneration. Since the incorporation of such a sock can
reduce this share by a factor of three, the total filtration cycle-
specific average clean gas particle concentration may be reduced by
half. In view of the continual environmental law intensifications,
this indeed presents a significant  advantage.

In view of the cause, this improvement is most probably derived from
the damping effect which such a sock provides as the bag re-ap-
proaches the cage. In order to verify the damping effect yielded by
the sock  (which is tightly stretched over the supporting cage), the
effective filter medium acceleration was measured for a number of
cycles. Two typical signals,  measured with and without the support-
ing sock as the filter fabric struck the supporting cage may be
viewed in Fig. 8  (the compressed-air pulse being introduced at
t = 0 ms) .

One can clearly observe that the sock reduces the maximal accelera-
tion, hence allowing the filter bag to settle in a damped fashion.
The cumulative number distributions and the number distribution den-
sities derived from all measured accelerations may be seen in Fig.
9. Here, it is evident that the use of the sock compresses the dis-
tribution and shifts the accelerations towards smaller values.
Hence, the clean-gas particle concentration reduction does indeed
arise from the damped bag movement.

Another method which also serves to dampen the bag's momentum was
presented by  |3 .  Here, the particle penetration through a pulse-jet
                                 18-6

-------
filter is minimized by gradually reducing the air pressure towards
the end of the cleaning pulse. This allows the inflated bag to re-
turn to its rigid supporting cage without such force, hence avoiding
dust which is still trapped by the filter medium from being driven
into the clean-gas stream. Such modified pulses reduced penetration
by up to 46 %, had no effect on the bag's pressure drop, but in-
creased the compressed-air consumption by more than 27 %.
Summary

The influence exerted by a protective, wide-meshed sock mounted over
the cage, and beneath the filter bag, can be summarized as follows:

Under the described filtration conditions, the total emission caused
by particle penetration during, and shortly after bag regeneration
may be reduced by a factor of three. This is accomplished by damping
the bag-cage-interaction. No negative influence could be determined
with regards to either the filter medium's residual pressure drop or
the residual areal dust mass.

Although the bag's life is presumably also extended,  this must be
verified by further investigations.
REFERENCES

1   R.  Klingel.  Untersuchung der Partikelabscheidung aus Gasen an
    einem Schlauchf ilter mit Druckstofiabreinigung.  Diisseldorf:  VDI-
    Verlag 1983
2   J.  Sievert.  Physikalische Vorgange  bei  der Regenerierung des
    Filtermediums  in  Schlauchfiltern mit  Druckstofiabreinigung.
    Dusseldorf:  VDI-Verlag,  1988.
3   D.  Leith, M. W. First,  D.  D.  Gibson.  "Effect  of Modified
    Cleaning  Pulses on  Pulse Jet  Filter Performance." Filtration &
    Separation,  15  (1978),  pp.  400-406
                                 18-7

-------
•£
                  5   Y   7  Ax
 1  dust  feeder
 2  filter  bag
 3  dust  hopper
 4  cleaning  system
    (pulse-jet)
 5  outlet  channel
 6  fan
 7  flow  rate control  device
 8  particle  counter
 9  device  to measure  areal
    density of dust deposit
pressure  transducers  (M-L - M4)
accelerometers (M^  -  M2)
         Figure 1.  Schematic of the pilot-scale  test  filter
          to
          0)
            1500
             1000
              500
                0
                          pressures averaged
                          over the given cycles
                    *  with  sock
         without  sock
                0,0    0,4    0,8     1,2    1,6    2,0    2,4
                              bag height / m

          Figure 2.  Maximum excess pressure within the bag
            during the cleaning pulse  (Oms foot  of  bag)
                                 18-8

-------
           a
           o
           0)
           M
           3
           CO
           CO
           0)
           M
           ft
           fO
           P
           T)
           -rH
           CO
           
-------
  CO
  >
co M
CD 0)
rH 4J
O G
•rH -rH
4->
S-l W
fO
£1,0
M a
0) -H
•9 ^
B -P
3 —I
C 3
     100000 F
10000 r
 1000 r
        100 ^
 Figure 5. Dust  penetration during and shortly after

                  the  cleaning pulse
               dust cake
                .xxxxxxxx,.
                                      I
  Figure  6.  Bag movement during the cleaning phase
                          18-10

-------
               10
          c
          o
          -H
          4-1
          0)
          G
          0>
          ft
               ,1
   ,01
             ,001
                                        cycles 58 -  60
without sock

  with  sock
                              10            20

                                  time  /  s
                                             30
Figure 7.  Specific penetration during and  shortly  after regeneration
      CO

      "e
      a
      o
      -H
      4->
      id
      M
      0)
      H
      0)
      U
      O
          2500
          2000
          1500
           1000
500
          -500
         -1000
                    A
                 without sock
                 .   .  .   <  i
                                               A
                                         with sock
              250
              300       350/250

            time / ms
                          300          350


                       time / ms
         Figure  8.  Typical acceleration transducer signals
                                 18-11

-------
      100
  c
  o
  -H
  -P
  -H
  M
  -P
0) OT
> -H
-H T3
-P
(C M
r-t 0)

3 "§
P 3
O C
                   with   sock

                 without  sock
              50
100
 150    200    250   300

acceleration  / g
350
400
      1,0
                                        with  sock

                                   °  without sock
              50    100   150    200   250    300    350    400

                         acceleration  / g
Figure  9. Measured acceleration distributions  (g = 10 m/s2)
                              18-12

-------
                          ACCELERATED BAG WEAR TESTING
                                 Larry G. Felix
                                Robert F. Heaphy
                           Southern Research Institute
                             2000 Ninth Avenue South
                              Birmingham, AL 35205
                                 Ralph F. Altman
                                 Ramsay L. Chang
                        Electric Power Research Institute
                              3412 Hillview Avenue
                               Palo Alto, CA 10412
                                 W. Theron Grubb
                   Grubb Filtration and Testing Services, Inc.
                              8006 Route 130 North
                                Delran, NJ 08075
ABSTRACT
A year-long accelerated bag wear test was recently concluded at compartment 1 of
the EPRI High-Sulfur Fabric Filter Pilot Plant (HSFP) located at the Gulf Power
Scholz station in Sneads, FL.  The purpose of this test was to subject a variety
of filter bags to operating conditions that are known to reduce their service
life.  These conditions included frequent cleanings (six periods of shake/deflate
cleaning per hour) and cyclic operation (daily passage through water and acid dew
points).  Most of the test bags in this compartment filtered flue gas.  However,
some test bags were attached to capped thimbles, to isolate the mechanical effects
of shake/deflate cleaning from those of filtration, and other test bags were hung
in the compartment so that they were exposed only to filtered flue gas.  Various
fiberglass fabrics and finishes (including unfinished fiberglass fabric) were
tested.  The test was concluded after 6181 h of operation with 29,548
shake/deflate cleanings (equivalent to 5 years of operation with a 90-min
filtering period) and 112 compartment isolations.  None of the test bags failed
during the test.   Tests carried out on samples of the test fabrics removed at
various times during testing, as well as tests carried out after the test was
completed, show that fabric strength was degraded most in areas where the bags
were exposed to the coolest temperatures (near doors and exterior compartment
walls), where bags had suffered internal abrasion, and at random drip stains (from
condensation of water and acid).

                                      19-1

-------
                          ACCELERATED BAG WEAR TESTING
INTRODUCTION
Expenditures to replace bags are usually the most significant maintenance cost
associated with the operation of a baghouse.  For a fourteen-compartment utility
baghouse with 250 bags in each compartment (a 250 MW boiler with a baghouse
operating at an A/C ratio of 2.1 acfm/ft2),  rebagging would cost $350,000 assuming
that bags could be purchased and installed for $100 each.  Thus, significant
savings are possible if bag replacement can  be delayed through extended bag life.
Thus, bag life is a topic of interest for baghouse research.

At the pilot scale, bag life remains a difficult fabric filter issue to address
because of the long time intervals involved.  Three-year bag life is commonly
assumed when comparing levelized costs for utility low-ratio baghouses and
electrostatic precipitators, but it is not cost effective to dedicate a pilot
facility to three years of operation solely  to study bag life.  The economic
inducements associated with bag life extension, in conjunction with the limita-
tions of pilot-scale research, led to the development of a method to accelerate
normal bag wear.  This method was tested at  the EPRI High Sulfur Fabric Filter
Pilot Plant (HSFP) (1,2).

This attempt to address the issue of bag life was both an effort to develop an
accelerated bag-wear testing method within the economic constraints of pilot-scale
operation and an attempt to quantify the durability of the fabrics that were
tested.  The test plan was based on the presumption that bag fabric is degraded by
both mechanical and chemical means.  Mechanical stresses are imparted to bag
fabric when bags are cleaned.  Thus, mechanical wear could be accelerated by
frequent cleaning.  Chemical degradation can occur when bag fabric passes through
acid or water dew points on startup and shutdown (3).  Thus, chemical attack could
be accelerated by frequent outages.  Finally, because it is probable that differ-
ent fabrics and finishes can respond in different ways to the effects of mechani-
cal wear and chemical attack, it was decided that a variety of fabrics and
finishes should be included in this test.
                                      19-2

-------
ACCELERATED BAG-WEAR TEST
Shake/deflate (S/D) cleaning is the most energetic bag cleaning method currently
used in utility baghouses.  Therefore, in the accelerated bag-wear test S/D
cleaning cycles were used at 10-minute intervals to accelerate fabric wear.
Chemical stress was imparted by consecutive overnight outages in which the bags
were first exposed to condensed sulfuric acid (by taking the test compartment off-
line without purging and letting the closed compartment cool to below the sulfuric
acid dew point) and later, to condensed water and acids (by purging and leaving
the compartment doors open so that the bags cooled to below the water dew point
before the compartment was brought back on line).

At the HSFP both hydrochloric and sulfuric acid are present in the flue gas.   The
dew point of hydrochloric acid vapor (120-150 ppm) is typically between 100 and
150°F and the dew point for sulfuric acid vapor (10-13 ppm) is typically from 250
to 260°F.  During startup , one to two hours are required for the temperature in a
compartment to rise from near ambient to above the sulfuric acid dew point.

Forty-two thimbles were available in the test compartment.   Thus, in order to test
as many bags as possible of each type of test fabric at different levels of
stress, only ten different types of fabrics and finishes were tested.  The details
of the construction of the test fabrics are provided in table 1.  Throughout  the
description of the accelerated bag-wear test, the test fabrics will be referred to
by the identification numbers shown in table 1 (bag types 1-1 through 1-10).   All
of the test fabrics were made of 100% DE-size filament E-type fiberglass except
for one special fabric.  The test fabrics are briefly described below.

Fabrics 1-1 and 1-2 were the "reference fabrics" for this test.  These two
fiberglass fabrics were actually the same material, a 54 x 30 count, 3x1 twill
weave with a 10% TEFLON B™ finish.  However, bags made of fabric 1-1 were made
with the warp side of the fabric on the interior of the bag (warp-in or 25%
exposed surface texturization, EST) and the fabric 1-2 bags were made with the
warp side of the fabric on the exterior of the bag (warp-out or 75% EST).  Test
fabrics 1-3 through 1-5 were the same as fabric 1-1 except that they had different
finishes.  Fabric 1-3 had no temperature or acid flue-gas-resistant finish (greige
or "loom-state" fabric).  Fabrics 1-4 and 1-5 had 7% TEFLON B and BGF 1-625
finishes, respectively.  The 7% TEFLON B finish was included because it was of
interest to determine if lesser amounts of the TEFLON B finish could provide
adequate protection for the fabric in a high-sulfur coal flue gas environment.
BGF 1-625 is a "standard" acid-resistant finish.  Fabric 1-6 was a 0% EST version

                                      19-3

-------
of fabric 1-1 with an 8% TEFLON B finish, fabric 1-7 was a singles warp-singles
reinforced fill  fabric with a 10% TEFLON B finish, and fabric 1-8 was a  14 oz/yd2
version of the reference fabric.  Fabric 1-9 was similar to fabric 1-8 except it
was woven with a special warp yarn that combined Polytetrafluoroethylene  (PTFE)
with fiberglass.  Fabric 1-10 was standard GORE-TEX™ membrane filtration fabric.

The test bags were subjected to three levels of stress.  The primary group
consisted of thirty bags; three bags of each of the ten fabric types.  These bags,
designated as filtering and shaking bags (F/S), filtered flue gas and were cleaned
every ten minutes.  The second group of bags, designated as non-filtering and
shaking bags (NF/S), were attached to capped thimbles so that the bags did not
filter flue gas but were shaken each time that the compartment was cleaned.
Fabric 1-9 was not included in the NF/S group because only a limited number of
these bags were available.  The third group of bags, designated as non-filtering
and non-shaking bags (NF/NS), were hung under a catwalk at the bag suspension
level of the compartment so that they were exposed to clean flue gas but were not
shaken.  Fabrics 1-2, 1-9, and 1-10, were not included in this group because of
space limitations within the compartment.  Figure 1 shows the location of each
test bag within the compartment, the area of the tubesheet containing each of the
three test groups, and which bags were removed for strength testing.  This test
matrix, in conjunction with detailed failure analysis and standard fabric strength
tests, was intended to help define failure mechanisms and to provide a relative
estimate of the useful filtering life for each of these fabrics in a high-sulfur
coal flue-gas environment.

Operations and Testing
Compartment 1 of the HSFP was dedicated to this test and used S/D cleaning, while
the other compartments in the HSFP were operated with sonic-assisted reverse-gas
cleaning.  The accelerated bag life test began on February 3, 1988 and continued
until December 7, 1989.  In this period the bags in compartment 1 accumulated
6181 h of operation at an air to cloth ratio of 2.5 acfm/ft2.  Throughout this
test the compartment was operated with a 90-min filtering time and a 5 s  shake
period at a frequency of 4 Hz (1.7 g acceleration).

For the first month of testing, the compartment was operated without an  acceler-
ated cleaning schedule and without regular compartment isolations to allow resid-
ual dustcakes to form; analysis of bag failures that occurred during earlier
testing of S/D cleaning at the HSFP indicated that the residual dustcake  provides
                                      19-4

-------
abrasion protection to the internal surface of the bag during cleaning (3).  Based
on the earlier evaluation of S/D cleaning at the HSFP, the deflation flow was set
to 1300 acfm for 20 s (0.4 acfm/ft2) during this break-in period.

At the end of the first month of operation (662 h) the accelerated cleaning
schedule commenced (six cleaning cycles per hour).  At this time the deflation
flow was reduced to 500 acfm for 8 s (0.06 acfm/ft2).  Deflation was set visually
by observing the bags during deflation through a view port in the access door in
the compartment roof.  The volume of deflation gas was reduced until the bags made
of greige fabric relaxed without flattening.  The supple greige-fabric bags tended
to flatten more than the other test fabrics.

On May 16, after 2055 h of operation (8444 cleaning cycles),  the accelerated
cleaning schedule was augmented with weekly compartment isolations.  Purging was
omitted from the isolation procedure to allow flue gas vapors to condense on the
dustcakes and bags to deliberately induce acid attack.  The compartment was
isolated after midnight and remained off-line until about eight a.m. the next
morning.  After remaining off-line for eight hours compartment temperature
typically was between 200°F and 220°F.  The sulfuric acid dew point at the HSFP
ranges from 250°F to 260°F.

The weekly compartment isolations without purging continued until September 7 when
the compartment had accumulated 4690 h of operation (21,561 cleaning cycles).  At
that time, daily isolations were started and lasted until the compartment had been
isolated without purging for 59 times.

On October 26, after 5500 h of operation (25,827 cleaning cycles), purging and
cooling to ambient temperature was added to the shutdown procedure.  From then on,
the compartment was isolated, purged, and cooled (by opening the doors) every
afternoon until testing ended on December 7 after 6181 h of operation  (29,548
cleaning cycles) and 54 additional isolations.  No bags failed through the entire
test period.

Residual dustcake weights for the ten test fabrics ranged from 0.25 to 0.39 lb/ft2
shortly after the deliberate compartment isolations began, from 0.31 to
0.62 lb/ft2  when compartments were isolated without purging,  and from  0.43 to
0.81 lb/ft2  when testing ended after a period when the compartment was isolated,
purged and cooled each night.  Table 2 gives average residual dustcake areal
                                      19-5

-------
loading for each of the ten fabrics included in the test at the conclusion of each
of the three operating periods.

It was not possible to measure the drag for each type of test fabric.  However,
the average drag in compartment 1 rose to 1.8 in. H20 min/ft by the end of the
initial period of operation with an accelerated cleaning cycle, increased 18% to
2.1 in. H20 min/ft while the compartment was isolated without purging, and
decreased 19% to 1.7 in. H20 min/ft when purging and cooling was added to the
shut-down procedure.  Reductions in average drag associated with cyclic operation
and purging have been observed before at the HSFP (1,1).

In view of the accelerated cleaning frequency,  numerous deliberate compartment
isolations, and minimal deflation, it is interesting to note that fabric filter
performance remained representative of a high-sulfur coal  baghouse throughout the
test.

Fabric 1-2 is essentially the same fabric used  during an earlier evaluation of S/D
cleaning at the HSFP (2).  A compartment of this same fabric was also tested
during a previous evaluation of the effect of fabric EST on baghouse performance
(2).  At the conclusion of the evaluation of S/D cleaning, the residual dustcake
areal  density and average compartment drag for  this fabric was 0.36 1b/ft2 and
2.0 in. H20 min/ft, respectively.  These values are close to the residual dustcake
areal  density and average compartment drag measured just before daily compartment
isolations began during the accelerated bag-wear test (0.39 lb/ft2 and
1.8 in. H20 min/ft, respectively).  At the end  of the earlier fabric texturization
tests  (when daily compartment isolations were used to accelerate residual dustcake
growth   with sonic-assisted reverse-gas cleaning), residual dustcake areal
density averaged 0.77 lb/ft2.  During the accelerated bag-wear test when a similar
compartment isolation procedure was used, the residual dustcake areal density of
fabric 1-2 was measured to be 0.81 lb/ft2.

RESULTS
The most surprising result was the absence of bag failures  (each test bag was
subjected to about 591,000 shakes).  Shake/deflate cleaning was chosen for this
test because it is widely believed that the cumulative effects of shaking directly
reduce bag life.  While this assumption may be  valid, the absence of bag  failures
was significant because it suggests the common  perception that fiberglass fabric
is too delicate to withstand a great deal of shaking  is incorrect.
                                      19-6

-------
Such durability is not unique.  Bag lives of over 7 years with S/D cleaning have
been reported at the Southwest Public Service Company's Harrington station Unit 3
baghouse (4).  The Harrington bag life is even more remarkable because this
baghouse uses a bag-cap acceleration that has been directly measured at 2.2 g,
0.5 g higher than the bag-cap acceleration used in compartment 1 and because the
duration of shaking was 10 s during most of the 7-yr operating period (4).
Clearly, when bag failures occur in a S/D cleaned baghouse, they are not
necessarily caused by the mechanical stresses of shaking.

Fabric Strength
Since no bag failures occurred, the task of resolving the issues that the test was
designed to address became more difficult.  In the absence of bag failures, fabric
strength testing became the only means of comparing the test fabrics.  Individual
yarn tensile strength tests were not performed because of the large number of
samples and the relatively high cost per sample.  Therefore, fabric strength was
measured by subjecting each fabric sample to Mullen burst testing.  Although the
Mullen burst test was originally developed for felted fabrics, and tends to
measure the bursting strength of texturized fill yarns (which are weaker than non-
texturized warp yarns), it is a better measure of overall fabric strength than the
other common fabric strength test (MIT flex) (5).  These tests were carried out by
Grubb Filtration and Testing Services, Inc. and were performed according to ASTM
Standard Test Method D3786 with clamping platens modified according to ASTM
Standard Test Method D3656, Section 17.1, for glass fabrics.  Care was taken to
avoid testing fabric that had been folded, creased, or obviously damaged during
removal and handling.

Samples of the NF/NS bags were removed on four occasions: before beginning weekly
compartment isolations without purging, before beginning daily isolations without
purging, before purging was added to the shut-down procedure, and at the conclu-
sion of testing.  Additionally, one of each of the F/S bags was removed after
3875 h of operation, immediately before starting daily compartment isolations
without purging.  Samples of all three sets of bags, NF/NS, NF/S, and F/S were
removed for strength testing when the accelerated bag life test ended.  Fabric
samples were taken from the middle of test bags to avoid the effects of "abnormal"
operating conditions (drip stains at the top, dust abrasion at the bottom, expo-
sure to cold external walls, and direct exposure to cold purge air).  The results
of these tests are summarized in table 3.  The locations of the bags removed for
testing are shown in figure 1.
                                       19-7

-------
The fabric strength tests revealed some unexpected differences among the three
groups of bags.   With a single exception, bags in the F/S group retained more
strength than the bags in the NF/S group or the bags in the NF/NS group.  This
exception was fabric type 1-6 (0% EST) which lost nearly identical fractions of
its original  strength in all  three groups of bags.  This observation further
indicates that shaking was not the primary cause of fabric strength loss.

Fabric Strength for F/S Bags.  The data shown in table 3 indicate that all of the
test fabrics lost less than 35% of their strength (relative to new fabric) in the
middle sections of the F/S bags.  The finished fabrics with 100% texturized
fiberglass filling yarns exhibited from 13% to 23% strength loss, and the
''reference-type" (10 oz/yd2)  finished fabrics had very comparable Mullen burst
values (from 420 psi to 509 psi).  The greige reference fabric, type 1-3,
exhibited a somewhat lower strength (359 psi) and higher strength loss (29%).
Test fabric 1-6, the all filament fabric, had the highest strength loss  (35%) of
any middle section sample of a F/S bag.  The 14 oz/yd2 fabric had an initial
strength that was roughly 70% greater than the 10 oz/yd2 fabrics, and although
this fabric lost 23% of its strength by the end of the test, it still was about
50% stronger than the 10 oz/yd2 fabrics.  Test fabric 1-9, with the glass/PTFE
filling yarn, also had a slightly higher strength loss (28%), but the failure mode
observed in the Mullen burst testing, splits across the warp yarns, indicated that
the strength of the warp yarns was actually determining the burst strength of the
fabric.

Comparison of middle section strength values after 6181 h of service to  those
measured after 3875 h of service, as shown in table 3, show little change.  Five
bag types changed less than 10%.  Three of the four bag types (test fabrics 1-3,
1-9, and 1-10) which exhibited greater than 10% loss from 3875 to 6181 h of
service were located near the compartment access door (see figure 1) and received
more direct exposure to ambient air during purged isolations.  The bag made of
test fabric 1-6, which had the highest strength loss from 3875 to 6181 h of
service, was more centrally located in the compartment.  The reference fabric bag
showed an apparent strength increase (+5%) during this period, yielding  the
highest strength of any 10 oz/yd2 fabric and the lowest strength loss of any
fabric after 6181 h of service.  Because of this unusual result the bag  section
was retested, yielding an average of 508 psi compared to the original result of
509 psi.  This fabric 1-1 bag was also located near the compartment door,  (see
figure 1).
                                       19-8

-------
Fabric Strength for NF/S Bags.  Excluding the all-filament glass fabric  (1-6), the
NF/S bags exhibited significantly lower strength in their middle sections  than the
F/S bags of the corresponding fabric type.  Strength losses of 23% to 47%  (from
new fabric strength), which is 9% to 23% additional compared to the strength  loss
of the F/S bags, were observed.  However, to make a valid comparison, it is neces-
sary to divide the NF/S bags into three categories based on their position within
the compartment.

NF/S bag types 1-1 and 1-2 were located along the common wall between compartments
1 and 2 of the HSFP.  These two bag types exhibited the lowest strength losses
versus new fabric (23% and 31%, respectively) and versus F/S bags (9% and  10%
additional strength loss, respectively) for all of the test fabrics and had the
highest strength values (452 psi and 402 psi, respectively) among the 10 oz/yd2
fabrics.

NF/S bag type 1-3 (greige fabric) was located along the same inner compartment
wall as were the bag types 1-1 and 1-2 mentioned above, but was in the corner at
the intersection of the "front" wall of the compartment.  This bag had the highest
overall strength loss (47%) and lowest strength (268 psi) of any NF/S bag.   It
cannot be determined conclusively whether this was solely due to the absence of
finish or whether it was located near a "cold" compartment wall (or both).

NF/S bags 1-4 to 1-8 and 1-10 were located along the "end" wall of the compartment
(see figure 1).  All of these bags exhibited comparable strength loss (compared to
new fabric) ranging from 35% to 45%.  Also, except for fabric 1-6,  all of these
test fabrics had strength losses (compared to F/S bags) of an additional 15% to
23%.  It is quite probable that bag position contributed to some general strength
reduction compared to other bags not adjacent to the end wall.  No NF/S type 1-9
bags were installed.

Fabric Strength for NF/NS Bags.  One bag each of types 1-1 and 1-3 through 1-8 was
hung in the compartment aisle in consecutive order from back to front (see figure
1).   Samples from these bags were taken after 6181 h of service, from 8 to 10 ft
above the original  bottoms of these bags.  This sample location is comparable to
the  middle sections of the F/S and NF/S bags.  Earlier samples were taken  from
lower sections of these bags.

All  of the NF/NS fabric strength data, including intermediate results after 1546,
3875,  and 5470 h of service are shown in table 3.  These data show that during the
                                      19-9

-------
last 711 h of operation (54 compartment isolations with cooling to ambient) the
finished,  texturized glass fabrics in the NF/NS bags suffered an additional 10% to
19% strength loss (relative to new fabric).  During this time the greige fabric
(type 1-3) lost only an additional 5% strength, and the all-filament fabric (type
1-6) exhibited an apparent strength gain of 13%.

After 6181 h of service,  fabric strength loss (relative to new fabric) for the
NF/NS bags were generally similar to the NF/S bags, ranging from 33% to 52%, with
the greige fabric (1-3) again exhibiting the lowest strength (242 psi) and highest
strength loss (52%).  Bag location, again,  appears to be a contributory factor in
strength loss.  The two bag types positioned closest to the access door, bag types
1-7 and 1-8, exhibited the greatest strength loss of any finished fabrics (46% and
48%, respectively).  The other finished fabrics lost from 33% to 42% of their
strength,  relative to new fabric.

The NF/NS fabric strengths were within + 15% of the NF/S middle section fabric
strengths except for fabric type 1-1 which was 25% lower.  This is probably due to
the unusually high strength measured for the 1-1 NF/S bag because of its favorable
placement in the compartment, rather than to any abnormal loss of strength by the
NF/NS bag.  These data also suggest that fabric strength, at least near the middle
of the bag, was unaffected by mechanical shaking.

Comparison of F/S, NF/S,  and NF/NS Operation.  Because of the proximity of some of
the bags to cold compartment doors and exterior walls, it is difficult to compare
directly the effects of compartment operation on the different types of fabrics.
However, the following general comments can be made:
     •    Generally, the F/S bags retained more strength than either the
          NF/S or NF/NS bags.  The only exception was the all-filament
          fabric (type 1-6).  Placement of many of the NF/S test fabrics
          along exterior compartment walls may be the primary reason for
          these differences.
     •    Based on the strength loss measured for the NF/S bags (rela-
          tive to the F/S bags), if more F/S bags had been located near
          the cold exterior compartment walls, some bag failures proba-
          bly would have occurred.
          The standard fabric with a 10% TEFLON B finish, both with 25%
          and 75% EST, appears at first to have retained more strength
          than the other test fabrics (except fabric type 1-5).  This  is
          misleading because of the "favorable placement" of the NF/S
          bags of this type and because a type 1-2 fabric bag was not
          tested.  In fact, the amount of TEFLON B finish applied to the
          standard fabric did not appreciably affect fabric strength
          retention.

                                      19-10

-------
     •    The greige fabric  (type 1-3) generally retained less strength
          than the other finished fabrics.
     •    Of all the "standard" fiberglass fabrics, the fabric with the
          acid-resistant finish (type 1-5) retained the most strength.
     •    Throughout the test, the strongest standard filtration fabric
          was usually the 14 oz/yd2, 10% TEFLON B finished fabric  (type
          1-8).
     •    Because few bags were available, the glass/PTFE filling yarn
          fabric (type 1-9) was evaluated only in the F/S mode, it is
          difficult to compare this fabric with the other test fabrics.
          Mullen burst testing of this fabric ruptured only the glass
          warp yarns, providing no information on filling yarn strength.
          This fabric also showed significantly better resistance to
          abrasion than the glass fabrics.

SUMMARY AND CONCLUSIONS
Filter bags can fail for a variety of reasons.  The causes for such failure
include manufacturing defects in hardware or fabric, improper installation or
tension, improper operation, and degeneration of the fabric.  Bag fabrics fail due
to the cumulative effect of either mechanical wear  (from repeated cleaning) or
from the result of chemical attack on the fabric or bag hardware (from repeated
passages through water and acid dew points).  The purpose of this test was to
accelerate bag fabric degeneration by subjecting bags made from a variety of
fabrics to frequent shake/deflate cleanings and compartment outages.  Ten types of
fabrics were tested.  Bags were installed on capped and uncapped thimbles (to
isolate the effects shaking from those of combined filtration and shaking) and
other bags were hung in the compartment aisle (for exposure to clean flue gas).
When this test ended, the bags had been through 6181 h of service, 29,458
shake/deflate cleaning cycles (approximately 591,000 shakes), and 113 compartment
isolations.   No bags failed during the test.

Fabric strength measurements (Mullen burst tests) were used to determine how well
the test fabrics withstood the mechanical stress of frequent cleanings and chemi-
cal attack from repeated passages through the water and acid dew points.  These
tests showed that the test fabrics lost strength throughout the test and that
measurable differences existed among the test fabrics at the end of the test.

Because the  filter bags in this test all withstood operating conditions intended
to precipitate fabric failure,  one important conclusion that can be drawn from
this  test is that:
                                      19-11

-------
     •    Fiberglass bag fabric is very durable.  Even greige fabric
          (without a finish designed to resist the effects of high
          temperature and acid flue gas) did not fail.


With respect to the wear imparted to the bag fabric by frequent shake/deflate

cleaning cycles:

     t    S/D Cleaning does not appear to impart enough cumulative wear
          to bag fabric to shorten bag life or cause bag failure,
          provided that bags are not over deflated before they are
          shaken.

     •    Because  S/D cleaning is more energetic than reverse-gas clean-
          ing or sonically-augmented reverse-gas cleaning, these results
          suggest  that proper use of either of these methods of cleaning
          should also not impart enough mechanical wear to the fabric to
          shorten  bag life or cause bag failure.


With respect to chemical attack on the glass fabric caused by repeated passages

through the water  and acid dew point and the effect of fabric finish on resisting

such chemical attack:

     •    Fabric finish does tend to protect bag fabric from chemical
          attack,  and some finishes appear to be more effective than
          others.   Within the group of fabrics that were tested, the
          best performing finish was the acid resistant finish (BGF
          1625).  Other acid resistant finishes may have done as well.

     •    Heavier  fabrics may withstand the effects of chemical attack
          better than lighter fabrics because these fabrics contain more
          fiberglass per square yard.

     •    Because  the minimum compartment temperature was lower after it
          was purged, it appears that cold startups may be the single
          most significant cause of fabric strength loss.

     •    Strength retention is a strong function of the location of
          bags within the compartment.  If a bag is located near an
          exterior compartment wall, it may suffer more from the effects
          of acid  and water dew point excursions than bags located in
          the compartment interior.


ACKNOWLEDGMENTS

This work was performed under contract RP1867-4 from the Electric Power Research

Institute.  The Project Manager is Dr. Ralph F. Altman.  The authors wish to thank

the Mr. Bob Gehri  of Southern Company Services, Inc. and the personnel at Gulf

Power's plant Scholz, the site of the HSFP, for their help and assistance
throughout this project.
                                      19-12

-------
REFERENCES

1.   K. M. Gushing, E. B. Dismukes, W. B. Smith, and R. R. Wilson.  Evaluation of
     the 10-MW High-Sulfur Fabric Filter Pilot Facility. Volume 1:  Reverse Gas
     and Sonic-Assisted Reverse-Gas Cleaning.  CS-6061, Volume 1.  Palo Alto
     California:  Electric Power Research Institute, February 1989.

2.   R. F. Heaphy, L.  G. Felix, R. R. Wilson, and K. M. Cushing.  Evaluation of
     the 10-MW High-Sulfur Fabric Filter Pilot Facility. Volume 2:  Shake/Deflate
     Cleaning, Flue Gas Conditioning, and Fabric Evaluation.  CS-6061, Volume 2.
     Palo Alto, California:  Electric Power Research Institute, In Preparation.

3.   A. G. Metcalfe and G. K. Schmitz.  "Mechanism of Stress Corrosion in E Glass
     Filaments."  Glass Technology, volume 13, 1972, pp. 5-16.

4.   K. M. Cushing and R. L.  Merritt.  Design, Operation,  and Maintenance of
     Fabric Filters in the Utility Industry.  Palo Alto, California:  Electric
     Power Research Institute, In Preparation.

5.   L. G. Felix, K. M. Cushing, W. T. Grubb, and D. V. Giovanni.  Fabric Filters
     for the Electric Utility Industry. Volume 3:  Guidelines for Fabrics and
     Bags.  CS-5161, Volume 3.  Palo Alto, California:   Electric Power Research
     Institute, 1988.
                                     19-13

-------
                                      OUTSIDE WALL
          O
          Capped

          O
          Capped
       Non Filtering/ Shake
       1-1
       1-3
                                          1-4
                                           1-5
                                           1-6
                                           1-7
                                           1-8
                                                   • Bags hung in walkway
Non Filtering / Non Shake
                                       Bottom Door
    o
    Capped
Non Filtering / Shake
                                     OUTSIDE WALL

                      O   Filtering Bags Removed after 3875 h of Service.
                      ©   Filtering Bags Removed after 6181 h of Service.

Figure 1.   Layout of  the test compartment  showing  locations of test bags.
                                         19-14

-------
                                                                 Table 1

                                                       DESCRIPTION OF TEST FABRICS1
UD
 i
I—i
cn
Fabric
ID No.
1-1
1-2
1-3
1-4
1-5
1-6
1-7
Warp Yarn
Construction2
150
150
150
150
150
150
75
1/2 4.0Z
1/2 4.0Z
1/2 4.0Z
1/2 4.0Z
1/2 4.0Z
1/2 4.0Z
1/0 l.OZ
x 3.8S
X 3.8S
x 3.8S
x 3.8S
X 3.8S
X 3.8S

Fill Yarn
Construction2
150 1/4 TEXT 4.0Z x 3.8S
150 1/4 TEXT 4.0Z x 3.8S
150 1/4 TEXT 4.0Z x 3.8S
150 1/4 TEXT 4.0Z x 3.8S
150 1/4 TEXT 4.0Z x 3.8S
150 1/4 4.0Z x 3.8S
50 1/0 TEXT l.OZ +
Weave3
3 x 1
3 x 1
3 x 1
3 x 1
3 x 1
3 x 1
3 x 1
LHTW
LHTW
LHTW
LHTW
LHTW
LHTW
LHTW
Thread
Count
54 x 30
54 x 30
54 x 30
54 x 30
54 x 30
54 x 30
54 x 30
Fiber
Content
ECDE
ECDE
ECDE
ECDE
ECDE
ECDE
ECDE
Finish4
10% TEFLON B
10% TEFLON B
Greige
7% TEFLON B
BGF I 625
8% TEFLON B
10% TEFLON B
EST
I%1
25
75
25
25
25
25
25
Weight
(oz/vd2 )
10
10
10
10
10
10
10
          1-8
37 1/0 l.OZ
  150 1/0 1.0Z/2.5S

2 @ 75 1/2 TEXT l.OZ +
  1 I? 75 I/O 0.7Z with
  final  2.5S Twist
3 x 1 LHTW   44 x 24
ECDE
10% TEFLON B   25
          1-9      37  1/0 l.OZ
          1-10     75  1/0 l.OZ
                      Glass/PTFE (Proprietary)   3 x 1 RHTW   46 x 30
                      75 1/2 TEXT 4.0Z x 3.8 S   3x1 LHTW   54 x 30
14
                                                    ECDE/   10% TEFLON B   75       16
                                                    PTFE

                                                    ECDE    10% TEFLON B   75       10
                                                              - GORE-TEX
        NOTES:   1A11  test bags were 12 in.  diameter and 415 in. long.   The total  area per bag was 104 ft2.
                 2Nomenclature after ASTM Standard D-578-83.
                 3LHTW = Left Hand Twill  Weave,  RHTW - Right Hand Twill  Weave.
                 4A11  of the fabrics except  fabric 1-9 were woven and finished  by  BGF,  Inc.
                  Fabric 1-9 was woven and finished by JPS Industrial Fabrics,  Inc.

-------
                           Table 2



                  RESIDUAL DUSTCAKE WEIGHTS
Fabric Description
Residual Dustcake Height, 1b/ft2
I.D. Weight,
Number oz/vd2 Finish
1-1
1-2
1-3
1-4
1-5
1-6
1-7
1-8
1-9
1-10

Date
No.
No.
10 10% Teflon
10 10% Teflon
10 Greige
10 7% Teflon B
EST
%
B 25
B 75
25
25
10 Acid Resistant 25
10 8% Teflon B
10 10% Teflon
14 10% Teflon
14 10% Teflon
10 10% Teflon
Gore-Tex
Bags Weighed
of Compartment Isolations
of Shake/Deflate Cleaning
0
B 25
B 25
B 75
B 0

Cycles
3063 h
in Service
0.33
0.39
0.36
0.31
0.32
0.31
0.38
0.38
0.25
0.38

6/28/88
13
13,620
5468 h
in Service
0.41
0.62
0.54
0.39
0.35
0.33
0.53
0.41
0.31
0.45

10/25/88
68
25,678
6181 h
in Servii
0.49
0.81
0.75
0.53
0.52
0.43
0.59
0.62
0.44
0.75

12/7/88
113
29,458
                            19-16

-------
                                     Table  3

                          RESULTS  OF  MULLEN BURST  TESTS
                                           Fabric  Strength
 Test
Fabric

 1-1
 1-2




 1-3




 1-4




 1-5




 1-6




 1-7




 1-8




 1-9


 1-10
Hours in
Service

  0
  1546
  3875
  5470
  6181
  0
  1546
  3875
  5470
  6181
  0
  1546
  3875
  5470
  6181
  0
  1546
  3875
  5470
  6181
  0
  1546
  3875
  5470
  6181
  0
  1546
  3875
  5470
  6181
  0
  1546
  3875
  5470
  6181
  0
  1546
  3875
  5470
  6181
  0
  3875
  6181
  0
  3875
  6181
Filter/Shake
Mullen
Burst
(psj )
586

407

509
586
456
454
506

408

359
534

405

420
589

528

494
897

710

585
567

474

467
985

694

758
866
707
624
573
522
440
Strength
Lost
m
0

30.5

13.1
0
22.2
22.5
0

19.4

29.1
0

24.2

21.3
0

10.4

16.1
0

20.8

34.8
0

16.4

17.6
0

29.5

23.0
0
18.4
27.9
0
8.9
23.2
Non Filter/Shake
Mullen
Burst
(psj )
586



452
586

402
506



268
534



302
589



381
897



578
567



334
985



612



573

314
Strength
Lost
m
0



22.9
0

11.5
0



47.0
0



43.4
0



35.3
0



35.5
0



41.1
0



37.9



0

45.2
Non Filter/Non Shake
Mullen
Burst
(psj )
586
430
378
393
338



506
417
304
266
242
534
423
436
411
342
589
514
442
483
394
897
884
662
498
579
567
474
354
364
304
985
734
518
614
514






Strength
Lost
(%)
0
26.6
35.5
32.9
42.3



0
17.6
39.9
47.4
52.2
0
20.8
18.4
23.0
36.0
0
12.7
25.0
18.0
33.1
0
1.4
26.7
44.5
35.5
0
16.4
37.6
35.8
46.4
0
25.5
47.4
37.7
47.8






                                      19-17

-------
     COLLECTION OF REACTIVE AND COHESIVE FINE PARTICLES
            IN A BAG FILTER WITH  PULSE-JET CLEANING

                             E.  Schmidt
                             F.  Loffler
                     Universitat  Karlsruhe (TH)
       Institut  fur Mechanische Verfahrenstechnik und Mechanik
            Postfach  6980, D-7500 Karlsruhe, West Germany
ABSTRACT

The collection of reactive and therefore extremely cohesive  fine
particles by means of a bag filter with pulse-jet  cleaning may  cause
trouble. The electron beam dry scrubbing (EBDS)  process,  used to  re-
move S02 and NOX in a power station  of  the Badenwerk AG,  Karlsruhe,
sets an example. To solve the appearing problems in order to achieve
a satisfactory filter performance extensive  know-how of process en-
gineering is necessary. This paper deals with the  following  sub-
jects :  Construction of the cleaning-system,  selection  of  the filter
medium,  precoating, and dosage of an auxiliary dust during filtra-
tion .
                                 20-1

-------
     COLLECTION OF REACTIVE AND COHESIVE FINE PARTICLES
            IN A BAG FILTER WITH  PULSE-JET CLEANING
THE  PROBLEM
The collection of fine dusts (characterized by a cumulative mass
distribution median below 1 |j.m)  can present a substantial problem
for a number of processes. Bessemer conversion,  the calcium carbide
production and fluidized-bed combustors are but  a few of the indus-
trial techniques which demand a reliable fine dust filtration. De-
pending on their physical and chemical characteristics in question,
some particulate materials reveal a tendency to  adhere. In such
cases,  it is quite possible that extensive measures become necessary
in order to obtain satisfactory operating characteristics from the
applied filter equipment and hence, from the complete plant.

The following reviews the problems which the collection of fine,
baking dusts present with the aid of a concrete  example.  This con-
cerns the trimming of a pulse-jet filter for a pilot EBDS-system in
block 7 of the steam generating station for Badenwerk AG in Karls-
ruhe. Care has been taken to present the experience gained and the
results derived to the reader in a fashion which will allow a gener-
alization, or even an implementation in other, similar applications.

METHOD  OF  APPROACH

Before expounding the systematics of the investigations conducted
with the aim of solving the given problem, the EBDS-process requires
an initial brief description. The EBDS-technique is a secondary,
dry, flue-gas cleaning process in which S02 and  NOX are simultane-
ously extracted, hence reducing the pollutant emission of fossil-
fired plants.  This basically involves the irradiation of the fil-
tered flue gas by high-energy electrons in the presence of ammonia
(see Fig. 1 |1 ).  in simplified terms, SO2 and NOX are oxidized to
                                 20-2

-------
sulphuric and nitric acids by the free radicals, released as a re-
sult of the irradiation. These gaseous acids are neutralized by the
ammonia and are hence converted to extremely adhesive and cohesive
ammonium sulfate and ammonium nitrate in the form of solid parti-
cles. It is these dusts which are then filtered by an existing
pulse-jet fabric filter  (1 chamber; 64 bags, each 5 m long) as sche-
matically illustrated in Fig. 1 as well.

The existing bag filter in need of modification was in quite a deso-
late condition  (choked and encrusted by centimeter-thick baked de-
posits) with a performance to match (high pressure drop and inade-
quate cake removal). The method of approach taken to redeem its ini-
tial efficiency involved the following measures: The initial step
was to analyse the effectiveness of the pulse-jet cake discharge
system, which was subsequently extensively improved, coupled with
the selection of a more adequate filter medium. Following this, a
comprehensive investigation into pre-coating and the dosing of an
additive during the product filtration was then conducted. The last
step was to determine the effects of a variation of specific opera-
tional parameters on the filter's behaviour together with the possi-
bility of increasing the product 's salt content

CLEANING   CONDITION  IMPROVEMENT
The  basic  principle  of  pulse-jet  cleaning
The pulse-jet technique is a filtration process in which a so-called
dust cake forms on the surface of filter bags supported by internal
metal cages.  It is this filter cake which is the true,  highly effi-
cient filter medium. Nevertheless, it must obviously be periodically
removed from the bags,  since the permeation resistance of the filter
continually increases.  The cleaning of a filter bag is executed in
such apparatus by the abrupt injection of compressed air into it.
The bag suddenly inflates,  hence throwing the cake from its surface.
After accomplishing this, the filtration cycle may then be recom-
menced.

The compressed-air required for the cleaning is pumped into a stor-
age vessel or tank. The cleaning pulse is released by a rapidly op-
                                 20-3

-------
erating diaphragm valve which allows the compressed air to flow into
a so-called blow-pipe.  This possesses an injection orifice for each
simultaneously cleaned filter bag which is centrally located above
each bag neck. The pressure pulse is injected from the orifice into
the bag, drawing surrounding air in with it.  The effect may some-
times be enhanced when an injector nozzle is  attached to the bag
neck. The complete system is schematically illustrated in Fig. 2.

Analysis  of  the  cleaning	conditions
In order to achieve a complete cake discharge,  the filter medium
must be subjected to a maximal deflection coupled with an adequate
internal excess pressure. Extensive research  conducted by Sievert
|2| on the regeneration of filter media has shown that the regional
cleaning intensity of a filter bag is mirrored by the maximal inter-
nal pressure generated during the cleaning cycle at the specific lo-
cation. If this application dependent critical pressure is exceeded,
then a cleaning efficiency in the region of 100 % can be expected to
result.

In order to define the cleaning conditions for the given applica-
tion,  investigations were conducted on a special filter bag with a
low permeability. The in.xirnum effective pressure which was attained
during the pressure pulse within the bag was  measured at four loca-
tions, as illustrated in Fig. 2. Fig. 3 shows the interpolated re-
sults  of the measurements for a tank pressure of ApT = 6 bar ("old
system"). A comparison between the experimental results with those
of comparable bag filter equipment indicates  that the maximal pres-
sure attained in the central and lower bag sections appears inade-
quate .

Megsvres—t&ken—1£	increase   the  effectivity

The aim of these measures was to enhance the  maximal pressure in the
central and lower bag sections during the cleaning pulse. For this
purpose, the individual cleaning system components were critically
examined.
                                 20-4

-------
Compressed-air feed  and tank. Each  of the  64  filter bags  were
subjected to successive pulse-jet cleaning in banks of four,  whereby
care was taken to ensure that the compressed-air tank responsible
for the cleaning of each bank attained the same nominal pressure
between each discharge. Furthermore, in order to ensure that  a large
volume flow is maintained by the blow-pipe during the whole cleaning
cycle, the pressure within the tank should not sink excessively. In
this instance, the tank pressure drop was at most 12 % of the re-
spective nominal pressures  (4 bar and 6 bar). As such, the tank ca-
pacity was obviously able to cope with the requirements imposed.

Blow-pipe  orifice.  The volumetric  gas  flow emitted from the blow-
pipe orifice may be approximately described by the physical laws
valid for a tapered nozzle  |3|. Provided the so-called critical
pressure ratio is exceeded, then the volume flow only depends on the
density of the gas within the blow-pipe and the orifice's cross-sec-
tional area.

An increase in the diameter of the blow-pipe orifices will reduce
the pressure within the pipe and consequently the density of the
emerged compressed air, hence ultimately reducing the total volume
flow. In the range of interest, however, this reduction is plainly
over-compensated by the enlarged cross-sectional area. For example,
whilst an increase of each orifice diameter from 10 mm to 16 mm re-
duced the maximal established excess pressure within the blow-pipe
from 2.9 bar to 2.2 bar  (tank pressure: 6 bar), the calculated vol-
ume flow for each orifice increased from 51 1/s to 106 1/s. This
inexpensive and relatively simple modification enabled the average
maximal excess pressure within the bags to be raised by approx.  20
%. A further widening of the orifices, however, is not only limited
by the fact that at some point the compressed air will no longer be
administered to all four outlets at the same pressure, but also
because the critical ratio at each outlet will no longer be main-
tained.  Pressure measurements directly above the four blow-pipe
outlets have revealed that under the given conditions (2" diaphragm
valve,  53 mm 0 blow-pipe),  similar pressures existed at all four
orifices (±5 %) .
                                 20-5

-------
Diaphragm  valve,  blow-pipe diameter.  Firstly,  the duration of
the compressed-air pulse is to be briefly discussed. The pneumati-
cally operated diaphragm valve (dv)  is actuated by an solenoid valve
(sv).  Measurements of the pressure rise within the filter bags have
repeatedly shown that pulse durations of tdv = 300 ms (tgv  =  100 ms)
quite sufficiently allow the maximum excess pressure, vital for the
cleaning effectivity, to be established within the bags. Longer du-
rations generally only lead to an unnecessary compressed air waste
without any significant influence on the cleaning effectivity.

The choice of larger valves with corresponding blow-pipes may sig-
nificantly improve the cleaning,  despite the initial cost.  Such a
modification (1.5" —> 2")  would theoretically allow the  maximal com-
pressed air volume flow to be increased from 106 1/s to 123 1/s in
this case.  As a result, the maximal  excess pressure within the bags
could be further increased by an average of approx. 15 %.

Blow-pipe/filter bag  distance.  The  distance between  the blow-
pipe orifice and the bag's neck was,  in this case, not varied. Nev-
ertheless,  the significance of this  additional parameter should not
remain unmentioned, due to the influence exerted on the quantity of
surrounding air sucked into the bags during the cleaning pulse (see
Fig. 4).

When the distance s between the orifice and the neck is too large,
the induced secondary volume flow,  which can be as large as five-
times that of the primary jet  4|,  will not fully enter the bag.
This is due to the widening of the primary jet characterized by the
angle of spread a. Should this distance be too small, then although
the jet is fully injected into the bag,  the secondary air volume
remains relatively small. Measurements have verified the anticipated
behaviour,  that the largest air volume is then drawn into the bag
when the jet just completely covers  the bag's periphery.

Alongside all this theory, one should naturally not forget that in
engineering practice, both linear (distance 5)  and angular (angle e)
misalignments can occur between the  orifice and bag axes. The actual
influence of inaccurate manufacture  and assembly becomes more evi-
dent with the following numerical example: Under the assumption of
                                 20-6

-------
an angle of  spread  of  a =  10°, a bag diameter of D =  160  mm  and  a
orifice diameter of d  =  16 mm, then the optimal distortion Sopt is
408 mm. If the axes should become shifted by 8 = 10 mm, then Sopt is
reduced to 352 mm,  whilst an  angular distortion of E  = 5° will re-
duce Sopt  to 269  mm.

Venturi Insert.  A profusion  of various  injector  nozzles are avail-
able on the market. Nevertheless, their installation  in the  upper
supporting cage section  is not always advisable. An elementary anal-
ysis and corresponding measurements  |2| show that the presence of
such an injector influences the maximum compressed air influx. As
previously stated,  a high cleaning efficiency demands a complete
filter medium inflation. If the bag is relatively long, then a cor-
respondingly large  gas volume must be injected into it at an ade-
quate rate. Above a specific  bag size, the restriction caused by the
venturi becomes substantial.  Measurements have verified that this
was exactly the case with the apparatus in question. Hence,  the in-
ternal bag pressure was  able  to be increased by approx. 10 % by sim-
ply removing the Venturis from the necks.

Fig. 3 finally demonstrates the results of the complete range of im-
provements made to the cleaning system (larger blow-pipe orifices,
larger diaphragm valves  and removal of the Venturis)  by comparing
the maximal internal excess pressures accomplished by the old and
the modified cleaning  system  at four locations along the bag. It is
obvious that the desired pressure increase in the central and lower
bag regions was indeed attained.

FILTER  MEDIUM  SELECTION

Today,  the bags of pulse-jet  filters tend to be more often made from
reinforced synthetic nonwovens and felts than of conventional woven
fabrics.  In order to fulfil its purpose (i.e. low dust emission and
good cleaning characteristics) the filter medium should allow as few
particles  as possible to penetrate into it and hence, immediately
form a  superficial filter cake. This is especially necessary when
fine,  adhesive particles are to be collected.
                                 20-7

-------
The non-calendered polyacrylonitrile needled felt from which the
bags of the filter were made, was not able to fulfil this require-
ment. Its large pores and protruding fibres allowed the dust-cake to
be quite firmly affixed. Scanning electron micrographs plainly dem-
onstrated how the individual adhesive particles accumulated on and
around the individual fibres. Not even the best pulse-jet cleaning
system could successfully remove a cake possessing such a founda-
tion. For this reason,  a new filter medium, consisting of a PTFE
membrane-coated polyester needled felt was installed and tested. Be-
cause the membrane is so fine,  even the smallest particles can hard-
ly permeate into the felt. Nevertheless, subsequent experiments re-
vealed that the dust in question still adhered so strongly to the
filter bag, that a satisfactory cleaning remained impossible. In
view of this, a stable filtration behaviour could obviously only be
attained by precoating the bags.

PRECOATING  INVESTIGATIONS
Protective  layer  application
The aim of precoating is to assist the separation of the filtered
dust  (often referred to as the product) from the filter medium. This
layer, which should consist of a material which is easily removed
from the filter bag, is usually re-applied after each cleaning cy-
cle. The selection of a precoating material should be made with re-
spect to its compatibility with the filtered product, taking care to
ensure that it does not impede its further processing or disposal.
In this case, so-called mineral powder was chosen. This is a mixture
of different volcanic constituents and is viewed as a so-called soil
conditioner. Hence, its presence in the ammonium salts, intended for
use as a fertilizer, is obviously of no disadvantage.

A further criterion for the choice of the precoating material is the
particle size distribution. If the substance is too coarse, unde-
sired deposits may either form between the dosing device and the
filter chamber, or within the filter chamber itself. When the pre-
coating material is too fine, then its dispersion within the gas
stream may prove difficult, in addition to it impeding the formation
of a light, easily removed and low permeability filter coating.
                                 20-8

-------
Another important precoating aspect is the required uniform distri-
bution of the precoating material in the filter chamber and upon the
bags. One can assume, however, that the differences in the precoated
layer thickness remain negligible small as a result of homogenizing
flow processes  (caused by the fact that the dust transport occurs
where the flow resistance is lowest).

Protective  layer  thickness

The question of the average precoating layer thickness must now in-
evitably arise. On the one hand, the product may not be allowed to
come into contact with the filter medium, whilst on the other hand,
thicker precoating layers do not only increase the consumption of
precoating substance but also the off-line duration and the filtra-
tion pressure drop. Hence, a number of fundamental aspects are to be
reviewed regarding the necessary minimal precoating layer thickness,
or respective precoating layer mass. Initially, the two filter medi-
um extremes are to be examined  (see Fig. 5). The one extreme is a
porous needled felt possessing a fibrous surface, i.e. a large num-
ber of individual fibres which protrude from the compact internal
complex  (A).  The other extreme is a laminate medium (B), for which
the applied layer  (e.g. a PTFE-rnembrane) is absolutely impervious to
the precoating substance.

If fine,  non-adhesive dust is to be filtered, the laminate medium
can generally be applied without the necessity of a precoating lay-
er. However,  in the case of the porous felt,  a thin precoating layer
which closes  the pores proves advisable  (C).  This ensures that the
finer particles may no longer permeate into the medium. Because the
precoating material lodges relatively deep within the fabric,  this
will not  be completely removed during the cleaning cycle, and as
such, a new precoating may not be specifically necessary after each
regeneration.  This does not, however,  apply to coarse, adhesive
dusts.  Here a new, thick precoating layer must be applied prior to
each filtration cycle,  since even contact with the protruding fibres
must be avoided at all costs (D).  A thin precoating layer must now
also be applied to the laminate media,  although a thickness of just
a few particle diameters will adequately ensure that the adhesive
                                 20-9

-------
particles do not touch the laminate  (E). The collection of fine, ad-
hesive dust requires thicker precoating layers  (F). Because of the
quantity of precoating substance necessary, coarse filter media
should not be used. In the case of the extremely fine and reactive
ammonium salt products which emerged from the gaseous reaction prior
to entering the filtration chamber, the protective layer had to, in
addition to coping with the tasks of a surface filter, also assume
the roles of both a granular-bed reactor and a deep-bed filter. This
led to a preferable precoating layer of at least 1 mm.
Protective  layer  stability
The gas flow necessary to apply the precoating layer may either be
extracted from the environmental air or from the clean-gas stream.
If a specific temperature is to be maintained within the filter
chamber, the operation with recirculated air is also advisable.
Fluctuations in the volume flow within the filter chamber cannot be
avoided when switching from the precoating phase to the product fil-
tration. It is therefore vital to ensure that during the transition,
none of the applied protective layer flakes from the bag.

The basic problem is to be briefly discussed with the aid of Fig. 6.
In the case of a dry, non-adhesive dust, the cohesive forces which
internally bind the protective layer, fundamentally stem from inter-
particulate van-der-Waals interactions. This also, applies to the ad-
hesive forces between the protective layer and the filter medium. In
addition, a contact force is exerted on the cake by the gas permea-
tion across the face area A. This assists the stabilization of the
loose, fragile dust cake formed by the precoating layer. When multi-
plied by a specific coefficient of static friction, this contact
force may be viewed as a frictional force which, together with the
adhesive forces, compensate the gravitational force acting on the
precoating layer. A reduction of the gas permeation will reduce the
precoating's frictional force. Should this drop below a specific
critical value, then the constant gravitational force could cause at
least some sections of the layer to flake from the bag surface.
Other investigations have shown |5 , that such layers are preferably
detached in the vicinity of the filter medium's surface, which, as
far as the precoating task is concerned, is of course extremely
disadvantageous.
                                 20-10

-------
In order to investigate the stability of the protective  layer,  two
different precoating thicknesses were repeatedly  filtered onto  the
bags  (1 kg/m2 and 1.9 kg/m2)  at a constant gas flow rate of
7000 m3/h. After this, the volume flow was reduced in increments of
1000 m3/h. At each new filter face velocity, the  dust which fell in-
to the hopper was extracted and weighed. Down to  a volume flow  rate
of 1000 m3/h,  the dust mass collected from the hopper remained  com-
paratively low. It is altogether possible that this dust was not de-
tached from the bags at all, but is simply the result of whirled-up
sediments which accumulated within the raw-gas feed. Larger dust
masses  (up to 25 kg) only flaked from the bags when the  raw-gas was
completely shut-off. It may therefore be established that when
switching-over from the clean-gas flow to the raw-gas flow  (or  in
general terms, from the precoating to the filtration mode), the gas
flow should not be completely shut down, otherwise the protective
layer may be damaged. In the case of the system under investigation,
relatively low face velocities of approximately 6 m/h adequately
maintained a stable layer.

INFLUENCE OF THE OPERATIONAL  PARAMETERS
Additive  mass  flow  rate
The availability of an efficient cleaning system, the choice of an
ideal surface-filtration assisting filter medium  and the application
of a thick precoating layer still do not guarantee a satisfactory
filtration of fine, adhesive particles. These dusts may form such an
impervious cake that even after a relatively short period, the pres-
sure drop abruptly increases to a point where the filtration must be
intervened.  In the case of the ammonium salts, such a rapid pressure
drop increase occurred for a product mass flow rate of approx.
10 kg/h after just 1 hour of continuous filtration. One can easily
estimate that  at  this point,  the filter cake which was formed on the
precoated layer was less than 30 |J,m thick i.e. had just began to
close the pores.  This tendency,  to form such an impermeable dust
cake may,  however,  be countered by permanently feeding a suitable
additive to  the filtration gas.  One such substance is the material
used for the  precoating.
                                 20-11

-------
The mass of mineral powder which had to be added to the raw gas in
order to attain a loose,  highly permeable dust cake was experimen-
tally investigated. Fig.  7 illustrates the individually measured
temporal pressure drop in dependence of the mineral powder feed rate
(dots).  In addition,  the  calculated increase has been plotted as a
function of the mineral dust alone (drawn line), which should be
approached by the measured values at large mineral dust dosages. One
can see that under the given conditions,  the lowest pressure drop
increase emerged for a mineral powder dosage of 10 kg/h (which cor-
responds to the average formation rate of the product).

The  temperature
Further parameters which  would influence  the operational character-
istics  of the filter unit are the ammonium concentration,  the humid-
ity and the temperature.  Fig. 8 illustrates the pressure drop in-
crease  under otherwise identical conditions,  as a function of the
filtration time for various temperatures. Although similar dust
quantities were filtered, the lowest pressure drop increase was at-
tained  at a temperature of 110 °C.  Apparently,  a looser,  not quite
as adhesive dust-cake then emerged. At 90 °C,  the pressure drop
could also be observed to increase slightly,  whilst at 80 °C the
pressure drop increased quite rapidly (especially during the first
6 hours).  The dust cake then obviously possessed a denser structure.
At 70 °C,  a stable filtration was,  under  such conditions,  out of the
question.  The investigation had to be aborted after approx.  4 hours
due to  a tremendous pressure drop increase which occurred within
just a  few minutes.

This example clearly demonstrates how a perfectly functioning dust
collector can be brought  to its knees by  simply exceeding a critical
limit of a specific variable which may be of no significance whatso-
ever up to this specific  value.

UPGRADING  THROUGH  DUST  RECYCLING

In order to reduce the necessary quantity of precoating material and
therefore the amount contained within the product, the filtered dust
removed from the bags can also be used for precoating. Should the
                                 20-12

-------
hopper dust be agglomerated, then this must be comminuted prior to
recycling. The question of whether the re-ground product is suited
as a precoating substance depends on the possibility of adequately
dispersing the dust within the gas flow. The application of the hop-
per dust resulted in this case in a looser, more permeable protec-
tive layer, which could be more easily detached from the filter me-
dium. As far as the ammonium salts are concerned, the product recy-
cling for the purpose of precoating was therefore of advantage. The
salt content of the product could, as a result, be increased from
approx. 25 % to 50 %.

A further possibility of upgrading the final product by means of
dust recycling is the use of the cleaned dust as an additive during
the filtration. Contrary to the precoating trends, denser cakes now
formed when the recycled mixture was used; the pressure drop in-
creased more rapidly. The cause of this may lie in the reduced reac-
tivity between the recycled product and the filtered dust. According
to this, the recycling of the product during the filtration can
prove to be disadvantageous with respect to the filtration pressure
drop. This may, however, be compensated by the advantage of a higher
product salt concentration.

SUMMARY
Bag filters can be used to collect fine adhesive dusts. Satisfactory
operational characteristics, however,  may only be attained at the
cost of elaborate process-engineering measures. The following items
remain vital:

  •  a  capaciously dimensioned cleaning system
  •  a  filter  medium promoting surface  filtration
  •  careful precoating
  •  a  supplementary additive feed during  the product  filtration.

When able to be accommodated by the process in question, a modifica-
tion of the reactivity of the product to be filtered by means of a
variation of specific peripheral condition (e.g. the temperature)
could significantly (or even decisively) assist the filtration.
                                 20-13

-------
REFERENCES

1   H. Angela, J.  Gottstein,  K. Zellner.  "Flue Gas Cleaning by the
    Electron Beam  Process at  the RDK Pilot  Plant." 4^ Symposium  on
    Integrated Environmental  Control. Washington, D. C., March 1988,
2   J. Sievert.  Physikalische Vorgange bei  der Regenerierung des
    Filtermediums  in  Schlauchfiltern mit  DruckstoBabreinigung.
    Diisseldorf:  VDI-Verlag,  1988.
3   K. Oswatitsch. Grundlagen der Gasdynamik.  Wien, New York:
    Springer-Verlag,  1976.
4   B. Eck. Technische  Stromungslehre. Berlin,  Heidelberg,  New York:
    Springer-Verlag,  1978.
5   W. Hoflinger,  A.  Hackl.  "Untersuchungen zur Schichtstabilitat
    bei der Staubabtrennung  in filternden Abscheidern mit Hilfe von
    Precoatschichten."  Chem.-Ing.-Tech. MS  1768/89.
    1  graphite tube heat exchanger
    2  X-ray shielding
    3  ammonia-air-mixture injection
    4  electron accelerators
     -f~7\ flue9as
       )d
5 baghouse
6 final product
7 induced draught fan
                                 to stack
        Figure 1. EBDS  pilot plant with bag  filter apparatus
                                  20-14

-------
        blow-pipe
 valve
pressure tank
              "*7L    AP
               /hi      max
               I	^^^^-»
 excess pressure Ap
       Ap
          max
                      time
tn
it)
M

-P
rH
*H
m
    Figure 2.  Pulse-jet filter
    bag cleaning system
                                          6 4
                                          o
                                          -p
                                          B
                                          O
0)
o
fi
nJ
-P
W  1
-H  1
T)
               old
               system
• new
  system
                                                    10
                         15
                        20
                                              max. excess pressure
                                                  in bag / hPa
         Figure 3. Comparison be-
         tween the maximal pres-
         sures attained in the
         bags by the standard and
         modified system
                                 20-15

-------
               blow
               pipe
                                 Figure 5. Different types of

                                 filter media  (some precoated)
                                protective
                                layer
                                filter
                                medium
Figure 4.  Blow-pipe and

filter bag configurations
Figure 6. Forces acting

on the precoated layer
                              20-16

-------
   CM
   0)
   M
   M
   O
   a
   -H
   0)
   to
   to
   a>
   M
      10
      10
      10'
51  10
H
T3
      10
                   product mass flow  rate  =  10  kg/h
                                  A(Ap)/At = minimal  :
                  5          10          15          20


             additive mass  flow  rate  /  (kg/h)
      Figure 7. Pressure drop  increase

      as a function of the mineral  powder feed rate
                    5          10          15

                       filtration time / h
Figure 8. Pressure  drop  increase for various temperatures
                             20-17

-------
Advanced Power System Participate Control Technology



          T.F. Bechtel (no paper provided)
                        P3-1

-------
FUTURE DIRECTIONS  IN
PARTICULATE  CONTROL
       TECHNOLOGY
          Sabert Oglesby
        President Emeritus

     Southern Research Institute
     Birmingham, AL 35255-5305
               P4-1

-------
     I APPRECIATE THE INVITATION TO TALK TO  THIS  CONFERENCE.   I HAVE ATTENDED




MANY OF THEM IN THE PAST, AND I FEEL IT IS ONE  OF THE MOST EFFECTIVE MEANS FOR




DISSEMINATION OF INFORMATION ON THE RESEARCH AND  APPLICATION  OF PARTICULATE




CONTROL TECHNOLOGY.









     THE SUBJECT OF MY TALK IS "FUTURE DIRECTIONS  IN  PARTICULATE CONTROL




TECHNOLOGY11,  SO I WILL GET OUT MY CRYSTAL BALL AND SEE IF  I CAN PREDICT SOME




DIRECTIONS FOR THE FUTURE.  AS DR.  CHARLES KETTERING  ONCE  REMARKED,  "WE SHOULD




ALL BE INTERESTED IN THE FUTURE SINCE WE WILL SPEND THE REST  OF OUR  LIVES




THERE".








     I DO NOT PROFESS TO BE THE PARTICULATE CONTROL COUNTERPART TO JEANNE JLTXON




WITH CLAIRVOYANT POWERS,  SO I WILL ATTEMPT TO EXTRAPOLATE  WHAT  MAY BE EXPECTED




IN THE FUTURE.









     WE HAVE  COME A LONG WAY IN CONTROL OF PARTICULATE EMISSIONS  IN  THE  PAST




FEW DECADES.   I REMEMBER QUITE VIVIDLY THAT ONE HAD TO TURN ON  HEADLIGHTS WHEN




DRIVING TO WORK IN THE WINTER IN MANY URBAN AREAS WHEN THERE WAS  A TEMPERATURE




INVERSION. TODAY,  IN MOST CITIES,  PARTICULATE LEVELS ARE MUCH  REDUCED DUE TO




CHANGE FROM COAL TO NATURAL GAS  FOR HEATING AND BETTER CONTROL  OF PARTICULATE




EMISSIONS FROM INDUSTRIAL SOURCES.









     AN ANALYSIS OF THE  CHANGES  IN  PARTICULATE EMISSIONS WILL SHOW THAT  THEY




ARE DRIVEN BY THREE FACTORS.   FIRST,  AND PERHAPS THE MOST  INFLUENTIAL IS STATE




AND FEDERAL EMISSION REGULATIONS.   SECOND,  IS PROCESS CONSIDERATIONS AND THIRD,




IS ECONOMIC REASONS.
                                     P4-2

-------
      PRIOR TO ENACTMENT OF CLEAN AIR LEGISLATION, PART1CULATE CONTROL DEVICES




 WERE INSTALLED PRIMARILY FOR PROCESS CONTROL.   PRECIPITATORS WERE USED FOR




 REMOVING LARGER PARTICLES FOR PROTECTION OF INDUCED DRAFT FANS.   CYCLONES AND




 PRECIPITATORS WERE ALSO USED TO  REDUCE DUST FALL-OUT IN THE IMMEDIATE VICINITY




 OF A PLANT,  AND SOME PRECIPITATORS  WERE INSTALLED TO RECYCLE COLLECTED




 PARTICULATE.   EXAMPLES  ARE COLLECTION OF SODIUM SULFATE FROM BLACK LIQUOR




 BOILERS  IN PAPER MILLS  AND DUST  FROM CEMENT KILNS.   FOR THESE APPLICATIONS,




 EFFICIENCIES  WERE NOT REQUIRED TO BE VERY HIGH AND  MOST INDUSTRIAL PLANTS HAD




 PLUMES COMPOSED OF FINE PARTICULATE THAT WERE  VISIBLE FOR MANY MILES.








      WHEN  REGULATIONS LIMITING EMISSIONS WERE  ENACTED,  THERE WAS  CONSIDERABLE




 PRESSURE TO UPGRADE,  ADD TO,  OR  REPLACE  EXISTING  PARTICULATE CONTROL  DEVICES




 AND THIS PROCESS CONTINUED AS MORE  STRINGENT EMISSION LEVELS WERE ADOPTED.




 THESE REGULATIONS WERE  THE PRIMARY  DRIVING  FORCE  FOR  RESEARCH AND DEVELOPMENT




 IN PARTICULATE  CONTROL  TECHNOLOGY DURING THE LATE 1960's  and 1970'8.








      THESE ARE  THE SAME DRIVERS  OF  TECHNOLOGICAL  DEVELOPMENT TODAY.




 REGULATIONS REQUIRING REDUCED EMISSIONS  OF  FINE PARTICULATE,  REDUCTION OF




 GASEOUS  POLLUTANTS, AND MEANS FOR REDUCING  COST OF  POLLUTION CONTROL ARE




 FACTORS  THAT WILL DICTATE  FUTURE  DIRECTION  FOR RESEARCH AND  APPLICATION OF




 PARTICULATE CONTROL EQUIPMENT.








     CONTROL OF  FINE  PARTICULATE  CONTINUES  TO  BE  A  CONCERN.   IT HAS BEEN




 POINTED OUT THAT  PARTICLES SMALLER THAN MICRON DIAMETER REPRESENT ONLY about  1%




OF THE MASS EMISSIONS FROM A COAL FIRED  POWER  PLANT,  BUT  99%  OF THE NUMBER OF




PARTICLES.   ALSO, THE SMALLER SIZE FRACTION CONTAINS A HIGHER CONCENTRATION OF
                                     P4-3

-------
HEAVY METALS WHICH MAY REPRESENT A HEALTH PROBLEM IF THEY ARE INHALED.   ALSO,




THESE PARTICLES ARE WASHED OUT OF THE ATMOSPHERE  BY  RAIN  AND  APPEAR IN  THE




WATER SYSTEM.









     THE FIRST ATTEMPT TO CONTROL FINE PARTICULATE IS THE SO-CALLED PM10




REGULATIONS WHICH LIMIT MASS EMISSIONS OF PARTICLES  BELOW 10  MICRON DIAMETER.




DEPENDING UPON HEALTH EFFECT STUDIES, FUTURE REGULATIONS  MAY  PLACE  EVEN  GREATER




RESTRICTIONS ON EMISSIONS OF FINE PARTICULATE WHICH  WOULD IMPACT DESIGN  OF ALL




OF THE PARTICULATE CONTROL DEVICES AND DICTATE ADDITIONAL RESEARCH  TO MINIMIZE




THE COST IMPACT OF SUCH REGULATIONS.








     RESEARCH ON ELECTROSTATIC PRECIPITATION HAS HAD AN INTERESTING  HISTORY.




FOLLOWING THE INITIAL DEVELOPMENTAL WORK BY LODGE AND COTTRELL, CONSIDERABLE




FUNDAMENTAL RESEARCH WAS PERFORMED BY HEWITT ON DEFINING  PARTICLE CHARGE




RELATIONSHIPS AND BY COOPERMAN ON DEVELOPMENT OF ELECTRIC  FIELD EQUATIONS.




DR. HARRY WHITE AND HIS ASSOCIATES PERFORMED THE MOST COMPREHENSIVE  RESEARCH OK




ELECTROSTATIC PRECIPITATION,  INCLUDING PIONEERING WORK ON  PULSE ENERGIZATION




FLUE CAS CONDITIONING,  RAPPING,  AND MANY OTHER TOPICS.









     RESEARCH ON PRECIPITATION WAS ACCELERATED IN THE 1970'S AND CONSIDERABLE




PROGRESS WAS MADE IN BETTER UNDERSTANDING OF THE THEORY, DEVELOPMENT OF




MATHEMATICAL MODELS,  CONTROL OF DUST RESISTIVITY BY FLUE GAS CONDITIONING AND




PRECHARGING, AND IMPROVED DESIGN AND SIZING PROCEDURES,    IN THE PAST SEVERAL




YEARS THERE HAS BEEN A SUBSTANTIAL REDUCTION IN RESEARCH AND DEVELOPMENT ON




PRECIPITATORS IN THE U.S.,  DUE IN PART TO A DIVERSION OF  FUNDING TO  GASEOUS




POLLUTION CONTROL AND A GENERAL FEELING THAT PARTICULATE CONTROL TECHNOLOGY HAD




MATURED TO THE EXTENT THAT NO MORE RESEARCH WAS NEEDED.
                                    P4-4

-------
      DEVELOPMENTS THAT ARE CONSIDERED TO BE NEW TECHNOLOGIES ORIGINATED OUTSIDE




 THE US,  PRINCIPALLY EUROPE AND JAPAN.  THESE INCLUDE PULSE ENERGIZATION,




 INTERMITTENT ENERGIZATION, AND WIDE PLATS SPACING. THESE TECHNOLOGIES OFFER




 ADVANTAGES  IN IMPROVED PERFORMANCE IN COLLECTION OF HIGH RESISTIVITY DUST AND




 COST REDUCTIONS  AS A RESULT OF SAVING OF ENERGY AND STRUCTURAL MATERIALS.









      HOWEVER,  THERE IS A NEED FOR FURTHER RESEARCH TO ESTABLISH A BETTER




 UNDERSTANDING OF THESE TECHNOLOGIES.   ONE CONCEPT IS GENERALLY BELIEVED THAT




 PULSE ENERGIZATION RESULTS IN A MORE  UNIFORM CORONA FOR A ROUND WIRE CORONA




 ELECTRODE RESULTING IN MORE UNIFORM CURRENT DISTRIBUTION THROUGH THE DUST




 LAYER.  THIS  PERMIT WOULD HIGHER AVERAGE CURRENT AND VOLTAGE,  AS WELL AS HIGHER




 PEAK VOLTAGE,  FIELD STRENGTHS AND PARTICLE CHARGE.   IT HAS ALSO BEEN SUGGESTED




 THAT THE SHORT PULSE DURATION PERMITS HIGHER ELECTRICAL OPERATING PARAMETERS




 BECAUSE THE VOLTAGE IS REDUCED BEFORE THE ION CLOUD REACHES THE DUST LAYER.




 THUS  PREVENTING  SPARK PROPAGATION.  THESE EFFECTS  NEED TO BE  EXAMINED AND




 QUANTIFIED TO  PERMIT A MORE RATIONAL  UTILIZATION OF THE TECHNOLOGY.  WE NEED TO




 KNOW,  FOR EXAMPLE,  THE EFFECTS OF PULSE  ENERGIZATION WITH CORONA ELECTRODE




 CONFIGURATIONS OTHER THAN SMOOTH WIRES.   WE ALSO NEED TO KNOW  HOW CRITICAL




 PULSE WIDTH IS.   IF CORONA UNIFORMITY IS  THE IMPORTANT  FACTOR,  IS  VOLTAGE RISE




 RATE OR PULSE  WIDTH THE CRITICAL PARAMETER?   STILL  ANOTHER QUESTION  IS WHETHER




 THE HIGHER RATIO  OF PEAK TO AVERAGE VOLTAGE  IS RESPONSIBLE FOR IMPROVED




 PERFORMANCE.  HIGHER PEAK VOLTAGES RESULT  IN HIGHER PEAK FIELD STRENGTH SO THAT




 CHARGE ON LARGER  PARTICLES IS  HIGHER.  RESOLUTION OF THESE QUESTIONS MAY PERMIT




A MORE RATIONAL APPROACH TO PULSE ENERGIZATION.
                                     P4-5

-------
      INTERMITTENT  ENERGIZATION  IS YET ANOTHER OF THE SO-CALLED NEW TECHNOLOGIES




THAT  HAS RECEIVED  CONSIDERABLE  ATTENTION.   ITS APPLICATION HAS BEEN TO REDUCE




ENERGY WHEN  COLLECTING LOW RESISTIVITY DUST AND REDUCE BACK CORONA WHEN




COLLECTING HIGH RESISTIVITY DUST.  THE PROCEDURE,  FIRST INTRODUCED IN JAPAN.  IS




TO SKIP VOLTAGE CYCLES OR HALF  CYCLES.  THE RESULT IS  HIGHER PEAK VOLTAGE




DURING THE ENERGIZED VOLTAGE CYCLE.  AVERAGE VOLTAGE AND AVERAGE CURRENT MAY




ALSO BE LOWER.  DATA REPORTED IN THE LITERATURE INDICATE A REDUCTION IN ENERGY




WITH NO APPRECIABLE CHANGS IN MASS EMISSIONS IN THE CASE OF LOW RESISTIVITY




DUST AND AN  INCREASE IN EFFICIENCY AT REDUCED  ENERGY LEVELS FOR HIGH




RESISTIVITY  DUST.









     AS WITH PULSE ENERGIZATION, THERE ARE  SOME UNCERTAINTIES  ABOUT  HOW




INTERMITTENT ENERGIZATION INFLUENCES PRECIPITATOR OPERATION.   FROM A




THEORETICAL  STANDPOINT,  HIGHER PEAK VOLTAGE RESULTS IN HIGHER  PEAK HELD




STRENGTH WHICH WOULD GIVE HIGHER CHARGE ON  PARTICLES ABOVE  ABOUT 0.5 /im




DIAMETER.  THE LOWER AVERAGE CURRENT, HOWEVER, WOULD RESULT IN LOWER CHARGE ON




PARTICLES SMALLER THAN ABOUT 0.5 ^m DIAMETER WHERE DIFFUSION CHARGING




PREDOMINATES.  THE LOWER AVERAGE VOLTAGE AND LOWER AVERAGE  CURRENT WOULD ALSO




RESULT IN LOWER FIELD STRENGTH AT THE PLATE, WHICH WOULD REDUCE COLLECTION




EFFICIENCY.  AN OFF-SETTING FACTOR WOULD  BE AN INCREASE IN REMOVAL OF  LARGE




PARTICLES BY THE INLET AND INTERMEDIATE FIELDS.  THIS WOULD REDUCE THE  RATE OF




DUST ACCUMULATION ON THE OUTLET FIELD AND THUS REDUCE THE RAPPING LOSS.









     IN THE CASE OF HIGH RESISTIVITY DUST,  THE SAME GENERAL ANALYSIS WOULD




APPLY. EXCEPT THAT THE OPTIMUM OPERATING POINT IS DETERMINED BY ELECTRICAL




BREAKDOWN IN THE DUST LAYER,  WHICH ESTABLISHES THE MAXIMUM  CURRENT.  IF THIS
                                     P4-6

-------
 CURRENT IS ESTABLISHED BY INTERRUPTING THE VOLTAGE CYCLES RATHER THAN REDUCING




 THE VOLTAGE WITH NORMAL ENERGIZATION, A HIGHER PEAK VOLTAGE WILL BE MAINTAINED.




 THE RESULT WILL BE INCREASED EFFICIENCY BECAUSE OF THE INCREASED CHARGE ON




 LARGE PARTICLES.  THE LIMITATION OF INTERMITTENT ENERGIZATION IS A POSSIBLE




 INCREASE IN EMISSION OF SMALL PARTICLES.  IN THE CASE OF LOU RESISTIVITY DUST,




 AND THE INCREASE IN EFFICIENCY THAT CAN BE REALIZED WHEN COLLECTING HIGH




 RESISTIVITY DUST COMPARED TO ALTERNATIVE TECHNIQUES.   ADDITIONAL RESEARCH IS




 NEEDED TO  ESTABLISH THE LIMITS ON THE USE OF.IE,  PARTICULARLY THE EFFECT ON




 COLLECTION OF FINS PARTICLES.









      WIDE  PLATS SPACING,  THE LAST OF THE  NEW TECHNOLOGIES  HAS BEEN RATHER




 WIDELY ADOPTED BY MOST MANUFACTURERS.   IT APPARENTLY  GIVES A HIGHER COLLECTING




 FIELD DUE  TO  INCREASED SPACE CHARGE ASSOCIATED WITH WIDER  SPACING AND MORE




 UNIFORM CURRENT DENSITY.   THESE FACTORS RESULT IN HIGHER MIGRATION VELOCITIES




 SO  THAT LESS  PLATE AREA IS REQUIRED TO ACHIEVE THE DESIRED EFFICIENCY.









      IN ADDITION TO THESE TECHNOLOGIES, OTHER  FACTORS  WILL INFLUENCE THE




 DIRECTIONS  OF PRECIPITATOR RESEARCH AND DEVELOPMENT.   THESE RELATE TO THE




 METHODS  THAT  WILL BE USED TO CONTROL GASEOUS POLLUTANTS.   IN THE  US, CONTROL OF




 SULFUR  OXIDE  EMISSIONS  WILL  BE THE MOST IMPORTANT CONSIDERATION,  AT LEAST FOR




THE NEAR TERM.   THERE ARE A  NUMBER OF METHODS  FOR REDUCTION OF S02  EMISSIONS




CURRENTLY UNDER DEVELOPMENT, AND  THE EFFECTS ON PRECIPITATOR APPLICATION  WILL




DEPEND ON THE  PARTICULAR  S02 CONTROL METHODS USED.  MOST LIKELY,  THERE WILL BE




A NUMBER OF METHODS ADOPTED DEPENDING  ON  THE PARTICULAR LOCAL SITUATION.
                                    P4-7

-------
     FOR  SOME  S02 CONTROL METHODS,  SUCH AS  IN DUCT INJECTION OF CALCIUM




COMPOUNDS, THE RESULT WILL  BE A GREATLY INCREASED INLET DUST BURDEN,  A SMALLER




DUST PARTICLE  SIZE, AND  PERHAPS A HIGHER RESISTIVITY.   ALL OP THESE FACTORS




ADVERSELY INFLUENCE PRSCXPITATOR PERFORMANCE.   IF THESE S0a CONTROL




TECHNOLOGIES ARE RETROFITTED TO EXISTING PLANTS,  RATHER DRASTIC METHODS  MUST  BE




APPLIED TO PREVENT EXCESSIVE PARTICULATS EMISSIONS.  AT PRESENT,  THE  ONLY




OPTIONS WOULD  BE TO ADD  PLATS AREA  OR GO TO AN ALTERNATE PARTICU1ATE  CONTROL




DEVICE, SUCH AS A FABRIC FILTER EITHER IN SERIES  WITH OR REPLACING  THE ESP.









     THERE MAY BE SOME METHOD FOR INCREASING ESP  PERFORMANCE  THAT MAY IN SOME




INSTANCES BE SUFFICIENT TO ACCOMMODATE THE INCREASED DUST BURDEN.   IN ADDITION




TO SOME OF THE MORE RECENT TECHNOLOGIES DISCUSSED, MODIFICATION OF  CORONA




ELECTRODE DESIGN MAY GIVE TWO BENEFICIAL EFFECTS.  BY RAISING THE AVERAGE




FIELD,  THE BUILD UP OF A DUST CONCENTRATION GRADIENT WOULD BE ACCELERATED.




THIS WOULD HAVE THE EFFECT OF INCREASING MIGRATION VELOCITY SINCE A HIGHER




PERCENTAGE OF THE DUST WOULD FALL IN THE BOUNDARY LAYER AND BE  COLLECTED.  IT




HAS BEEN SHOWN ON PILOT SCALE TESTS THAT LARGER DIAMETER  ELECTRODES INCREASE




COLLECTION EFFICIENCY BY INCREASED FIELD STRENGTH AT THE  COLLECTION ELECTRODE




AND BY AN INCREASED AVERAGE FIELD.   FURTHER RESEARCH IS NEEDED  TO ESTABLISH THE




RELATIONSHIP BETWEEN AVERAGE ELECTRIC FIELD AND TURBULENCE ON PARTICLE




CONCENTRATION GRADIENT,  TO QUANTIFY THE INCREASED COLLECTION  EFFICIENCY




RESULTING FROM A HIGH CONCENTRATION GRADIENT,  AND THE INCREASED COLLECTION




FIELD RESULTING FROM ALTERNATE ELECTRODE DESIGNS.









     SO THESE REPRESENT SOME OF THE DIRECTIONS FOR ESP RESEARCH THAT  ARE NEEDED




TO IMPROVE PERFORMANCE AND REDUCE THE COST OF PARTICULATE  CONTROL.
                                     P4-8

-------
 FABRIC FILLERS




      FABRIC FILTERS REPRESENT AN ALTERNATIVE TO ESP'S FOR PARTICULATE CONTROL.




 THEY HAVE BEEN USED FOR MANY YEARS TO COLLECT PARTICULATE IN LOW GAS VOLUME




 APPLICATIONS,  BUT IN RECENT YEARS THERE HAS BEEN INCREASING INTEREST AND




 APPLICATION OF FABRIC FILTERS FOR PARTICULATE CONTROL ON ELECTRIC UTILITY




 BOILERS.   AS OF THE END OF 1989,  THERE WERE ABOUT 100 BAGHOUSES IN OPERATION ON




 ELECTRIC  UTILITY BOILERS WITH AN INSTALLED CAPACITY OF ABOUT 21,000 MW,









      THE  MAJOR IMPETUS FOR FABRIC FILTERS FOR UTILITY SERVICE WAS THE INCREASE




 IN THE USE OF  LOW SULFUR COALS,  PARTICULARLY IN THE WESTERN PART OF THE  US.   TO




 MEET EMISSION  LIMITS,  ESP'S MUST  EITHER USE FLUE GAS CONDITIONING TO REDUCE




 RESISTIVITY OR INSTALL VERY LARGE UNITS.   SINCE FABRIC FILTERS  ARE NOT




 SENSITIVE TO ELECTRICAL PROPERTIES OF THE ASH,  THEIR USE HAS  INCREASED




 SIGNIFICANTLY  FOR COLLECTION OF HIGH  RESISTIVITY ASH FROM LOW SULFUR COALS,




 THE  PRIMARY DIFFERENCE BETWEEN ESP'S  AND  FABRIC FILTERS  IS  THAT EFFICIENCY OF




 ESP'S  CAN VARY WITH PHYSICAL AND  ELECTRICAL PROPERTIES OF THE DUST BEING




 COLLECTED,  BUT ENERGY  CONSUMPTION AND PRESSURE  DROP ARE  INDEPENDENT.




 EFFICIENCY OF  FABRIC FILTERS  ON THE OTHER HAND,  ARE NOT  SO  SENSITIVE TO DUST




 PROPERTIES BUT PRESSURE DROP  IS.









     IN RECENT YEARS,  RESEARCH ON FABRIC  FILTERS HAS BEEN CONCERNED WITH




TECHNIQUES  TO  PREDICT  PRESSURE DROP AND TO REDUCE IT.  ONE  CHARACTERISTIC OF




FABRIC FILTERS  IS THAT  PRESSURE DROP  INCREASES  WITH TIME FOR  REVERSE GAS




CLEANING  SYSTEMS.   STUDIES  ON  PILOT SCALE  UNITS  INDICATED THAT  USE OF SONIC




HORNS ON  REVERSE CAS UNITS  GAVE BETTER BAG CLEANING AND  REDUCED PRESSURE DROP.




TODAY, ABOUT HALF OF THE REVERSE GAS UNITS  USE  SONIC HORN ASSISTED BAG




CLEANING,  EITHER CONTINUOUSLY  OR INTERMITTENTLY OPERATED.
                                    P4-9

-------
      ANOTHER PAST AREA OF RESEARCH HAS BEEN USE OF ELECTRIFIED BAGS.




 PRECHARGERS  OF VARIOUS KINDS WERE USED TO CHARGE DUST PRIOR TO ENTERING THE




 BAGS.  THE RESULTING  DUST CAKE WAS MORE POROUS AND THE PRESSURE DROP  LOWER.




 HOWEVER,  INTEREST IN  APPLICATION  OF THIS TECHNOLOGY HAS DECLINED SINCE IT




 APPEARS THAT OTHER METHODS OF DUST CAKE MODIFICATION WOULD BE MORE COST




 EFFECTIVE.   FUTURE RESEARCH ON FABRIC  FILTRATION WILL BE A CONTINUATION OR




 EXTENSION OF PRESENT  STUDIES, MAINLY OF FACTORS THAT REDUCE PRESSURE  DROP,




 IMPROVE BAG  LIFE,  INCREASE AIR TO  CLOTH RATIOS,  AND IMPROVE THE ABILITY TO




 PREDICT PERFORMANCE BASED ON COAL  AND  ASH PROPERTIES.









     A CURRENT RESEARCH EFFORT IS  THE  USE OF CONDITIONING AGENTS TO AFFECT BAG




HOUSE OPERATION.  AMMONIA ADDITIONS  TO  THE FLUE GAS  PRODUCES  AMMONIUM BISULFATE




AND AMMONIUM  SULFATE  IF THE LEVEL  OF S03  IS SUFFICIENT.   AMMONIUM SULFATE




INCREASES ASH ADHESION AND PRODUCES  A HIGHER POROSITY  DUST CAKE AND




CORRESPONDING REDUCTION IN PRESSURE  DROP-  ANOTHER USE OF AMMONIA CONDITIONING




IS TO REDUCE PARTICULATE  EMISSIONS DURING  THE CLEANING CYCLE.   SOME ASHES  WITH




SMOOTH SPHERICAL SURFACES TEND TO  PENETRATE THE  DUST CAKE AND  INCREASE




EMISSIONS AND OPACITY ESPECIALLY DURING THE CLEANING CYCLE.  ANOTHER  PROBLEM




EXPERIENCED WITH FABRIC FILTERS IS ACID ATTACK WHEN COLLECTING  ASH  FROM HIGH




SULFUR COALS.  THE SO, PRESENT IN THE FLUE GAS COMBINES WITH WATER  IN LOW




TEMPERATURE REGIONS AND ACID IS CONDENSED ON PARTS OF  THE  BAG.   CALCIUM




HYDROXIDE AS  WELL AS AMMONIA ADDITIONS  TO THE FLUE GAS REDUCE  S03




CONCENTRATIONS AND PREVENT BAG FAILURE  DUE TO ACID ATTACK.
                                   P4-10

-------
      RESEARCH IS ALSO CONTINUING ON BAG FABRIC.  IN THE US, FIBER GLASS HAS




 BEEN USED ALMOST EXCLUSIVELY ON SAGS FOR ELECTRIC UTILITY SERVICE BECAUSE OF




 ITS ABILITY TO WITHSTAND FLUE GAS TEMPERATURE.   IN AUSTRALIA, POLYMERIC FABRICS




 HAVE BEEN USED BECAUSE THE EXTREMELY LOW SULFUR COAL AND HIGHLY ALKALINE ASH




 PERMIT OPERATION AT MUCH LOWER TEMPERATURES THAN UTILITIES IN THE US.   RECENT




 DEVELOPMENTS OF HIGHER TEMPERATURE POLYMERS IN THE US HAVE PROVIDED NEW FABRICS




 THAT OFFER PROMISE OF HIGHER EFFICIENCIES OF BETTER CLEANING PROPERTIES, AND




 LOWER PRESSURE DROPS THAN CURRENTLY USED FABRICS.   THE NEWER FABRICS CAN BE




 MORE EASILY TEXTURI2ED TO DECREASE PARTICLE PENETRATION AND YET PERMIT BETTER




 CLEANING.








      PULSE JETS FOR BAG CLEANING ARE A NEW TECHNOLOGY TO ELECTRIC UTILITY




 APPLICATIONS ALTHOUGH THIS TECHNOLOGY HAS BEEN  USED FOR MANY YEARS IN  OTHER




 APPLICATIONS.   THE BETTER CLEANING BY PULSE JET UNITS OFFER THE POSSIBILITY OF




 HIGHER AIR TO CLOTH RATIOS AND SMALLER BAG HOUSE VOLUMES WITHOUT SIGNIFICANTLY




 HIGHER PRESSURE DROPS.








     A CONTINUING  AREA OF RESEARCH IN DEVELOPMENT  OF DIAGNOSTIC PROCEDURES  THAT




 WILL PERMIT ANALYSIS OF ASH CHARACTERISTICS AND PERMIT A PREDICTION OF PRESSURE




 DROP BASED  ON ASH  AND COAL PROPERTIES.   BETTER  MEASURES OF  THE  PHYSICAL




 PROPERTIES  OF THE  ASH AND CORRELATION WITH POROSITY,  AND ADHESIVELY WILL PERMIT




A MORE RATIONAL BAG  HOUSE DESIGN.   FOR NEW INSTALLATIONS WHERE  ASH SAMPLES  ARE




NOT AVAILABLE, A NEED  EXISTS TO RELATE  ASH PROPERTIES  TO COAL CHEMISTRY.








     ONE PROBLEM AFFECTING  BAG LIFE  IS  BUILD-UP  OF BAG WEIGHT OVER A PERIOD OF




TIME.  INCREASED BAG WEIGHT CAN CAUSE  CUFFING WHICH  FATIGUES  THE BAG NEAR THE
                                     P4-11

-------
ATTACHMENT POINT AND REDUCES BAG LIFE.   THE  PROBLEM APPEARS MORE FREQUENTLY IN




UNITS OPERATING IN A MODE THAT CYCLES BELOW  THE MOISTURE DEW POINT AND CAUSES A




CEMENTATIOUS MATERIAL  THAT CANNOT BE REMOVED BY NORMAL CLEANING METHODS.









     THE EFFECTS OF SOj AND NOX CONTROL  ON OPERATION OF FABRIC FILTERS IS




ANOTHER AREA OF RESEARCH.  DEPENDING ON  THE  METHOD OF CONTROL OF THESE




POLLUTANTS. PROPERTIES AS WELL AS QUANTITIES OF PARTICULATE MATERIAL WILL VARY.




EFFECTS OF SORBENT ADDITIONS ON PRESSURE DROP AND  OTHER PHASES OF BAG HOUSE




OPERATION NEED TO BE INVESTIGATED SO THAT OPERATING PARAMETERS CAN BE




PREDICTED.








     CLEAN UP OF HIGH TEMPERATURE, HIGH  PRESSURE COMBUSTION GASES  IS  A




TECHNOLOGY DRIVEN BY PROCESS AND ECONOMIC RATHER THAN BY AIR POLLUTION




CONSIDERATION.  IN PRESSURIZED FLUID BED COMBUSTORS,  THE CONCEPT IS THAT HIGHER




EFFICIENCIES CAN BE OBTAINED BY USING A  GAS  TURBINE TO EXTRACT ENERGY FROM  THE




GASES PRIOR TO A CONVENTIONAL STEAM CYCLE.   THE OTHER NEED  FOR HOT GAS  CLEAN UP




IS FOR HIGH TEMPERATURE HIGH PRESSURE GASIFIERS.   THIS  PROCESS  PRODUCES METHANE




CARBON MONOXIDE AND OTHER GASES THAT ARE FIRST PASSED THROUGH A GAS TURBINE




FOLLOWED BY COMBUSTION OF THE GASES, A SECOND TURBINE,  AND  FINALLY A




CONVENTIONAL STEAM CYCLE.  BOTH PROCESSES CAN INCREASE  OVERALL EFFICIENCIES BY




SEVERAL PERCENT OVER THOSE OPERATED AT ATMOSPHERIC  PRESSURE.









     SEVERAL TECHNOLOGIES ARE CURRENTLY BEING CONSIDERED FOR HOT GAS  CLEAN UP.




MOST OF THE PRESENT RESEARCH IS DIRECTED TOWARD CERAMIC  FILTERS OF VARIOUS




CONFIGURATIONS.   CANDLE FILTERS UTILIZE A CYLINDRICAL CONFIGURATION WITH THE




EXTERNAL SURFACE COMPOSED OF A SMALL PORE SIZE CERAMIC AND A  LARGER PORE SIZE




CERAMIC SERVING AS THE SUBSTRATE FOR MECHANICAL SUPPORT.
                                     P4-12

-------
      THE SMALL PORE SIZE CERAMIC SERVES AS THE FILTER MEDIUM IN CONTRAST TO LOW




 TEMPERATURE FABRIC FILTERS WHERE THE DUST CAKE IS THE PRIMARY FILTERING MEDIUM.




 CLEANING OF THE CERAMIC FILTER IS BY HIGH PRESSURE PULSES THAT ARE INTENDED TO




 CLEAN DOWN TO THE BARE CERAMIC SURFACE.  THIS APPROACH IS PRACTICALLY




 UNAVOIDABLE SINCE THE CERAMIC MATERIAL CANNOT FLEX OR BEND TO BREAK UP THE DUST




 CAKE.









      AN ALTERNATIVE CONFIGURATION IS THE CERAMIC CROSS FLOW FILTER IN WHICH GAS




 ENTERS THROUGH A SET OF PARALLEL CHANNELS,  PASSES THROUGH A SMALL PORE SIZE




 CERAMIC, AND OUT THROUGH CHANNELS PERPENDICULAR TO THE DIRECTION OF INLET GAS




 FLOW.









      THE MAJOR PROBLEM WITH CERAMIC FILTERS SO FAR HAS BEEN MECHANICAL OR




 THERMAL FAILURE OF THE FILTER MEDIA.   TO DATE,  THERE HAS BEEN INSUFFICIENT




 OPERATING TIME TO ASSESS THE LONG TERM FILTERING PROPERTIES OF THE CERAMICS,




 PARTICULARLY THE PRESSURE DROP BUILD  UP OVER A LONG PERIOD OF OPERATION.








      A SECOND APPROACH TO HOT GAS CLEAN UP  HAS  BEEN MOVING GRAVEL  BED  FILTERS.




 THIS  TYPE FILTER INVOLVES USE OF  CERAMIC BEADS  AS THE FILTER MEDIUM WITH




 CIRCULATION  TO AN OUTSIDE FACILITY FOR REMOVAL  OF DUST PARTICLES FROM  THE




 BEADS.   BECAUSE THE  FILTER MEDIUM IS  CIRCULATED TO A LOWER TEMPERATURE ZONE  FOR




 CLEANING,  FLUE GAS TEMPERATURE DROP ACROSS  THE  FILTER BED  HAS BEEN UNACCEPTABLY




 HIGH.









     A THIRD APPROACH HAS BEEN USE OF  ELECTROSTATIC  PRECIPITATION.  MOST OF  THE




RESEARCH TO DATE HAS BEEN TO DETERMINE WHETHER A  STABLE CORONA COULD BE




ESTABLISHED AT THE OPERATING TEMPERATURE AND PRESSURE.  DEVELOPMENT HAS BEEN
                                      P4-13

-------
HAMPERED BY PROBLEMS OF  INSULATOR DESIGN AND  STRUCTURAL PROBLEMS.   IT HAS BEEN




FELT THAT THESE WERE MORE  FORMABLZ THAN THOSE OF  CERAMIC FILTERS,  AND




CONSEQUENTLY, LITTLE HAS BEEN  DONE TO  OPTIMIZE AN ES? DESIGN FOR THIS SERVICE.









     RESEARCH FOR THE IMMEDIATE  FUTURE WILL PERHAPS  BE CONSTRAINED TO




DEVELOPMENT OF MATERIALS AND DESIGN OF CERAMIC FILTERS CAPABLE OF WITHSTANDING




THERMAL AND MECHANICAL LOADS FOLLOWED  BY STUDIES  OF  LONG TERM FILTERING




PROPERTIES.








     AT PRESENT, THERE ARE NO ACTIVE PROJECTS I AM AWARE OF ON USING  ESP'S FOR




HOT GAS CLEANUP. SHOULD  CERAMIC  FILTERS NOT BE MADE  TO FUNCTION SATISFACTORILY.




THERE ARE SOME APPROACHES TO ESP  DESIGNS THAT PROMISE  TO GIVE THE  REQUIRED




DEGREE OF GAS CLEANING.









     THESE THEN, REPRESENT MY ASSESSMENT OF FUTURE DIRECTIONS  OF PARTICULATE




CONTROL TECHNOLOGY.  I HOPE THAT  THIS HAS POINTED OUT  THAT  THERE ARE A NUMBER




OF IMPORTANT POSSIBILITIES FOR IMPROVEMENT IN PARTICULATE REMOVAL  DEVICES.  THE




CONCEPT THAT WE KNOW ALL THE ANSWERS ON THE CONVENTIONAL DEVICES IS SHORT




SIGHTED,   CHANGES IN REQUIREMENTS FOR CONTROL OF SMALL PARTICULATE, AS WELL AS




CHANGES IN THE NATURE OF THE PARTICULATE TO BE CONTROLLED, MAKE  IT NECESSARY




THAT WE HAVE THE KNOWLEDGE TO PROPERLY DESIGN EQUIPMENT TO MEET THE NEW DEMANDS




AS THEY OCCUR.
                                       P4-14

-------
                     DU FONT'S ENGINEERING FIBERS
                         FOR HOT GAS FILTRATION
                             CASE HISTORIES

                              P. E. Frankenburg
                     E. I. Du Pont de Nemours & Co., Inc..
                                  Oak Run
                            Chestnut Run Plaza
                         Wilmington, DE  19880-0701
Abstract
Among the many fibers produced by Du Pont, are several which have the properties
necessary for application in hot gas filtration.  The range of recommended operating
conditions for using these fibers are discussed, and are then exemplified further by the
presentation  of typical operational experience for these materials under a wide variety of
conditions. Data presented will  include case histories on:

        Stoker industrial boiler burning high  sulfur coal
        Fluidized bed industrial boiler
        Pulverized coal  utility boiler burning a variety of coals from different
        sources
        Hazardous waste incinerator removing hydrocarbons from
        contaminated soil
                                      21-1

-------
                      DU FONT'S ENGINEERING FIBERS
                         FOR HOT GAS FILTRATION
                             CASE HISTORIES
BACKGROUND
Du Pont, founded in 1802 as a small gun powder manufacturing company, has grown to
become a highly diversified, chemistry based, international enterprise, filling a broad range
of human needs. Products are used in food production, crop protection, packaging,
clothing, transportation, health care, recreation, and shelter.

Specializing in high technology, Du Pont develops and manufactures products based on
the chemical, biological, and physical sciences, and sells them to every branch of industry.
It is also active in the life sciences, and since its merger with Conoco in 1981, has
expanded its activities into energy and feed stock exploration, production, and processing.

Du Font's annual worldwide sales exceed 33 billion and dollars, and the company's total
assets approach the $30 billion dollars.  Over 140,000 employees work in offices,
laboratories, plants,  mines, oil fields, and refineries throughout the world.

A large segment of the Company's expertise is in the area of polymer chemistry,  and in
synthetic fiber in particular.  In addition to being a world scale producer of fibers for textile
use, Du Pont is also a major manufacturer of speciality fibers for industrial applications.
These are marketed through the  Industrial Products Division of the Fibers Department.
Products include fibers with exceptionally high strength, good to excellent chemical
resistance, as well as high thermal resistance.  They include such products as Kevlar®  and
Nomex® aramids, Teflon® felt and Tefaire® felt fluoropolymer products, useful in pollution
control, as well  as Tyvek® spunbonded  olefin, Sontara® spunlaced fabric and Cordura®
nylon for envelopes, medical gowns and drapes, and sporting goods.

Engineering Fibers for Hot Gas Filtratign

Since the commercialization of Nomex® aramid fiber in the late 1960's, Du Pont has
become a major supplier of fibers for dry paniculate filtration media in both woven fabric
and felt form. Today there are three filtration products designed to operate continuously at
temperatures between 100 and 260°C and under a variety of chemical conditions. Du  Pont
offers: Nomex® aramid, Teflon® TFE  fluorocarbon fiber, and since 1986, Tefaire®, a felt
                                       21-2

-------
made from a blend of Teflon® and glass fibers, as well as Dacron® and Orion® acrylic
fibers for lower temperature use.

Each one of these products has an effective position in a spectrum of properties ranging in
temperature from 100 to 260°C, and in a chemical environment that can cover a wide range
of conditions in terms of pH, humidity, and gas components.  A diagram showing the
position of the Du Pont products over a range of conditions is shown in the first slide (Slide
#1).
It can be seen that the fiber capable  of operating under the most severe operating
conditions namely Teflon®, occupies that portion of the diagram which encompasses the
region with the most severe operating conditions in temperature, and especially in chemical
terms. Next to it, with the same temperature capability, but with a slightly lower chemical
resistance due to the sensitivity of glass to hydrogen fluoride, is Tefaire®, a fiber blend of
uniformly inter dispersed Teflon® and glass fibers.  Woven glass filters, with which we are
all familiar, also fall into this general  applications area. However, they are more sensitive to
both chemical and mechanical conditions. Finally,  capable of operating at sustained
temperatures up to 200°C, and under less severe chemical conditions, is Nomex®.
Poly(acrylonitrile) and poly(ester)  are added for reference.  One can easily see how, at a
temperature above 210°C, only Teflon® or Tefaire® are expected to have an acceptably
long useful life. The delineation between these is a matter of gas chemistry and filtration
requirements.  As the temperature drops below 210°C, Nomex® comes into play as well,
but only under mildly aggressive chemical/humidity conditions.  Finally, as the temperature
drops to less than about 130°C, PAN comes into play, and below 120°C PET.   It should be
remembered that the diagram represents a generalized picture.  Selection of the optimum
product is best arrived at by consultation with fiber producers, filter media manufacturers,
filter equipment manufacturers, bag makers, or consultants.
We will now discuss recent histories  of installations in which these fibers were  used.

Case Histories.
      A) Fluidized Bed Boiler- Nomex®.
            1) Description and history. A circulating bed boiler, represents one approach
to control paniculate and acidic components, while offering a route to good NOx control
and high combustion efficiency. This control methodology has the advantage that its output
is an easily disposable, neutral, dry dust.  However, many boilers of this type, when fitted
with bag houses containing woven glass bags, have encountered particulate control and
                                      21-3

-------
pressure drop problems.  This was also the case at the location described in the next slide
(#2).

The boiler, by Lurgi/Combustion Engineering Company, burned 800 MM BTU of fuel per
hour, and generated 650,000 Ibs of steam during the same time interval. CaCOS
consumption was 93% of stoichiometry.

The fuel, Eastern Pennsylvania anthracite culm, had a fuel value of 6,000  9,000
BTU/pound and contained 0.45 - 0.50 percent sulfur. The ash content averaged 40
percent. The ash was extremely fine, with a median diameter of 36 microns, but with 83%
of the population below microns and 50% below 0.5 microns.

Gas generated by the boiler (Slide #3) passed through a heat exchanger and a number of
cyclones before entering the bag house. Conditions at the bag house entrance were:
360°F (180°C); low water content; and adust load of 4   13 grains/actual ft3 (9  30
grams/m3). The gas flow was 950,000 actual ft3/min (440 m3/sec).

The bag house  (Slide #4)  is a Standard Havens pulse jet design, contains 16
compartments of 210 bags each, and is operated with off-line cleaning. It operates with an
air-cloth ratio (net) of  4.0 ft3/ft2/min (1.2 m3/m2/min). The bags are 5.5 inches in diameter,
and have a length of 241 inches (140 mm x 6.1 m).  The bags initially were Gl_ 04 Textured
x Textured woven  glass, 16.5 oz/yd2 (560 g/m2).

These bags were initially installed in June 1986, but proved to be unsatisfactory.  They
blinded, resulting in a high pressure drop and excessive opacity, which, in  turn, limited the
capacity of the installation. (Slide #5 & 5A).  In February1987, all of the bags were replaced,
half with 14 oz/yd2 (475 g/m2) Nomex® needlefelt bags and the other half with 16.5 oz/yd2
(560 g/m2) woven glass. The glass bags performed poorly and were replaced in October
1987.
            2 Current operation.  After 23 - 30 months' operation with Nomex bags, the
unit  continutes to operate near design efficiency and with a regulated pressure drop of 5.0
inches water (125 mm H2o).  Each bag is being pulsed approximately every 20 minutes.
Emissions are less than 0.05 Ibs MM BTU (0.005 grains/ACF; 11 mg/m3). There is some
evidence of abrasion  of the fabric in the bottom disk of the bags.  Analysis of bags removed
from the bag house indicated an expected bag life of 48 months.
                                      21-4

-------
Summary

This installation exemplifies the necessity of matching filter medium characteristics with
operational requirements.  In this case, the combination of temperature requirements and
dust characteristics governed the preference.  In addition, the difference in performance
characteristics between a woven structure and a needlefelt demonstrate the substantial
difference in results one obtains with these structures.  In a woven structure, the filtration  is
primarily accomplished by a filter cake which is supported on a plate with small openings
through which the air passes. These openings represent about 20% of the total surface of
the medium.  As a result,  the gas velocities through the pores can be very high, and filtration
efficiency can suffer, or, if particulates are trapped in the pores, high pressure drops result.
Needlefelts, on the other  hand, represent a continuous open surface, with virtually 100% of
the surface permitting air  passage. This reduces the local air velocity, improves paniculate
capture efficiency, and reduces pressure drop. In  addition, since a needlefelt is made up of
individual fibers,  rather than fiber bundles, in an array,  the surface area available for
particle capture is much greater than in woven fabrics.

       B) Industrial Boiler - Tefaire®

             1)  Description and History   One of the first installations of Tefaire®
needlefelt has been running continuously since December 1983 at one of our own plant
sites.  At the time of installation, a second bag house, on a similar boiler, and fitted with
100% Teflon® bags was also in service.  One year after installation, the Tefaire® bags
typically operated at a pressure drop of about 1/2 that of the Teflon® bags.  With time,
however, the two filter media reached about the same  pressure drop,  primarily, we believe,
due to the high percentage of un-burned hydrocarbon in the Stoker fly ash.  The Teflon®
bags were removed at least once since their installation for cleaning by washing.  The
Tefaire® bags have never been removed, in spite of periodic boiler excursions which
resulted in hydrocarbon deposition.  A summary of the pertinent data is given in (Slide #6 &
6 A).

             2)  Cleanability  Studies on the cleanability of the Tefaire® bags
demonstrated two commercial  processes for achieving the desired result.  Dry cleaning with
a solvent for the un-burned hydrocarbons, in an industrial laundry yielded clean bags with
essentially unchanged physical properties. Washing of the same bags in a commercial
                                        21-5

-------
laundry, by using the cycles: rinse, rinse, detergent wash, spin dry, air dry, resulted in
equally clean bags, again without a change in properties.

            3) Durability  After 4.5 years' service, a few bags were removed to assess
their properties. Details are given in (Slide #7).  The complete retention of felt properties for
this period of use, indicates that the bags of Tefaire® are essentially unaffected by flue gas
conditions resulting from the combustion of high sulfur coal.

Summary

On the basis of this study, and data obtained since then, we estimate the average Tefaire®
felt life to be in excess of four years or 150 thousand pulse events, whichever is sooner. Of
course operating parameters, such as particle size distribution, dust cohesion and
adhesion tendencies, and mechanical problems will affect bag life as well, and will also
have to be taken into consideration.

      C. Pulverized Coal Utilities Boiler - Tefaire®

            1) Description  Differing fly ash characteristics from various coals often
present more of a problem than can be handled by a single filter medium. At a municipal
power station in West Germany, filter bags made of Du Font's Tefaire® needlefelt
demonstrated the ability to operate satisfactorily under such conditions,  while meeting very
tight emission requirements. The boiler in this installation was a 160 MW slag-tap boiler,
utilizing coals from Poland, West Germany, South Africa, and Australia.  The slag-tap
system was installed to melt the fly ash and to extract it as a liquid, so as to produce large
ash  particulates which were ecological more acceptable, and had value as a highway
ballast material. This resulted in very high dust loads, ranging from a maximum of 40 g/m3
(17 grns/ft3) to a more normal 15 g/m3 (7 grns/ft3). The fly ash obtained in the combustion
of coals from W. Germany and S. Africa were relatively easy to handle, but the use of Polish
and  Australian coals presented  problems. These two ashes were of the non-agglomerating
type, and were slow to settle under standard bag house operating conditions.  Particle size
distribution was not a problem, with a mean particle size of about 20 microns observed.

            2) History   Initially the bag house had been fitted with bags of glass felt, and
these operated acceptably.   However, after 6 months' operation the bags failed due to
degradation of the binder resin,  which in turn caused de-lamination of the filter from its
                                        21-6

-------
support.  Filter bags of 100% Teflon® fiber were then installed, but these were not
acceptable because of excessive paniculate leakage.

Finally, bags of Tefaire® were installed. The filter medium basis weight was modified for
optimum  filtration and pressure drop performance by evaluating various basis weight
materials over a range of air/cloth ratios, using fly ash obtained from the more difficult coals
from the power station.  A diagram of the equipment used to define the filter medium is
given in Slide #8.

After the  bags of Tefaire® medium were installed, we still encountered some problems with
pressure  drop, especially with the more difficult coals.  The problems were traced to two
causes:
       a) The air/cloth ratio was substantially higher than specified.
       b) The fly ash had a tendency to be re-entrained, since it was
           so light and non-agglomerating.

In time, conditions were altered to incorporate an off-line cleaning operating mode. This
permitted the very light fly ash types to settle without being re-entrained, even though the
average air cloth ratio was raised an additional 20%. Typical steady state operating
conditions are shown in (Slide #9 & 9A.)
            3) Steady - State Operation

      This mode of operation was maintained for about 1.5 years with only one major
upset. That upset was caused by a  malfunctioning of the acid neutralization scrubbing
system then being installed. This caused excessive condensation in the bag house and
lead to blinding.  One compartment of bags was removed, and the surface of the bags was
brushed clean using stiff bristle, long handle brooms.  The remaining compartments were
pulsed while still warm, and eventually cleaned themselves.
            4) De-sulfurization

      Finally, the de-acidification system was  brought on-stream two years after the
Tefaire®  bags had been installed. This system lowered the SOx content and the gas
temperature to the point where chemical conditions were much more benign (reduced SOx,
65°C). As a result, bags of  polyester needlefelt were installed.  The Tefaire® bags have
been retained for future use.
                                       21-7

-------
Summary

      This installation demonstrated the versatility of the filter medium, its ability to recover
from extreme operating upsets, demonstrated the cleanability of the filter medium, and
finally demonstrated that the filter medium will operate under conditions of high
temperature, high dust load.  All of this while maintaining the low emission levels required
in todays' new installations.
      D  Circulating Bed Combustor - Site Remediation

      1)    Description and History   A turnkey site remediation  project, removing oil
seepage from over 11,000 tons of soil by combustion, had been in  operation since late
1988. A diagram of the unit is given in Slide #10. The circulating bed combustor (CBC) is
reported to be capable of removing most hazardous RCRA wastes to a level of over
99.99% (1).

The system uses high velocity air to entrain circulating solids in a highly turbulent
combustion loop.  Solid waste is introduced between the cyclone and combustion chamber,
and is exposed to uniformly high temperature, resulting in rapid heating of the feed and
highly efficient combustion. Limestone can be fed to remove acidic components,
eliminating the need for wet scrubbers down-stream of the combustor. The hot gas,
containing substantial amounts of finely dispersed soil, passes through a gas heat
exchanger, where it is cooled to about 190°C (380°F), and passed  on to the  bag  house.
Essentially any type of auxiliary fuel can be used to maintain the required combustion
temperature of 870°C (1600°F).  The site can be seen in Slide 10A.

The bag  house, supplied by Flex Kleen, contains 72  filter bags on 21 wire mild steel cages.
Operating conditions are given in Slide #11.

            2) History   The initial soil cleaning operation  was hampered by bag failures.
The bags used were PTFE membrane on a woven glass substrate.   Failure was due to the
mechanical action aggravated by the need to pulse very frequently (about
once/minute/bag) to maintain acceptable pressure drop.  This resulted in an  average bag
life of about 10 days.  All other operating requirements were  met. The frequent need to
change bags  represented an  obvious efficiency problem to the operation, and the search
for an alternate filter medium  was undertaken.
                                      21-8

-------
In consideration of the operating conditions, the choice of materials was limited to those
which could filter with the required level of efficiency and under the conditions of
temperature and chemistry imposed by the combustor system.  A listing of possible media,
and their acceptability is given in Slide #12.

The long list of filter media, and their properties, which were considered, points out the level
of information required to reach a satisfactory decision. Not only does one need to choose
between construction variations (such as weave design for woven media, felt weights,
surface treatments), but response to basic temperature and chemical environment
conditions needs to  be known as well.

Eventually, bags of Tefaire® PTFE//glass needlefelt were chosen for evaluation.  Results
with this medium were excellent, with acceptable performance for 72 days. Stack
emissions evaluation at that time indicated that the dust leakage rate was approaching the
allowable limit for dust  emissions and bag replacement was indicated.

"Visilite" testing at the conclusion of the operating period indicated that there were no bag
failures, and that only one bag appeared to have a slight  leak.  Subsequent analysis found
that the latter was related to bag installation, and not related to bag structure failure. Cross-
section photomicrographs of bags removed from service showed substantial penetration of
dust through the filter medium, indicating that the increase in leakage appeared to have
resulted via a sifting mechanism,  aggravated by the need for frequent pulsing.  Testing  of as
received fabric in our laboratory panel tester, using fly ash from  a pulverized coal boiler,
showed leakages (under less frequent pulsing conditions) to be about 0.002 grains/ft3 (5
g/m3), which is essentially the same value obtained with virgin, conditioned filter medium.
Mullen burst strength, as well as other properties, were unchanged from an un-used control
indicating that no degradation of strength parameters had occurred (Slide #13).  We
recommended to the customers that the bags be washed  and held in  reserve for future use.
Tefaire® needlefelt continues to be in use at this site.

REFERENCE

(1) OGDEN Environmental Services Literature
                                       21-9

-------
Product Positioning • Du Pont Fibers

  •np. D«Q. C
                            Temp. D*fl. f
  Dry/Neutral
            Chemical Environment
                                         Circulating Bed Boiler
                                   	Operating Data

                                    Initial operations:  Installed glass bags
                                                      June 1986
                                                      • Excessive pressure drop
                                                      • Reduced capacity

                                    Interim operation:  Replaced half of bags with
                                                      NOMEX®
                                                      • February 1987
                                                      « 14.0oz/yd2 (475 g/m2)
                                                       needlefelt
Circulating Bed Boiler
Installation Description
Type: Lurgi/Combustion Engineering
Capacity: 800 MM Btu/hr.
Fuel: Eastern Anthracite Culm
• 0.45-0.50% sulfur
• 6,000-9,000 Btu/!b.
• 40% ash
• Generates approximately
200 ppm SOx

Circulating Bed Boiler
Operating Data
Current operation: Replaced remainder of
glass bags
• October 1987
Pressure drop: 5.0 In. H2O (125 mm H2O)
Cleaning frequency: 20 mln. interval/bag
Emissions: < 0.05 Ibs/Btu
(1 1 mg/m3=0.005 grns/tt3)
No detectable SOx
Problems: Abrasion of some bottom disks
Estimated Bag Life: 48 months
i
    Circulating Bed Boiler
         Gas Conditions
Temperature: 360°F (180°C)

H2O content: Low

Dust load:    4-13 grns/ft3 (9-30 g/m3)
                                            Operating Data

                                       Stoker-Spreader Boiler

                                          Application
                                          • Stoker-Spreader boiler
                                            -100,000 Ibs/hr. steam capacity
                                            - 3% sulfur coal
                                            - Continuous operation
                                          • Standard Havens bag house
                                            - Pulse-Jet - off-line
                                            - 90 psl pulse
                                            -15 minutes Interval/bag
                                            - 2 grns/ft3 load
                                          • Rlter medium
                                            - TEFAIRE* - 21 OZ/yd2
                                            -TEFLON«-23oz/yd2
     Circulating Bed Boiler
      Bag  House and Bags
 Bag house
 Manufacturer:
 Gas flow:
 Operating mode:
 A/C ratio:
 Bags
 Dimensions:

 Material:
Standard Havens
• No venturles
• 16 compartments
• 210 bags/compartment
950,000 ACFM (450 CMS)
Off-line cleaning
4.0 ft'/ft2/min. (1.2 mVnWmin.)


5.5 in. dla. x 241 In. long
(140 mm dla. x 6.1m long)
GL 04 woven glass (text x text.)
16.5 oz/yd2 (560g/mJ)
                                            Operating  Data
                                       Stoker-Spreader Boiler
                                                Operation

                                                « Emissions
                                                                TEFAIRE*
                                                                                 TEFLON®

                                                            - 0.018 grns/tt3 (actual)    0.017 grns/ft3
                                                            - 6.3 In. pressure drop    4.7 In.
                                                            - Air cloth ratio = 3.5 fpm  2.8 fpm
                                                            -Drag = 1.8           1.7

                                                 Results
                                                 • Properties
                                                  retention - No loss of properties
                                          21-10

-------
Durability Data • TEFAIRE®
Stoker-Spreader Boiler
Exposure
Permeability
> As received (dirty)
• Vacuumed
• Washed
Tensile strength
• Washed
Mullen burst
• Washed
4.5 years
ft3/rt2/mln.
@ 0.5 In.
5.6
11
29
Ibs./in.
70
Net (psl)
260
Unused
H2OdP
-
-
23

68

210
                                                        Circulating Bed Combustion
                                                            (2) Combustion chamb

                                                        Limestone   1
                                                         fnort    '
                                                       : oaoBi EnvkMwnUi tunic**, inc.  Cooling water   Ash conveyor cys
    Filtration Tester Schematic
                               ROW
                               rotomater
      Dual canlatar —»*•
       Turbo-compr*«*of
    Site Remediation Bag House
     Bag House Description and
   Typical Operating Conditions
  Number of bags
  Bag dimensions
  Cage design
  Dust load
  Temperature (max)
  Pressure drop
  Pulse frequency
  Permitted leakage

  Dust characteristics
72
-6.0' dia. x 120' long
21 wire, mild steel
- V. high
-410°F(210°C)
 < 10 in. H?O (< 250mm H2O)
 > 1 per minute
 < 0.08 grns/ft3 - (180mg/m3)
 No visible emissions
High population below 1 micron
Non-agglomerating
Angular
              Operating Data
Typical Bag  House Operating Conditions
  W. German Power Generating Station

    Application
    • 2,688 bags, pulse-Jet
      8 compartments
      Various coals:
      - W. German, Polish, S. African, Australian
    • Buehler bag house
      Pulse-jet - on/oft line
      1 minute Interval/bag
      17 g/std.m3 (7grns/std.tt3)
    • Filter medium
      -TEFAIRE*-22 oz/yd2
Selection of Filter Media
for Circulating Bed
Medium
Woven structures
Glass/membrane
m-Aramld
Poly(imide)
Sulfar
Poly (tetrafluoro-
ethylene) (PTFE)
PTFE/glass
Acceptability
No
Limited
No
No
No

Limited
Yes
Combustor
Reason
Excessive leakage
Short bag life
Hydrolysis concern
Hydrolysis concern
Temperature concern

Leakage concern
Bag life?
              Operating Data
Typical Bag House Operating Conditions
  W. German Power Generating Station

     Operation
     • Emissions
       < 30 mg/std.m3 (< 0.013 grns/std.ft3)
     • A/C ratio
       - 1.5-2.24 m/min. (4.9-7.3 fpm)
     • Pressure drop
       -150-250 mm H2O (6-10 in H2O)
     • SOx concentration
       - 1700  mg/m3
     • Temperature
       -210°C (410°F)
           Comparison of Filter
      Medium Properties TEFAIRE®
    Bags After 103,700 Pulse Cycles

                  TEFAIRE*      Properties after
Parameter Units   specifications  72 days' use

Mullen burst  psi       210±50          230
           kPa       > 500

Tensile      Ibs/in     63±20          60
strength     Newtons   108±24          104
Basis
weight

Leakage*
Ibs/in
Newtons

oz/yd2
g/m2

grns/fl3
mg/m3
 21±1.5
 710±50
  < 0.010
  < 23
2.0.4
690
                                   0.0004
                                   0.9
                                                    •Not • •tmltkJVun. y«»«J B UK nrto of 7 *nt (2.1nVnin.) mfPC By nf% 13 mbL dMnlng pUM IrTtarata.
                                              21-11

-------
            OPTIMIZING BAGHOUSE PERFORMANCE AT THE MONTICELLO STATION
                             WITH AMMONIA INJECTION
                                  Kent Duncan
                                  Robbie Watts
                                   TU Electric
                        Monticello Steam Electric Station
                                  P.O. Box 1266
                            Mount Pleasant, TX 75455
                                Randy L. Merritt
                                  P. Vann Bush
                           Southern Research Institute
                             2000 Ninth Avenue South
                                 P.O. Box 55305
                            Birmingham, AL 35255-5305
                                Walter V. Piulle
                                 Ramsay L. Chang
                        Electric Power Research Institute
                              3412 Hill view Avenue
                               Palo Alto, CA 94303
ABSTRACT

A project began in December 1988 to demonstrate the ability of ammonia injection
to reduce emissions and decrease flow resistance through a baghouse at the TU
Electric Monticello Station.  Ammonia injection systems, flow monitoring devices,
new deflation system controllers, and a data acquisition system were designed and
installed in the Unit 1A and IB baghouses.  New bags were installed in the 1A and
IB baghouses prior to the project.   The flue gas entering the Unit 1A baghouse is
treated with approximately 25 ppm of ammonia, and the Unit IB baghouse is main-
tained as a control with no ammonia injection.

Ammonia reaction with S03 on the fly ash increases the ash cohesivity and the
porosity of the dustcake, which reduces the flow resistance and ash penetration
through the baghouse.  Suppression  of ash bleedthrough permits the use of more
energetic cleaning, which further decreases flow resistance through the baghouse.
Ammonia injection has resulted in 20% more gas  flow at a given pressure drop
through the 1A baghouse than through the untreated IB baghouse.  At the same flow
rate the 1A baghouse has approximately 2 in. H20 lower pressure drop than the IB
baghouse.
                                       22-1

-------
            OPTIMIZING BAGHOUSE PERFORMANCE AT THE MONTICELLO STATION
                             WITH AMMONIA INJECTION
INTRODUCTION & BACKGROUND

The Monticello Station,  operated by TU Electric has three units fired by Texas
Lignite coal.   Units 1 and 2 are identical.   Each is a 575-MW unit retrofitted in
1978 with Wheelabrator-Frye shake/deflate-cleaned (S/D) baghouses which are in
parallel  with  Research Cottrell  electrostatic precipitators (ESP).  The 36-
compartment baghouses were originally sized  to take 80% of the boiler exhaust (net
air-to-cloth (A/C) of 2.9 acfm/ft2).   Figure 1 shows that each baghouse is made up
of three  12-compartment  baghouses,  and that  the middle baghouse for each unit
functions as two separate 6-compartment baghouses.  Thus the A and B air heaters
each exhaust into 18 compartments of bags.  Each set of 18 compartments is oper-
ated and  cleaned independently.   Since each  set of 18 compartments has a dedicated
booster fan, the flow splits between the ESPs and baghouses can be varied over a
wide range.

Since these baghouses came on line  in 1978-1980, they have operated with much
higher than expected pressure drops (10-12 in. H20) and take only about 50% of the
boiler exhaust (measured net A/C -  2.0 acfm/ft2).  This results in high flow
through the ESPs, reducing their specific collection areas, and causes higher
opacities in the stack exhaust.

Testing with a small, portable Fabric Filter Sampling System (FFSS) confirmed that
the Monticello lignite ash is significantly  different from many other coal ashes,
and is not filtered effectively by  many standard bag materials (1,2).  FFSS test-
ing showed that injection of low concentrations of ammonia gas (NH3) significantly
reduced the emissions of ash (2_).  Compartment-scale testing was conducted at the
Unit 1A baghouse for 18-months in 1985-1986  to evaluate four different fabrics,
with and  without ammonia injection.  Results from this phase of testing were
favorable and  led to full-scale demonstration (£).

Full-scale demonstration of ammonia injection began in December 1988.  Ammonia was
injected  into  the Unit 1A 18-compartment baghouse and the Unit IB baghouse was
maintained as  a control  with no ammonia injection.


EQUIPMENT PREPARATIONS

Ammonia Injection

Ammonia injection systems were fabricated by the Monticello plant and were in-
stalled at the inlet to  the unit 1A and IB baghouses. (Provisions for ammonia
injection were made for  the unit IB baghouse, although injection will not begin
until the current test plan is concluded.)  The ammonia headers for the 1A and  IB
baghouses were connected to the existing piping supplying ammonia to the  electro-
static precipitators (ESPs).
                                        22-2

-------
A flow controller for the pure ammonia gas was installed downstream of the ammonia
evaporators and storage tank.  A flow controller module and a strip-chart record-
ing for the pure ammonia flow rate was located in the boiler control room.  The
ammonia flow rate is periodically monitored by control room operators to assure
proper ammonia flow rate is maintained.

For each baghouse, four ammonia injection probes were inserted into the inlet duct
at a location 15-20 feet downstream of the air heater.  Each probe consists of 24
injection nozzles equally spaced along the 13-foot depth of the duct.  Nozzles are
positioned 60 degrees normal to the direction of flue gas flow, alternating from
one side of the probe to the other along the probe length.  The top 12 nozzles are
5/32 inches in diameter and the bottom 12 nozzles are 7/32 inches in diameter.
This design improves the uniformity of dispersion along the length of the injec-
tion probe.  Injection velocity was calculated at 60-100 fps at a flow rate of 180
scfm.

Instrument-grade air from the soot blower system provides the carrier air to
deliver the ammonia through the injection nozzles at duct velocity. Separate
programmable flow controllers were installed to independently control the combined
flow rate of the carrier air and ammonia gas to the 1A and IB ESPs and baghouses.
The flow controller modules are located in environmental enclosures midway between
the injection headers for units 1A and IB.


Flow Monitoring Devices

To adequately evaluate the effects of ammonia injection and changes to various
baghouse cleaning cycle parameters, determination of drag through the bag filters
is required.  (Drag is defined as the tubesheet pressure drop per unit gas veloc-
ity through the bags.) Two methods for determination of flow were implemented.


Annubars.  Flow measuring annubars were installed in the 1A and IB baghouse inlet
ducts.  The annubars were inserted in a center test port on the ducts which ser-
vice 12 of the 18 compartments in each baghouse,  as shown in Figure 1.   Support
guides were welded to the floors of the ducts to secure the annubars.

Pressure lines for each annubar probe were run to pressure transducers in an
enclosure midway between the two devices.  The transducers are equipped with
square-root extractors to facilitate conversion of the pressures to flow rates.
A timer and a series of solenoid valves were assembled to periodically isolate the
pressure transducers and purge the pressure lines with 100 psi air.  Signal cables
were run to the baghouse control room for monitoring flow rates on the computer
data acquisition system and a chart recorder.

The annubar devices were calibrated with manual  pitot velocity measurements in
May,  1989.

Individual  Bag Flow Monitors.  An individual  bag flow monitor system (IBFM),
manufactured by ETS,  Inc.,  Roanoke, Virginia, was purchased and installed at the
unit  1 baghouse.  The IBFM system measures flow rate through individual bags via
orifice plates mounted on bag thimbles.  Measurement of the tubesheet differential
pressure and the individual  bag flow rates allows calculation of drag through the
bags.

Orifice plates were installed in the bag thimbles of 10 bags located in four
compartments;  three compartments in the 1A baghouse (K2, K3, and K4) and one
                                        22-3

-------
compartment of the IB baghouse (L3).  An instrument box penetrates each of the
four compartment walls and allows connection of pressure lines and thermocouple
cables to a signal conditioning module (SCM), installed in the vicinity of com-
partment K3 at the tubesheet level.  A signal cable connects the SCM to the  per-
formance monitor unit located in the baghouse control room.  Orifice AP, tubesheet
AP and compartment temperature signals from the SCM are also connected to the SRI
computer data acquisition system.

Baghouse Maintenance

Compartment Rebagqings.  The project to evaluate ammonia conditioning was coordi-
nated with the rebagging of the 36-compartments of unit 1.  The bags for all but 3
compartments are a standard fiberglass fabric with a 10% teflon-B finish,
10 oz/yd2, 3X1 twill, and 75% exposed surface texturization (EST).  The other
three compartments (12, J2, and J4) have bags of the same design except they have
25% EST.

The 18 compartments in the Unit IB baghouse were rebagged during the period be-
tween September 1 and December 7, 1988.  Rebagging of 14 of the 18 compartments in
the Unit 1A baghouse began on December 8 and was completed on February 17, 1989.
Two compartments (16 and J6) were rebagged in May-June 1988 and two compartments
(12 and J2) were rebagged in August 1987.

Automatic Deflation System.  There are separate deflation systems for units 1A and
IB baghouses.  The systems include fans,  adjustable dampers on the inlet and
outlet sides of the fans, recirculation dampers, and damper controllers.  When no
compartments are in the deflation period of the cleaning cycle, the deflation gas
taken from the outlet duct is returned to the outlet duct through a recirculation
damper.  During the deflation period in a compartment cleaning sequence, the
recirculation damper closes and the gas is passed through the compartment defla-
tion poppet damper.  The resultant deflation pressure can be observed on the
compartment tubesheet magnehelic gauge.

The deflation system has historically been operated by manually setting the defla-
tion fan outlet dampers.   The position of the outlet deflation fan damper can be
observed by an indicating pointer on the motor drive shaft.  Tubesheet pressure
drops due to deflation have often fluctuated from 0 to 3 in.  H20 through the
course of normal baghouse operation.  Programmable controllers were installed on
the unit 1A and IB baghouses to modulate the deflation fan outlet damper and
deliver a stable deflation pressure drop as measured at the compartment tubesheet.
Honeywell Universal Digital  Controllers,  Model  3000,  were selected for this appli-
cation.  These controllers replaced the existing system designed for automatic
deflation control, although this system had never functioned properly.

Kurz flow meters were installed in the deflation ducts of the units 1A and IB
baghouses and their outputs were connected to the programmable controllers for the
deflation fan outlet dampers.   Timers were configured to allow feedback from the
flow meters to control  the deflation damper for a period of 10-15 s within the
deflation period.  The result was control  of deflation pressure to within ±0.5 in.
H20 of the desired setting.

Shaker Operation.  The shaker system was  inspected for proper operation.  Inspec-
tion of the shaker tubes  inside the compartment found no apparent wear at the J-
hook assembly.   This assured that movement of the shaker exterior to the compart-
ment was evidence that the bags were actually shaking
                                       22-4

-------
Causes of shaker failure are slippage of the pulley belt or motor failure, both of
which can be observed outside the compartment.  An inspection of shaker operation
for all 36 compartments (1A and IB baghouses) found approximately 10 shakers
inoperable due to belt slippage.  Repairs were performed and all shakers were
placed in operation.  Reinspection of the shaker operation is made periodically to
assure proper operation.


MONITORING AND DATA ACQUISITION SYSTEM

A location was selected in the unit 1 baghouse control room to place the various
monitoring equipment used in the project.  An instrument cabinet was acquired to
enclose the components.  A customized face plate was fabricated by plant personnel
to efficiently arrange the various components in the instrument cabinet.  The
layout is shown schematically in Figure 2.  An adjacent cabinet, which previously
housed the opacity monitors and deflation controllers, was used as a junction box
for the cables from the various instruments to the new cabinet.


Opacity Monitors

The opacity monitors for the 1A and IB baghouses were transferred to the equipment
cabinet and were mounted on a panel at the top of the cabinet.  Output signals
from the opacity monitors are charted on recorders and are input to the data
acquisition system.  The unit 1A and IB baghouse outlet opacity monitors are
periodically calibrated by plant personnel.


Flow Monitors

Digital controllers for automatic deflation control were installed in the instru-
ment cabinet.  The controllers provide continuous readout from the Kurz flow
meters in the deflation ducts, and output for charting the measured flows.   A
Yokagowa 3-channel chart recorder was installed adjacent to the controllers.

The raw flow rate data from the annubar devices are continuously charted for the
1A and IB baghouses.  The data are also input to the computer data acquisition
system where they are adjusted based on the pitot calibration of the annubars.
The corrected flow rates are displayed and recorded by the computer.


Compartment Tubesheet Pressure

The tubesheet pressure lines for four compartments (Kl, K3, K6, and L3) were run
to the baghouse control room.  The compartments were selected to correspond to two
of the compartments with the IBFM sensors (K3 and L3) and two compartments at
opposite ends of the baghouse (Kl and K6).  Individual  magnehelic gauges were
installed on the equipment panel.  These gauges allow observation of normal com-
partment operation during filtering and cleaning periods.


IBFM Performance Monitor

The Performance Monitor is located on a shelf in the instrument cabinet and is
connected by cable to the SCM at the compartment location.  The Performance Moni-
tor allows a selection of modes for the retrieval  of IBFM data.  Measurements of
IBFM orifice differential  pressures,  compartment tubesheet differential pressures,
                                       22-5

-------
and compartment temperatures are output to an integral chart recorder and the SRI
computer data acquisition system.


Data Acquisition System

Provisions were made to sample data values from multiple sources and record these
values at prescribed intervals (default data storage rate is once per minute).  A
Dell System 200 PC/AT computer was integrated with a Strawberry Tree Data Acquisi-
tion System.   The data acquisition system can accommodate 16 analog inputs.  In
addition to the output of the monitors described above, a multi-signal cable
provides data from the boiler control  room to the instrument cabinet.  Four con-
trol room parameters were made available from the control room computer system.
The parameters being recorded by the SRI data acquisition computer are:


  BAGHOUSE DATA                  UNIT DATA	         	IBFN DATA	

flange-to-flange AP        boiler load                    bag ID
outlet opacity             stack opacity                  bag flow rate
flow rate   Annubar        coal  BTU calculation           tubesheet AP
                           stack S02  concentration        compartment temperature


The computer data acquisition software was developed specifically for this proj-
ect.  A pictorial schematic drawing of the baghouse can be displayed with the
various parameters and their values updated on the monitor screen each second.
Alternately,  a tabular data summary of the input parameters can be displayed.

A dedicated modem and phone line and Carbon Copy software allow access to the data
acquisition system from remote computer systems including computers at SRI in
Birmingham.  Data are recorded locally on a 20 MB hard disk drive and can be
retrieved on demand.


STARTUP HISTORY & CALIBRATIONS

Implementation of this project was performed in stages.  Ammonia injection at the
unit 1A baghouse began on December 8,  1988 coincident with the first compartment
rebagging in unit 1A.  Rebagging of the unit 1A 18-compartment baghouse was com-
pleted on February 17, 1989.  The IBFM system and computer data acquisition system
began operation on February 16.   Annubar flow sensor probes were installed during
the 1989 Unit 1 spring outage (April  19-29).


Initial Operation

Pure ammonia flow rate was initially set at 17 scfm for the entire unit 1 system.
The three ammonia controllers (1A baghouse, 1A ESP,  and IB ESP) were set to
equally divide the combined ammonia and air flow rate (corresponding to -15 ppm
NH3 concentration in the baghouse inlet duct).  The deflation dampers for the unit
1 baghouses were manually set to deliver a deflation pressure of 1.0-1 5 in  H20
at the compartment tubesheet.  Shaker  duration was initially set at 0-1 seconds.

Baseline opacities averaged 4-5% for each baghouse.   Opacity spikes for individual
compartments  (coincident with the resumption of service following cleaning) were
typically 10-15% above the baseline.   The shaker durations for unit 1A and IB were
                                        22-6

-------
 increased to -3  s on January  10,  1989.  The  increased  shaker duration  of  3  s
 reduced the baghouse differential pressure by  1.5-2.0  in. H20  and  increased
 opacity spikes to 30-60% for  the  first several baghouse cleaning cycles.  Eventu-
 ally, the opacity spikes were reduced to near  previous levels.  The  shaker dura-
 tion was increased to 5 s on  February 2, 1989.

 The pure ammonia flow rate was  increased to  30 scfm and the proportion of ammonia
 and air flow rate was increased to the 1A baghouse on  February 9,  1989 (estimated
 NH3 concentration of 25 ppm in  the baghouse  inlet duct).


 Annubar Calibrations

 The annubar devices are located in the center  of the ducts which direct flow to 12
 of the 18 compartments in each  baghouse (see Figure 1).  Consequently, the total
 duct flow rate varies as a function of the cleaning cycle.  The minimum flow
 corresponds to when the first of  the twelve  compartments associated  with  that
 inlet duct (II or Ml) is taken  out of service  for cleaning, and the  maximum flow
 corresponds to when the first of  the six compartments  associated with  the other
 inlet duct (Kl or LI) is taken  out of service  for cleaning.  Typical data from the
 annubar devices  is shown in Figure 3.  The corresponding flange-to-flange pressure
 drops are shown  in Figure 4.

 Special measurement techniques  were required to precisely calibrate  the annubar
 measure of flow  rate.  For example, during the velocity measurements on the unit
 1A baghouse, the flow rate to the I-J rows of  compartments was held  constant by
 adjusting the baghouse booster  fan damper position.  The annubar flow  rate reading
 was monitored with a chart recorder at the inlet duct  location.  The pitot probe
 was monitored by a pressure transducer and recorded along with the annubar flow
 rate.

 Velocity measurements were made at two different flow  settings at the  1A  baghouse
 and at one flow  setting for the IB baghouse.  A data summary is presented below.


     Annubar     Annubar AP     Pitot Flow    Annubar  Flow       Ratio
     Location    (in. H20)        (kacfm)        (kacfm)      Pitot/Annubar

       1A           0.54           312             270            1.16
       1A           0.67           347             303            1.14
       IB           0.37           274             236            1.16


 For the three calibrations, the flow rate as determined from the pitot measure-
 ments was 14-16% higher than the  annubar reading.  This calibration  factor was
 applied to the value of annubar flow rate recorded by the computer data acquisi-
 tion system.


 Flue Gas  Chemistry Measurements

The concentrations of S02  and S03  upstream of the baghouse and the concentrations
of free ammonia gas  at the baghouse outlet (K-row duct) were measured  in  May 1989.
SOX  measurements  were made at a sampling port location immediately downstream of
the unit  1A air heater.   The point of sampling was -10 feet upstream of the
ammonia injection location.  S03 concentrations ranged from 0.53 to  0.69  ppm and
S02  concentration ranged 749 to 762 ppm.   The S03 concentration is difficult to
quantify  at Monticello due to the alkaline ash which tends to adsorb the  S03.


                                        22-7

-------
The pure ammonia injection rate was set at five different values from 11 to 34
scfm.   Ammonia concentration in the flue gas at the outlet of the baghouse de-
creased as expected when the ammonia injection rate was decreased.  Ammonia was
detected in the flue gas at the outlet of the baghouse at all five injection
settings at concentrations ranging from 11 to 3 ppm.


IBFM Problems

A portion of the data from the IBFM sensors exhibited unrealistic behavior which
showed the expected decrease in A/C after the compartment returns to service, and
then an unexpected increase in A/C approximately halfway through the filtration
cycle.  Each of the IBFM sensors were removed and inspected.  Several leaks on the
inlet side of the orifice were discovered; they were due to dents in the top lip
of the thimbles to which they were mounted, and the felted gasket material was
eroded away in the area of the leak.  Some leaks were found in the pressure lines
connecting the sensors and signal conditioning module.  The dented bag thimbles
were reshaped to improve the sealing surface.   A more durable gasket material was
applied to the sensor, and sufficient high temperature RTV was applied to the top
of the thimble to assure a complete seal with the sensor.  The sensor assemblies
were reattached to the appropriate bag thimble.  Leaks in the pressure lines were
corrected.

Erratic A/C data were still evident following the elimination of leaks at the IBFM
sensors.  An independent pressure transducer monitored by a chart recorder was
connected to individual sensors at the pressure taps at the compartment penetra-
tion module.  This diagnosis revealed that the upstream pressure tap was gradually
being plugged during a filtration cycle, and often being adequately cleared by
subsequent purging to ambient or during the compartment cleaning cycle.

Multiple leaks at the mating of the IBFM sensors to the thimbles have reoccurred
and all sensors have been removed from the compartments.  A modification was made
to shield the upstream pressure tap from impaction, to eliminate plugging.  The
sensors will be welded to the respective thimbles, since all gasket and sealing
materials tried have been ineffective.


OPTIMIZATION

Adjustments were made to the timing sequence of the cleaning cycle operations.
The cleaning cycle sequence of events and a typical timing schedule are illus-
trated  in Figure 5.  The timing durations for each of the operations were evalu-
ated and refined based on their effects upon cleaning effectiveness.

In addition to the sequence and timing schedule for cleaning cycle parameters,
prerequisite to successful baghouse optimization is the integrity of the system
hardware.  The malfunction of isolated components (poppets, shakers, bag fabric,
bag tensioning) can result in a chaotic effect to the baghouse cleaning effi-
ciency.  Intensive maintenance inspections of all system components was initiated
as a part of the optimization program.


Compartment Inspections

The optimization of baghouse operation must include the elimination of bag leaks
and failures which are manifested in ash accumulations in compartments and outlet
gas opacities.  Some problems can be located with careful observation of the
                                        22-8

-------
baghouse opacity charts (high opacity spikes or an increase in the opacity base-
line).  Bag failures in certain locations of the compartments (toward the floor
and opposite the outlet poppet shelf) may not be detected by the opacity monitor
due to insufficient velocities to loft the ash to the outlet duct.  Undetected,
these failures cause piles of ash to accumulate on the tubesheet floors.  Periodic
inspections are the only viable means of identifying and correcting these prob-
lems.

A task force group was formed to investigate the opacity problems on Units 1
and 2.  This group coordinates efforts between plant Operations, Maintenance, and
Technical Services departments.  The Maintenance Department has implemented
routine preventative maintenance inspections of the Units 1 and 2 baghouse com-
partments.  The objective of this project is to eliminate major maintenance prob-
lems such as severely damaged bags, floor leaks, wall leaks, and damper problems.
The compartment floors are usually vacuumed following any necessary corrective
actions.  A written inspection record is completed by the maintenance crew for
review by the Technical Services Department.


Shaker Duration

In parametric evaluations during the initial phase of this project, a 2 in.  H20
reduction in flange-to-flange pressure drop (and an increase in baghouse opacity
spikes) was noted when the shaker timers were increased from a nominal  1 to  3 s
duration.

Subsequently the shakers on both baghouses were set at 5 s duration.   The shaker
timers were then increased from 5 to 10 seconds.  For the 1A baghouse the opacity
spikes increased 5-10% for individual compartments with the increased shaking
duration.  For the IB baghouse the opacity spikes increased from -10% to 25-40%
over the next 24 hours.  The 1A baghouse flange-to-flange pressure drop was
reduced by 0.5 in. H20 over a 3-4 hour period following the increase in shaker
duration.  No discernable improvement in flow rate (or reduction in flange-to-
flange baghouse pressure drop) was observed with the increased shaking duration
for the IB baghouse.

From these evaluations, the majority of the improvement in baghouse flow rate (and
increase in baghouse opacity) attributable to shaker duration occurred between 1
and 5 seconds.  There was an additional improvement in baghouse flow rate from 5
to 10 seconds for the 1A baghouse, although this improvement was not evident for
the IB baghouse.   Although there is an increase in baghouse opacity following an
increase in the shaker timers, the magnitude of these opacity spikes generally are
reduced after a day of operation.

A 10-s shake duration is recommended for effective bag cleaning. The contribution
of the increased baghouse opacity spikes to overall stack opacity is minimized
considering that a baghouse opacity spike occurs at five-minute intervals and the
duration of the spike is only 10-15 seconds.


Deflation Gas

Proper function of the deflation system is critical to effective bag cleaning.
With  no deflation the flange-to-flange pressure drop will typically increase -2
in.  H20 in one filtration cycle and the compartment opacity spikes will be signif-
icantly reduced,  both indicative of ineffective cleaning.
                                       22-9

-------
Observation through a lighted viewport on the lower access door of compartment K5
confirmed that deflation results in a measurable compartment tubesheet differen-
tial  pressure only when all  of the bags have been completely deflated (usually
30 s  after the opening of the deflation poppet damper).  Within practical limits,
complete deflation can be accomplished by either increasing the deflation flow
rate  (deflation fan inlet damper adjustment) or increasing the deflation period
(timer adjustment).

Reports from the Harrington  Station concluded that it was critical to operate at
less  than 0.5 in.  H20 of deflation pressure (3).  To determine if a relationship
existed between cleaning effectiveness and the amount of the deflation pressure
the deflation inlet damper was varied to produce deflation pressures in the range
of 0.5 to 4 in. H20 at the tubesheet.  No effect on baghouse pressure drop or
opacity was observed for the 1A or the IB baghouses over this range of deflation
pressure.  The flange-to-flange pressure increased (and opacity spikes decreased)
only  when the deflation gas  volume was not sufficient to collapse all of the bags
and yield a measurable deflation pressure at the tubesheet.


Filtration Period

The length of the  filtration period can be varied by adjustment of the interval
and settle timers.  Adjustments to reduce the total filtration period from 85 to
65 minutes resulted in a reduction in flange-to-flange pressure drop of 0.5-0.8
in. H20.  The opacity spikes were not affected by this reduction in the filtration
period.

Due to the high inlet grain  loading (-10 gr/acf), it is estimated that each bag
receives -15 pounds of ash per hour of filtration.  Shortening the filtration
period reduces the average amount of ash in the bag and thereby reduces the
average pressure drop through the dustcake.


First Shake

There are provisions for a first shake (within the deflation period) and a second
shake which follows the deflation period.  Historically, the first shake has not
been  used.  The use of the first shake delay and first shake timers can be de-
feated by a controller on the baghouse control panel.  As an exercise to determine
the effectiveness  of a combination of the first shake and second shake, the con-
troller was set to enable the first shake timers.

The first shake timer was set at 5 s, to occur 20 s after the beginning of the
deflation period.   Following the deflation period, the normal 10 s shake was also
performed.

The first shake produced a marginal reduction in flange-to-flange pressure drop
and no increase in the baghouse opacity.  It is suspected that the normal cleaning
cycle (deflation followed by shaking) is sufficient to clean the bags without an
additional shake period.


MEASUREMENTS AND RESULTS

Baghouse Flow

Since early in the project,  the 1A baghouse has exhibited -20% more flow at a
given pressure drop than the IB baghouse.  This is evident in Figures 3 and 4.


                                        22-10

-------
The ammonia injected into the 1A baghouse inlet duct, reacting with sulfur
trioxide in the flue gas and on the particles, results in ash that has higher
cohesivity.  This ash produces a more porous dustcake than the unconditioned ash.
Higher porosity results in lower drag.  Thus, for a given pressure drop, the
conditioned ash can accommodate a higher filtering velocity.  It is suspected that
the low levels of naturally-occurring S03 may be limiting the benefit achievable
with ammonia injection, and further reductions in drag may be possible by in-
jecting S03 into the flue gas (4).


Baghouse Opacity

Ammonia conditioning of the 1A baghouse serves to reduce the penetration of ash
through the fabrics and the resultant opacity spikes following compartment clean-
ing.  Although the opacity baseline and spikes are generally less for the 1A
baghouse conditioned with ammonia, bag failures and leakages can overwhelm any
benefit caused by the ammonia injection.  Based on results reported by the Univer-
sity of North Dakota Energy and Mineral Research Center, more emissions reductions
would be expected with higher concentrations of S03 in the flue gas (4).


Bag Weights

Bags were weighed in several compartments to gauge the effectiveness of the clean-
ing cycle parameters and the effect of fabric texturization.  Each of the compart-
ments had received ammonia conditioning since December 8, 1988.   The data are
summarized in the following table.

Data
Group
1
2
3
4
5
1,4,5
2,3
3,4,5

1,2
Average
Comp. Weight Months in
Date ID (Ib) Operation
8/18/89 K4 52.0 8.7
11/17/89 12 58.0 24.9
11/17/89 J4 24.4 9.7
11/30/89 J2 41.1 28.1
12/18/89 K3 65.0 12.0
10 s shake duration prior to measurement
5 s shake duration prior to measurement
Compartment was taken out of service immediately
the cleaning cycle.
Fabric
EST
(%)
75%
25%
25%
25%
75%


after

Compartment was taken out of service with an unknown
period of filtration since cleaning. The measured bag

weight could be high by as much as 15 pounds.

For the same time in service bags with 75% EST weigh significantly more than the
25% EST bags.   Compartments bagged with 25% EST fabrics historically have higher
opacity spikes than compartments with 75% EST fabrics.   These higher opacity
spikes  are consistent with the lower bag weights measured in these compartments.
                                       22-11

-------
Ash Analyses

Bag ash samples were obtained from several  compartments  and were analyzed for
soluble ammonia and sulfates.  The following table presents the data for these
samples and also bag ash samples obtained during compartment-scale testing of
ammonia injection performed at Monticello from May 1985   December 1986.
       Date
       7/17/85
       7/18/85
       7/16/85
       7/19/85
       7/17/85
       7/18/85
       7/16/85
       7/19/85

       12/7/86
       12/7/86
       11/30/89
       11/30/89
       11/30/89
       11/30/89

       12/18/89
       12/18/89
Compartment

    13
    14
    15
    16
    J3
    J4
    J5
    J6

    15
    J5
    N4
    N4
    J2
    J2

    K3
    K3
With Ammonia
S04*
wt %
0.34
0.28
0.34
0.39
0.30
0.28

1.35
0.39
          3
        ppm
        157
         61
         97
        142
0.69   1500
         97
         98

        150
        150
Without Ammonia
S04*
wt %
                                                                    3
                                                                  ppm
                  0.32  <5
                  0.33  <5
                  0.30  <5
                  0.34  <5
                  0.48

                  0.22  <20
                  0.22
                             4.9
       * present as sulfate products on the ash particles
       + present as ammonium bisulfate or ammonium sulfate in the ash
The data from December 1986 show higher sulfate concentrations than the other
data.  Variations in the sulfate (S04) concentrations on the ash samples suggest
that either a change in the fuel or the combustion process has resulted in varying
amounts of S03.  This suggests that there were higher concentrations of S03 in the
flue gas immediately prior to collecting ash samples in 12/86 than prior to col-
lecting either 1985 or 1989 data.  That would explain the much higher baseline
sulfate concentrations in the ash (0.48% vs -0.2-0.3%) and the much higher sulfate
and ammonia concentrations in the ammonia-conditioned ash.

Low levels of sulfates measured in the ash samples collected in 1989 indicate a
deficiency of available S03, which limits the beneficial effects of ammonia condi-
tioning.  These data confirm the advantage of independently injecting a steady
supply of S03 into the flue gas to provide higher concentrations of the NH3 + S03
reaction product to condition the ash for better filtration performance.
SUMMARY
          Ammonia injection has resulted in lower opacity spikes and lower
          drag in the baghouse at Monticello.
                                       22-12

-------
     •     The performance benefits,  though representing significant improve-
          ments  in emissions,  may be limited by low levels of S03 present
          in  the flue gas.

     •     Bags with 75% EST have significantly higher residual  dustcake
          weight than bags  having 25% EST.

     •     Better emissions  control  (with concurrent S03 and ammonia injec-
          tion)  should should  permit the use of bags with 25% EST, result-
          ing in even lower drag.


ACKNOWLEDGMENTS

The authors would like to thank personnel  at the Monticello Station including Mr.
Lee Hyman and other Maintenance Department personnel, and Mr. Zack Glover and Mr.
Jimmy Justice of the Instrumentation and Control Department.   These individuals
have provided considerable  help and  assistance which have made  this research
effort  possible.  This project was  funded  under EPRI RP-1129-23.


REFERENCES

1.  L.  G. Felix  and R. L. Merritt.   "Fabric Screening Studies for Utility
    Baghouse Applications."  In Proceedings: Fifth Symposium on the Transfer and
    Utilization  of Particulate Control Technology (August 27-30, 1984), Volume 3.
    EPRI Report  CS-4404, V.3., February 1986.

2.  L.  G. Felix, R. L. Merritt, and  K. Duncan.  "Improving Baghouse Performance
    at  the Monticello Generating Station."  Journal  of the Air  Pollution Control
    Association   36: 1075-1085 (September  1986).

3.  R.  L. Chambers, 0. C. Plunk, and S. L. Kunka.  "Fabric Filter System Study:
    Fourth Annual Report."   Prepared for Industrial  Environmental Research Lab,
    Research Triangle Park, NC, August 1984.

4.  S.  J. Miller and D. L.  Laudal .   "Real-Time Measurement of Respirable Particu-
    late Emissions From a Fabric Filter."   In Proceedings: 16th Annual Meeting of
    the Fine Particle Society  (April 1985).
                                       22-13

-------
                                  BOOSTER FAN
                     ANNUBAR
UNIT 1
   1B
   AIR HEATER '
   EXHAUST
   1A
   AIR HEATER
   EXHAUST
       EXISTING
--1
\
!-— J



1N6

1M6
1N5

1MB
1N4

1M4
1N3

1M3
I
1N2 ' 1N1
1
1
1M2 | 1M1
                                      L6 I 1L5 I 1L4 | 1L3 | 1L2  | 1L1
1 K6
] 1K5 . 1K4 | 1K3 | 1K2 | 1K1
I     I    I     I    I
                                                                 I STACK l

                                                                  \     /
                     ANNUBAR

l; —
"~ --J

A -


f



1J6

116

1
1J5 I 1J4
I
i
115 , 114
!

1J3

113


1J2

112


1J1

111

                                  BOOSTER FAN
     Figure 1.   Layout of the  Monticello  Station  Unit 1 Baghouse
                                    22-14

-------
                   OPACITY
                  DEFLATION
                TUBESHEET AP
     DATA ACQUISITION SYSTEM
                IBFM MONITOR
 D

 D


ODD
 a I
 a I

LTD
                                             D

                                             D
                                           la
                                           la.
Figure 2.   Layout of the  Monitoring and  Data Acquisition  System
                                 22-15

-------
   350
   325
  300
 . 275
O
  250
  225 -
  200
   16:00
                                                                                  Baghouse 1A
                                                                                Baghouse IB
16:30
17:00         17:30        18:00

            TIME ON JUNE 21, 1989
                                                   18:30
                                                               19:00
                                                                            19:30
               Figure 3.   Flow rate in rows  I  and J (baghouse  1A)  and flow rate
               in rows M and  N (baghouse  IB)  for June 21, 1989.

-------
ro
i—>
~-i
                                 BAGHOUSE1A
                                                                                                  BAGHOUSE1B
           15
           14
       O
        (N
           13
       I  12
           11
                    J	L
                                   _L
                                                  J	L
                                                                           15
                                                                           14
                                                            O
                                                             (N
                                                            1   13
                                                             c
                                                            CL
                                                            UJ
                                                            in
                                                            o
                                                            £   12
                                                                           11
10
16:00    16:30   17:00   17:30    18:00   18:30   19:00   19:30
                    TIME ON JUNE 21. 1989
                                                                                    i
                                                                                                           I
                                                                                                                          I
10
16:00    16:30   17:00   17:30   18:00    18:30   19:00   19:30
                  TIME ON JUNE 21, 1989
                                      Figure 4.   1A  and IB baghouse pressure  drop for  June 21,  1989.

-------
65s
                                                  65s

INTERVAL
TIMER QN

COMP OUTLET CLOSE
DAMPER
1ST SETTLE ON
TIMER


DEFLATION OPEN
DAMPER C|_OSE

2ND SETTLE ON
TIMER

SHAKER ON
MOTOR

FINAL ON
SETTLE
TIMER Uhh
RESET
TIMER

REINFLATION CLOSE
DAMPER
OPEN

^_ 	














->


•^

40s
_/

60s

/ \
/ \
30s

| 10s

! 20s

1s->i
n
20s |-<- i
/

•^ 	

35s











•*-
35s
< >
\

	 •>-


\_














Figure 5.   Cleaning Cycle Timing Schedule
                  22-18

-------
      ENHANCING BAGHOUSE PERFORMANCE WITH CONDITIONING AGENTS:
                    BASIS, DEVELOPMENTS, AND ECONOMICS

                            S.J. Miller and D.L. Laudal
                     Energy and Environmental Research Center
                            University of North  Dakota
                            Box 8213 University Station
                         Grand Forks, North Dakota 58202
ABSTRACT

Key ash properties that influence fabric filter performance are identified and quantified.
The theoretical and empirical bases for reducing particulate emissions and pressure drop
by modification of these ash properties are presented.  Results show that fine
particulate emissions are reduced from 100 to 10,000 times and baghouse pressure
drop is reduced up to 75% when small amounts of ammonia and S03 are injected
upstream of a baghouse.  The conditioning makes the fly ash more cohesive and results
in a more porous  dust cake but  does not appear to cause bag cleaning problems.
Conditioning reduces particulate emissions by enhancing the ability of the ash to bridge
large pores and pinholes and by inhibiting the seepage of ash through the fabric.  It
appears that conditioning will reduce particulate emissions and baghouse pressure drop
for a variety of coals with  shaker-cleaned,  pulse-jet, and reverse-air baghouses.
Conditioning can  be implemented at a cost less than 10% of the levelized cost of a
baghouse, but this cost can be  more than recovered if the pressure drop and/or the size
of the baghouse is reduced.


INTRODUCTION

While fabric filters have been successfully applied to utility boilers for a variety of coals,
their performance is dependent  on key ash properties.  Baghouse performance can be
defined in terms of particulate removal efficiency, tubesheet pressure drop, bag
cleanability,  and bag  life.   Particulate emissions must be low enough to meet current
regulatory standards  and to alleviate concerns over visibility impairment and potential
adverse health effects caused by fine  particles in the atmosphere.  Pressure drop should
not only be within design specifications but should be kept low to minimize fan power
consumption. Bags must be easily cleaned to avoid high residual dust cake weights and
excessive bag cleaning.  Bag life must be long enough to ensure the economic benefit of
fabric filters and to maintain reliability, avoiding excessive maintenance.  Any effects
that particulate  properties have  on these performance parameters are of interest to
boiler operators and baghouse designers.  Knowledge of the relationships between ash
properties and baghouse performance is important to optimize performance for specific
coals.
                                       23-1

-------
It is well known that ash properties affect baghouse performance, but predicting
performance from ash properties is difficult.  Bush et al. (1) has evaluated ash properties
and corresponding fabric filter results and concluded that the fundamental properties of
dust cake ash that determine baghouse performance are particle morphology, particle
size, and cohesivity ( i.e., cohesive characteristics).  Fly ash particle size  distribution is
known to have a major effect on baghouse pressure drop; smaller particle size causes
higher pressure drop for the same amount of dust on the bag and same porosity.
However, the fly ash particle size distribution by itself is not sufficient to  predict
pressure drop.  Additional dust characteristics that relate to pressure drop are those that
describe the ability of the dust to maintain a high porosity or resist packing and that
facilitate dust cake release.  This implies  that the cohesive characteristics of the dust
are important. It would appear that the particle size distribution would have an effect
on particulate emissions, but emissions are also affected by dust characteristics that
describe the ability of the dust to bridge large pores or pinholes and the ability of the
dust to resist seepage or reentrainment.  Again, this implies that the cohesive
characteristics are important, but the exact characteristics that affect  pressure drop may
not be the same characteristics that control  particulate emissions.  Furthermore, the
question exists whether it is possible to alter these characteristics to enhance baghouse
performance.

The Energy and Environmental Research Center (EERC) at the University of North Dakota
has conducted research for a number of years, sponsored by the U. S. Department of
Energy, to investigate how fine particle emissions depend on fly ash properties  (2,3).
Further work showed that fly ash properties could be changed to reduce  fine particle
emissions and baghouse  pressure drop by injecting small amounts of ammonia (NH3) and
sulfur trioxide (S03) into the flue gas upstream of a baghouse (4-8). The conditioning
agents increase the cohesive strength of  the fly ash and facilitate the  formation of a
more porous dust cake (9).  Early work was conducted with shaker-cleaned and pulse-
jet-cleaned bags.  Further experimentation has been completed in the  last two years
investigating the use of ammonia and S03 as conditioning agents with a reverse-air-
cleaned fabric filter (10-13).  One purpose of these tests was to determine whether the
increase in cohesive strength of the fly ash might lead to bag cleaning problems with
more gentle bag cleaning.  Since reverse-air-cleaned bags are known to develop high
residual dust cake weights, there was concern that conditioning might aggravate the
problem.  On the  other hand, there was the  possibility that conditioning might reduce
pressure drop with reverse-air cleaning by the formation of a more  porous dust cake and
by facilitating dust cake release. This paper presents results of 500-hour tests  with a
pilot combustor and baghouse in which ammonia and SO3 were injected upstream of a
reverse-air-cleaned baghouse and explores the effect of conditioning on ash properties.


EXPERIMENTAL

A schematic of the Particulate Test Combustor and baghouses is shown  in Figure 1.
The 0.16 MW combustor is preheated firing natural gas, and the baghouses are
preheated electrically to prevent condensation during startup.  Once the combustor  is
heated to normal  operating temperature on gas, it is switched to firing pulverized coal
while the flue  gas is filtered by the pulse-jet baghouse.  After achieving stable
combustion on coal, the flue gas is introduced to the reverse-air baghouse for the start
of a  test. Particulate emissions and baghouse pressure drop are then monitored as  a
function of time to document the process of dust cake buildup on the fabric. Flue gas is
continuously  monitored for 02, C02, S02, and NOX and,  along with system temperatures
and pressures, is  recorded on a data logger.
                                       23-2

-------
The reverse-air baghouse contains a single, full-scale 12 inch by 30 foot bag.  The bag
fabric was a Midwesco 601E, which was employed  in previous conditioning tests at
EERC.  This fabric is typical of fabrics used in large-scale, reverse-air baghouses filtering
flue gas from utility coal-fired boilers.  The 10-oz/sq yd fabric is 3 X 1  twill woven glass
construction with 10% Teflon B finish. The fill yarns are texturized while the warp
yarns are not texturized.  The dirty side (inside) of the bag has a surface of
approximately 75% texturized yarns.  The reverse-air bag was cleaned every two hours
by first directing the flue gas through the pulse-jet baghouse, which employed Ryton
felted bags, and then backwards through the reverse-air baghouse for  a period of 45
seconds. The filtering air/cloth ratio (face velocity) and the cleaning velocity for the
reverse-air tests were 2.1  ft/min.  The baghouse was preheated electrically prior to the
tests and held at a constant 300°F during testing.  There were no dew point excursions
for either test.  The bag was suspended with a load cell that permitted continuous
monitoring of the bag weight.  The initial bag tension for the tests was 75 Ib.  The bag
suspension cap contained a 10-inch-diameter glass sight port which, along with  sight
ports in the top of the baghouse, permitted viewing  of the inside of the bag during bag
cleaning. Sight ports were also located along the side of the baghouse which allowed
viewing the clean side of the bag.

Particulate emissions were measured by three methods:  1) total mass particulate
emissions were measured by conventional dust loadings according to EPA Method 5,
2) emissions as a function of time and particle  size in the range  of 0.5 p,m to 30 Mm
were measured with a TSI APS 33 particle sizer, and 3)  the submicron particle size
distribution was measured with a TSI Differential Mobility Particle Sizer and
Condensation  Nucleus Counter. The amount of ammonia in the flue gas downstream of
the baghouse  was determined by bubbling a flue gas sample through a dilute solution of
sulfuric acid and then measuring the ammonia concentration in the solution. Tests also
included a determination  of the amount of ammonia  adsorbed on the fly ash. Baseline
and conditioning tests were conducted with two coals - Monticello, a  7000 Btu/lb
Texas lignite and Pittsburgh #8 seam, a 12000 Btu/lb bituminous coal  (13). This paper
includes results for  only the 500-hour tests with  the Monticello coal.  Conditions for the
baseline and conditioning tests were held constant except for the 25 ppm ammonia and
12 ppm S03,  that were injected into the flue gas at separate locations  just upstream of
the baghouse.  Three composite baghouse ash samples were taken for each of the 500-
hour baseline and conditioning tests.  These composite samples were each made up of
smaller baghouse ash samples taken every two days.  The tensile strength of these
composite samples  was measured as a function of porosity by a Cohetester, and the
aerated and packed porosities were measured with a Powder Characteristics Tester.
Both of these  instruments are manufactured by Hosokawa Micron International.  The
dust resistance coefficient, K2, was also measured in the laboratory as a function of
porosity. Extensive coal  and ash analyses were also completed but are not reported
here (13).
RESULTS

Pressure Drop

A comparison of baghouse pressure drop between the conditioned and baseline 500-
hour tests is shown in Figure 2.  The bags were cleaned every 2 hours except for an
initial 4-hour period at the start of the test. Values plotted include the pressure drop
just before and  after bag cleaning.  For the baseline test, the before-bag-cleaning
pressure drop was in the range of 9 to 11 inches of water, while the after-bag-cleaning


                                       23-3

-------
pressure drop increased somewhat with run time and was about 3.3 inches of water at
the end of the test.  With conditioning, the before-bag-cleaning pressure drop was in the
range of 2 to 2.3 inches of water and after bag  cleaning, the pressure drop was about
0.7 to 1  inch of water.  It is important to note that there is no significant increase of
pressure drop with time for the conditioned test, which indicates that conditioning does
not cause bag cleaning  problems or an increase  in  residual dust cake weight.

The actual residual dust cake weight at the end  of the 500-hour tests was  in the range
of 6 to 7 Ibs for both the baseline and conditioning tests which is quite light. However,
there  was a difference in the dust cakes for the  two  tests.  Without conditioning, at the
end of 500 hours, the clean side (outside) of the bag was quite dirty with a noticeable
dust layer.  With conditioning, at the end of the  test, the outside of the fabric was
completely clean with no evidence of fly ash penetrating the fabric. This difference in
dust cakes affects bag tension during cleaning as shown in Figure  3.  The initial bag
tension at the start of both tests was 75 Ib.  As dust accumulates during normal
filtration, bag weight will increase, but part of the  increased weight on the  bags is
balanced by the increased upward force on the bag cap as the differential pressure
across the bag  increases. The net result is that  the bag tension will increase by several
pounds over a 2-hour filtration cycle.  During bag  cleaning, the bag collapses into a star
shape and billows inward to the extent allowed by the anti-collapse rings.  This pulls the
ends  of the bag together and results in greatly increased bag tension.  The bag tension
during cleaning is dependent on  pressure drop across the bag in the reverse direction
and the reverse-air face velocity. At the end of  the 500-hour baseline test,  the reverse
pressure drop during cleaning was about 2.5 inches of water compared to about 1 inch
of water with conditioning.  After the first 250 hours, bag tension  during cleaning for
the baseline test was about 260 Ib compared to  only 150 Ib with conditioning (see
Figure 3), which demonstrates an additional benefit of conditioning. The high bag
tension without conditioning would likely lead to reduced bag life and may cause high
strain on the baghouse structure.  The reduced tension with conditioning should
facilitate longer  bag  life and allow a lighter support structure for the bags resulting in
additional economic  benefits.
Presssure Drop  Analysis

An explanation for the dramatic reduction in baghouse pressure drop as a result of
conditioning can be derived  by looking at the main ash properties that affect pressure
drop across a  dust cake.  Pressure loss across a fabric filter is the result of viscous flow
through a porous medium. The magnitude of the pressure drop is dependent on gas
properties such as viscosity and velocity, and on filter media properties  such as filter
thickness, porosity (void volume fraction), and pore size.  Assuming incompressible
flow, small pressure  drop compared to ambient pressure, and steady viscous flow,
according to Darcy's law, the pressure drop increases linearly with dust cake thickness
and velocity:

                                  A P  = K2 W V
where
   AP  = differential  pressure across baghouse tubesheet (inches  of water)
    K2 =  specific dust cake resistance coefficient (inches of water-ft-min/lb)
   W  = areal dust cake weight (Ib/ft2)
   V  =  face  velocity or air-to-cloth ratio (ft/min)
                                       23-4

-------
Prediction of K2 in terms of measurable dust parameters has been attempted, but
accurate prediction of the actual operating K2 for any given baghouse is difficult. The
Carman-Kozeny relationship is derived  from a theoretical capillary model and, assuming
spherical particles, takes the form  (14):

                          K,  = 36 k M  (1   e) / e3  Pn D2
                            d.                       i p

where
   K2 =  specific dust cake resistance coefficient (sec/ft); note: K2 can be converted to
     inches of water-ft-min/lb by multiplying by a factor of 311.6
   k  =  Carman-Kozeny constant  (-5) (dimensionless)
   \i  =  gas viscosity (Ib-sec/ft2)
   e  -  porosity  (dimensionless void volume fraction)
   pp =  particle density (Ib/ft3)
   D  =  particle diameter (ft)

Bush et al. and Gushing et al.  (15,16)  have reported an empirical relationship between
K2 and porosity for coal fly ash:

K2 =  (4  M / D2 pp) [(1   e)/e] [7.5  + 9.1(1  e)  35.8(1  e)2  + 560(1   e)3]

where the D term  is referred to as the  drag equivalent diameter. In the Carman-Kozeny
equation, D refers to the actual physical diameter for  monosized spheres.  For fly ash
that has  a broad particle size distribution,  the mass median diameter generally cannot be
used for  D for either equation.  The value of the characteristic diameter is dependent on
the particle  size distribution, specific surface area, and particle shape.  These equations
show that K2 is most sensitive to particle  size  (or characteristic particle size term) and
porosity.  Any  attempt, then, to alter K2 should focus on these  properties  and any
explanation  of a change  in K2 must include particle size and porosity.

K2 was measured for each of the three composite baghouse hopper ash samples for the
baseline  and conditioning tests.  To determine K2, a 150 g sample  of ash was placed in
a cylinder with a porous bottom and the pressure drop across the ash layer was
measured at constant air flow rate through the dust for several levels of dust
compaction. The porosity of the ash layer was calculated by measuring particle density
by helium pycnometry  and by measuring the dust layer thickness and cylinder diameter.
Results of the K2 measurements are shown in  Figure 4, along with the Carman-Kozeny
and Bush models.  A curve for each of the models was fit to  the measured K2 and
porosity  values for both the baseline and conditioned  data.  It appears the baseline  data
follow both  models closely while the data from the conditioned test seem to more
closely fit the  Carman-Kozeny relationship. Both the baseline and conditioned results
represent data  from three separate samples.  For a single sample, the K2  measurements
should define a smooth curve with minimal data scatter, and  such was the case for the
individual samples.  All data from the three baseline samples fit a nice curve with little
variability.  While the three conditioned samples showed more variability, their
composite data still define a distinct curve separate from the  baseline data.  The reason
why the  baseline and conditioned data formed separate curves  is not clear.  If the
particle size distributions and specific surface areas are unchanged, it is expected that
the two data sets would define the same  K2 curve.  Plausible explanations are that  the
particle size distribution for the conditioned samples was somewhat smaller than the
particle size distribution for the baseline samples, or the conditioned  samples had an
increased specific surface area.  Previous  data have not clearly indicated  any shift in the
fly ash particle size distribution as  a result of conditioning (9).  Coulter Counter data did


                                       23-5

-------
show the volumetric median diameter of one of the conditioned samples to be 11
compared to 13 jum for the baseline samples, but extensive data were not taken, and
specific surface area measurements have not yet been completed.

The explanation why the baseline data in Figure  4 were over a porosity range of 43% to
60% (void fraction of 0.43 to 0.60), while the conditioned data cover a range of 58%
to 75% is that the baseline ash has a much  greater tendency to compact. Procedures
were the same for all tests in that the same  approximate  compaction force was used to
obtain the low porosity measurements and no external compaction force was used to
obtain the maximum porosity measurements. Ash porosity as a function of compaction
force appears to be an important property of the dust which is also evident  from other
measurements to be discussed later.

Several important observations are obtained  from the K2 data and models in Figure 4.
First, both the data and models demonstrate that a small  increase in porosity can
significantly reduce K2.  At constant dust cake weight and face velocity, this would
correspond to a proportional decrease in baghouse pressure drop.  Second,  conditioning
caused a distinct difference in the measured porosity range.  These curves by
themselves do not define the porosity of the baghouse dust cake, but it would appear to
be a safe assumption that  dust cake porosity for the baseline and conditioning tests
would be somewhere between the respective minimum and maximum porosity values
shown.  The actual K2 values of the dust cake during operation can be determined from
dust loading and pressure  drop data.  The 500-hour tests were started with new bags
and the first 4 hours were  conducted without bag cleaning.  After the initial 4 hours, the
tubesheet pressure drop was 10.5 inches of water for the baseline test and 2.15 inches
of water for the conditioned test, which corresponds to a K2 of 17 inches of water-ft-
min/lb for the baseline test and 3.5 for the conditioned test.  Looking at Figure 4, this
implies that the dust cake  porosity was about 47% for the baseline test and  71 % for
the conditioned test. K2 can also be approximated by the increase in pressure drop
between bag cleanings.  From Figure  2, pressure drop increased about 6.5 inches (from
about 3 to  9.5 inches) between the 2-hour bag cleaning intervals for the baseline test,
compared to about 1.4 inches (from about 0.8 to 2.2 inches) for the conditioning test.
These data result in somewhat higher K2 values of 21 for the baseline test
corresponding to a dust cake porosity of 45% and 4.5 for the conditioned test
corresponding to a dust cake porosity of 68%.

A question exists whether  a valid K2 measurement can be attained by measuring the
increase in pressure drop between bag cleanings.  As long as the assumptions-constant
acfm, constant dust cake porosity, and no significant flow through large pinholes-are
valid, this should result in  a valid K2 measurement.  It doesn't matter if dust cake
loading over the bag's surface is uneven as  long as each part of the bag obeys the
Darcy's pressure drop equation.  With a multiple compartment baghouse, the flow rate
through a  given compartment will increase after  bag cleaning to  maintain constant
baghouse  pressure drop independent of flow path, and the first assumption would not
be valid.  With the single-bag  baghouse, however, constant flow was maintained after
each bag cleaning by adjusting the ID fan suction, implying that this is a valid
assumption. The second assumption of constant porosity is probably not valid and
offers an explanation of why the measured K2 between bag cleanings was higher than
the K2  measured during the first 4 hours of testing. Viewing the bag cleaning process
through sight ports in the top of the baghouse and bag cap showed that the bag
generally did not clean  uniformly, but large patches of dust cake were sometimes left.
This means that, after bag cleaning, the actual local face velocity for highly cleaned
areas is much  greater.  If porosity depends on face velocity, this could account for


                                      23-6

-------
differences in porosity at various locations on the bag and offers an explanation why the
overall effective porosity of the newly added dust could be greater than that measured
in the first 4 hours.  It is also likely that the porosity of residual dust at the dust/fabric
interface may decrease due to the shock of reinflation after bag cleaning and by
repeated subjection to the compaction force caused by the bag pressure drop.
However, this should affect only the  pressure drop  contribution from the residual dust
cake and not cause an error in the K2 measurement of the added dust. The third
assumption is also valid  for these tests, because no pinholes  were detected for the
conditioning test.  For the baseline test, there was  some evidence  of pinholes, but the
reasonably high particulate removal efficiency would indicate that only a  small portion of
the total flow went through pinholes.  Some pinhole penetration can be tolerated
without a significant increase in particulate emissions (9).

From the bench-scale K2 data and models shown in Figure 4 and the measured K2 values
from the baghouse data, we can conclude that conditioning reduced K2 by about 4.5
times by increasing the dust cake porosity from a range of 45% to 47%  for the baseline
test to a range of 68% to 71 % for the conditioned  test.  An obvious question is, "What
is the possibility of reducing K2 even  further by injecting  higher concentrations of
conditioning agents?"  The K2 versus porosity curves indicate that  K2 could be reduced
by an additional 50% by increasing the porosity from 70% to 80%.  The concentrations
of 25 ppm ammonia and 12  ppm S03 have proven  to effectively reduce pressure drop
and particulate emissions for several  coals and conditions; however, these
concentrations have not been optimized Tor any specific coal or set of conditions.
Therefore, the possibility exists for even further improvement, but  determination of the
effect of conditioning agent concentrations remains as a research need.

Porosity characteristics of the baseline and conditioned ashes were also measured with
a Powder Characteristics Tester that  performs several different mechanical
measurements of  bulk powder, such  as fly ash (13). Two of the more useful
measurements appear to be the aerated and packed density, which, along with particle
density, provide aerated and packed  porosity.  The  aerated porosity is obtained by
sifting an ash sample through a vibrating 60-mesh screen into a  100-cc cup so that dust
overflows the cup edge.  The excess dust is scraped off with a knife edge and the
weight of the known volume of dust  is measured to determine the bulk density. The
packed density is  determined by adding an extension to the cup and filling the extension
with additional sifted ash. The cup with extension  is then placed in a mechanism that
raises the cup about 1/2 inch and lets the  cup fall against a stop.  This is done once per
second for a period of 3 minutes. The cup extension is then removed and the excess
dust scrapped off as before.  There is no external compaction force on the dust layer.
Compaction is caused by the natural  settling that occurs as the dust is shocked.
Results of these tests are shown in Table 1 for the  baseline and conditioned samples.
Three or four repeat tests were completed on each  of the three baseline  and
conditioning baghouse ash samples.  Standard deviations shown in Table 1 include all
baseline results grouped together and all conditioned results grouped together.
Although, there is slightly more data  spread for the conditioned samples  compared to
the baseline samples  (similar to the K2 results), the  effect of conditioning on the aerated
and packed densities is very clear.  These  data again demonstrate  that the baseline ash
has a high tendency to compact, and that  conditioning imparts to the ash a resistance
to  packing.  It would  appear that dust cake porosity might be predicted by these
measurements, but enough data are not available to correlate with actual dust cake
porosity. In addition, actual dust cake porosity may depend on other factors such as
face velocity, fabric type, and cleaning method. Nevertheless, the aerated and packed
                                       23-7

-------
porosity measurements would appear to be useful methods in helping to predict
baghouse  pressure drop.

Cohesive strength of the ash samples was determined with a Cohetester, which directly
measures tensile strength.  An  ash sample is placed in a 5-cm diameter cell split into
two halves.  One half of the cell is stationary and the other half is suspended such that
the cell can be pulled  apart with minimal force when no powder is in the cell. When an
ash sample is placed in the cell, the force required to pull apart the sample is plotted as
a function of displacement for each test.  Multiple tests at different compaction  forces
provide information to plot cohesive tensile strength as a function of porosity for a given
ash.  Tests were again conducted with the  three composite ash samples for each  500-
hour test;  results are shown in Figure 5.  In this case, there was more data scatter with
the baseline results, but the two defined curves are  greatly different. From these  results
we can conclude that conditioning significantly increased the cohesive tensile strength
for a given porosity. The range in porosities was determined by the range in compaction
force, which was the  same for both conditioning and baseline tests. The maximum
compaction force allowable with the Cohetester resulted in a porosity of 39% for the
baseline samples and  53% for the conditioned tests. Similarly, the minimum
compaction force resulted in a porosity of only 51% for the baseline samples compared
to 67% for the conditioned samples. These results  again show that a major effect of
conditioning is to greatly reduce the packing tendency of the ash.

From the bench-scale and baghouse K2 data, we concluded that the actual dust  cake
porosity for the baseline test was in the range of 45% to 47% and for the conditioned
test in the range of 68% to 71 %.   Looking at the tensile strength values for these
porosity ranges provides an interesting  result.  The corresponding tensile strength for
the baseline tests is in the range of 0.7 to 1.0 gf/cm2 compared to 0.4 to 0.6 gf/cm2
for the conditioned tests.  While there is some data  scatter in this porosity range for the
baseline tests and extrapolation of the conditioned data was necessary to obtain the
tensile strength value for the highest porosity, the results indicate that the actual tensile
strength of the dust cake decreased with conditioning rather than increased.  This result
was not predictable, because previous measurements of ash pellet strength (8) and
effective angle of internal friction (17) showed that  conditioning causes an increase in
the cohesive strength of the ash. However, this result is highly desirable because it
would appear that  bag cleanability would be directly related to the actual dust cake
tensile strength.  A reduction in dust cake tensile strength should facilitate bag cleaning.
This may explain why the bag cleaned very well for the conditioned test and why there
was no increase in residual dust cake weight.  These results should  be considered
preliminary and need to be verified with other tests.  The Cohetester tensile strength
measurement, however, appears to be a good method to evaluate fly ash for fabric filter
performance and possibly predict bag cleanability.

To summarize the effect of conditioning on baghouse pressure drop, several
measurements show a significant increase in ash porosity,  which directly translates to
increased  dust cake porosity and  reduced baghouse pressure drop.  The conditioning has
a double effect in that it increases porosity, which allows operation at a lower pressure
drop.  The lower pressure drop  in turn reduces the compaction pressure on the  dust
layer allowing  a  high porosity to be maintained.  The reverse is true for the baseline ash
or any ash that has a high tendency to compact.  The tendency to compact causes high
pressure drop, which  results in  a  greater compaction force leading to even lower
porosity and higher pressure drop.  Therefore, a treatment, such as  ammonia and S03
conditioning, that reduces the compaction tendency of the ash, can be effective in
reducing baghouse pressure drop.
                                       23-8

-------
Particulate Emissions

EPA Method 5 dust loading measurements (Table 2) show that particulate emissions
were greatly reduced with conditioning, which was expected because of previous
results (4-8).  The EPA Method 5, however, is probably not sensitive enough to measure
collection efficiencies better than 99.999%, and it provides no information about
emissions as a function of time from bag cleaning.  Much more information is provided
from the APS data shown in Figures 6 and 7.  The  APS results are reported  as
respirable mass, which is a weighted sum of particles from 0.5 /urn to 10 urn (4).  The
APS measurements were taken only during the day, which explains why there are
distinct groups of data.  The spikes, especially evident for the conditioning test,
represent emissions just after bag cleaning;  however, most of the data groups include at
least two complete  bag cleaning cycles over a 4-hour period.  In Figure  7, the first 10
minutes after bag cleaning has been deleted, which greatly reduces the  spikes and gives
a better picture of the emissions for over 90% of the time throughout the 500-hour
tests.  Bags were cleaned every two hours for both 500-hour tests, except for an initial
4-hour filtration period.  After bag cleaning, data were taken every few  minutes for the
first 20 minutes to establish how rapidly the emissions dropped with dust buildup on the
fabric.  After the first 20-minute period, measurements were generally taken every 10
minutes until the next bag cleaning. Looking at Figure 6, it is apparent that  emissions
dropped somewhat  during the first 75  hours for both tests  and then remained fairly
steady for the remainder  of each test.  For the baseline test, respirable mass emissions
were about 1  mg/m3 compared to about 10~4 mg/m3 for the conditioned test. Respirable
mass values measured with the APS at the inlet to the baghouse were in the range of
850 mg/m3 to  1050 mg/m3, which when compared to the outlet emissions, corresponds
to a respirable mass collection efficiency of about 99.9% for the baseline test and
99.99999% for the conditioned test.  Conditioning  reduced particulate emissions by
about four orders of magnitude.  It is apparent that a substantial portion of the
emissions occurs in the first few minutes after bag  cleaning. An attempt was made to
integrate the respirable mass emissions over several cleaning cycles to determine the
collection efficiency based on an entire 2-hour filtration period. Based on the last two
respirable mass data sets from the 500-hour Monticello conditioning test, the average
respirable mass collection efficiency was 99.99996%, with  about half of the emissions
occurring in the first minute after bringing the bag back on line.  Looking at Figure 6  for
the conditioning test, most of the instantaneous spikes are below 10~1 mg/m3, which
corresponds to a respirable mass collection efficiency of  99.99%.  This  means that even
worst-case instantaneous collection efficiency is still better than 99.99% when
conditioning is employed.  Extensive measurement of submicron particle emissions also
confirm the extremely high collection efficiency (13).   Conditioning apparently improves
collection efficiency by enhancing the  ability of the dust to bridge large  pores or pinholes
and enhancing the ability of the dust to resist seepage or reentrainment.

Ammonia-Slip

Determinations of the amount of  ammonia in the flue gas downstream of the baghouse
(ammonia-slip) and in the fly ash  resulted  in a mean ammonia-slip of 3.1  ppm for six
samples with  a standard deviation of 1.1.  Fly ash ammonia analysis showed that 22.5
ppm of the original 25 ppm ammonia injected was transferred to the fly ash.  The two
measurements should add up to the 25 ppm of injected ammonia  for complete closure.
The ammonia-slip values are reasonably low and would appear to be of  little
environmental concern.  Nevertheless, the ammonia-to-SO3 molar ratio has not been
optimized, which implies that the ammonia-slip  could  be further reduced  by fine tuning
of the ammonia and S03 concentrations.


                                      23-9

-------
BENEFITS of CONDITIONING and EFFECT on ECONOMICS

Results show that a significant reduction in baghouse pressure drop and particulate
emissions are a direct benefit of conditioning. The impact of these benefits on the
economics of power production is highly site-specific, but a brief evaluation is
considered.  EERC has previously evaluated the cost of installing an ammonia and S03
conditioning system for a 500-MW plant (8).  The total  levelized cost of conditioning
was 0.18 mills/kWh (1986 $) compared to 3.02 mills/kWh  for a conventional reverse-air
baghouse (based on EPRI estimates (18)).  More recently, Burns and Roe Service
Corporation evaluated the economics of conditioning for DOE (19).  While the
assumptions were not all identical to the EERC evaluation, they stated that total
levelized  costs, including a conditioning system, were about the same as a base case
without conditioning when the size of the baghouse was held constant and the
tubesheet pressure drop was reduced by 50% (from 5 to 2.5 inches) with conditioning.
Further, they concluded  that there  appears to be good economic justification  for
pursuing  conditioning  in  conjunction with  fabric filtration for fine particulate control.

A summary of the benefits and effect on economics is presented below, where the
benefits from a reduction in  pressure drop and reduced  particulate emissions  are
considered independently.


Benefits from Reduced Pressure Drop

   •     Reduced baghouse pressure drop with the same cleaning cycle results in a
         direct savings  in fan power, which  may, in some  cases, more than pay for the
         cost of conditioning.

   •     Lower pressure drop and/or less frequent bag cleaning implies that bag life
         might be extended and maintenance costs reduced.

   •     In  cases where load  has been limited  by unacceptably high pressure drop, the
         economic benefit of conditioning would be substantial.  An improvement of
         only 1 % load capacity would likely more than pay for the  cost of conditioning.

   •     For new plant  installations,  results  imply that the size of the baghouse could
          be reduced with conditioning to achieve an overall economic benefit.  Further,
         conditioning provides backup reliability that should encourage vendors to
          promote less conservative designs.


 Benefits  from Reduced Particulate  Emissions

   •     The economic benefit in meeting emissions regulations is obvious.  Again,
          conditioning provides backup reliability which should encourage vendors to
          promote less conservative,  more economical designs.

   •      Emissions trading or banking may be  a  reality in the future, which provides an
          economic benefit  and incentive to reduce emissions beyond current
          regulations.

   •      For a new plant, PSD requirements may restrict plant location. The lower
          emissions with conditioning should facilitate  meeting PSD requirements.


                                       23-10

-------
   •     Enhanced particulate collection with conditioning encourages the use of
         economical fabrics and should promote smaller baghouses with overall
         savings.

   •     Reduction of particle bleed-through and bag blinding should  extend bag life and
         reduce maintenance.

   •     Enhanced public relations because of reduced emissions  (possibly cleaner than
         ambient air) has significant intrinsic benefits.


SUMMARY

Results show that fine particulate emissions and  baghouse pressure drop are greatly
reduced when ammonia and S03 are injected upstream of a woven-glass,  reverse-air-
cleaned filter bag.  The conditioning changes the cohesive characteristics  of the fly ash
resulting in a more porous dust cake.  Conditioning does not appear to cause  bag
cleaning problems, which indicates that this technology is applicable to reverse-air-
cleaned baghouses.  Results are consistent with  previous conditioning tests with other
coals, fabrics, and cleaning methods.  It appears that conditioning will reduce particulate
emissions and baghouse  pressure drop for a variety of coals with shaker-cleaned, pulse-
jet, and reverse-air baghouses.

A review of economic factors presents a strong case for the use of conditioning to
reduce pressure drop and particulate emissions.  For current operating plants, the
advantages are very site-specific, but, for new plants, there appears to be a general
economic incentive for pursuing this technology.

Research needs include demonstrating the technology on a larger scale for an extended
period, evaluating the potential to operate at higher air/cloth ratios with conditioning,
optimizing conditioning agent concentrations,  evaluating the  effect on ash disposal and
utilization, testing with pulse-jet baghouses, and  further evaluation of predictive ash
characterization methods.
ACKNOWLEDGEMENTS

The authors wish to acknowledge that the described work was funded by the U.S.
Department of Energy Pittsburgh Energy Technology Center, and that the test coal was
provided by the TU Electric Monticello Power Station.
REFERENCES

1.   P.V.  Bush, T.R. Snyder, and R.L. Chang.  Determination of Baghouse Performance
    from Coal and Ash Properties: Parts I and II.  J. of the Air Pollution Control Assn.
    39: 228 and 39: 361, 1989.

2.   D.R.  Sears and S.J. Miller.  Impact of Fly Ash Composition Upon Shaker Baghouse
    Efficiency.  Paper 84-56.6 presented at 77th Annual Meeting of the Air Pollution
    Control Assn., San Francisco, CA, June 24-29,  1984.
                                       23-11

-------
3.   S.J. Miller and D.R. Sears. The Influence of Coal-Specific Fly Ash Properties Upon
    Baghouse Performance:  A Comparison of Two Extreme Examples,  in: Proceedings
    of the Fifth Symposium on the Transfer and Utilization of Particulate Control
    Technology, EPRI CS-4404 Vol. 3, p. 22-1, February 1986.

4.   S.J. Miller and D.L. Laudal.  Real-Time Measurement of Respirable Particulate
    Emissions from a Fabric Filter, in:  T. Ariman (ed.) Particulate and Multiphase
    Processes, Vol. 2.  Contamination  Analysis  and Control, p. 663,  Hemisphere Pub.
    Corp., 1987.

5.   S.J. Miller and D.L. Laudal.  Particulate Removal Enhancement of a Fabric Filter
    Using Flue Gas Conditioning. Paper presented at the Third EPRI  Conference on
    Fabric Filter Technology for Coal-Fired Power Plants, Scottsdale, AZ, November 19-
    21, 1985.

6.   D.L. Laudal and S.J. Miller.  Flue Gas Conditioning for Improved  Baghouse
    Performance,  in:  Proceedings of the Sixth  Symposium on the Transfer and
    Utilization of Particulate Control Technology,  EPRI CS-4918 Vol. 3, November
    1986, p. 14-1.

7.   D.L. Laudal and S.J. Miller.  Flue Gas Conditioning for Baghouse Performance
    Improvement With Low-Rank Coals,  in: Symposium Proceedings, Fourteenth
    Biennial Lignite Symposium on the  Technology and Utilization of Low-Rank Coals.
    Dallas, TX, May 18-21, 1987, University of North Dakota  Energy Research Center,
    Grand Forks, ND.

8.   S.J. Miller and D.L. Laudal.  Flue Gas Conditioning for Improved  Fine Particle
    Capture In Fabric Filters:  Comparative Technical and  Economic Assessment,  in:
    Low Rank Coal Research Final Report, Vol.  II Advanced Research and Technology
    Development.  DOE/FC/10637-2414, Vol. 2 (DE87006532), April 1987.

9.   S.J. Miller and D.L. Laudal.  Mechanisms of Fabric Filter Performance Improvement
    With Flue Gas Conditioning,  im  Proceedings of the Seventh EPA/EPRI Symposium
    on the Transfer and Utilization of Particulate Control Technology.  EPRI GS6208
    Vol. 2, February 1989.

10. S.J. Miller and D.L. Laudal.  Flue Gas Conditioning Applied to Fabric Filtration, in:
    Proceedings of the European Symposium on the Separation of Particles from Gases,
    Nuremberg, West  Germany, April  1989.

1 1. S.J. Miller and D.L. Laudal.  Flue Gas Conditioning: A Method to Improve
    Collection Efficiency and Reduce Pressure Drop With Fabric Filters, in: H.M. Ness
    (ed) Proceedings of the Fifteenth Biennial Low-Rank Fuels Symposium. DOE/METC-
    90/6109 (DE90000427) p. 177 May 1989.

12. S.J. Miller and D.L. Laudal.  Flue Gas Conditioning Applied to Fabric Filtration, in:
    Proceedings of the Sixth Annual International Pittsburgh Coal Conference.
    Pittsburgh, PA, Vol.1  p.  330, September 1989.

13. S.J. Miller. Flue Gas  Conditioning  for Fabric Filter Performance Improvement.  Final
    Project Report for  contract no. DE-AC22-88PC88866 for Pittsburgh Energy
    Technology Center.  December 1989.
                                      23-12

-------
14.  A.E. Scheidegger. The Physics of Flow Through A Porous Media. The MacMillian
    Co., New York, 1957.

15.  P.V. Bush, T.R. Snyder, and W.B. Smith.  Filtration Properties of Fly Ash from
    Fluidized Bed Combustion.  J. of the Air Pollution Control  Assn.  37:  1292, 1987.

16.  K.M. Gushing,  P.V. Bush, and T.R. Snyder.  Fabric Filter Testing at the TV A
    Atmospheric Fluidized-Bed Combustion (AFBC) Pilot Plant.  EPRI CS-5837 May
    1988.

17.  F.G. Pohl.  A Novel Ring  Shear  Device for the Purpose of  Classification of Fine
    Powders.  Paper presented at the 17th Annual  Meeting of the Fine Particle Society,
    San Francisco, CA, July  1986.

18.  R.W. Sheck, R.R. Mora, V.H. Belba, and F.A.  Horney.  Economics of Fabric
    Filtration and Electrostatic Precipitators--1984.  EPRI CS-4083, June  1985.

19.  J. Ratafia-Brown.  Technical/Economic Evaluation of Flue  Gas Conditioning for Fine
    Particulate Control.  Prepared for U.S. Department of Energy  Pittsburgh Energy
    Technology Center under Burns and Roe Service Corporation  Contract No. DE-
    AC22-89PC88400, Subtask 45.09,  July 1989.
    FD Fan
                         fl
J


'
^
KHI
Heat * 1
Exchange
? 1
\
\
                                                 NH3+ S03 injection
              Figure 1.  Particulate test combustor and baghouses.
                                      23-13

-------
                   12
                 ex
                 o
                 cr
                 a

                 LjJ
                 a:
                 ID
                 CO
                 CO
                 L±J
                 cr
                 Q_
O
5
m
   6 -
                    4 -
                    2H
-o- BASELINE
    (before  and  after bag cleaning)

-•- CONDITIONED
    (before  and  after bag cleaning)
                             I 00
                  200
                    RUN
        300     400
       TIME  (HOURS)
                                                         500
600
Figure 2.  Baghouse pressure drop for 500-hour reverse-air baghouse tests with
          Monticello coal. Upper values are pressure drop before bag cleaning; lower
          values are after bag cleaning. Bags were cleaned every 2 hours.
                   300
                                     i  i i  i i  : i i  i i  I
                                    200    300    400
                                     RUN  TIME (HOURS)
                                        500
                               600
Figure 3.  Bag tension during bag cleaning for 500-hour reverse-air baghouse tests with
          Monticello coal.  Bags were cleaned every 2 hours.
                                        23-14

-------
                    25-
                    20-
                  c

                 1
                     5-
                  O
                  IN
                  I 10-

                  Cfl
                  CD

                  o

                  »  5-
. _ Carman —Kozeny
— SoRI/Bush

o  Monticello Baseline
•  Monticello Conditioned
                     0.40    0.50    0.60    0.70    0.80    0.90    1.00
                              POROSITY (void  fraction)
 Figure 4.  Specific dust cake resistance coefficient, K2, as a function of ash porosity
          with Carman-Kozeny and SoRI/Bush models fit to data.
                                                 -A- BASELINE
                                                 -— CONDITIONED
                               40       50       60       70
                                   PERCENT  POROSITY
                        80
Figure 5. Cohesive tensile strength as a function of ash porosity as measured by the
         Cohetester method.
                                       23-15

-------
                   io
                                          -o-  BASELINE
                                          -•-  CONDITIONED
100     200     • 300     400
       RUN TIME  (HOURS)
                                                                500
Figure 6.  Respirable mass particulate emissions for 500-hour reverse-air baghouse tests
          with Monticello coal.  Bags were cleaned every 2 hours.
                                          -o- BASELINE
                                          -•- CONDITIONED
                   10 -5-
                              100
          I i M i i i i i i i
         200      300
       RUN TIME (HOURS)
400
500
Figure 7.  Respirable mass particulate emissions for 500-hour reverse-air baghouse tests
          with Monticello coal, ignoring the first 10 minutes after bag  cleaning.  Bags
          were cleaned every 2 hours.
                                       23-16

-------
                           TABLE 1

               AERATED AND PACKED POROSITY
                   Aerated           Packed
                   porosity           porosity
      Ash type       %      aa   nb
      Monticello    62.6    0.6  9     40.1   0.8   9
       baseline

      Monticello    75.8    1.5  10    55.0   1.2  11
       conditioned

      3 a =  standard deviation
      b n =  number of tests
                             TABLE 2

         DUST LOADINGS AND BAGHOUSE COLLECTION EFFICIENCY
                    MEASURED BY EPA METHOD 5
                              Dust loading           %
                                Grains/scf        Collection
   Test                       inlet    outlet       efficiency
Monticello baseline           5.83     0.0054      99.907a
                                                 n = 8
                                                 a =  .082

Monticello conditioned        6.45     0.000019   99.999a
                                                 n  = 9
                                                 a  = .0004

3 n  = the number of valid outlet dust loadings; a  = the standard  deviation of
 the collection efficiency values for the given  number of tests.
                             23-17

-------
                     BAGHOUSE PERFORMANCE ADVISOR
             A KNOWLEDGE BASED BAGHOUSE  OPERATOR  ADVISOR
                         James  P.  Eckenrode
                         Gary P-  Greiner
                         ETS,  Inc.
                         3140 Chaparral Dr.,  SW
                         Roanoke,  VA 24018-4394
                         Earl  Lewis
                         Baltimore Gas  and Electric Company
                         C.  P.  Crane Power Plant
                         P-  0.  Box 1475
                         Baltimore,  MD  21203
                        Dr.  Ramsey  Chang
                        Electric  Power  Research Institute
                        3412  Hillview Avenue
                        Palo  Alto,  CA 94303
ABSTRACT
Recognizing the lack of integrated baghouse  instrumentation  and
continuous performance feedback available  at  most  baghouse  sites,
the Electric Power Research Institute has  funded the  development  of
a software package that provides  the user  a  means  of  monitoring his
system for stable operation and identifying  the root  cause  of
problems as they develop.  Easy access by  operators to  current  and
historical operating data is a feature of  the software  along with
the built in expertise of baghouse consultants for the  analysis and
troubleshooting of common problems.  This  paper describes  the
purpose and features of the developed software package  that  runs  on
a desktop personal computer and can be applied to  any baghouse
system.
                                24-1

-------
                   BAGHOUSE  PERFORMANCE ADVISOR
           A KNOWLEDGE BASED BAGHOUSE OPERATOR ADVISOR
INTRODUCTION

Instrumentation of baghouse systems in North America  is  often
incomplete and not integrated for easy operator usage.   Although
many baghouses have a computer data acquisition system incorporated
in the porcess control instrumentation, they do not provide  adequate
capabilities and sensitivity for diagnostic monitoring.   In  many
installations only a limited interface between the baghouse  and main
plant controls exists.  Often the baghouse will be operated  from  a
physically separate control room.  ETS, Inc. has spent the past six
years developing instrumentation and techniques specifically de-
signed to monitor and troubleshoot baghouse operation.   This effort
has resulted in the development of the Individual Bag Flow Monitor
(IBFM) and the Baghouse Performance Monitor(l).  The  advent  of the
microcomputer has made the collection, storage, analysis, and
display of information much more widely available.  Now,  under
contract to the  Electric Power Research Institute (EPRI). ETS is
developing a Baghouse Performance Monitor Expert Software (BPMES)
package to integrate all baghouse instrumentation signals and
incorporate its expertise in monitoring and troubleshooting  tech-
niques.  The BPMES system is intended for use on any  type of bag-
house system where the user wishes to maximize the continuous
performance of their baghouse while  minimizing the job  of evalua-
tion and interpretation of baghouse operating data.   The objective
of this effort is to develop a knowledge and graphics based  micro-
computer system for assisting in the detection, diagnosis and repair
of baghouse operating problems.  Although the BPMES is not an expert
system that reasons for itself by inference, it is a  knowledge based
operator advisory system which provides some of the features associ-
ated with expert systems (2).  It makes expert advice available 24
hours per day, provides assistance in analysis of graphical  data,
helps users identify probable causes of failures, and provides
suggestions for remedial action.   It helps to transfer the knowledge
of baghouse diagnostic experts to less experienced personnel who
need it for daily operations. It is designed to operate  on an
IBM-compatible personal computer with an EGA quality  monitor, 20
megs of hard disk storage,  and an Epson-compatible printer.
                                 24-2

-------
When problems such as high emissions or abnormal pressure  drop  are
detected, the program leads the user through analysis  and  solution
procedures based on extensive ETS baghouse experience.  Specific
maintenance instructions can then be generated  for  corrective
action.  The result is consistent baghouse operation,  lower  operat-
ing and maintenance costs, fewer unscheduled outages,  and  added
protection against process de-ratings or compliance  violations.

The BPMES utilizes commonly available baghouse  operating parameters,
along with some novel specialized parameters, collected in a data-
logger and displayed on a computer for easy monitoring of  current
operation.  Such a system is shown in Figure 1.  The data  transfer
is accomplished by telecommunications which allows   the program user
to monitor one or more systems from a single remote  site or  provide
access to the data from more than one analysis  station.  This will
provide engineers and managers, remote from the control room, easy
access to current and historical operation.  The datalogger  will
handle twelve inupts which can be processed to  give  up to  20 parame-
ters for the BPMES.  The data will be stored in the  form of  one-hour
averages for trend analysis, and in the form of user adjusted
one-second to one-minute averages for troubleshooting  purposes.  A
list of typical baghouse system parameters is shown  in Table 1.  The
parameters selected for use at any one baghouse site   are  chosen by
the user depending on need and  availability of instrumentation.


TREND ANALYSIS

Catastrophic baghouse problems are usually detected  and presented to
operators by the control system supplied with the baghouse.  Long
term, slow and  gradually developing problems are frequently not
detected by that control system, and the identification of the  root
causes of problems are not addressed.  An example of a typical  trend
analysis graph is shown in Figure 2. In Figure  2, the  full load
pressure drop is steadily rising throughout the period indicating
higher flow resistance has developed.  Since the load  during this
period is not at full capacity, the actual pressure  drop will not
rise as much, if at all,  and will probably stay below any critical
value.  Since the typical baghouse instrumentation  only displays
actual pressure drop, the operators will not be alerted to the
developing problem until full load operation is resumed.   Recogniz-
ing that the most important element in this system  is  information
control, the software allows the user easy access to both  current
and historical data sets; provides a variety of graphical  displays
for visual analysis;  and incorporates expert help screens  to assist
the user in the diagnosis of problems.  When instability is  detect-
ed, additional trend analysis  graphs of other  key  parameters,  along
with specialized troubleshooting displays, are  available to  assist
the user in defining the root cause of the problem.  Figure  3
presents the list of available trend analysis parameters,  shown in
the user interface format. To assist operator analysis, these
parameters can be viewed in four different ways:
      •   Single Graph - Data for a single parameter will  be
          displayed versus the time the data was collected.
                                 24-3

-------
          Double Graphs - Data for two different parameters will be
          displayed,  on separate graphs, for the same time period.
          Dual Operating Periods - Data for a single parameter
          will be displayed on the same graph, for two different
          time periods.
          Parameter vs. Parameter - Data for two different
          parameters will be displayed, on the same graph, one
          on each axis, for the same time period during which
          the data was collected.
STABILITY CHECK

In a special separate function, the system collects, calculates and
analyzes those trend analysis parameters which the operator
pre-selects to define whether the baghouse is stable.  By stable, we
mean acceptable and constant gas flow, opacity and pressure drop
levels.   Key parameters of average opacity, peak opacity, full load
G/C ratio,  and full load pressure drop are collected and displayed
in graphs of duration as little as one day to as much as one year.
Predetermined limits and bands of acceptable operating levels allow
the user to define individual system stability.
TROUBLESHOOTING

One of the premises upon which the BPMES is built is that  all
baghouse system  operating characteristics are unique and  that
absolute parameter levels are not  as important as relative  and
trended values. A basic feature widely utilized in this  system is
the parameter template which describes an operating parameter, for
that specific site, during good, normal operation.

The normal operational parameters are an integral part of  the
problem solving process.  These parameters are set during  the
initialization of the software at a plant site.  Figure  4  presents
the list of available troubleshooting parameters, shown  in the user
interface format.  The initialization consists of collecting operat-
ing data and storing it in a database to be called on by the program
for data comparison.  From this database, acceptable operating band
limits and parameter templates are derived. Such a template  is shown
in Figure 5 which shows the flange-to-flange pressure drop of a
modular reverse air baghouse through a cleaning cycle.   With the
proper resolution and presentation of this parameter, every  system
damper  movement can be observed and analyzed.  By establishing  a
template in the computer of how this parameter's acceptable  opera-
tion looks, periodic comparisons of current conditions can be made
by overlaying the current data  trace with the template.   Through
use of visual help windows and screens, the user is guided in the
analysis of several troubleshooting and trend analysis screens to
identify the root cause of problems. The help screens are  set up to
correspond with the potential problems that might occur  for  a given
parameter.  When a parameter is plotted, the appropriate help
screens for that parameter are available.  Each help screen  leads
                                 24-4

-------
the user through  instructions  that  identify  primary and/or secondary
causes of the problems.  The help screens  are  interrelated,  program
context-sensitive, and prompt  the user  to  the  next  help  screen or
data plot.   The help screens explain  what  to analyze on each plot,
list in order of  probability the potential root  cause of the prob-
lem, and prompt the user to the next  graph or  help  screen to further
isolate the identification  of the  problem.  The key troubleshooting
aids are:
              Damper operation check
              Source of opacity
              System flow check
              System drag check
              Sonic horn check

In addition combinations of any two collected  parameters can be
displayed together and compared to  prior collected  data sets for
analysis.   Another very useful function of the BPMES is the  source
of opacity graph  which allows  early detection  of bag failures and
identifies their  location in the baghouse.   It identifies the
module, where the failed bags  are located.   The  technique used is
shown in Figure 6 and requires the  coordinated display of opacity
with  the cleaning cycle.  This analysis is  based on the fact that
opacity spikes occur from just cleaned bags  or modules.   This
information is often available at plant sites  but normally in
different areas,  making the required  coordinated analysis of the
information difficult. Using this troubleshooting screen or  a hard
copy printout, maintenance personnel  can be  easily  directed  to
correct the problem before opacity  violations  occur or further bag
damage is created.  One of the novel  parameters  utilized by  the
system is generated by ETS's patented Individual Bag Flow Monitor
(IBFM)(1).   The IBFM produces  gas flow or  G/C  levels and fabric drag
values for individual bags within the collector-  This reliable and
accurate flow characteristic information can be  utilized to  detect
increase in inlet flow levels, loss of bag cleaning effectiveness,
and increase in fabric drag or flow resistance.   In reverse  air and
shake deflate baghouses, the key parameter of  reverse gas flow can
be easily monitored.  In pulse jet  collectors  the relative emissions
from individual bags have been monitored using the  IBFM sensor
lines.

Through the use of IBFM sensors in  a  matrix  configuration within a
common compartment,  alternative analysis is  made easy(3)(4).
Alternative test bags are distributed within a common compartment
and three bags of each type are outfitted  with Individual Bag Flow
Monitors as shown in Figure 7-  (For  pulse jet baghouses the IBFM's
would be located  at the top of the  bags).  Temperature and pneumatic
pressure signals which detect both  flow and  pressure drop are
delivered through the pass-through  box to  the  signal conditioning
module (SCM)  which is located  local to the baghouse.   The tempera-
ture and pressure signals are  converted to electrical signals in the
SCM and transmitted via cable  to the  control and monitoring  modules'
datalogger  which  is located in a control room  environment. Through
simultaneous  measurement of flow and  pressure  drop,  alternative
fabric flow properties can be monitored and  plotted vs.  time for
                                 24-5

-------
comparison,  as shown in Figure 8.  The.best fabric  from  a  pressure
drop standpoint will be the one producing  the  lowest  flow  resistance
or drag,  over the life of the bag.  This information,  coupled  with
bag cost  and bag life, allows each user to determine  the best
overall bag choice for their specific site.

Similarly,  baghouses considering alternative analyses  of cleaning
energy, bag tensioning, sonic horns, pre-coating, etc.,  will find
that the  analyses can be performed at very low cost using  the  BPMES
software.  The retro-fit conversion and monitoring of  a few bags  can
generate  the required performance information, through improved  data
collection  and evaluation techniques, before large  investments in
full scale  modifications are made.  The evaluation  time  is also
greatly reduced since many alternatives can be evaluated simulta-
neously using these techniques.
EXPERT HELP

In addition to providing context-sensitive help screens  for  graphs
of specific operating parameters, an advisor function  is  provided by
the program under Expert Help selection.  A logic tree of possible
malfunction indications is presented and help screens  at  each branch
of the tree provide information to  assist the user who  is diagnos-
ing an unfamiliar problem.  Figure 9 depicts the logic tree  for  a
pressure drop problem.  This expert help function acts as a
computer-based training aide for those  less  familiar  with baghouse
operation and the analysis of operating data.  The less  experienced
user can respond to the logic tree presentations, seek assistance
from the help screens, and learn which parameters are  appropriate
for troubleshooting specific types of problems.  The help screens
are text-based messages which the user may edit.  Because of this
feature, the expert knowledge contained in the program can be
modified to customize the message to an individual baghouse  system.


REAL-TIME DATA

The BPMES system also permits the display of real-time,  unprocessed
data from the datalogger.  This real-time data has not been  convert-
ed from the datalogger's ASCII file format to a binary file  for  use
by BPMES. Because it  is still in ASCII  format, it is available  for
viewing to confirm the datalogger's proper operation.  In addition,
if a computer workstation is dedicated  to the BPMES system,  another
data collection mode, Initiate Automatic Datalogging,  permits
continual, periodic on-line monitoring  of system operation.  At
user-set predetermined intervals, the computer automatically calls
the datalogger, retrieves the most recent data,  analyzes it for
comparison with preset alarm limits, and presents a list  of  any
alarm messages that occurred in that most recent data  set.

FIELD DEMONSTRATION

A field trial of EPRI's Baghouse Performance Monitor Expert  Software
                                 24-6

-------
System is taking place at Baltimore Gas and Electric's  C.  P.  Crane
Station during the twelve months of 1990.  Additional demonstration
sites are being considered.  C. P. Crane Station has  two  units.
Each unit has a 188 MW net generating capacity and utilizes  crushed
coal to fire Babcock & Wilcox cyclone boilers.  The commercial
operating dates of Unit 1 and Unit 2 were  1961 and 1963,  respective-
ly.  They were converted to oil in 1970 and converted back  to coal
in 1983.   Identical QEESI baghouses were retrofitted  onto  each  unit
during the 1983 coal reconversions.  Some  basic design  details
regarding these baghouses are presented in Table 2.   In 1988  and
1989, respectively, IBFM's and BPM's were  installed on  Units  1  and
Unit 2.  For the EPRI demonstration project,  the IBFM's were  relo-
cated to provide two in each compartment.  This will  provide  the
capability for maximum collection of troubleshooting  data.   The EPRI
BPMES system was installed on a plant personal computer by  the  Plant
Performance Engineer in December 1989 and  January 1990  for  Units  1
and 2, respectively.  The EPRI demonstration  project  for  the  BPMES
at C. P.  Crane Station in 1990 consists of the following  tasks:

Task 1: Start-up and Initialization (2 months)

Task 2: Program Set-up and Check out (4 months)

Task 3: Testing (5 months)  - After the above tasks are completed, a
final report will be issued documenting the results and the  final
program software incorporating modifications  and updates  that
occurred during the demonstration.  Figure 13 displays  some  prelimi-
nary outputs of the BPMES system from C. P. Crane Station.

Much of the capacity of the BPMES software is currently incorporated
in the ETS's "Trend Analysis Baghouse Software" (TABS)  which  is a
part of its Baghouse Performenace Monitor  (BPM) instrumentation
system.  When BPMES becomes a released product, users of  TABS will
be provided a copy to update their systems.
CONCLUSION

The baghouse has been developed to the point where it  is considered
the best available control technology for many applications.   This
certainly is true when the systems are in peak operating condition
as they normally are during certification and compliance testing.
The problem and challenge has been to maintain the baghouse to that
high performance level throughout its operating  life.   The  installa-
tion of EPRI's Baghouse Performance Monitor Expert Software system
will allow the user to come much closer to that  objective by  making
available to novice operators the expertise of baghouse diagnostic
specialists.  It is the first knowledge based particulate control
operations advisor available to the electric utility  industry.
                                24-7

-------
REFERENCES

1.    Greiner,  Gary P- and John C. Mycock and David S. Beachler.
     "The IBFM - A Unique Tool for Troubleshooting and Monitoring
     Baghouses."  In Proceedings of the 77th Annual Meeting of the
     Air Pollution Control Association, San Francisco, CA., June,
     1977.

2.    Musgrove, John G.  and Di Domenico, Peter N. "Expert Systems -
     Operator Advisors  for Particulate Control Equipment." In
     Proceedings of the Joint ASME/IEEE Power Generation Conference
     Philadelphia, PA., September, 1988.

3.    Sturtevant, John E.  and Gary P.  Greiner. "Optimizing  and
     Controlling Baghouse Operations at BG&E C. P. Crane Station."
     In Proceedings of  the 7th EPA/EPRI Symposium on the Transfer
     and Utilization of Particulate Control Technology, Nashville,
     TN.,  March, 1988.

4.    Mycock,  John and John Ross and Gary Greiner- "Lab  Analysis as
     a Tool for Improving Baghouse Operation and Maintenance."  In
     Proceedings of the 76th Annual Meeting of Air Pollution Control
     Association, Atlanta, QA., June, 1983.
                                24-8

-------
                            Table 1

         TYPICAL BAGHOUSE PARAMETERS UTILIZED BY  BPMES
Input Signals
  Calculated Parameters
Inlet flow volume
Selected individual bag flow
volumes
Inlet and outlet temperature
Flange-to-flange pressure drop
Module pressure drop
Reverse gas flow volume
Compressed air pressure
Stack opacity
Unit Load
                              Actual G/C ratio
                              Projected full  load  G/C
                              Projected full  load  pressure  drop
                              Fabric drag
                              Reverse gas  flow G/C  ratio
                           Table 2

           C.P. CRANE STATION BAGHOUSE PARAMETERS
        Number of Compartments                     10
        Number of IBFM's per compartment           2
        Number of Bags per Compartment             522
        Bag Dimensions               12 in.  (diameter)  x  35  feet
        Bag Material                 Acid Resistant Fiberglass
        Bag Cleaning                 Reverse Air
        Operating Temperature        350 degrees  F
        Design Pressure Drop         Less than  10  i.w.g.
        Overall Dimensions           87'8"W  x 133'9"L  x 76'0"H
                           24-9

-------
                   BAGHOUSE

                   dP. IBFM ff
                                                      Air Pressure
                                                              	^.i

                                                     SON 1C HORN/
                                                       —-.-L-^1^
                                               Tolccommuni Gallons
                                     Compulor and

                                    Expert Software
                                      Figure 1
             Proj. F1-F1 dP & Unit Load  (HU) - ouer  Monthly Period
Station:  EPFI                            Fl-Help
       c
P
r
o
L
o
a
a

n
u
          -(-——]-<  '  '  •  --y——-] •   •  •  '
          10/06/89      10/13/89      10/20/89       10/27/89      11/03/89
          10/06/89      10/13/89      10/20/89       10/27/89      11/03/89
                                      Figure 2
                                        24-10

-------
                  Baghouse Performance Monitor Expert System
Stability
Check
Trend
Data
Troubleshoot
Data
Report
Generation
Expert
Help
Special
File
             View Present Data
             View Archived Data
               Single Graph
               Double Graph
                                   Parameter vs Parameter
                                   Dual Time Graph
                         Choose Two Options.
    1)  Proj.  Gas to Cloth
    2)  Gas to Cloth Ratio
    3)  Module Temperature (F)
    4)  Fabric Drag
    5)  Module Delta P
    6)  Proj.  F1-F1 dP
    7)  Flange to Flange Delta P
    8)  Unit Load (MW)
    9)  Opacity
    10) Max.  Opacity
                    11)
                    12)
                    13)
                    14)
                    15)
Compressed Air, psi
Outlet Temperature (F)
Inlet Flow
IBFM t
IBFM dh
r
-------
Start Time
A P ', :
" H20 ' •
i •
7 -
A P 6 °
" H2O 5 .
3 -
2 -
12:00:00
,—m
Damper
rwYYY"
Dpe
rv
ration Check
s

Expanded View
Re
r- D£

J J R«v»ra* Air
J1-HU.
	 Actual Data

; i N<
2nd Hull ^
RA -Inflate
12:54:30

r-lnflatlon
mper Old
>t Open

I

	 Template
Highlight Tkne
7 A P
I "H20
. 7
• 6 A P
. 5 " H2O
• 4
- 3
• Template
Figure 5  Damper Operation Check Graph
Start Time
eo -
„ % Opacity eo -
to *
20 •
% Opacity
20 —
15 -
10 -
5 -
0 -
12:00:00 Source of Opacity Problem 12:18:00
rvfSMvrv l^/^l^Jr^^
A
ViJU/vAAJW




V

Expanded View
	 r\j~u~u~\ru~\ru~ii_ri_ri_



... ^A^AJ\

1


MU\AA

Highlight TJrne
: A P

— 6 »
A P
-4
-2

Avq. @
Cleaning
Avq. @
Normal
Operation
Figure 6  Source of Opacity Check Graph
               24-12

-------
              1
IBFM
                                          (.	BAGHOUSE WALL
                                             <-PASS-THRU
                                               INTERFACE TERMINAL
  CENSOR AP

    ^.TUBESHEETAP


-MODULE TEMPERATURE
                                             e-SIGNAL CONDITIONING
                                              MODULE
                                                          CONTROL  & MONITORING
                                                          MODULE
                         Figure 7   IBFM™  INSTALLATION
                                    24-13

-------
                             PERM.  AFTER  30"  UAC .
Stat ion :  Demo
       3E
 P
 E
 R

 E
 A
 B
 I
 L
 I
 T
 V
                                                                        F9-Print
—13
D MENARDI 601C
_MENARDI 681T
  JPS 648-AR
* BUR. 427-1625
• BUR. 454-1625
^MID KNIT GLAS
o GORTEX-UOU GL
                                   14   1 b   1 13

                                       MOMTHS
                                                             ?!EI  ^ O  3 S   34
                                     Figure  8
                   Baghouse  Performance  Monitor Expert System
Stability
Check

Trend
Data
Troubleshoot
Data

Hiah Gas Flow
Report
Generation

Expert
Help
Special

High delta P
High Opacity
File

           Bad  Reading
           High Resistance
                               Cleaning  System
                               Blinded Bags
                               Inlet  Loading
                                                   Check Reverse Gas G/C Ratio
                                                   Check Damper Operation
                                                   Check Sonic Horn
 pcstationi  EPRI
 IMoves Cursor    <->  Selects Item
                                     ESC Back Up
              Fl Help
                                   Figure 9
                                      24-14

-------
                        EFFICIENCY OF FABRIC  FILTERS AND  ESPs
                        IN CONTROLLING TRACE  METAL  EMISSIONS
                           FROM COAL-BURNING  FACILITIES
              Roger C. Trueblood   Particulate Collection Specialist
            Christopher Wedig - Air Pollution Control Device Specialist
              Richard J. Gendreau - Supervisor, Fluid Bed Combustion
                      STONE & WEBSTER ENGINEERING  CORPORATION
                                 245 SUMMER  STREET
                                 BOSTON, MA   02210

 INTRODUCTION
 Trace  metal  emissions  from  fossil  fuel  combustion  facilities have only  recently
 been subjected to  regulatory scrutiny.   Lacking federal  guidance, many states  have
 implemented,  or  are  developing,  their own  Air  Toxics policies  or guidelines.
 These  are as  varied  as  the  states  themselves,  from  a vague  "nothing  that  is
 harmful"  standard  to  specific numerical  values for modelled offsite concentrations
 of  hundreds of substances.
 Various metals are usually included in  any such list of  regulated emissions.   Lead
 (a  criteria  pollutant)  and mercury  and beryllium  (non-criteria  pollutants) are
 regulated  for  PSD sources  under  the  Clean  Air   Act.    State  programs  frequently
 lower  the emissions  quantities  that  trigger reviews and add  other metals to the
 list.
 This is not  to  imply that metal emissions associated with fossil fuel burning for
 power production  present  a  health hazard or should even be a significant concern;
 only that  the  regulatory  climate is shifting  toward  inclusion of metals analysis
 as  part  of the  permitting  process.  In fact, applying BACT  for particulates  in
 power plants  is probably more than adequate control of trace metals.
 Some states  have  categorically  determined that the risk of air toxics from fossil
 fuel combustion is small.   Maryland, for example,  specifically exempts fossil  fuel
 combustion  sources  from  its  Air  Toxics  program.   Other  states,  in  practice,
 require little  analysis of  this  class  of  sources.   Nevertheless,  the  impact  of
 trace metal  emissions  from  fossil  fuel combustion, especially  coal,  is more and
 more frequently examined in permitting  new facilities.
 There are  very  limited  data available  on the  fate of  various metals from coal  in
 modern  power  plants and air pollution control devices and, in fact, on the concen-
 tration of these  metals in  coals from  various sources.   Although we are not aware
 of  any  power plant  that  has yet been  denied a  permit  due to  trace  metals emis-
 sions,  we  have  seen  a broad range of assumptions  used by applicants in  estimating
 their  impacts,  in  some cases  leading  to  permit conditions  that were not  well
 founded.   Thus we  want  to examine the  available  data  on the fate of these metals
within  the power plant and associated air pollution devices.

                                        25-1

-------
Stone & Webster  has  recently  performed  technical  reviews  for  several  projects
where the construction  and operating permits have included limits on the emissions
of non-criteria pollutants, including a number of trace metals.  In a few of these
projects the plant owner  and the engineer made a good faith effort to predict the
expected level  of trace  metal  emissions; however, the lack of data and experience
with these  considerations,  by  the  owner and the regulatory  agencies  resulted in
unreasonably low limits  being  specified  for  several  trace  metals.   Subsequent
analysis of pilot plant  test  results  indicated  that some  of these  limits  were
probably not achievable.   The  limits for two trace metals  on one project were so
low that the specified  EPA test method would have required sampling times of 20 to
100 hours.   The  mercury  emission limit,  on that project, was actually 500 times
less than the  maximum  allowable working place concentration specified in Germany,
where pilot plant testing was performed.

In all  these projects,  had conservative assumptions and  analyses been used, along
with a  more rigorous examination and  analysis  of the sources  and  disposition of
the trace metals, the  resulting emission limits would still  have been well below
state hazardous  pollutant  regulations.   Therefore,  the problem is  not  that power
plants  are  a  major  source of  trace  metal  emissions,  quite the opposite  is true;
but rather,  in  today's  regulatory environment,  how do  we  establish realistic plant
emission limits using the extremely  limited data available today?

SOURCES AND BEHAVIOR  OF TRACE METALS IN POWER PLANTS
It is clear that a better understanding is required  of the factors that contribute
to and  influence  the emissions of trace metals  from power plants.   All  the parti-
cipants  in  the permitting process  must develop this  knowledge  so  that decisions
are  based   on  the best  available  information.   Factors  that must  be  considered
i nclude:
     1.   The  sources and  concentrations  of trace metals entering the boiler
     2.   The  behavior of  trace  metals in the combustor/furnace and back end
          of the boiler
     3.   The  type and characteristics of  the  particulate  collection system

Understanding  these  factors  will make it possible  to better  predict trace metal
emission from  new projects  without  flue gas desulfurization  (FGD) systems.   The
behavior of trace metals  within FGD systems is outside the scope of this paper.
                                        25-2

-------
Sources and Concentrations of Trace Metals Entering a Boiler
The primary source  of trace metals entering  the  boiler is the coal or other fuel
burned in the  unit.   In the case  of  fluidized bed boilers, the limestone used to
capture sulfur dioxide may also be a minor source of trace metals.

Estimating the concentration  of trace metals in coal, on a general basis (by coal
type, region,  etc)  is difficult and potentially problematic.   The published trace
metal concentration  data are  limited and can  range over two  to  three orders of
magnitude.   Using the mean values from these sources may result in serious errors.
Even  within  specific  coal  seams,  trace  metal  concentrations  can vary  several
orders of magnitude.   It is therefore important to  thoroughly  sample  and analyze
the  specific  coal  supply before  establishing emission values.   If this  is  not
possible, conservative  values  should  be  used to provide preliminary estimates of
trace metal  emissions.  Estimates should be subject to revision when more specific
fuel data are available.

These data  are not typically part  of coal analysis, which is  why available data
are severely limited.   If the anticipated coal supply  is  from  a specific source,
representative  data may  be  obtainable.   However,  it is more common that coal will
be purchased via brokers and may come from a variety of sources  at  any  given time.

Behavior of  Trace Metals in the Combustion/Furnace and Back End of  the  Boiler
To a  great extent,  the chemical and  physical  behavior  of  the trace metals  during
combustion and the  subsequent  cooling of the flue gas by transferring  heat to the
superheat,  reheat,  economizer and  air  preheater  surfaces  in the  boiler  and back
end of the  plant will determine where,  and  in what form,  these elements will  be
collected or emitted.   With the exception of mercury, it is expected  that trace
metals will  form  a  stable oxide under the conditions  normally  expected in either
PC  or fluidized  bed  boilers.   The  relatively high  vapor pressure of elemental
mercury presents  a different situation,  which is addressed  below.

The amount  of  each  metal  that is   emitted  from  the  power  plant  stack will  be
determined predominantly by whether its oxide remains a solid during combustion or
volatilizes.   Examination of  the boiling points and vapor  pressures of the trace
metal  oxides may be  helpful  in identifying  whether volatilization  is likely to
occur.
                                        25-3

-------
Metal oxides that  remain  a solid at combustion  temperatures  can be assumed to be
uniformly  distributed  throughout  the  various  ash  streams,   i.e.,   bottom  ash,
economizer/air heater hopper ash, fabric filter/ESP hopper ash and the ash emitted
with  the  stack  gases.   The  emission  of  the  trace  metal in  the stack  gas  can
therefore be  estimated  simply  as  a function  of the percentage  of  the  total  ash
entering the fabric  filter/ESP  and  the collection efficiency of the fabric filter
or  ESP  (99.9  percent plus).   Since there  is  no enrichment of the  trace  metal  in
the  fine  particulate   fraction,  the   different  collection  characteristics  of  a
fabric filter versus  an ESP  should  not influence the emission of the trace metal.

Some  trace  metals,  e.g.,  lead  and  arsenic,  form  oxides  that  volatilize  during
combustion and  subsequently  condense  as  the  combustion  gases  are  cooled  by  the
downstream heat  transfer surfaces ahead of the particulate collection system.   The
condensing vapor tends to form new submicron  particles or deposit on the  very fine
ash  particles  in  the   flue  gas, thus  significantly  enriching  the  smaller  size
fraction  of  the ash  with  the  trace  metal.   The  volatilization also results  in
concentrating  most of the  trace metal  in  the  fly ash  fraction  entering the parti-
culate collection  system,  rather than  it being uniformly  distributed throughout
all the ash streams.  Enrichment factors  of  10  to  1  or more in  the  ash  particle
size  range  0.1  to  1.0  microns  have been  reported for  lead,  arsenic, and  other
elements  (Ref 1).   The  enrichment  process significantly increases  the difficulty
of predicting  trace metal  emissions  since,  at  this  time, there  are inadequate data
on  which  to  base  estimates  of  enrichment factors  and  therefore  emission  rates.
The  emission  rate will   also  be somewhat  influenced  by  the  type of  particulate
collection device  being  used, since  ESPs  tend to emit  a higher  percentage of  the
very  fine  ash  size  fraction  than  do  fabric  filters, as discussed  later  in  this
paper

The following  is a general  discussion  of the  expected  behavior of some of the most
common trace metals (beryllium,  lead,  mercury,  arsenic,  and chromium).

1.   Beryllium
Beryllium metal  oxidizes in  the combustor/furnace  to form  beryllium  oxide (BeO).
It  may  also be  present as the mineral "beryl"  (beryllium silicate).   Both  the
oxide and  the  mineral  are  refractories and  do  not vaporize except  at  very high
temperatures (b.p.  7052°F).   Therefore, any beryllium present in  the coal  would be
                                       25-4

-------
expected to  be uniformly  distributed  throughout all the  ash  streams  and removed
from the flue gas at the average efficiency of the fabric filter or ESP

2.   Lead
Lead is  normally  present  in coal as lead sulfide (PbS).   Lead sulfide oxidizes in
the combustor/furnace to form lead oxide (PbO).   Lead oxide has a boiling point of
2795°F  (Refs  2 &  3)  indicating  that only  a  small   percentage  of the  lead  oxide
should be volatilized in a PC boiler and essentially none in a CFB.

However, the presence of chlorine in the combustion  process, forming lead chloride
(PbCl), has been  shown  to greatly increase the  volitization of lead (Refs 2  & 3).

Stone & Webster has been involved in coal burn test  at a  fluidized bed pilot  plant
in which the  distribution  and  emission of  lead were determined.   A  significant
amount of lead was  present in the clean flue gas  (after the baghouse) indicating
significant enrichment of  lead  in the  fine particulate.   Until  much more informa-
tion  is  available  on  the  emission of  lead from   power  plants,   a  conservative
approach should be followed.   Initial estimates  of 10 to  20 percent  emitted in the
stack gas appear  to be  reasonable.

3.   Mercury
Mercury is  the most  volatile of all  the metals.   It is  found in coal  principly as
mercury sulfide (HgS),  cinnabar.   Upon  heating,  the HgS  is converted to elemental
mercury (Hg).   A  small  percentage of the Hg may  be oxidized, condensed in the back
end  of the plant,  and  removed  with  the  fly  ash; however,  because  of  its  low
boiling point  (674°F),  it  is expected that  80 to  100 percent of the  mercury in the
coal  may be emitted  as Hg vapor   Tests at several power  plants in  Europe  indi-
cated emission of 25 to 100 percent of  the  Hg in  the coal.   For initial estimates,
80 to 100 percent is  recommended.

4.   Arsenic
Arsenic is  found  in nature in its elemental form  and combined in a large number of
compounds.   At  combustion  temperatures,   arsenic,  regardless  of  form,  rapidly
volatilizes  and  is  oxidized  to  arsenic oxide  (As203).   The As203  vapor is con-
densed in the  back  end of the boiler.    Like  lead,  the  arsenic is concentrated in
the fine particulate  fraction.    Because  of its  high  volatility,  the enrichment
                                        25-5

-------
factor for arsenic  is  high.   Conservative estimates of arsenic emissions would be
10 to 20 percent of the arsenic in the fuel.

5.    Chromium
Chromium is  present in coal primarily  as chromite (FeCr204 and  Cr203).   Both of
these forms  remain  solid  at very high temperatures.   Chromite is used as a common
refractory and  the  boiling point  of Cr203  is  7232°F,  with  a melting  point of
4415°F.   Therefore,  chromium  is not  expected to  volatilize  in  any significant
amount and should  be  evenly distributed throughout the mass of ash in the boiler.
Like beryllium,  the  emissions  of chromium will be a function of the percentage of
ash entering the particulate collection system and the average efficiency of that
system.

THE TYPE AND  CHARACTERISTICS OF THE  PARTICULATE COLLECTION SYSTEM
The most common particulate  collection  devices in use today are the electrostatic
precipitator  (ESP)  and the  fabric  filter (baghouse).   The fabric  filter  can be
characterized as a  constant emission device with a varying efficiency, whereas the
ESP  is  a  constant efficiency  device  with  a  varying emission.   Both  of  these
devices  have  the  capability of  operating in  the  99.9+ percent  collection  range.
Either a fabric  filter or  an ESP is  highly effective in controlling the sub-micron
prticulate,  although  it  appears that  the  filter is  slightly more  effective.

Stenby reported that  the  ESP has a  reduction in efficiency in the 0.1 to 2 micron
range while  the fabric filter  has  a reduced fractional  efficiency  in  the  0.1 to
0.5  micron  range  (Ref 4).   The  ESP fractional efficiency  is  about  95 percent at
0.1 micron and  at  2 microns.  The  fabric filter,  on the other  hand,  operates at
above 99.5 percent  in  this  range.   Other sources  have obtained similar results.

The  ESP tends to  act as a classifier in  that the larger particles are removed in
the  inlet  section  and  the  average   particle  size diminishes  as  the gas proceeds
through the unit.   The other phenomena that occurs in the ESP is the agglomeration
of  some of the  fine particles  entering the  unit  and their removal  in  the  ESP at
the  larger agglomerated size.   This is  the probable  explanation  of the reduction
in  efficiency   in  the  0.2 micron  range  and  then  an increase  in  efficiency on
particles smaller than 0.2 micron.

Since the fabric filter functions with all components in parallel with each other,
it  does not  exhibit this  classifying characteristic.  There  can  be variations in
                                        25-6

-------
particle size  distribution  due to inlet duct take-offs into the individual filter
compartments.

The  Electric   Power  Research  Institute  (EPRI)  has  compiled a  large  amount  of
information on trace  elements  in  fly  ash (Ref 5).   Data  are available  on the
particle size  distribution  of  ash  being emitted to  the  atmosphere.   Little work
has  been  performed to correlate  fly  ash trace element concentration  with a size
distribution.   We  do  know  that some of the trace metals such as arsenic and lead,
show enrichment  factors  of  10 or more  in  the ash particle  size  range of 0.1 to
1 micron,  as noted earlier (Ref 1).

Further, research is  needed in the area of the chemical and physical properties of
particulate matter discharged from  power  plant  stacks before   more  objective
quantification or  prediction  of metals  emissions can  be  done.   Stack  test filter
samples can  be used  to  determine the  particle size  and trace element chemical
analysis of the particulate  matter.   Samples can be prepared  by making  dispersions
of the  filter  load using  an aqueous or  non-aqueous  wetting  agent  and  mild ultra-
sonics.  These dispersions  can be filtered onto 0.2 micron polycarbonate membranes
and  photographed  in  a Scanning Electron Microscope  at  calibrated magnifications
for  subsequent  computer-aided  size   analysis.   An   electron  microscopic  X-ray
analysis can  be  performed  on  each  filter to  determine the elemental  chemical
composition (e.g., trace  elements such  as arsenic though zinc).

Computer software  can  be employed  to  reduce  the  data from  such  an analysis and
produce numerous graphical  presentations of particulate matter equivalent physical
spherical  diameter versus  elemental  chemical  composition, expressed  on a weight
basis.   Data  are  needed  from  coal burning  units  with ESPs and  fabric filters in
order  to  quantify any differences  between  the  chemical/physical properties  of
particulate matter for these different  devices.

CONCLUSIONS
Trace metal emissions  from  fossil-fired power plants have recently been subjected
to  increasing  regulatory  scrutiny.   However,  there  are presently   insufficient
quantitative data  on the  sources  and  behavior of trace metals in power boilers to
accurately  predict the levels of these  emissions.

All  participants  in the plant permitting process need to develop  a better under-
standing of the  factors  that  contribute  to,  or influence, the  emission of trace
                                        25-7

-------
metals from power plants and to recognize the lack of high quality data underlying
such analyses.

The  physical   and  chemical  behavior of  the  trace  metals  under  the  conditions
present in the boiler can be used to help predict emissions.   Boiling point, vapor
pressure,   and   other  characteristics  can  provide  a  general  indication  of  the
potential  level of emissions.

ESPs are expected to emit slightly greater amounts of the volatile metals that are
concentrated in  the  submicron  particulate portion than  fabric  filters.   However,
this difference is not substantial.

Trace metal  emissions should not be  a  problem when permitting  new power plants.
Conservative estimates  of  metal  emissions  are expected  to  result  in  predicted
ground-level concentrations that are well  below allowable limits.

-------
REFERENCES

1.    Bjegman  & Els,  "Thoughts on  Recovery and Environmental  Implications  of
     Trace  Elements  in  South  African  Coal  Combustion Residues".   Presented  at
     Ash-Valuable  Resources,  Pretoria Republic of South Africa,  February 2-6,
     1987.

2.    Niessen,  W.R.,  "Air  Emissions  from  Thermal  Processing of  Wastewater
     Sludges,  Conflicts  in  Priorities",  Proceedings of the  National  Confer-
     ence   on  Municipal  Sewage  Treatment  Plant  Sludge  Management,  Boston,
     Massachusetts,  May 27-29, 1987.

3.    Gerstle,  R. ,  and  Carvitti,  J. ,  "Sludge Incineration and Hazardous Air
     Pollutants",  Proceedings of the National  Conference  on  Municipal  Sewage
     Treatment Plant  Sludge  Management,   Boston,  Massachusetts,  May 27-29,
     1987.

4.    Stenby,   Schrech,  Severson,  Herney  and  Teixeria,   "Fabric  Filters  vs
     Electrostatic  Precipitators"  presented  at  Second   Symposium  on the
     Transfer and  Utilization of  Particulate  Control  Technology",  Denver,
     Colorado,  July  23-27,  1979.

5.    EPRI-EA3236    Physical-Chemical  Characteristics of Utility Solid  Wastes,
     date September  1983.
                                        25-9

-------
             THE  STRUCTURAL ANALYSIS OF  DUST CAKES

                             E.  Schmidt
                             F.  Loffler
                     Universitat Karlsruhe (TH)
       Institut  fur  Mechanische  Verfahrenstechnik und Mechanik
            Postfach 6980,  D-7500  Karlsruhe, West Germany
ABSTRACT

Despite their excellent collection efficiencies,  however,  intensive
research is still in progress to improve the operational character-
istics of periodically cleanable cake forming surface filters.  One
of the paths in the direction of this goal is the attempt  to de-
scribe the influence of the cake structure on the filter perform-
ance .

The initial section of this paper is therefore concerning the de-
scription of the cake's preparation. This includes the two different
stages of the process necessary in order to consolidate the fragile
dust layers and the subsequent treatment of the stabilized sample,
up to  the micrographic imaging of specific cross-sections.  The  sec-
ond section is concerning the analysis of the cake structure, and is
focussed on the digitization of the micrographs,  together  with  the
evaluation of the porosity and the particle and pore size  distribu-
tions  by quantitative image analysis. Finally a number of  results
derived from a true filter cake are presented.
                                 26-1

-------
             THE STRUCTURAL ANALYSIS OF DUST  CAKES

INTRODUCTION

Cleanable surface filters,  in particular bag filters using nonwoven
fabrics as filter media,  are of great importance for dust separa-
tion. In spite of the high collection efficiency which they normally
achieve, intensive work is being carried out to improve the perform-
ance of this type of filter. A step in this direction is the attempt
to understand the influence of the dust cake structure on capture
efficiency, pressure drop and regeneration or filter cleaning. Each
of these depends on the specific conditions of filtration. Due to
its fragile texture, microscopic examination of the structure of the
dust cake is relatively difficult.  If electron microscopy is used,
the energy and the charge of the electrons suffice to eject parti-
cles from the dust cake and thus change its structure. Generally, it
is impossible to expose a cross-section of the dust cake, in order
to reveal its inner structure, without disturbing the structure
itself. Therefore a preparation technique for fixation of the dust
cake is required.

One possibility is the preservation of dust cakes by embedding them
in synthetic resin. Felix and Smith  1| have developed such a tech-
nique and have applied it to comparatively strongly bound residual
dust layers. This procedure was developed further by Morris  2] who
was able to use it to obtain a qualitative description of the dust
cake structure.

Here we describe a preparation technique in which the cake is em-
bedded in a curing resin too  |3|. But the preparation technique ena-
bles a cross-section to be exposed, from which quantitative evalu-
able scanning electron micrographs can be obtained. The basic system
for digitizing such photographs is relatively elementary, and hence
inexpensive. It incorporates a video digitizer,  a micro computer and
                                 26-2

-------
essentially self-developed software. With it, the photographs, which
may require manual treatment, are digitized, checked  (if necessary
corrected), and stored as black-and-white images.

The dust cake structure is defined according to the local porosities
and the independently determinable local particle and pore size dis-
tributions. The applied theories and methods of approach used to at-
tain the data are comprehensively described. The actual application
of the discussed technique to the structural analysis of a limestone
dust cake, formed on a laboratory filter disc, is finally presented.

PREPARATION OF  THE  DUST  CAKES

Before sections from a dust layer can be cut, the cake must possess
an adequate stability. This can be attained by embedding it in a
curable resin.  With this process, the dust cake which was filtered
onto a specific medium should be enveloped by the chosen substance
as comprehensively as possible without destroying its structural
configuration.

In order to ensure that the particles are not displaced as a result
of the capillary forces which arise whilst the embedding agent seeps
into the structure,  the cake must be suitably pre-stabilized.  This
is accomplished by means of a cyano-acrylate based single-component
reactive adhesive vapour,  which is mixed with dry nitrogen and gen-
tly forced through the filtered cake. The corresponding experimental
arrangement is  shown in Fig.  1. If the adhesive vapour mixture is
kept dry but the cake moist,  than at contact, a thin adhesive film
will form around the particles and the fibres, strengthening the
points of contact within the whole dust cake/filter medium complex.

Upon accomplishing this,  the stabilized cake can then be wetted by
an extremely low viscosity embedding substance.  One suitable materi-
al is  a mixture of four components:  Epoxy resin,  plasticizer,  hard-
ener and accelerator I 4| .  In order to embed the pre-stabilized spec-
imen,  this should be laid upon an approximately 1 cm thick highly
porous fibrous  layer within a suitable mould. Droplets of the syn-
thetic resin compound are  then added to the mould's edge,  gradually
                                 26-3

-------
filling it with the liquid, as shown in Fig. 2. As a result of the
capillary forces, the dust cake/filter medium  complex will become
thoroughly saturated. Finally, the specimen is cured in a drying
oven at 75 °C for 24 hours.

Since the cured block is too large for the subsequent grinding and
polishing process, smaller cubic sections with an edge length of
approx.  1 cm are cut. These cubes are then again embedded in a two-
component resin/ceramic filler mixture in a cylindrical synthetic-
rubber mould. Units are available with which the grinding and pol-
ishing may be facilitated by mounting five such cylindrical speci-
mens in a corresponding holder for simultaneous treatment.

The initial process involves a three-stage grinding of the specimen
face with silicon carbide grinding discs, for which a finer grade of
abrasive is used at each stage. The grade increments and grinding
parameters are to be chosen with respect to the prevailing condi-
tions, so that the grinding traces of the previous stage are com-
pletely erased by the next. When ground,  the specimen must then be
polished. Since the dust cake structure analysis demands an exten-
sively smear-free surface, a diamond polishing compound in three
different grades has proven most effective.

When viewing such polished specimens in a scanning electron micro-
scope (SEM)  the actual cake structure will still not be revealed
since the SEM chiefly resolves the local superficial relief.  The fi-
nal stage of the preparation therefore involves dissolving the lime-
stone particles with a suitable etching agent.  Since this process
does not attack the resin of the block,  the cavities where the indi-
vidual particles were once situated become clearly visible. The
specimens prepared in this way now allow images to be made in the
SEM upon which the quantitative analysis of the dust cake sections
is based.

DIGITIZATION  OF  THE  CROSS-SECTIONAL  PHOTOGRAPHS

The quantitative analysis of such cross-sectional specimen images is
not feasible without electronic data processing. Hence, the first
                                 26-4

-------
step is to read the images into a micro-computer  (Macintosh™ SE)
with the aid of a video digitizer  (Magic Eye™ System with serial
interface and video camera).  The software developed at the author's
institute additionally allows a program-controlled video image in-
put .

Before the images can be processed, however, a so-called light level
must be defined. Image locations which are darker than this value
will appear black on the computer's monitor  (dust particles i.e.
cavities), the other image locations hence being displayed as white
regions (pores i.e. resin matrix).  Care must therefore be taken to
ensure that the SEM photographs possess an adequate contrast. Should
the SEM conditions or photographic limits not provide the contrast
required,  then a subsequent manual treatment of the photographs is
essential. The complete video image of the specimen section photo-
graph  (which must be illuminated as uniformly as possible)  possesses
a maximal size of 720 x 512 pixels. From this, a sub-section of 512
x 342 pixels is read into the computer, hence masking any peripheral
camera aberrations. If the black-and-white image stored by the com-
puter should not quite comply with the master, the developed soft-
ware allows corresponding corrections to be easily conducted direct-
ly on the monitor. Fig. 3 illustrates such a digitized image taken
from a photograph of a limestone dust cake section. The black areas
represent the cut limestone grains whilst the white background de-
picts the surrounding pore void.

EVALUATION OF  THE  DIGITIZED  IMAGES

The  structure  of  solid,  porous  systems

A solid porous system is defined as a solid phase which is more or
less homogeneously interspersed with macroscopic pores. The basic
structure (matrix)  consists of a solid phase which is either uniform
or composed of individual elements. The hollow spaces may either be
isolated inclusions or combined pores. Hence, one differentiates
between the total pore volume and the combined or effective pore
volume.
                                 26-5

-------
The list of parameters which serve to define the structure of such
solid porous systems incorporates:  the porosity, the specific sur-
face area,  the capillary pressure function, the pore size distribu-
tion and pore shape, the particle size distribution and particle
shape, the electrical resistance of the system upon being saturated
with an ionic fluid, the homogeneity, the isotropy, et cetera. The
choice of the most suitable parameter depends on the predominant
conditions of each respective application.

A filtered dust cake is composed of irregularly shaped particles of
different sizes,  which comprise a discrete disperse, solid porous
system with interconnected pores. As previously stated, one of the
methods with which such dust cake structure-defining data can be ac-
quired, involves  the evaluation of specially prepared cake sections.
The actual data can be specifically derived from three-dimensional
specimens by analyzing enlarged two-dimensional sectional photo-
graphs and applying mathematical models and statistical concepts.
This method is the subject of the following sections.

Porosity

The  (volumetric)  porosity of a solid porous system is defined as the
ratio of void to  the total volume.  If a certain degree of homogenei-
ty can be assumed, the quantification of this parameter character-
izes, to some extent, such a system's structure. If sections are
randomly cut from a porous specimen, then in analogy to the volumet-
ric porosity, one can also determine from the plane areas the so-
called areal porosity, defined as the ratio of void area to the
total area. It can be verified |5  that for solid particle packings
(e.g. a dust cake) the volumetric porosity is usually equal to the
areal porosity, which is easily determined from the digitized image
as the ratio of the white pixels (pore space)  to the total number of
white and black pixels (pore space and particles) .

At this point, the question of the size of each respective image
section becomes relevant. The section must obviously be large enough
to restrain any influence which might be caused by local inhomoge-
neities as a result of the particle and pore distribution widths. On
                                 26-6

-------
the other hand, however, the image section must be adequately con-
fined in order to detect any significant porosity changes, e.g.
across the dust cake height. This optimization problem is graphical-
ly illustrated in Fig. 4 by the areal porosity 6,  plotted as a func-
tion of the image section area A.

Relevant measured porosity values will only then emerge when the
area A is selected between Amin and Amax. The enlargement  ratio from
the actual section to the image can then be correspondingly adapted
to the requirements of the utilized digitization equipment. Although
it is obviously a matter for a separate discussion, it might never-
theless be worth noting that the parameters Amin and  Amax  may also
serve to describe the dust cake homogeneity.

Particle  size  distribution

The mathematical description of a three-dimensional space with just
two-dimensional sections through it is the subject to which quanti-
tative stereology is devoted  |6|. A number of different methods ex-
ist with which stereological problems can be approached.  The statis-
tical-geometrical method stems from the measurement and classifica-
tion of a large number of two-dimensional objects  within a single
plane. This technique proves of advantage when the individual ob-
jects, may they be pores or particles, happen to be randomly dis-
tributed across the cut plane. In such cases,  the  investigations
conducted on one single cross-section will, provided a statistically
adequate quantity of elements are contained, yield information which
can be applied to the complete three-dimensional system.

The various stereological methods of determining the particle size
distributions of spherical objects constituting a  discrete disperse
porous system,  may be sub-divided into two categories: section ana-
lyses and chord analyses. The former involves determining the dis-
tributions of particular features of cut objects on the plane,
whereby the size of a cut sphere (i.e. circle)  on  the section in
question may either be characterized by its diameter or its area.
The chord analyses is based on chords or sections  created by ficti-
tious lines randomly drawn through the system |7|, which  in practice
                                 26-7

-------
may also be applied to cut sections. Fig. 5 illustrates the  differ-
ences of each approach when applied to the stereometric determina-
tion of the size distributions of spherical objects.

The chord analyses also delivers good results for particles  which
slightly deviate from a spherical shape. Moreover it can,  despite
certain confinements, also be employed when the particles  are
linked. As such, this method was applied to the given problem of
dust cake analysis. Since a comprehensive description already exists
I 9, 10|, the procedure is only to be briefly reviewed.

The procedure variant applied involves drawing a number of randomly
intersecting lines of cumulative length L over the cross-section
concerned. The object cross-sections (particles or pores)  which hap-
pen to be cut by the lines will be sub-divided into a number of seg-
ments possessing different chord lengths 1. Hence, the number dis-
tribution density q0 s(l)  of  the  different  chord lengths can then be
determined. A multiplication with the quotient derived from  the to-
tal number of chords ms and the cumulative length L according to eq.
(1) will deliver the function ns(l). The product ns(l)-dl specifies
the number of chords which are between 1 and 1+dl long, per  unit
length.
                                m.
                        n-(1) =7- ' Vs'1'                       (!)

Upon confining this approach to an adequately large number of spher-
ical and completely randomly distributed particles, then the func-
tion nR(x)  may be calculated by eq.  (2). The product nK(x)-dx then
specifies the number of spheres which possess a diameter between x
and x+dx per unit volume.
                                         -l    dna(l)
                                nQ(l) - 1
                                                dl
                                                      l=x          (2)
The implementation of this equation assumes that ns(l) is a continu-
ous, dif ferentiable function. This can, however, easily be attained
by an adaption with so-called cubic splines. The number distribution
%,k(x)  of  the  sphere  diameters  can then  be calculated with the aid
of eq.   (3)  .

-------
                                 nk(x)
                                 nk(x)  dx
                               ^nii,                                 (3)

 If one wishes to convert  from the  quantifying  parameter  "number"
 (index 0) to another  (e.g.  "area"  (index 2)  or "mass"  (index  3)),
 then this may be conducted  using eq.  (4)

                        x^-q.U)
              ^   =  -^,	
                       f *--!
                      r
q.(x)  dx
                                         i,  i e  {0,  1,  2,  3}       (4)
According to definition, this method  of  determining  the  distribu-
tions qj_(x)  from two-dimensional cross-sectional images,  is actually
only valid for spherical objects.  The  relevance of the  information
nevertheless gained from non-spherical entities must  therefore  be
clarified. The results delivered are, by all means,  a well defined
number of various spheres with  a specific  diameter distribution,
possessing both the same chord  length distribution and the same vol-
umetric share as the investigated  irregularly-shaped  objects .

If the peripheries of the cut objects within the plane  (in this case
particles or pores) happen to contain many edges, indentations  or
other irregularities, then the  chord analyses will deliver a higher
proportion of smaller chords, which effectively shift the size  dis-
tribution towards smaller values.  An implementation of an average
particle and pore shape-describing function into the  above equation
will,  however,  comprehensively  eliminate the inaccuracies  |11|.
Whether the additional information which this laborious  correction
delivers is worth the effort, must be decided for each specific ap-
plication in question.

Pore  size  distribution

A representative description of the pore structure within a discrete
disperse,  porous system requires the definition of a  geometrical pa-
                                  26-9

-------
rameter which plainly characterizes the pore size. Nobody, however,
has yet derived a generally valid and practically applicable axiom.
Terms such as "pore diameter" or "pore size" are generally merely
intuitive simplifications of the real circumstances. All algorithms
applied in view of determining the pore size distributions will
hence,  only define the pore size with respect to some pore model
system which is generally individually adapted to a single specific
medium.

One such model describes a discrete disperse porous system as being
a random packing of solid and hollow particles,  i.e. the merged pore
volume network is substituted for adjoining fictional hollow parti-
cles I 12| .  This model reduces the problem of the pore size defini-
tion to that of the particle size definition, which hence, allows
the application of the chord analyses to derive  the pore size dis-
tributions. The inherent error which arises from the fact that the
hollow particles are actually merged and not in  point-contact with
each other is negligible small  6|.  The influence of the pore shape
has already been discussed.

PROCESS APPLICATION

This section presents the results gained from the analysis of a true
filter cake with the described technique. The filter medium was a
flat 14 cm diameter disc of calendered polyester needled felt. The
dust used was limestone  (X^Q 3 =  3.8  |0.m determined by  sedimentation)
which was filtered onto the disc at a face velocity of 165 m/h up to
a pressure drop of 1500 Pa. Fig.  6 schematically demonstrates a pre-
pared section of the approx. 300 |im thick cake.  The location of the
photographed elements are depicted by rectangles. Eleven photographs
were respectively evaluated from an upper and lower cake plane
 (200 |im - 240 |im and 40 |im - 80 M-m from the filter medium) .

From the obtained data, the 95 % confidence intervals included in
Figs. 7 and 8 were derived. Furthermore,  a succession of 7 photo-
graphs across the cake height was evaluated, from which the follow-
ing results emerged: The areal porosities, averaged for each photo-
graph,  have been plotted in Fig.  7 from which the total porosity
                                 26-10

-------
change across the cake height becomes evident. Hence, one can clear-
ly observe that the cake porosity increases from approx . 80 % at the
vicinity of the filter medium to 89 % at the upper cake regions. The
source of this might be traced back to local cake compression,
caused by an increase in the pressure-difference during filtration.

Some selected parameters of the particle distributions, determined
across the cake height are displayed in the left diagram of Fig. 8.
In order to characterize the respective areal distribution  (q2 (x) ) ,
the x50 2 median values  and the  distribution parameters  XIQ 2 and
x90 2  have been included.  The  distribution  parameters are  defined by
eq. (5) .
                              ,
                               qr (x) dx = n %
                                                                  (5)
One can observe that as expected, the particle size does not signif-
icantly change across the cake height. The right diagram of Fig. 8,
however, clearly illustrates that the pore size  (in particular the
larger pores)  increases towards the cake surface. The lower porosity
values in the lower cake regions therefore originate from the corre-
spondingly smaller quantity of larger pores.

CONCLUSIONS

The described technique yields quantitative information concerning
the internal structure of cakes formed by dust filtration. Specific
examples include the locally-determinable porosities and the parti-
cle and pore size distributions.  The task which now arises is to use
the acquired structural data to either verify, supplement or invali-
date existing mathematical models which describe the permeability,
the cleaning behaviour and the collection efficiency, etc. or to de-
rive new concepts.
                                 26-11

-------
REFERENCES

 1  L.  G.  Felix,  W.  B.  Smith.  "Preservation of Fabric Filter Dust
    Cake Samples." J.  Air Pollut.  Control Assoc.f  33 (1983), pp.
    1092-1094.
 2  B.  A.  Morris.  "The  Pressure Drop Across Fibrous Filter Media in
    the Presence  of Electric Fields." Ph.D. Dissertation Princeton
    University  (1985).
 3  E.  Schmidt,  F. Loffler.  "Preparation of Dust Cakes for Micro-
    scopic Examination." Powder Technology, 60 (1990) 2, pp. 173-
    177 .
 4  A.  R.  Spurr.  "A Low-Viscosity-Epoxy Resin Embedding Medium for
    Electron Microscopy." J. Ultrastructure Research, 26 (1969),
    pp. 31-43.
 5  M.  Delesse.  "Pour  determiner la composition des roches." Annales
    des mines,  13  (1848), pp.  379-388.
 6  F.  A.  L.  Dullien.  Porous Media, Fluid Transport and Pore
    Structure. Academic Press,  1979.
 7  A.  Rosiwal.  "Uber  geometrische Gesteinsanalysen.  Ein einfacher
    Weg zur ziffermassigen Feststellung des Quantitatsverhaltnisses
    der Mineralbestandtheile gemengter  Gesteine."  Verhandlungen d.
    k.  u.k. geolog.  Reichsanstalt.  5 u.  6 (1898),  pp. 143-175.
 8  E.  E.  Underwood. Quantitative  Stereology.  Addison-Wesley
    Publishing Company,  1970.
 9  J.  W.  Cahn,  R. L.  Fullman:  "On the  Use of Lineal Analysis for
    Obtaining Particle  Size  Distribution Functions in Opaque
    Samples." Journal  of Metals.  (1956),  pp.  610-612.
10  G.  Bockstiegel.  "Eine einfache Formel zur Berechnung raumlicher
    Grofienverteilungen  aus durch Linearanalyse erhaltenen Daten."
    Zeitschrift  fur Metallkunde,  57 (1966), pp. 647-652.
11  F.  A.  L.  Dullien,  G. K.  Dhawan. "Characterization of Pore Struc-
    ture by a Combination of Quantitative Photomicrography and Mer-
    cury Porosimetry."  J. of Colloid and Interface Sci.f 47   (1974),
    pp. 337-349.
12  S.  Debbas,  H.  Rumpf. "On the randomness of beds packed with
    spheres or  irregular shaped particles." Chem.  Eng Sci. ,  21
    (1966) , pp.  583-607 .
                                 26-12

-------
    nitrogen
                   air
Figure 1.  Experimental equipment for prehardening procedure
              embedding agent
                        dust cake
                           filter medium
                              fiber layer
1                -   - f '-' ' ' -''   Jl   L   'J.
              »\ S   S  S  S  fi  S  / \
                  '"'*'*'
 Figure  2. Experimental equipment for embedding procedure
                  V
' *•<
                           v *
   Figure 3. Digitized image of a dust cake photograph,
                         26-13

-------
                                  image section area  A
                     Amin   Amax
Figure  4. Respective porosity as a function of the selected area
          diameters
                            areas
        0
                 e    o    o
           Figure  5. Different methods of evaluating

                  a cross-sectional image I 8|
   CD  CD
   CD  CD
        s s s /
 300 (im


CD  CD  CD
CZ3  CD  CD


 0 |J.m
                    CD  CD  CD  CD
                                         dust cake
                    CD  CD  CD   CD
                  s\\\
                 \ \ \ \ .
                s s s s s
                           10 mm
                                         filter medium
     Figure 6.  Locations of the photographed and evaluated


            elements within the cake cross-section
                             26-14

-------
     g
     -p
     XI
     in
     -H
     0)
     id
     o
          240
160


120


 80


 40


  0
                                            i
                  78     80     82    84     86

                            porosity /  %
                                               90
      Figure 7. Porosity  as  a  function of the cake height
o A n
• » A m-t
o A n
• o A n

O A D
® AH n D
o A n
I . 1 . I . I .
^ou
X4 U
zUU
g
^ 160
-P
XI
Cn
•H IzO
 A B-. A x5Q^2 .
— 	 	 	 Ox
« I 10, 2J
o A n v J
i.i.i
      12345

     particle  size / (J.m
                                  10     20      30


                                  pore size  /  |J.m
Figure 8. Particle  and pore size trends across the  cake  height
                               26-15

-------
                  EFFECTS OF ADDITIVES AND CONDITIONING AGENTS
                     ON THE FILTRATION PROPERTIES OF FLY ASH
                                  P.  Vann Bush
                                 Todd R.  Snyder
                           Southern Research Institute
                             2000 Ninth Avenue South
                                 P.O. Box 55305
                            Birmingham, AL 35255-5305
ABSTRACT
Fly ash characteristics strongly influence filtration performance.   More and more
variations  in these characteristics have been observed as fabric filters have been
used to collect fly ash from alternate combustion methods,  sorbent  injection
processes,  and various coal  types.   Field surveys and laboratory studies funded by
the Electric Power Research  Institute have identified many  key characteristics of
fly ash particles that influence filtration performance.   The U.S.  Department of
Energy is  extending this work to investigate means of enhancing the filtration
performance of fly ashes with conditioning agents.  This  investigation includes a
survey of  scientific literature on  related subjects,  laboratory screening tests to
identify potential  conditioning materials, and slipstream filtration tests to
evaluate the effectiveness of specific agents.  Conditioning agents evaluated
include dry powders,  moisture content of the filtered gas,  and agents that can
chemically  react with the flue gas  stream.  Results indicate that the addition of
relatively  small  amounts of  conditioning agent can significantly enhance filtra-
tion performance.
                                       27-1

-------
                  EFFECTS OF ADDITIVES AND CONDITIONING AGENTS
                     ON THE FILTRATION PROPERTIES OF FLY ASH
INTRODUCTION
Although many fabric filters operate efficiently with reasonable pressure drops,
the collection of certain types of ash can generate poor collection efficiencies,
or excessive pressure drops, or both.   Because of the increasingly broad applica-
tion of fabric filters to flue gas cleanup at coal-fired power plants, and the
potential for application to new combustion processes,  it is important to insure
that fabric filter technology is optimized to provide the most effective particu-
late collection at the lowest operating cost.  Flue gas conditioning has demon-
strated the potential to enhance baghouse performance.

Tests of fly ash collected at the Electric Power Research Institute (EPRI) High-
Sulfur Fabric Filter Pilot Plant have shown that the inherent flow resistance of
the dustcake ash is lowered by the passage of the filtering compartment through
dew points (1).  This is a kind of natural flue gas conditioning of the ash.

Ammonia  injection has been shown in laboratory and full-scale trials to cause a
reduction in the flow resistance of the dustcake and the penetration of the ash
through  the dustcake and fabric.  EPRI sponsored testing beginning in 1982 at the
Monticello Station of TU Electric to evaluate ammonia injection to increase ash
cohesivity in order to reduce ash bleedthrough.  The tests showed a more than 10-
fold reduction in emissions (3.2% to 0.2% penetration)  as a result of ammonia
injection (2).  Compartment-scale tests of ammonia injection have been completed,
and a full-scale demonstration program is under way at  Monticello.  Southern
Research Institute (SRI) has also tested ammonia injection at the EPRI High Sulfur
Fabric Filter Pilot Plant located at Gulf Power's Scholz Steam Plant and in side-
stream and compartment-scale tests at the Pennsylvania  Power & Light Brunner
Island Station.  In all cases, the flow resistance of the ash has been reduced as
a result of ammonia injection.  These effects of ammonia injection have also been
confirmed at the University of North Dakota  in testing  sponsored by the U.S.
Department of Energy (3J .
                                       27-2

-------
Consistently, fabric filters collecting ash downstream of spray dryers have shown
lower flow resistance and cohesivity than might be expected from the relatively
high surface area of the dust.  Whether this performance is due to changes in the
flue gas environment, the addition of the sorbent particles, or a combination of
the two is not known.

All ashes do not perform equally in fabric filters, and we know of some materials
or processes that alter the filtration properties of ash.  Therefore, it may be
possible to modify the undesirable characteristics of certain ashes to improve
their filtration performance.  An understanding of the key properties of ash that
govern filtration and the mechanism(s) whereby these properties are modified or
overcome is required before effective conditioning can be implemented.

Two approaches to conditioning are illustrated in Figure 1.  One approach is to
add agents that modify the ash/fabric interface in order to optimize the filtering
surface.  The other is to add agents that modify fly ash properties in order to
increase the ash cohesivity.  Both types of conditioning can modify filtering flow
resistance, dustcake retention, and collection efficiency.

An optimized filtering surface would have a high density of relatively small  pores
that are easily bridged over. Their presence in relatively large numbers would
decrease the local air velocity through each pore.  As a supplement to fabric
design, precoat materials can be applied to the clean fabric surface to provide
the desired pore structure.  Characteristics of the ideal precoat material  include
the ability to bridge relatively large pores, and a tendency to produce a highly
porous structure with relatively low resistance to gas flow.

Increased ash cohesivity causes the particles to form a more porous dustcake
structure.   Small increases in the porosity of a dustcake greatly reduce resis-
tance to gas flow through the cake.  When ash particles are made stickier through
conditioning, any tendency of particles to break loose from the dustcake structure
and slip through the dustcake and fabric is reduced.  Ash cohesivity can be in-
creased by increasing the attractive forces between particles.  These forces
depend on the inherent characteristics of the ash particles, and the additional
effects introduced by conditioning processes or materials.
                                       27-3

-------
LITERATURE REVIEW

A review of technical  literature was undertaken to locate descriptions of the
mechanisms governing particle-to-particle interaction, and identify prospective
ash conditioning agents.   Searches of titles and abstracts from NTIS, Searchable
Physics Information Notices (SPIN),  and COMPENDEX PLUS (engineering and technical
literature) were made through the DIALOG Information Services.  More than 5600
abstracts were reviewed.   A total of 134 articles were selected from the ab-
stracts, received, and reviewed for information applicable to this project.

Many of the articles reviewed emphasize the importance of particle shape, particle
size, and moisture content in the determination of bulk powder cohesivity.  Parti-
cle shape and size influence the relative magnitude of the interparticle forces
determining bulk cohesivity.  These forces include electrostatic and van der Waals
forces, and bonds formed when adsorbed surface films create liquid or salt bridges
between particles.  Adsorbed surface films may increase particle adhesion by as
much as five-fold (4).

The generic importance of particle size and shape in governing powder cohesivity
was anticipated from the results of the study of dustcake ash properties conducted
for the Electric Power Research Institute (J>).   The existence of liquid or crys-
talline bridges between particles has been implied by the relationships seen in
fabric filter field data between acid and water dew points and dustcake proper-
ties, and is an important aspect of cohesivity to investigate.

The literature reviewed suggests there is an optimum mixing ratio of conditioning
agent to substrate on a weight-to-weight (w/w)  basis.  Though the majority of the
articles dealing with this subject did not specifically discuss fly ash, the
optimum w/w values cited for conditioning agents applied to other materials were
usually between 0.1 and 2.0%.  In general, the addition of particulate condi-
tioning agents in this proportion to the substrate particles tends to place asper-
ities on the surfaces of the substrate particles.

A basic question about the effects of conditioning agents is whether the predomi-
nant mechanism responsible for the change in cohesivity is a change in particle
shape or a change in the chemical bonding of particles.  It is assumed that the
predominant mechanism may differ depending on the conditioning agent.  Agents that
have been identified as glidants in the literature purportedly produce the desired
change in cohesivity by attachment to the particle surface of the treated
                                        27-4

-------
material, thereby altering the particle shape.  Studies of fly ash have shown that
increased ash surface roughness is related to increased ash cohesivity.  In addi-
tion, the exposure of dustcakes to acid and water dew points generally increases
the residual dustcake thickness, which is an indication of increased cohesivity.
Therefore, both mechanisms of modifying cohesivity are implicated in fabric filter
experience.

From the literature review we identified some prospective agents:  fine powders
(e.g., fine silica, lime) and particle 'wetting' agents (e.g., ammonium bisulfate,
water vapor).  Some categories of chemically active compounds were suggested by
Dr. E. B.  Dismukes as potential agents:   amine sulfates, surfacants/dispersants,
flocculating or coagulating agents, silanizing agents, and deliquescent com-
pounds.

LABORATORY SCREENING TESTS
Ammonia + Sulfur Trioxide
The relative importance of the physical surface of particles (size and shape) and
the chemical bonds formed between particles was the subject of the initial experi-
ments we conducted.  Experience at SRI and at the University of North Dakota
indicated that ammonia and sulfur trioxide can be used to increase ash cohesivity
(1)3).  An experiment was designed to investigate the mechanism by which the
reaction products of ammonia and sulfur trioxide modify fly ash cohesivity.  It
was not known whether the products simply altered ash particle shape by attachment
as asperities to the substrate particle surfaces, or the chemical composition or
phase of the products otherwise modified the bonding strength of the particles.

The prior work with ammonia as a flue gas additive has been predicated on pro-
ducing an ammonium sulfate reaction product with the sulfuric acid vapor  (or S03)
in the flue gas.  This was based on the properties of ammonium sulfate and the
other possible products.  The reaction to yield ammonium sulfate proceeds in two
steps:
               NH3 (g) + S03 (g) + H20  -->  NH4HS04 (l,s)
               NH4HS04 (l,s) + NH3  -->  (NH4)2S04 (s)

The first reaction produces the bisulfate that is a liquid above 293°F.   Fur-
thermore, the bisulfate is highly acidic, with a pH less than 3.  The bisulfate
will  tend to continue to react with available NH3 to form the stable ammonium
                                       27-5

-------
sulfate,  which is a nearly neutral  solid up to 437"F.   The reaction rates of
ammonia and sulfuric acid vapor are not known.

Figure 2 shows the effect of conditioning with ammonia and sulfur trioxide on the
characteristics of dustcakes comprising Monticello ash particles.  In this case
Monticello ash was exposed to S03  and water vapor in a mixing chamber held at
300°F and then exposed to ammonia gas with a NH3:S03 stoichiometry greater than
2:1.  The bulk porosity of the ash was measured to assess the change in ash
cohesivity.  Bulk porosity is a good indicator of cohesivity, and also provides an
estimate of the dustcake porosity that would be expected in a fabric filter
collecting the ash.  The uncompacted bulk porosity of the ash, which we have used
to approximate the dustcake porosity, increased from 66% to 74% as a result of the
ammonia reaction with S03 on the particles.  This change in porosity would
translate to a change in flow resistance of a factor of two.

We found that exposing ash to premixed ammonia+S03 did not result in a change in
porosity.  This result, along with results from other laboratory analyses and
field tests, and in light of the literature review, suggests that the mechanism by
which ammonia gas modifies the flow resistance of ashes is particle surface
wetting: creating liquid bridges where particles contact.  Presumably, the ammonia
gas reacts with S03 on the surfaces of particles to form ammonium bisulfate which
is a  sticky liquid at SOOT.  The reaction may continue to completion, forming
ammonium sulfate, after the particles have stuck together in the dustcake.

Fine  Powders
Tests were conducted mixing ashes with various fine powders.  The fine powders we
selected for evaluation were:

                                                                   Specific
	Powder	  	Source	  Surface Area  (m2/gl
EH5                  fumed silica, Cabot Corp.                        380
M5                   fumed silica, Cabot Corp.                        200
L90                  fumed silica, Cabot Corp.                        100
Gasifier Char        pilot-scale gasifier candle filter               280
Hydrated  Lime        Longview, Dravo Co.                               13
Sodium  Bicarbonate   grade 3 DF fine powdered, Church & Dwight          6.7

Figure 3  shows  results of uncompacted bulk porosity tests on  blends  of Monticello
ash and various w/w fractions of these powders.   In most cases,  the  fine  powders
act as  lubricants, reducing the ash  cohesivity as measured  by bulk porosity.   In
                                        27-6

-------
general, the higher the surface area of the additive the larger the change in
porosity.  The largest effect was seen in blends with Monticello ash, which has
the lowest inherent specific surface area of the ashes we tested.

Screening Tests Conclusions
Results from the screening tests provided the following conclusions:
     •    Exposing ash that has adsorbed sulfur trioxide to ammonia gas
          increases bulk ash cohesivity.  The mechanism for this effect is
          ostensibly liquid bridges between particles:  ammonium bisulfate,
          the intermediate reaction product, is a liquid that forms on the
          surfaces of particles, increasing the interparticle adhesive
          forces.
     •    Blending very small amounts of fine powders with ash generally de-
          creases bulk ash cohesivity.  They act as lubricants, or glidants,
          increasing the separation and decreasing the contact area between
          particles which reduces the van der Waals force of attraction.

The best approach for increasing fly ash cohesivity is to employ the liquid bridge
mechanism that increases interparticle adhesion.  Fine powders do not appear
practical for increasing the cohesivity of fly ash.  Powders may be useful  for
altering the filtering substrate; in fact, some powders that claim to improve
filtration are commercially available.

PROOF-OF-CONCEPT FILTRATION TESTS
Based on the findings from the literature review and screening tests, a series of
filtration tests were performed to evaluate candidate conditioning agents.   One
group of tests evaluated candidate agents for modifying the cohesivity of ashes.
Another group of tests evaluated candidate agents for use as precoat materials to
modify the filtering surface.

A Fabric Filter Sampling System (FFSS) was used in these tests.  The FFSS is a
small,  portable fabric filter developed by Southern Research Institute for EPRI to
sample a slipstream of gas and simulate the operation of a reverse-gas-cleaned
baghouse (2).   In our tests the FFSS sampled from a wind tunnel into which fly ash
had been redispersed.   Conditioning agents were either mixed with the ash before
the ash was injected into the wind tunnel, injected into the wind tunnel concur-
rently with the ash, injected into the FFSS inlet to condition the ash, or in-
jected  into the FFSS to coat the fabric filter before sampling the ash.
                                       27-7

-------
Analyses of Filtration Test Results

The standard way to compare filtering data is to normalize the pressure drop by
the filtering face velocity (or air-to-cloth ratio), yielding the quantity called
drag.   Drag is then presented as a function of the dust loading (or dustcake areal
density) on the filtering fabric throughout the filtration test.  Two characteris-
tics of this relationship between drag and areal density are shown on Figure 4, an
idealized data set.

One of the two principal characteristics of the drag-areal density relationship is
the specific drag coefficient, K2, which is defined as the slope of the curve
after ash particles have filled or bridged the pores in the fabric or dustcake.
The relationship is expected to become linear after these pores have been filled.
Differences in K2 between tests are expected to reflect the differences in ash
properties:  size, shape, and cohesivity.  Increasing the ash cohesivity will
result in an increase in dustcake porosity, a key determinant of the flow resis-
tance through the dustcake, which will lower K2.

The other principal characteristic of the drag-areal density relationship is the
effective residual drag, Se, which is the value of the linear part of the curve
extrapolated to the beginning of the filtration period.  Differences in Se between
tests are expected to reflect differences in fabric design and ash properties.
The degree to which ash penetrates into the pores of the fabric depends on the ash
cohesivity and the characteristics of the fabric pores.  The flow resistance
depends on the fabric pore distribution and the porosity and shape of the ash in
the pores.  Adding a precoat to the fabric that improves the uniformity of the
pore distribution or prevents ash from penetrating into the pores will lower Se.

Advantages in filtering drag can be obtained by reducing either Se or K2.  Lower
Se or K2 of the collected ash results in lower average drag over a given filtering
period.  Alternatively, lower Se or K2 permits longer filtration cycles between
cleaning cycles for a given average drag.  Maximum advantage for filtering drag
can be achieved by lowering both Se and K2.

Increasing ash cohesivity and improving the pore distribution on the filtering
medium will also improve particulate collection efficiency (2).
                                       27-8

-------
Modifying Ash
Fine Powders.  The laboratory screening tests indicated that most fine powder
additives would lower the cohesivity of ash.  This would lower dustcake porosity
and increase K2.   Filtration tests were performed to confirm this.  These test
results are presented in Table 1.

Initial testing showed a reduction in K2 for the Monticello ash conditioned with
0.2% w/w EH5 silica.  Key properties of the collected dustcake ash were measured.
The specific surface area of the conditioned dustcake ash was greater than the
unconditioned dustcake ash.  The increase in specific surface area is due to the
high specific surface area of the EH5 silica (380 m2/g).  Furthermore, the ash
cohesivity, as indicated by the porosity values, was lower for the conditioned
Monticello ash than for the unconditioned ash.  These changes in the ash proper-
ties were expected to result in a higher K2, but this was not the case.  The lower
K2 was accompanied by a dramatic decrease in collection efficiency, which further
indicates lower ash cohesivity.  The dustcake collected after this test contained
many pinholes of approximately 120 /^m diameter.  Unobstructed gas flow through the
pinholes masked the higher gas flow resistance of the conditioned dustcake ash.

Ash from Scholz Steam Plant was also mixed with EH5 silica (0.2% and 0.5% w/w).
K2 for the mixtures was higher than for the unconditioned Scholz ash.  The silica
decreased the bulk cohesivity of the ash, resulting in a decrease in dustcake
porosity.  In this case, the cohesivity was not so reduced that pinhole penetra-
tion was significant.

One of the powder additives resulted in a decrease in K2.  Addition of 1% by
weight of hydrated lime, Ca(OH)2, to the Monticello ash caused about 25% reduction
in K2.   This was  not anticipated from the screening tests (ref.  Figure 3).  More
tests  would be required to verify this result and its cause.

Various Liquids.   Several  alternatives to the NH3 + S03 conditioning agent were
considered to evaluate the concept that the formation of liquid bridges between
particles increases bulk ash cohesivity and dustcake porosity.  Triethylamine  (as
a substitute for  ammonia), oil  mist,  and water vapor were tested.

Triethylamine was evaluated as a substitute for ammonia because of the differences
in thermodynamic  and kinetic properties of the reactants of the two gases with
S03.   There is reason to expect these differences yield a normal ethyl-ammonium
                                       27-9

-------
sulfate that is in liquid phase at typical cold-side flue gas temperatures.  This
would be desirable if liquid bridging is an effective mechanism for increasing ash
cohesivity.

Triethyl amine was injected into the wind tunnel with the ash.  S03 was injected
into the FFSS to mix with the ash and triethylamine.  The test data are included
in Table 1.  It appeared from the coloration and qualitative cohesiveness of the
ash collected on the fabric that there had been incomplete mixing of the condi-
tioning agents in the FFSS.  Ash collected in the region of the fabric directly
opposite the S03 injection port was noticeably more cohesive than ash in other
regions.  Nevertheless,  both K2 and Se were reduced by the increase of ash cohes-
ivity caused by the triethylamine and S03.  It is presumed that better mixing of
the agents and ash would yield greater reductions in K2 and Se.

Proof-of-concept tests were performed to evaluate the effectiveness of depositing
a thin layer of oil on the surface of the fly ash particles prior to their collec-
tion on the fabric.  The oil chosen for these tests was di(2-ethylhexyl) sebacate
because it is non-toxic, a supply was readily available, and it had been success-
fully generated as a fine mist with an existing spray nozzle.  The nozzle was used
to spray the oil mist into a large settling chamber.  The oil droplets exiting the
settling chamber had a median diameter of -0.8 p.m.   These fine sebacate particles
were injected into the FFSS through a port adjacent to the inlet tube through
which the fly ash particles enter the FFSS.  Oil injection began after enough
unconditioned Monticello ash had been sampled to coat the fabric.

The FFSS filtration data in Table 1 show an improvement in K2 due to conditioning
the ash with the oil droplets.  The degree of mixing and the uniformity of deposi-
tion of oil on the Monticello ash particles is not  known for this trial.  K2 may
have decreased even further if the oil droplets had been smaller, or if more
mixing had been possible between the oil and the ash particles.  We believe oils
other than the sebacate should produce similar effects if they have approximately
the same viscosity.

Tests were performed to determine the effects of increased relative humidity on
the filtration characteristics of the Monticello ash.  Moist air produced by
bubbling metered air through a controlled-temperature water bath was injected into
the FFSS chamber to mix with the ash-laden air sampled from the wind tunnel.
Different humidity levels were produced by maintaining different water bath
                                       27-10

-------
temperatures.   Filtration data, shown in Figure 5, indicate that increased levels
of relative humidity cause significant reductions in Se and K2 values.

In full-scale  applications the reduction in flue gas temperature that would be
associated with water injection to increase relative humidity would also decrease
total  flue gas volume.  Reducing flue gas volume flowrate would reduce filtering
pressure drop.  An additional benefit would result from a reduction in the viscos-
ity of the flue gas due to the drop in temperature (pressure drop is proportional
to gas viscosity).  Although the full implications of conditioning with increased
humidity levels need to be examined in more detail, this approach holds signifi-
cant promise.

Optimizing Filtering Surface by Precoatinq Fabrics
Precoating of the fabric surface to improve the pore distribution could reduce Se
and reduce emissions of difficult-to-collect ashes.  When highly cohesive ashes
are being filtered, a precoat material may also improve dustcake release for the
control of residual dustcake area! density.  Precoating new bags has been prac-
ticed by many of the fabric filter vendors, with the intent of protecting the bags
from oil ash or other potentially harmful particulate matter or flue conditions
that might be encountered during startup.  Fly ashes have been commonly used for
this purpose.   Some commercially-available powders have also been used, including
an aluminum silicate product.

We evaluated an aluminum silicate powder that had several features desirable in a
precoat material: very low specific surface area (1.4 m2/g) and very high bulk
porosity (95%).  Proof-of-concept tests of the aluminum silicate were performed
with the Monticello and Scholz ashes.  The fabrics were precoated by withholding
ash feed from the screw feeder while feeding the aluminum silicate into the wind
tunnel through an eductor.  This procedure was continued until the chart recording
of filtering drag indicated that the fabric surface had been completely coated
(the rate of change of the drag had become constant).  At this point the feeding
of precoat material into the eductor was discontinued and the feeding of fly ash
from the screw feeder was begun.  Results of these tests are presented in Table 1,
and in Figure  6.   These data indicate that the aluminum silicate powder produced a
large  reduction in Se.
                                       27-11

-------
CONCLUSIONS

     •    At least part of the explanation for the effect on the flow resist-
          ance of ashes of passages through acid and water dew points is
          the formation of liquid bridges between particles that increase
          cohesivity and result in more porous dustcakes.

     •    Liquid bridges formed by ammonium bisulfate,  the intermediate
          reaction product of NH3 + S03,  cause the increased porosity
          responsible for the change in flow resistance when ammonia is
          injected into flue gas.

     •    We speculate that the principal factor in spray dryer ash perform-
          ance is the increased interparticle forces between the wet sorbent
          particles and the ash.   This results in high  dustcake porosity.
          The mechanical strength of the  cake is apparently low, for rea-
          sons not yet clear.

     •    None of the fine powder glidants we have tried have desirable
          effects.  No reduction  in cohesivity is desirable.

     •    Appropriate modification of the ash can result in substantial
          improvements in filtration performance.  Furthermore, modifica-
          tion of the fabric filtering surface by precoating can potentially
          yield significant improvements  in filtration  performance.

     •    The filtration of low cohesivity ashes calls  for the use of a
          precoat material to prevent the penetration of the ash into the
          pores of the fabric, and the addition of a conditioning agent to
          form liquid bridges between the particles collected in the dust-
          cake.  These liquid bridges increase ash cohesivity and dustcake
          porosity, resulting in  a decrease in filtering drag.


ACKNOWLEDGEMENTS

This work was supported by the U.S. Department of Energy under Contract No. DE-

AC22-88PC88868.  Dr. Perry D. Bergman was the Project Manager.  The authors are
grateful for the advice and consultation  of Dr. E. B. Dismukes of Southern

Research Institute.  John Hester, Brenda  Rinehart, and  Marvin Steele performed

some of the analytical tests we have reported.  Michael Robinson carried out the

proof-of-concept filtration tests and assisted in the analysis of the results.

BHA Group provided the aluminum silicate  powder used in the precoat tests.


REFERENCES

1.   L. G.  Felix, G. E. Kenniston, and R. L. Chang.  "The Combined Effect of
     Fabric and Ash on Utility Baghouse Performance: Part I." In preparation.

2.   L. G.  Felix, R. L. Merritt,  and K. Duncan.  "Improving Baghouse
     Performance at the Monticello Generating Station."  JAPCA, September,
     1986,  36:1075-1085.
                                       27-12

-------
3.    D.  L.  Laudal  and S.  J.  Miller.   Flue Gas Conditioning for Improved
     Baghouse Performance.   In Proceedings of the Sixth Symposium on the
     Transfer and  Utilization of Particulate Control  Technology,  EPRI CS-4918
     Vol.  3,  1986,  p. 14-1.

4.   S.  T.  Johansen.   "Fouling of Walls in Systems Where Hot, Dusty Gases are
    Transported."   Sellskapet for Industriell og Teknisk Forskning, Trondheim
    (Norway), 1988.

5.    P.  V.  Bush,  T.  R.  Snyder, and R. L.  Chang.  "Determination of Baghouse
     Performance  from Coal  and Ash Properties: Part I." JAPCA, February, 1989,
     39:228-237.
                                       27-13

-------
          COAL
     COMBUSTION
       Combustor
       Design
COMBUSTION

        t
                         FLY
                         ASH
        Coal
Coal
Chemistry
Coal
Preparation
                           ASH
                    MODIFICATIONS
                  Mechanical Collection
                                 DUSTCAKE
                                 ASH
                             Conditioning
                             (Ash Properties)
                                                                ASH
                                                          COLLECTION
                                                          Flue Gas Chemistry,
                                                          Baghouse Operating Conditions
                                                          (Fabric, Cleaning Method,
                                                          Startups/shutdowns)
           .^^ RESIDUAL
           ^^ DUSTCAKE
Conditioning
(Ash/Fabric Interface)
           Figure 1.  Two approaches to  improving filtration performance by conditioning.

-------
-Q
o
 CN
I
 cu
 o
 c
 o
-t-1
 en
'en
 0)
 en
 o
O

 CD
 D
3  \-
1   h
      0.60
                      0.65
        0.70


Dustcake Porosity
0.75
0.80
           Figure 2.   Effect  of  NH3  + S03  conditioning of the porosity and resultant gas
           flow resistance  of Monticello ash.

-------
                                     o — o M5 silica
                                     • —• L90 silica
                                     A —A EH5 silica
                                     A —A chemical hydrated lime
                                     D —a gasifier  char
                                     • —• sodium  bicarbonate
                  Additive  in Mixture, % wt.
Figure 3.   Results  of uncompacted bulk porosity tests on blends of Monticello
ash and various powders.

-------
CD
<
DC
Q
                               Filling of Fabric and

                               Dustcake Pores
                    DUSTCAKE AREAL DENSITY (W)
    Figure 4.  Drag versus dustcake areal  density showing  the definitions  of  K2
    and  S0.

-------
o
 CM
 CP
 a
      0.7
      0.6   -
      0.5  -
      OA  -
      0.3  -
Q    0-2  -
      0.1  -
      0.0
                                                          conditioned (54% RH)
                                              conditioned (69% RH)
        0.00
0.01
0.02
0.03
0.04
0.05
                                         Areal Density.  Ib/ft'
                Figure 5.  Filtration of Monticello ash on 25% EST fabric at three levels

                of relative humidity (RH).  Unconditioned refers to a baseline test with a
                relative humidity of 7-10%.

-------
o
 CN
 cr>

 D
 i_

Q
       1.2
       1.0  -
       0.8  -
       0.6  -
0.4  -
       0.2   -
       0.0
                                                                                  Scholz unconditioned
                                                          Monticello unconditioned
                                                                                    Scholz on

                                                                               aluminum silicate precoat
                                                                                    Monticello on


                                                                               aluminum silicate precoat
         0.00
                     0.01
0.02



Areal Density, Ibs/ft
0.03


   2
0.04
                                                                                               0.05
                    Figure 6.   Effects  of precoating 25%  EST fabric  with aluminum silicate.

-------
                                                                       Table 1
                                                    Results  of Proof-of-Concept Filtration  Tests
ro
O
ASH FILTERED
Monticello
Monticello
Monticello
Monticello (325°F)
Monticello + 0.2% EH5 silica
Monticello + 1 .0% NaHCC>3
Monticello + 1.0% Ca(OH)2
Monticello + 1.0% aluminum silicate
Monticello + Triethylamine (325°F)
Monticello at 54% RH
Monticello at 69% RH
Monticello + oil droplets
Monticello on aluminum silicate precoat
Scholz
Scholz
Scholz
Scholz + 0.2% EH5 silica
Scholz + 0.5% EH5 silica
Scholz + 0.5% dodecyl sulfate
Scholz + 1.0% aluminum silicate
Scholz on aluminum silicate precoat
FABRIC
EST, %
0
25
25
25
0
0
25
25
25
25
25
25
25
0
25
25
0
25
25
25
25
K2
in. H2O-min-ft/lb
14.8
1 1.8
12.3
6.9
1 1 .4
14.8
9.0
9.2
5.4
8.7
5.3
7.9
8.5
28.2
15.8
14.6
20.8
22.5
14.9
13.4
15.3
Se
in. H2O/(ft/min)
1.02
0.32
0.29
0.22
0.66
1.21
0.35
0.23
0.16
0.25
0.16
0.28
0.05
1.12
0.45
0.38
1.60
0.40
0.36
0.33
0.06
COLLECTION
EFFICIENCY, %
99.0
99.9
99.1
94.2
74.0
98.0
99.8
99.9
97.3
99.4
99.9
99.9
99.7
99.7
99.0
98.9
99.6
99.8
99.4
97.8
99.8
                          Tests performed at room temperature except where noted

-------
           PARTICLE  SIZE  EFFECTS  ON HIGH-TEMPERATURE DUST CAKE FILTRATION
                 FROM A COAL-FIRED ATMOSPHERIC FLUIDIZED-BED COMBUSTOR
                                  Richard A.  Dennis
                                 Larry D.  Strickland
                                   Ta-Kuan Chiang
                         United States Department Of Energy
                         Morgantown Energy Technology Center
                          P. 0. Box 880 Collins Ferry Road
                         Morgantown,  West Virginia 26505-880
ABSTRACT

Four tests were conducted to study the cleanup of particulates from the effluent
of pressurized fluidized-bed coal combustion in a combined cycle process.  These
tests examined the effects of particle size on high-temperature dust cake filtra-
tion from a coal-fired fluidized-bed combustor.  An atmospheric fluidized-bed
combustor with an inside diameter of 6 in. was used to generate the particulate-
laden process streams.  Three different particulate size distributions were ob-
tained with cyclones;  a fourth size distribution was obtained by using the combus-
tor exhaust gas without cyclones.   Dust cake filtration performance was quantified
by measuring the specific flow resistance (K2)  of  the  dust  cake,  which ranged  from
7.4 to 115.5 in. H20 ft min/lb  for  mean particle  sizes  of  11.4  to 4.2  fun.   Results
indicate that particle size has a strong effect on K2  values  in a non-linear  func-
tional relationship.
                                       28-1

-------
        PARTICLE  SIZE  EFFECTS  ON HIGH-TEMPERATURE DUST CAKE FILTRATION
            FROM  A COAL-FIRED  ATMOSPHERIC FLUIDIZED-BED COMBUSTOR
INTRODUCTION

The Morgantown Energy Technology Center (METC) of the U.S. Department of Energy is
supporting research on several advanced technologies that can use the large U.S.
coal reserves.  One of these near-term advanced technologies is the pressurized
fluidized-bed combustion (PFBC)  combined-cycle process.  The PFBC combined-cycle
process can increase the overall efficiency by about 4% over conventional coal-fired
power plants.  The process also meets the New Source Performance Standards  (NSPS)
established by the U.S. Environmental Protection Agency.

In a conventional sense, the PFBC combined-cycle process uses an in-bed heat ex-
changer to raise steam as the prime mover in a steam turbine cycle.  The second
cycle in this dual-cycle system utilizes a gas turbine with the high-temperature and
high-pressure (HTHP) combustor exhaust gas as the working fluid.  Conditions of the
exhaust gas for the first generation of PFBCs will be about 1,600 °F and 165 psig,
with an unaltered particulate loading of 10,000 to 20,000 ppmw.  The most formidable
problem presently facing this technology is related to the gas cycle, specifically,
the compatibility of a gas turbine with the particulate loading in a combustor's
effluent.

Several approaches have been taken to clean the gas stream of dust prior to the gas
turbine inlet.  These techniques, which have been shown to reduce particulate load-
ing to acceptable limits, include fixed-bed filters,  moving-bed filters, and ceramic
barrier filters.  High efficiency cyclones have also been proposed for gas cleaning,
but the exiting particulate loading could still be on the high end of inlet concen-
tration limits for conventional gas turbines.  While the turbine could be "rugged-
ized" to possibly withstand these conditions, additional downstream cleanup equip-
ment would still be needed to meet NSPS requirements.  These stipulations make the
high-efficiency-cyclone approach unattractive.

Assuming that barrier or bed-type filters become the champions of HTHP gas stream
cleanup, it would then be possible to use unaltered,  commercially available gas
turbines;  this would also eliminate the need for additional gas stream cleanup
equipment while still meeting NSPS requirements.  This assumption raises questions
concerning the performance of HTHP filter systems with the given inlet particulate
concentrations and the particle size distributions in the filtrate.  The specific
question to be answered is, what is the overall effect on HTHP dust cake filtration
when the particulate concentration and size distribution in the filtrate are altered
by placing cyclones upstream of the filter?  This study provides information that
may be used to answer this question.


EXPERIMENTAL APPROACH

The particulate size distribution and concentration emitted from an atmospheric
fluidized-bed combustion (AFBC)  process were varied using cyclones, and effects on
barrier-filter dust cake filtration were examined.  Since dust cake filtration is


                                        28-2

-------
the operative mechanism for removing particles from the  filtrate among the HTHP
filters considered for PFBC applications, the results may be applicable to both bar-
rier and bed-type filters.  Theory predicts that dust cakes formed  from particles
with a smaller mean size will have a higher specific flow resistance  (K2)  and a
higher pressure drop.  A main objective of this work was to evaluate  the magnitude
of this trend at high temperature.  The effect of the altered combustor effluent on
dust cake filtration was quantified by measuring the K2  value  and the total pressure
drop through the filter system.

To generate a representative particulate sample for the  filtration  tests, an AFBC
system was utilized.  Although it would have been possible to re-entrain particles
from a PFBC system and to conduct the tests at simulated HTHP conditions, it was
believed that using actual particles emitted from an AFBC system would be more im-
portant to the test objectives than if high-pressure test conditions  with simulated
particulate loadings were used.  The absence of high pressure in the  test series was
not detrimental, since the Reynolds number for flow through the porous dust cakes
considered here was less then 1.0 for either a high or low pressure case  (1) .  Fur-
thermore, flow through a porous media with a Reynolds number of less  than 1.0 is
governed by viscous drag  (2,3) .  Consequently, the increase in filtrate momentum
caused by an increase in pressure does not influence the filtration phenomenon as it
relates to the experimental objectives.  In addition, the difference  in pressure
between PFBC and AFBC operation would not substantially change the  filtrate viscos-
ity, since an increase in pressure, especially at high temperature, has little ef-
fect on gas viscosity (4J  .


TEST FACILITY AND OPERATIONAL EXPERIENCE

The relevant information needed to evaluate the experimental results  in terms of de-
sign and operation of the test facility is presented here; a more complete descrip-
tion is presented elsewhere (5.) .  An AFBC system with an inside diameter of 6 in.
was utilized to generate a particulate-laden gas stream; the gas stream was fed to a
vessel housing a full-size candle filter.  The cyclone system was positioned between
the combustor and filter vessel, facilitating various cyclone configurations.

The 6-in. AFBC system was fired with Pittsburgh No. 8 coal, ground  to a size of
12 x 10 mesh and fed at a rate of 10 Ib/h.  Greer limestone, ground to 12 x 0 mesh,
was used as the sulfur sorbent and was fed at a rate of 1.5 Ib/h.   These fuel and
sorbent feed rates provided a calcium-to-sulfur molar ratio of 1.6.   Approximately
1,200 scf/h of combustor effluent was sampled in the freeboard, which provided the
particulate-laden gas stream for the tests; the remaining exhaust was cleaned and
vented to the atmosphere.   Because of the large heat loss of the small combustor,
the sampled freeboard gas temperature was typically 550 °F.   Freeboard gas sampling
occurred 60% above isokinetic conditions for all of the tests.  Gas transported from
the combustor to the filter vessel was guard-heated and was carried in 1-in.,
20-gage stainless steel tubing.  Table 1 presents the combustor operating conditions
for the tests.

The cyclone system was used to alter the particle size distribution and the resul-
ting particle concentration leaving the combustor and entering the  candle filter
vessel.   The cyclone system was designed so that different cyclone  configurations
could be installed within the guard-heated enclosure.  The two cyclones used here
had designed d50 cut sizes of 2.3 (Jm  (cyclone 1) and 1.5  (am  (cyclone 2).   The
respective inlet design velocities were 30 ft/s and 54 ft/s.  Standard guidelines
were used to design the inlet areas,  internal diameters, and exit diameters of the
cyclones,  but modifications were employed that simplified the fabrication process.
These modifications included (1)  circular inlet cross-sections instead of the
                                        28-3

-------
typical rectangular shape, and  (2) modification of the typical conical  transitions
to straight outlets to the bottom solids exit.

During the four tests, particulate size distributions for the filter vessel  were
obtained by (1) using the single primary cyclone  (cyclone 1, Test One),  (2)  using
the straight unaltered combustor exhaust gas  (no  cyclone, Test Two),  (3)  using the
secondary cyclone alone  (cyclone 2,  Test Three),  and  (4) using the  primary and sec-
ondary cyclones in series (cyclones 1 and 2, Test Four).  In application, the actual
inlet velocities were 28 ft/s for cyclone 1 and 52 ft/s  for cyclone 2,  based on in-
let conditions of 1,210 scf/h at 25 psig and 600  °F.   The cyclone outlet dust con-
centrations were 1,826 ppmw for cyclone 1,  1,504  ppmw for cyclone 2, and  291 ppmw
for cyclones 1 and 2 in series.  The inlet dust loading  to the cyclone  system was
about 7,725 ppmw for Tests One, Two, and Four.  Heaters  on the cyclone  system were
designed to maintain the gas temperature.  In practice,  however, the gas  stream was
typically reheated.  Inlet temperatures were typically 550 °F;  exit temperatures
were about 800 °F.

The same Refractron Corp. RI-20 candle filter was used in all tests.  The outside
diameter of the filter was 2.32 in., the inside diameter was 1.57 in.,  and the
overall length was 39.37 in.  These dimensions provided  a filter surface  area of
2.0 ft2.   Silicon carbide was  used to  fabricate  the filter,  which was  isostatically
pressed to the basic shape and then machined to the final dimensions.   The hemis-
pherical filter flange, used to mount the filter  in the  tube sheet, and the  bottom
filter plug were cemented in place.   For an RI-20 filter grade,  mean and  large pore
sizes were measured by an ASTM bubble test to be  43 and  77 (im, respectively.  The
slope for pressure drop versus face velocity was  determined to be 0.7 in. H20 min/ft
for air at ambient conditions.  A maximum working temperature of 1,800  °F and a room
temperature modulus of rupture at about 1,950 psi were suggested for filters of this
material in this configuration.

The candle filter vessel was fabricated from a 20-in., schedule-40  pipe;  150 Ib
blind and slip-on flanges formed the closures on  each end.  A schematic of this
vessel is shown in Figure 1.  Cast in-place insulation was provided to  reduce heat
loss from the vessel.  Guard heaters were also installed to maintain a  high  fil-
tration temperature.  A shroud was placed between the candle filter and the  vessel
guard heaters to permit uniform heat transfer to  the filter and to  prevent dust from
settling on the guard heaters.  The shroud was fabricated from 5-in.,  schedule-10,
304 stainless steel pipe with an inside diameter  of 5.29 in.  Filtration tempera-
tures were measured in the filtrate inlet nozzle,  midway along the  length of the
filter 0.5 in. from the filter surface,  and above the tube sheet as the cleaned gas
left the filter.  The inlet, midpoint,  and exit temperatures,  when  averaged  over
time for the first three tests, were 1,280 °F,  1,321  °F,  and 1,274  °F.   While this
vertical variation in temperature would not be characteristic of an actual filtering
system, the difference between the temperatures in the three tests  was  never greater
than 80 °F.   The respective  inlet and midpoint temperatures, when averaged over time
for Test Four, were 1,088 °F and 1,128  °F.   A problem with the thermocouple preven-
ted accurate reporting of the gas exit temperature.   The inability  to operate the
filter vessel at 1,500 to 1,600 °F as  planned was  primarily a result of the  low gas
temperature in the combustor freeboard.

The filter was cleaned on-line every hour during  the tests with a 1 s pulse  of gas
from a high-pressure nitrogen reservoir.   The blow-back  cleaning gas was  piped from
a reservoir vessel and exited a 0.375-in. x 20-gage stainless steel tube, which was
inserted along the center line 1 in. into the filter.  Dust removed in  the cleaning
process was caught in a pan positioned at the bottom of  the vessel.  Also, the fil-
ter tube sheet assembly was  designed so that the  candle  filter could be removed
without dislodging any residual dust,  which was weighed  separately.  The  reservoir


                                        28-4

-------
pressure was 207 psig  for  Test  One,  was  increased to 225  psig in the middle of Test
Two, and was held at 225 psig for  Tests  Three  and Four.


RESULTS

The main objective of  the  four  tests was to  evaluate the  effect of particle size on
K2.   Consequently,  the  only parameters varied were mean particle size and the  resul-
ting dust concentration.   The discussion illustrates how  K2 and a  calculated value
for the mass of the residual dust  cake were  evaluated from the candle-filter dust
cake filtration data.  Figure 2  shows the pressure drop signature  across the candle
filter versus time for Test Three, which can be  used as a reference in  the  discus-
sion.

As reported by others  (jo) , the  specific  flow resistance of the dust cake K2 is
defined in Equation I  to be the  pressure drop  through the dust cake Pd,   divided by
the face velocity Vf times  the areal dust density Wc.  In  this relationship,  the
areal dust density is  the  average  dust cake  mass per unit area:

                  K2 =  Pd/(V£ • wc)  .                                               (1)

In Equation I, Pd is  the difference between the candle-filter pressure drop just
before and just after  a cleaning cycle.   The areal density of the  dust  cake  was
difficult to determine since the cleaning cycle  periodically  removed the accumulated
dust.  To determine this quantity, the dust  concentration must be  determined  and  the
filtration face velocity must be known.   The areal dust density can be  written in
terms of the filtration face velocity Vf, dust  loading Cd,  and filtration time  T::

                  Wc =  Cd  • Vf •   Tt  .                                               (2)

By substituting Equation 2 into  Equation 1,  an expression for the  specific  flow re-
sistance was developed, based on experimentally  determined quantities:

                  K, =  Pd/(Vf2  •  Cd  •  T2)  .                                         (3)

The dust concentration Cd was  calculated  by dividing the total collected mass of
dust by the total amount of gas  passing  through  the  filter.   The total  dust  collec-
ted was determined by  weighing the dust  from the  candle-filter catch pan and  the
residual dust cake that was still on the filter  at  the end of a test.   A total gas
volume sent through the filter for the period of  the test  was determined by multi-
plying the average candle-filter flow-rate by the  length  of the test period.

Equation 1 can also be used to estimate  the  mass  of  the residual dust cake Wr.  As
the name implies,  the  residual dust  cake  is  dust  that  remains on the filter  after
multiple cleaning cycles.  An indication that the  residual dust cake has reached  a
steady state occurs when the pressure drop across  the  filter  after cleaning  repeat-
edly returns to the same value.   The amount  of dust  involved  in the residual  dust
cake can be estimated  by assuming that the specific  flow  resistance K2 calculated
for the steady-state cleaning cycles is  the  same  for the  residual  cake,  and  that  the
residual dust cake is  uniform along  the  length of  the  filter.   If  these assumptions
are made,  Equation 1 can be solved for the areal  density  of the dust cake,  which is
assumed to be equal to  the mass  of the residual  dust cake  divided  by the filter
area.   The pressure drop Pr used in this  equation is now the difference between the
pressure drop of the initial clean filter at the  beginning of a test, and the
steady-state pressure  drop across the freshly cleaned  filter:

                  Wr =  (Pr  • Af)/(Vf  • K2)   .                                        (4)
                                        28-5

-------
Test Results

Candle-Filter Test One was conducted for approximately 34 h with the  single  primary
cyclone in-line.  During the test, the filter vessel pressure averaged  23.7  psig,
with a standard deviation of 1.6 psi.  Based on the flow rate, vessel pressure,  and
candle filter midpoint temperature, an average face velocity was calculated  to be
13.1 ft/min with a standard deviation of 1.2 ft/min.  The total gas passing  the
filter was measured to be 41,639.8 scf and the resulting dust collected was  measured
to be 5.7 Ib.  The residual dust cake was measured to have a mass of  0.21  Ib.  The
pressure drop versus time signature for this test showed the characteristic  increase
in the pressure drop of the freshly cleaned filter until the developing residual
dust cake reached a steady state.  A steady-state residual dust cake was formed  at
approximately 24 h.  Beginning with hour 25 of the test, the next nine  cleaning
cycles were used to determine an average Kj  value.   The  last  filtration cycle was
not used since it was cut short of a full hour.  The average pressure drop during
the filtration cycles was 36.3 in. H20,  with a  standard  deviation  of 10.4 in. H20.
An average K2 value was  calculated to be  33.0 in.  H20 ft min/lb, with  a  standard
deviation of 11.7 in. H,0 ft  min/lb.  The average  particle  diameter  based on volume
for the collected dust was 6.8 |im, with  an overall particulate concentration of
1,826 ppmw.  The mass of the residual dust cake was calculated to be  0.16  Ib, and
the maximum pressure drop across the filter was 4.6 psig.

Candle-Filter Test Two was conducted without a cyclone in-line, allowing the  unal-
tered AFBC particulate-laden gas stream to reach the candle filter.   This  test was
broken into two time periods since the candle filter vessel was taken off-line
because of a plug in the downstream flow control-loop.  However,  since  the filter
was down for a short period,  and since the test at that time had only lasted  4.5 h,
the candle filter vessel was not cleaned out, and the dust collected was averaged
into the second part of the test.  The second part of the test lasted 31 h,  with an
average vessel pressure of 23.8 psig and a standard deviation of 1.8 psi.  Temper-
ature, pressure, and flow rate conditions produced a candle-filter face velocity of
13.1 ft/min, with a standard deviation of 1.0 ft/min.  The total dust collected  and
the gas passing the filter were measured for the combined parts of the  test  and
found to be 25.6 Ib and 44,178.5 scf, respectively.  Unfortunately,  the residual
dust cake mass in this test was not kept separate from the pan dust.  Increasing the
cleaning pressure during the test effectively removed the residual dust cake, which
had reached a steady state 17 h into the test.   Subsequent to the formation  of the
initial residual dust cake, five filtration cycles were used to calculate  K2. The
average pressure drop for these cycles  was 34.7 in. H20,  with  a  standard deviation
of 4.4 in. H20.   The average  K2 value was determined to be 7.4 in. H20 ft min/lb,
with a standard deviation of 1.6 in.  H20  ft  min/lb.   The  average particle diameter
based on volume was 11.4 |im for the collected dust, with an overall particulate
concentration of 7,725 ppmw.   The residual dust cake mass was calculated to  be
0.78 Ib; the maximum pressure drop across the filter was 3.9 psig.

Candle-Filter Test Three was conducted for 64 h.   The filter vessel pressure  aver-
aged 23.5 psig, with a standard deviation of 1.3 psi.  Using the temperature, pres-
sure, and flow rate conditions, a filter face velocity of 13.2 ft/min was  calcu-
lated, with a standard deviation 0.8 ft/min.  During the test, 77,548.8 scf  of gas
passed through the filter, depositing 8.75 Ib of dust.  At hour 41 into the  test,
the stable, residual dust cake became detached from the filter.  When this occurred,
the candle-filter pressure-drop returned to its initial clean value.    No  incident
could be attributed to the detachment of the initial residual dust cake.   Following
the upset, a new residual dust cake formed,  taking approximately 24 h to reach a
steady state.  This dust cake weighed 1.15 Ib.   The subsequent 18 filtration cycles
were used to calculate an average K2  value of 40.7  in. H20 ft min/lb, with  a  stan-
dard deviation of 7.7 in. H20  ft  min/lb.   The average  pressure  drop  for these 18
cycles was 36.8 in. H20,  with  a standard  deviation  of  3.8  in.  H20.  The  average


                                        28-6

-------
particle diameter based  on  volume  was  8.2  (nn for the collected dust,  at a total
dust concentration  of  1,504 ppmw.   The calculated mass  of  the residual dust cake was
determined to be 0.20  Ib for the first period and 0.24  Ib  for the second period.
The maximum pressure drop across the filter  for  this test  was 4.8 psig.

Candle-Filter Test  Four  was conducted  with cyclones  1 and  2  in series.   Again,  this
test was run in two segments because of problems with the  flow control valve.   The
first segment lasted for approximately 15.5  h and the second segment  lasted for
34.8 h, producing a total test time of 50.3  h.   The  vessel pressure averaged
22.3 psig, with a standard  deviation of 2.5  psig.  The  total flow of  gas through the
filter was determined  to be 56,736  scf.  Since the filtration flow rate  varied
during the initial  part  of  this test,  the  last 16 filtration cycles were used to
calculate K2.   The  average pressure drop for these cycles was 22.5  in. H20,  with a
standard deviation  of  5.8 in. H20.   During this period,  the  face velocity averaged
13.0 ft/min, with a standard deviation of  0.1 ft/min.   During the entire test,
1.17 Ib of dust were collected: 0.86 Ib from the catch  pan,  and 0.31  Ib  from the
residual dust cake.  With this dust loading,  a concentration of 291 ppmw was deter-
mined.  An average  K2 of  115.5 in.  H20  ft min/lb  was  calculated,  with  a standard
deviation of 29.8 in.  H20 ft min/lb.  The average particle diameter for this dust
was determined to be 4.2 (im. A residual dust cake was  not calculated,  since a
distinct formation  period could not be determined.   The maximum pressure drop across
the filter for this test was recorded  as 3.6 psig.


DISCUSSION OF RESULTS

The most significant result of this test series  is the  quantification  of the effect
of particle size on the  specific flow  resistance (K2) for various dust cakes.  In
the candle filter test block, when  the effect of particle  size  on Kj was compared
among the tests, a  reduction in mean particle diameter  by  volume  produced a  large
increase in K2,  in  a functional  relationship that appeared to be non-linear.  A
reduction in the mean  particle size by 62% increased the K2 value by a factor of 15.
A summary of the results and operating conditions  for the  test  series  is  presented
in Table 2.

While the trend appears  strong,  there  was  some variability in the  data.   This was
true for measurements  of particle  size and K2.  The variability in K2  was  quantified
using a standard propagation of uncertainty  analysis.   In  Tests One and  Four, most
of the uncertainty  in  K2 was produced by the  variation in pressure drop between  fil-
tration cycles.  This  fluctuation  could be because of variations  in the  volume  of
dust carried over from the  fluidized-bed combustor through the  cyclone system and
delivered to the candle  filter.   Higher dust  loadings will produce  a  higher  pressure
drop within equal filtration periods.   An  alternate  theory for  this fluctuation
could be that the residual  dust cake was changing  after a  quasi-steady state had
been reached.   This theory  is supported by the somewhat similar  slopes of the pres-
sure drop versus time  cycles  while the  residual  dust  cake  was  forming  (indicating
uniform dust loading).    Once  the residual  dust cake  seems  to have  reached a  steady
state,  it then begins  to  outgrow its own supporting  mechanisms  and  consequently de-
cays.   This reforming  and decaying procedure  could produce filter  flow versus pres-
sure drop characteristics that would produce  different  pressure drop  versus  time
slopes.   The highest uncertainty in K2  was  associated with Test Four and was en-
hanced further by the  low dust concentration.  The lowest  uncertainty  was associated
with Test Two because of  the  low standard  deviation  in  the average  pressure  drop
during the filtration  cycles, and the  high dust  concentration.

An additional source of  error that affects confidence in this  trend is associated
with particle size  measurement.   While  the size  distributions  from cyclone  1 (Test
One)  and cyclone 2   (Test  Three)  were expected  to be  close, it  was  assumed,  based on
the  cyclone design,  that  the particulate size  distribution leaving  cyclone  2 would

                                        28-7

-------
be smaller than that leaving cyclone 1.  However, coulter counter measurements  in-
dicated that this was not the case.  (This could possibly be because  of  inconsis-
tencies in the grab samples taken from the candle-filter vessel  catch-pan  for the
two tests.)   Also, high temperature effects on gas viscosity have been known to
change cyclone performance, thus altering the predicted d50  size.  All other indi-
cators, such as the calculation of K2,  the  dust  concentration,  and the designed d50
cut size of cyclone 2, indicate that a smaller particle size should have been de-
livered from cyclone 2.  The mean particle size of the collected dust was  based on
the average of the mean particle size by volume for three grab samples taken for
each test.

If theory is used to examine the effect of particle size on K2  and a  simple expres-
sion for the areal density of the dust cake, shown in Equation 5 is substituted into
Equation 1 and solved for K2,  Equation  6  then  demonstrates  that K2 will vary by  the
inverse cubed power of the particle diameter.   When some simplifying  assumptions are
made regarding the porosity of the dust cake and the number of particles per unit
area,  it can be seen that the experimental data has a similar trend.

                  Wc = (N •  p •  TT •  D3)/6 ,  and                                    (5)

                  K2 = (6 Pd)/(Vf •  p •  71 •  N  • D3)  .                               (6)

When the effect of mean particle size and dust concentration on the average pressure
drop during the filtration cycles is evaluated,  it can be seen that Test Four,  with
the highest K2,  had the  lowest  average  pressure  drop  during the filtration cycles.
This is because of the non-linear effect of particle size and the confounding linear
effect of dust concentration on pressure drop through a porous dust cake.   Darcy's
Law shows that dust concentration and the resulting dust cake thickness  are directly
proportional to and linear with pressure drop through a porous dust cake.   So,  even
though the particle size is reduced, which would tend to raise the pressure drop,
the large reduction in dust concentration overcomes the particle size effect; the
net result is a reduced pressure drop.   The confounding of these test parameters
results from an inability to reduce the mean particle size with cyclones without
reducing the particulate concentration.  This interaction can be illustrated by
comparing Tests Two and Four.  When the mean particle size in Test Four  is  reduced
by 62%, dust concentration is reduced by a factor of 26, which is sufficient to re-
duce the average pressure drop during the filtration cycles from 34.7 in.  H?0 to
22.5 in. H,0.   Since both tests  had  equal filtration  intervals,  the  areal dust  cake
density and thickness would also be reduced by a factor of 26.

The reduced values in the average pressure  drop during a filtration cycle  and the
maximum pressure drop in Test Four can also be attributed to the lower filtration
temperature.  When the average filtration temperature of Test Four is compared  to
that in the first three tests, a difference of about 200°F is observed.   The lower
temperature produced a viscosity value that is 6% lower.  However, even  though
Darcy's law predicts the linear dependence  of pressure drop on viscosity,  this  can
only account for a small portion of the reduced values.  In addition, if Test Four
had been conducted at the same temperature  as the first three tests,  the 1C, value
would also have been slightly higher.

Another phenomenon related to the low dust  concentration that appeared in  Test  Four
was the inability to detect a distinct formation period for the residual dust cake.
Since the residual dust cake never reached  a steady state during Test Four, the low
maximum pressure drop cannot be compared to the other tests.  If Test Four had  been
allowed to continue, and if a steady-state  residual dust cake had formed,  the maxi-
mum pressure drop through the filter would probably have been higher  than  for the
other tests.
                                        28-8

-------
In calculating the  residual  dust  cake,  two  assumptions  were made:   (1)  the K, value
calculated during the  filtration  cycles was applicable  to the residual  dust cake,
and  (2) the residual dust  cake  was  uniform  over  the length of the  candle filter.
However, the distribution  of the  dust  cake  was clearly  not uniform over the length
of the  filter.  Typically, a bulbous dust cake was  found at the end of  the filter,
and a short cylindrical  segment was found midway along  the length  of the filter.
Between these two irregular  sections and above the  cylindrical section,  there  typ-
ically  was a thin  (approximately  0.125  to 0.25 in.  thick)  cracker-like  cake.   The
irregularity in the distribution  of the residual dust cake was most likely caused  by
the vessel and filter  configuration, the resulting  filtrate flow patterns,  and  the
filter  cleaning procedure  used  during  the tests.  These results also suggest that
the assumption of equal  K2 values for the filtration dust cake  and the residual dust
cake may not be valid.


CONCLUSIONS

Based on results of this testing, the  following  observations can be made:

  •  For the particulate size distributions tested  and  the associated dust
     concentrations, a 62% reduction in mean particle size produced an  increase
     in the specific flow  resistance of the dust cake  (K2) by a  factor of 15.

  •  In HTHP filtration, with all other parameters  equal,  unaltered AFBC
     combustor exhaust provides an  effluent that is more amenable  to dust  cake
     filtration than an  effluent  modified with cyclones.   This is  substantiated
     by the lower K2 value and lower maximum pressure drop for  filtration of
     unaltered combustor exhaust.

In conclusion, for HTHP  filters that rely on dust cake  filtration,  allowing the
unaltered combustor exhaust  to  reach the filter  may improve  the operational perfor-
mance of the device.   Improved  performance  has been quantified by  a lower  K2 value
and a lower maximum pressure drop through a filter.  These  results  suggest  (1)  up-
stream cyclones should be  removed from  the  flow  stream,  or  (2)  the  operational  per-
formance should be turned  down, so that particles with  a larger mean  size  distribu-
tion will be received  at the filter.  This  design philosophy has been supported by
others  (7.) .   As shown  in the test series, this will help to  reduce  the  K2 value and
will reduce the overall pressure  drop through the particulate cleanup system.   Fur-
thermore,  if upstream  cyclones  are  removed,  an additional  pressure  drop  would not  be
expended,  and the need for separate ash handling equipment  would be  eliminated.


NOMENCLATURE

K, -  Specific  flow  resistance,  in. H20  ft min/lb
Pd -  Pressure  drop  during a filtration  cycle, in. H20
Vf -  Filter  superficial face  velocity,  ft/min
Cd -  Dust concentration,  lb/ft3
T!   Length  of  filtration cycle, ruin
T,   Time to form residual  dust  cake, h
A£ -  Filter  filtration  area,  ft2
Wr - Mass of residual dust  cake, Ib
Wc - Areal dust cake density, lb/ft2
Pr -  Pressure  drop  because  of residual  dust  cake, in. H20
N  -  Number  of spheres per unit area of  diameter D
p  -  Particle density,  lb/ft3
D    Diameter of  sphere,  ft
                                       28-9

-------
REFERENCES

1.   J.P.K. Seville, R. Clift,  C.J.  Withers and W. Keidel.  "Rigid Ceramic Media for
    Filtering Hot Gases."   Filtration and Separation,  July/August 1989,  p. 265-271.

2.   R.E. Collins.  Flow of  Fluids  Through Porous Materials.   New York:   Reinhold
    Publishing Co., 1961, p.  51-52.

3.   D. Hillel.  Fundamentals  of  Soil Physics.  New York:   Academic Press,  1980.

4.   R.W. Miller.  Flow Measurement  Engineering Handbook.   New York:   McGraw Hill
    Co., 1983, p. H1-H5.

5.   R.A. Dennis.  Evaluation  of  Dust Cake Filtration at High  Temperature With
    Particulate From an Atmospheric  Fluidized-Bed Combustor,  Technical  Note,  U.S.
    Department of Energy, Morgantown Energy Technology Center (in press).

6.   D. Leith and R.W.K. Allen.   "Dust Filtration by Fabric Filters."   In  Progress
    in Filtration and Separation 4.   R.J. Wakeman,  Editor, 1986,  p.  1-51.

7   M.W. First.  High Temperature  Gas Filtration Research  Needs.   Harvard  School of
    Public Health, for the  U.S.  Department of Energy, Morgantown Energy  Technology
    Center, Contract No. DE-AC01-83FE60365,  1984, 89 p.
                                       Filter Hold Down Flange
                  Pulsed Blow Back
                   (Cleaning Gas)~
Clean Gas Outlet
                                                         Tube Sheet
                                                        Shroud
                                                        Poured Insulation
                                                        Candle Filter
                                                       20"
                                                           -Dirty Gas Inlet
                  Figure 1.  Schematic  of the Candle Filter Vessel
                                        28-10

-------
150
                                                    Maximum Pressure Drop
                                                         Across Filter
12    18    24    30     36    42
                 Time, (hours)
48
                                                        54
60
66
72
         Figure 2.  Pressure Drop Versus Time  for Test Three
     Table 1.  Nominal Operating Conditions for the  6-in.
                 AFBC System During the Test  Series
                    AFBC Operating Parameter: Value

                           Pressure:  25 psig
                        Bed  Temperature:   1,575°F
                            Excess Air:   100%
                      Coal Type:  Pittsburgh No. 8
                        Coal Size:   10 x 12 mesh
                     Sorbent:  Greer Limestone Sand
                       Sorbent Size:  12 x 0 mesh
                        Coal Feed Rate:  10 Ib/h
                      Sorbent Feed Rate:  1.5 Ib/h
                         Ca/S Molar Ratio:  1.6
                      Fluidizing Velocity:  5 ft/s
                      Air Flow  Rate:   2,500  scf/h
                            Bed  Depth:   21  in
                          Bed Diameter:   6  in
                    Particulate Emission: 7,725 ppmw
                                 28-11

-------
Table 2.  Summary of Results From Candle Filter Test Series
Test Parameter
Date
Run Time, h
Cyclone Configuration
Filter Face Velocity, ft/min
Particle Density, g/cm3
Average Particle Diameter
by Volume, Jim
Average Surface Area of
Sampled Particulate (BET) , mVg
Dust Concentration, ppmw
Average Midpoint Candle
Filter Temperature, °F
Average Vessel Pressure, psig
K2, in. H20 ft min/lb
Calculated Residual Dust
Cake Mass, Ib
Measured Residual Dust Cake
Ma s s , Ib
Maximum Total pressure Drop
Before Cleaning, psig
Average Pressure Drop During
Filtration Cycles, in. H20
Test One
5/19-20/89
34
Cyclone 1
13.1
2.79
6.8

5.3
1,826
1,318

23.7
33.0
0.16

0.21

4.6
36.3
Test Two
5/21-23/89
35.5
No Cyclone
13.1
2.79
11.4

11.0
7,725
1,309

23.8
7.4
0.78

See Text

3.9
34.7
Test Three
5/23-26/89
64
Cyclone 2
13.2
2.80
8.2

7.5
1,504
1,331

23.5
40.7
0.24

1.15

4.8
36.3
Test Four
9/21-23/89
50.3
Cyclones
1 s 2
13.0
2.86
4.2

5.2
291
1,128

22.3
115.5
See
Text
See
Text
3.6
22.5
                          28-12

-------
               GENERALIZATION OF LABORATORY DUST CAKE CHARACTERISTICS
                             FOR FULL-SCALE APPLICATIONS
                                   Ta-Ruan  Chiang
                                   Richard A.  Dennis
                                  Larry  D. Strickland
                                    Charles  M. Zeh
                         United States  Department  of  Energy
                         Morgantown Energy Technology Center
                                    P.O. Box  880
                                Morgantown, WV  26505
ABSTRACT

The characteristics of dust cakes formed on porous ceramic filters were investigated
to examine high-pressure and high-temperature gas filtration in pressurized fluid-
ized-bed combustion (PFBC).   Information on filter cake porosity was obtained from a
laboratory-scale coal-fired atmospheric fludidized-bed combustion  (AFBC) system and
a porous, ceramic disc-filter setup.  A laboratory-scale natural-gas pressurized
combustor laden with re-suspended PFBC particulate and two full-size ceramic candle
filters were used to collect application data on the specific resistance coefficient
of a filter cake at elevated temperatures and pressures.  Dimensionless particulate
and flow characteristics were used to establish a relationship between filter cake
porosity, particle shape, particle volume-surface mean diameter, and face velocity.
The functional dependence of a filter cake's specific resistance coefficient and the
flow Reynolds number were also established.  Operating data from full-scale utility
fabric filter installations  were introduced to examine this approach as a possible
means for extending laboratory data to full-scale applications.
                                        29-1

-------
               GENERALIZATION OF LABORATORY DUST  CAKE  CHARACTERISTICS
                             FOR FULL-SCALE APPLICATIONS
INTRODUCTION

Previous laboratory research on barrier filters at high temperatures  and  pressures
indicates that barrier filters may provide a viable particulate  cleanup method  for
pressurized fluidized-bed combustion  (PFBC) applications.  Actual  test results  in
gasification and PFBC applications  (1.) confirm the technological feasibility.   How-
ever, traditional questions on the risks of scale-up and the adequacy of  laboratory-
simulated testing remain unanswered.  Therefore, to accurately evaluate barrier
filters under actual operating conditions, the U.S. Department of  Energy  through a
cooperative agreement with American Electric Power, will test barrier filter devices
on a PFBC slipstream at the Tidd Clean Coal Demonstration Project  (2.) .  The study
described here was initiated to support future testing at the Tidd site.

To verify the adequacy of laboratory-simulated particulate testing, two approaches
were used:  (1)  a particulate dust cake generated from actual coal combustion and
sorbent injection in a laboratory-scale test facility, and  (2) a dust cake generated
from the re-injection of sub-pilot PFBC particulate in a separate  laboratory-scale
test facility.  The purpose was to compare these two basically different  dust cakes
to see if laboratory-simulated particulate testing is representative  of actual  oper-
ating conditions.  These data were then compared with operating  data  available  from
full-scale and pilot-scale fabric filter installations for pulverized-coal  (PC)
utility boilers to see if the laboratory data could be extended  to full-scale appli-
cations.  Using dimensionless groups that characterize dust cake filtration and
viscous flow through a long pipe,  a consistent trend was developed for all three
sets of data obtained at different scales and under different operating conditions:
(1) the laboratory-scale dust cake formed by actual coal combustion and sorbent
injection under atmospheric pressure and elevated temperature, (2)  the laboratory-
scale dust cake formed by re-injection of sub-pilot PFBC particulate  under elevated
pressure and temperature, and (3)  the full-scale dust cake formed  on  utility fabric
filters at atmospheric pressure and 300°F.


EXPERIMENTAL METHODS

Two test apparatuses were configured to evaluate the performance of ceramic candle
filters under various operating conditions:  (1) a laboratory-scale coal-fired  at-
mospheric fluidized-bed combustion  (AFBC)  system (3)  comprising  a  ceramic disc  fil-
ter made from ceramic candle filter materials to provide fundamental  information on
filter cake porosity under actual coal combustion and sorbent injection,  and  (2) a
laboratory-scale pressurized natural-gas combustor (^) comprising  two full-size cer-
amic candle filters with re-injected particulate to provide application data under
elevated temperature and pressure.  The re-injected particulate  was primary cyclone
dust from the New York University (NYU)  subpilot-scale PFBC test facility (5) ground
to a mass median diameter (MMD)  of 4.68 (Xm, with a geometric  standard deviation of
1.62.
                                       29-2

-------
Ten tests were  conducted  in an AFBC  disc-filter apparatus  to provide filter cake
porosities and  specific resistance coefficients for various  face velocities and par-
ticle size distributions.   The particle  size  distribution  was varied by using dif-
ferent size cyclones  (or  no cyclone)  upstream of the disc  filter.   A virgin disc
filter was used for each  test.   Prior to each test,  the clean disc weight,  and the
clean disc pressure drop  versus  face velocity were  established.   Filter cakes were
formed under various  steady flow conditions but no  pulse-jet cleaning cycles.  A
wide range in test conditions  were used  as  shown below.

                       Gas  Temperature:   1,000 to 1,300 °F
                           Gas Pressure:   10 to 33 psig
                          Face Velocity,  vf:   16  to  27  fpm
                  Particle  Size  Distribution,  volume-surface mean
                          diameter,  dvsrn:   2.76 to 4.86 (am
                   Particle Size Distribution, geometric standard
                             deviation:  1.44 to 1.54
                   Particulate Specific Gravity:  2.68 to 2.81
              Particulate Specific Surface  Area, S:  2.28  to 4.85  m2/g

Particle size distribution  was obtained  from  Coulter  counter measurements,  assuming
log-normal distribution.  Particulate  specific  surface  area  was  obtained  by  Brun-
aeur-Emmett-Teller (BET)  analyzer; particulate  specific  gravity  by helium pycno-
meter.

After a filter  cake formed  with  the  desired thickness  under  a specific  face velo-
city, the disc  filter was by-passed  from the  AFBC flow.  A hot nitrogen flow was
then directed through the established  filter  cake and  the disc filter to  examine the
specific resistance coefficient  of the filter  cake  at  various  face  velocities.  To
avoid possible  alteration of the  established  filter cake, face velocities during
this hot nitrogen flow test were  kept  below the  face velocity under  which the filter
cake was formed.  At the  completion  of the test, the disc filter and filter cake
were carefully  removed from the  test  apparatus  for  final weighing  and cake  thickness
measurements.    The filter cake porosity  was then calculated  from the particulate
bulk density,  which was derived  from  the cake weight and volume.

Four tests were conducted in the  pressurized  natural-gas combustor  to examine the
effects of face velocity  and cleaning  technique  on  the  specific  resistance  coeffi-
cient of the ceramic candle filter.   Virgin candles were used for  each  test.   Prior
to each test,  the clean candle weight, and the  clean candle  pressure  drop versus
face velocity were established.   Filter  cakes were  formed and cleaned repeatedly.
Test conditions are shown below.

                        Gas Temperature:   1,350 to 1,550°F
                            Gas Pressure:  8.5 to 9.9 a tin
                       Face Velocity,  v£:  7.5 to 19 fpm
                    Particle Size Distribution,  volume-surface mean
                          diameter, dvsm:  3.56 to 3.96 (im
                  Particle Size Distribution, geometric standard
                             deviation:  1.66 to 1.74
                   Particulate Specific  Gravity:  3.07  to 3.18
            Particulate Specific Surface Area,  S:   5.04 to  6.62  m2/g

Continuous testing was aimed for  each test.  A total of  641  h of testing  were con-
ducted,  including two tests lasting for  192 and  196 h.   A total  of  2,166  data points
were recorded  and analyzed, or 998 cleaning cycles  (including 346  off-line  cleaning
cycles).
                                       29-3

-------
RESULTS

Differences in testing conditions between  the  two test apparatuses precluded direct
comparison of the two sets of dust  cake  filtration data.   Dimensionless groups that
characterize dust cake filtration and  gas  flow properties were thus used to estab-
lish correlations.  Following the general  analysis of flow through a packed bed  ((5),
dimensionless groups of interest included  the  flow Reynolds number Re£,  the particle
Reynolds number Rep,  the  particle shape factor X,  the  dust  cake  porosity £,  and the
skin-friction coefficient c£.   If pg is assumed to be the gas density, (X the gas
viscosity, pp the particle density,  Tp  the  particle relaxation time constant, K2 the
specific resistance coefficient of  the dust  cake,  APC  the dust  cake pressure drop,
and <3C the dust  cake areal density,  by analogy with  flow  through a long pipe and the
defining Fanning equation, it can be shown that

                  Ref =  (2/3)Rep/[(l -  £) X]   ,  and                                (1)

                  cf = (6 £3A) (K2 Tp/Rep)  ,                                         (2)
where

             Rep = pg vf dvj[l ,  \ =  (1/6) d™  pp  S  ,

             TP  = (1/18)  pp dL,Yn ,  and K2  =  APC/(0C vf)  .

Temperature and pressure information are contained in  the gas density and gas
viscosity terms.

According to the empirical Blasius  law,  a  logarithmic  plot  of Equations 1 and 2
leads to a linear relationship in the  viscous  flow region for a  given dust cake
porosity.  Figure 1 is a plot using the  experimental data obtained from the labora-
tory AFBC test series; the dust cake porosity £ and  the specific resistance coeffi-
cient Kj  were  derived from the dust  cake weight, volume,  and  areal  density.  Figure
1 demonstrates this linear relationship.   When  Ref was 3.36X10"4  to  l.OOxlO"2, the
observed pressure drop of the dust  cake  implied a  value of  the  skin-friction coeffi-
cient cf,  given  by

                  cf = 8.766  Re,'0'615  .                                              (3)

Equation 3 can be used to predict barrier-filter pressure drops  for a given particle
size distribution and specific surface area,  provided  that  the  dust cake porosity is
known.  To develop an empirical relationship for dust  cake  porosity,  dimensionless
characteristic groups were again used.   It is physically  plausible that the modified
dust cake solidity  (1 - e)A, is a function  of the normalized particle residence time
2EA/7STK,  where  STK  is the  modified  Stokes  number,  based  on the particle's  volume-
surface mean diameter.  The normalized particle residence time  is defined as the
ratio of time tr taken for  a particle entrained in a gas  stream to  go through a dust
cake channel of an effective particle  diameter, and  the particle relaxation time
constant Tp.   If de is assumed to be an effective particle diameter in the local flow
direction, and ve the  effective  flow velocity,  then the normalized  particle
residence time is

                  Tr/tp = (d./v.)/[pp d_2/(18 \L)]

                        =  [de/(vt/e)]/[Pp O'dS ji)]  .                            (4)
                                        29-4

-------
Assuming that de is proportional to dvsm, A. and S account  for  various  particle shapes
and surface areas:  de <*=  A/d^,  since S <>;  A,,  the normalized particle residence time
defined in Equation 4  reduces to

                   tr/Tp oc  2 e X2/STK ,

where the modified Stokes  number STK is based on the particle's volume-surface mean
diameter dvsm:

                   STK  = pp v£ d_2/(9 p.  dV8J .

Figure 2 is a logarithmic  plot  of  the  modified dust cake solidity  (1   e)A, versus
the normalized particle residence  time 2eA//STK, using  the  experimental data ob-
tained from the laboratory AFBC test  series.   Data from operating full-scale and
research pilot-scale utility fabric-filter installations (2)  are also plotted in
Figure 2 to validate this  approach and increase its applicable range.  The linear
relationship in Figure 2  is  given  analytically by

                   (1 - e)£-°-293 = (0.558/X)  (A,2/STK)0'293  .                             (5)

Experimental data  on full-size  ceramic candle filters obtained from the laboratory
PFBC test series were  used in Equation 5 to establish dust cake porosities.  Using
the known dust cake porosity and experimental K2 values,  values  for c£ and Ref from
the laboratory PFBC test  series were  obtained;  these data were plotted with the data
from laboratory AFBC test  series in Figure 3.  Figure 3 also includes data from
operating full-scale,  and  research pilot-scale,  utility fabric-filter installations.
Figure 3 shows a general trend  in  accordance  with the characteristics of Equation 3
for all data points, regardless of the origin or scale.  A noticeable departure
exists only for the fabric filter  installation at  Monticello.

Notice that Equation 5 describes dust  cake porosity using characteristic particle
parameters that are only measurable in the laboratory.   These characteristics
include particle size  distribution,  specific  gravity,  and specific surface area.
Operating parameters include the face  velocity and the  temperature, and thus, the
gas viscosity.  However, according to  Equation 5,  dust  cake porosity is not gas
density or pressure dependent.

Equation 3 can also be written  explicitly in  terms of particle and flow character-
istics :

                   K2 = 1.869 e~3[(l - e)|I pp S2]°'615(pg v£)°'385  .                       (6)

However,  Equation  6 indicates that the  specific resistance coefficient is gas
density or pressure dependent.   Equation 7 is an alternate representation of
Equation 3 in the  familiar form of the  Carman-Kozeny equation where the Carman-
Kozeny specific resistance coefficient  K2|C_K) is defined by Equation 8:

                   K2 =  K2(c.K){0.187[Rep A/'d -  err385} ,                             (7)

where

                   K2(C_K) =  180 [(1 -  E)/e3]A,2[(l/(pp Ol  •                           (8)
                                        29-5

-------
CONCLUSIONS

Dimensionless grouping and normalization of particle and flow characteristics  in-
dicate that dust cake filtration data can be compared, regardless of the  operating
condition  (PC, AFBC, PFBC),  the scale (full, pilot, laboratory), the dust  source
(coal-burning, coal-burning with sorbent-injection, injected flyash),  and  the  type
of filter used  (woven glass-fiber,  porous ceramic candle) .  Empirical  relationships
for porosity and the specific resistance coefficient of the dust cake  were estab-
lished.  Thus, the means for predicting barrier filter performance under  actual con-
ditions is available, if a representative particle sample can be analyzed  in the
laboratory or if a design specification is given.

Dust cake porosity is found to be independent of gas pressure or density;  the  spe-
cific resistance coefficient of a dust cake is, however, gas pressure  dependent.

The particle shape factor and its Stokes number affect the dust cake porosity  in a
complicated way.  In general, a smaller particle size constitutes a smaller poro-
sity; a larger particle shape factor, however, leads to an increased porosity, thus
counter-balancing the diminishing porosity caused by smaller particles.  A higher
face velocity would generally lead to a dust cake with a higher porosity.   However,
this dust cake would also have a larger specific resistance coefficient,   resulting
in a higher pressure drop.  Therefore, as a general rule, utility fabric  filters are
invariably sized for operation at a low face velocity.

Although the dust cake porosity and specific resistance coefficient can now be pre-
dicted using Equations 5 and 6, a full-scale barrier filter application would  not be
complete without information on areal densities of the residual dust.


ACKNOWLEDGMENTS

The perseverance and hard work of Mr. Richard Griffith, Mr. Mark Tucker, Mr. Charles
Carter, Mr. David Turner, Mr. Larry Kisner, Mr. Gary McDaniel, Mr. Rick Pratt, Mr.
John Rotunda, Mr. John Trader, and other members of the Morgantown Energy  Technology
Center Project Support Staff, made this work possible.


REFERENCES

1. V.P.Kothari and J.R. Longanbach,  Editors.  Proceedings of the Eighth Annual
   Gasification and Gas Stream Cleanup Systems Contractors Review Meeting,  Vol. I,
   U.S. Department of Energy, Morgantown Energy Technology Center,
   DOE/METC-83/6092, NTIS/DE88010253, May 1988, 414 p.

2. M.J. Mudd and D.A. Bauer.  "TIDD PFBC Demonstration Plant: Combined Cycle PFBC
   Clean Coal Technology."  In Proceedings of the Ninth International  Conference on
   Fluidized Bed Combustion, Vol. I,  1987, p. 256-260.

3. R.A. Dennis.  Evaluation of Dust Cake Filtration at High Temperature With
   Particulate From an Atmospheric Fluidized-Bed Combustor, Technical  Note.
   U.S. Department of Energy, Morgantown Energy Technology Center  (in  press).

4. C.M. Zeh, T-K. Chiang and W.J. Ayers.  Evaluation of Ceramic Candle Filter
   Performance in a Hot Particulate Laden Stream, Technical Note.  U.S. Department
   of Energy, Morgantown Energy Technology Center  (in press).
                                        29-6

-------
5.  V.  Zakkay, et al.   Construction and Utilization of a Larqe-Scale  Coal  Fired
   Pressurized Fluidized  Bed Facility.  New York University,  for  the  U.S.  Department
   of Energy, Morgantown  Energy Technology Center, DOE/MC/14322-1800,
   NTIS/DE85013685, August  1985,  245 p.

6.  J.M. Kay.  An Introduction to Fluid Mechanics and Heat Transfer.  London:  The
   Cambridge University Press,  1957,  p. 253-257.

7.  P.  Vann Bush, T.R.  Snyder and R.L.  Chang.  "Determination  of Baghouse
   Performance from Coal  and Ash Properties: Part I."  JAPCA, Vol. 39,  No.  2,
   February  1989, p.  228-237.
           8.0


           7.5


           7.0


           6.5


           6.0


           5.5


           5.0


           4.5


           4.0
            -8.0    -7.6     -7.2    -6.8    -6.4    -6.0
                                      In (Ref)
                                                          -5.2
-4.8
         Figure 1.  Laboratory ATBC Test  Series,  Skin Resistance Coefficient
                             Versus Flow  Reynolds Number
                                        29-7

-------




















I.O
-t c
I .O
1.4
1.2
1.0
X
07 °'8
1
<=, 0.6
£ 0.4
0.2
n n
\j,\j
-0.2
-0.4
-0 6
• 11
• Laboratory AFBC ,/
• Utility Fabric Filters s'
/^
ma ./^
.^ B10
-^
B.8
*s^
^^
•5^^B7
H2 »4/B6 -^
• *s^
/^
*5-?r"« m3
s' •
•
•
I I I I I I

















23456789
In (26X2/STK)
Legend
1-Monticello; 2-Intermountain Power Project Unit 1; 3-Brunner Island; 4-Arapahoe
Pilot
clone
AFBC






















; 5-Arapahoe Unit 4; 6-Arapahoe Ecolaire; 7-Arapahoe Pilot Downstream of Cy-
; 8-Scholz High-Sulfur Pilot; 9-Nixon; 10-EPRI High-Sulfur Test Center; 11-TVA
Pilot
Figure 2. Modified Filter Cake Solidity Versus Normalized Residence Time
for Laboratory AFBC and Utility Fabric Filters
9 0

8.5

8.0

7.5

"? 7.0
	
£ 6.5

6.0

5.5
5.0

4.5
4.0
• 1

• 7

.11 • *6
-\^ «5 H2
^~~\ B3
^^--. BIO f
^"~\|9*
^^~~^\ •* * *
« V\^ ** .»
* • ^V^* • ** •
•••""•-• *• ^ •*"
•** ^^? /****•
•• * /". *^Tr^*tf*
• Laboratory AFBC »>• • ^~~~~
A Laboratory PFBC * * • *•
• Utility Fabric Filters *
	 1 	 1 	 i_ i


















-9-5 -8.5 -7.5 -6.5 -5.5 -4.5
                                        In (Ref)

                                       Legend
1-Monticello; 2-Intermountain Power Project Unit 1; 3-Brunner Island; 4-Arapahoe
Pilot; 5-Arapahoe Unit 4; 6-Arapahoe Ecolaire; 7-Arapahoe Pilot Downstream of Cy-
clone; 8-Scholz High-Sulfur Pilot; 9-Nixon; 10-EPRI High-Sulfur Test Center; 11-TVA
AFBC Pilot
         Figure 3.  Skin Resistance Coefficient Versus Flow Reynolds Number
          for Laboratory AFBC, Laboratory  PFBC, and Utility Fabric Filters
                                       29-8

-------
                     HIGH  TEMPERATURE  FILTRATION
                        USING  CERAMIC  FILTERS
                     L. R. White  and  S.  M.  Sanocki
                          3M  Co.,  218-3S-03
                    St. Paul,  Minnesota 55144-1000
ABSTRACT

Filters must satisfy the following requirements:  they must collect dust
at high efficiency, they must be  cleanable,  and they must be durable.
Greater than 99.9% of  the particles  must be removed from most dusty gas
streams in  order to  meet  NSPS.   To  avoid unacceptable  energy loss,
pressure drop across the filters must  be less than a  few percent of the
gas stream  pressure.    Finally,  filters  must  last  one to  five years
depending on their cost  in  order  to  satisfy  economic  constraints.
Ceramic fabric filters have been tested for the past ten years and are
now in  commercial  service.   A brief  history of their  development is
given along  with  a report  on current status.   Test  results on tough,
porous,  and  shock  tolerant  structures  produced  by  chemical  vapor
infiltration  of  silicon carbide  into woven,  braided,  and non-woven
substrates will also be presented.   These structures  have shown promise
as hot gas filters and a report on their status is given.
                                  30-1

-------
                             INTRODUCTION

  High  temperature filters  are needed  for advanced  coal conversion
processes such as pressurized fluidized bed combustion  (PFBC) and coal
gasification/gas turbine combined cycle systems.  Filters would remove
particulates  from  hot gases before  expanding them  through turbines.
Several  approaches  have been  taken  to  hot gas  filtration including
baghouses, ceramic cross-flow filters,  fixed and moving granular beds,
etc.   In this paper  the focus  is on  baghouse  filtration using both
fabric  and  rigid filters with  cleaning by  pulse-jet.   Two  kinds of
filters  are  discussed,  ceramic  fabrics  and porous,  rigid structures
called candles.

Fabrics and finishes

  In  fabric filtration at  temperature  less than about  500 F,  fabric
finishes  are  used to minimize  abrasion.   At high  temperatures such
finishes cannot be used.  However, if the  fiber used in the filter is
harder than the particulate being collected, then the particulate acts
like a solid lubricant.

In weaving ceramic fabrics yarn  must  be selected  carefully.  Some yarns
are too  stiff  to weave  or are damaged during weaving.   Key parameters
are tensile modulus and diameter of the individual fibers  in the yarn.
The higher the modulus,  the stiffer the yarn and the harder it will be
to weave.  But,  the ability  of  fibers to tolerate sharp bends depends
strongly on filament diameter and stiff yarns can be woven  if filament
diameter is small enough.
  The ceramic fabric discussed here is woven  from  an alumina-boria-
silica yarn (Nextel 312™-available from 3M Co.).

Seams

  Conventional fabric filters are cylindrical in shape and  prepared by
sewing an axial seam and early versions of ceramic bags were made with
seams.  However, seam failures occurred, and woven seamless tubing was
developed.   Seamless tubing  is flattened  during weaving  and  takeup
processes and the edges  suffer some yarn damage,  nevertheless, seamless
tubing  is stronger than tubes with seams.

Hardware

   Ceramic fabrics are damaged during  cleaning if they are supported over
wire  cages.   On the other hand, hundreds  of  thousands of pulses have
been  applied  to ceramic fabrics disposed  over  perforated metal cages
with  no  failures observed.   Most filter bags have a  sewn-in bottom and
the cage is completely covered by the bag.  Although ceramic fabric bags
can be  produced  with  one end closed  they are abraded by the bottom of
the cage.  As an alternative bags have been made which  are  open at both
ends.   Installation has  been done by clamping to the cage  at both top
and bottom.   Cages  had  to be made with  solid bottoms of course.  Good
life  has  been observed  with  this method of installation.   It has been
                                  30-2

-------
necessary to use temperature-compensating clamps since ordinary clamps
will not hold fabric in place after a few cycles in temperature.

  Axial  expansion  of  cages must  be  accounted  for when bags  are
installed.    Ceramic fabrics can  be  exposed to temperature swings in
excess  of  1000  degrees  and the  mismatch  in  coefficient  of  thermal
expansion between  cage and  fabric  can stretch the fabric.   If cage
growth is excessive then bags must be installed with  longitudinal slack.

  Radial fit of fabric to cage must be good with little slack.  Diameter
of seamless tubing must be controlled during weaving and cages must be
fabricated with close  tolerance on outside  diameter.   If fit is poor,
creases  form  in the  fabric and  these  creases are  potential  failure
points.

  It  might  appear that use of metal cages  and  tube sheets leads to a
restriction  in  filtration  temperature.    From  the  standpoint  of
filtration  alone,  the  upper limit  in temperature  is  determined  by
stickiness of the particulate.  When particles begin to fuse filtration
becomes difficult  because the cake  blinds and is  hard to remove by
pulse-cleaning.   In  filtering  flue  gas from  coal  this  temperature
depends on whether oxidizing or reducing conditions  are present and the
composition of the  ash.   It  can be as  low as  1700  F.   In incineration
where a variety  of  inorganic constituents are  present and low-melting
eutectics can be expected to form, the  upper  limit  is much lower.   In
coal gasification, there is a temperature  window defined on  the low end
by the dew point of tars and on the high end by the softening point of
the ash.  Therefore, the temperature at which a filter can be operated
depends on the service more than on the metals used in the bag house.

  In  filtration  of  flue  gas  from  combustion systems it is important to
avoid collecting large glowing or  burning  particles.  These "sparklers"
can damage  fabric  if  they fuse  the dust  cake.  Molten ash can either
react with the fabric  and weaken it or simply embrittle it by rigidizing
it.    After  all the fuel has burned out  of a particle  it  will cool,
solidify and if the ash  has  penetrated  the  fabric,  rigidize it.  Then
when the fabric  is  pulse-cleaned  the  edge of  this  brittle region will
tear.    It  is good practice  to maintain a thick dust  layer since  the
thicker the  dust layer  the  larger the  glowing particle that  can be
collected without  damage to the  filter.  Thus  overcleaning should be
avoided.  When high concentration of sparklers exists in the flue gas,
mechanical separation of the largest of them is highly recommended and
a variety of devices can be used for that purpose.

Chemical Attack

  Ceramic  fabrics  used  in  filtration  of  flue gases are  subject to
chemical attack  in two  different ways,  primarily.   At temperatures
greater than about 1400 F, alkalis have significant vapor pressure, will
be present in the vapor phase,  and will attack alumina-based ceramics.
To  avoid  destruction   of  the ceramic  fabric  temperature must  be  low
enough so that alkalis are present as solids only.  Dust-covered  fabrics
                                  30-3

-------
which pass through the dew point are subject to attack after the fabric
is reheated.  The mechanism  is  not understood fully but this  has been
observed several times.  A potential attack mechanism  is reaction with
low-melting  eutectics  which  are  leached   from   the ash  and  then
concentrated on fiber surfaces.

                     PERFORMANCE OF CERAMIC FABRIC
Life test

   An EPRI-sponsored life test1 was
run  on  ceramic  fabric  and  test
conditions along with  results are
in Table 1.  Test  conditions  were
set to simulate hot gas cleanup in
the turbo-charged boiler concept of
PFBC.   Testing was  begun in  the
fall of 1985  and ended in spring of
1989.  Operating time on the fabric
was slightly less  than  two years.
The most important results are in
Figure 1 which  shows modest loss in
fabric strength after 17000 hours.

Application
                                     Table  1     Typical   operating
                                     conditions
                                          ceramic fabric filter test

                                     Temperature, degrees C    400-450
                                     Air/cloth ratio,  m/min    1.8-2.0
                                     Baghouse  delta P,  kPa     1.4
                                     Flue gas  rate,  cu m/min   18-20
                                     Collection efficiency
                                     >99.95%
                                    Fig. 1  % Strength  retention vs hours
   In  petroleum refining cat
cracking    catalyst    is
regenerated in a fluidized bed
which operates at about  1300
F.  Periodically,  catalyst is
discharged to a storage hopper
and the vent from that hopper
must be  filtered to  capture
catalyst dust.   The  stream is
hot and  the simplest way to
filter  it  requires  a   high
temperature    filter.
Comparison    of     a   low
temperature baghouse with air
bled in to cool the gas showed
that cost for the air blower
and controls exceeded that for
the  hot  baghouse   and,  of
course,  operation  would be
much simpler.  A baghouse with
ceramic  fabric  filters was
installed in the first guarter of 1988 and has been operating  since.
                                             V~7\
                              Figure  1
                                  30-4

-------
                            CANDLE FILTERS


  Candle  filters are being  developed  which will be  made by chemical
vapor infiltration  (CVI)  of  silicon carbide  on fibrous  substrates.2
Conventional candle  filters  have been made by  sintering a mixture of
ceramic fibers and granules  and  a  clay binder.  Typical shape has been
like a large test tube, with dimensions about five feet long by about
three inches in diameter.  Wall  thickness  has been about 1/2 inch.  A
goal in making CVI  candles  is to  reduce wall thickness thus reducing
weight and also pressure drop.  Less weight and pressure drop simplify
tube sheet design and the thinner the wall, the shorter the filtration
path and the more cleanable the filter.

  Making  CVI candles involves  the following steps,  making a ceramic
preform, coating  it with a  carbon interface, and  applying a ceramic
coating by chemical vapor infiltration.   The  carbon barrier is needed
to promote toughness.

  At this writing,  a number  of preforms have  been fabricated and a few
CVI  runs have  been  completed but no tests  on finished specimens have
been made.

REFERENCES
   1.    Weber,   G.  F.  and  Grant  Schelkopf,  PERFORMANCE/DURABILITY
   EVALUATION  OF 3M COMPANY'S HIGH-TEMPERATURE NEXTEL™  FILTER BAGS,
   Final  project report,  EPRI RP 1336-16

   2.  Stinton, D. P., R.  A.  Lowden, Ramsey Chang,  Ceramic Engineering
   &  Science Proceedings,  vol.  9,  no.  9-10,  p.  1233-1244,  1988
                                  30-5

-------
                HIGH TEMPERATURE  FILTER MEDIA  EVALUATION

                             D. J. Helfritch
                              P. L. Feldman

                            Research-Cottrell
                 Environmental Services and Technologies
                              P.O. Box 1500
                      Somerville,  New Jersey  08876
ABSTRACT
Four commercial high temperature  filter  media have been operated
in parallel under typical  flue  gas  conditions.   The filters were
a self supporting ceramic fiber mat, a ceramic fiber woven cloth,
a laminated  stainless  steel wire  cloth, and a  sintered Inconel
felt.  Testing was  carried out  at 850°F, utilizing on-line pulse
jet  cleaning.    The  effects of  air  cloth  ratio and  cleaning
intensity  were examined.    Significant  differences  were  found
between the  filter  media with  respect to collection  efficiency,
pressure drop, and  durability.   These results  are presented and
discussed.
                                 31-1

-------
INTRODUCTION
     The  highest  allowable  gas   temperature  has  generally  been
considered  to  be  500°  when  fabric  filtration  is  to be  used for
particulate  cleanup;   however,   the  need  for  high  efficiency
particulate removal does not cease for  higher temperatures.  This is
especially true for the metallurgical,  mineral processing and power
generation  industries.   Often a  clean,  hot gas  is  needed for the
process, such as for gas turbines.
     In  response  to  this  need   for  high  temperature filtration,
several new filter media have recently  been  developed.  These filter
media   are   of   metallic  or  ceramic   construction,   capable  of
withstanding 1000° and  2000°F  respectively.    For all  the filter
media,  fibers   are  produced from  the  base  material  and  are
subsequently processed  to form a  filtration surface.   This is done
either by weaving the fibers  to produce a cloth, or by forming a felt
material by laying down an interlocked bed of fibers.
     Four  such  high   temperature filter   media are  commercially
available.  They are:


     •  Bekipoi® ST25GFX - A  web of randomly laid  Inconel
        fibers,  forming  a solid porous plate by means of sintering.
        Maximum recommended temperature = 1020°F.   Supplier:  N. V.
        Bekaert, S.A.,  Zwevegen,  Belgium
     • Dynapore^ 405705  Laminated Filter Plate - Layers of stainless
       steel wire cloth and perforated plate sintered into a
       monolithic porous structure.  Supplier:  Michigan Dynamics,
       Inc., Garden City, Michigan.


     « Nexte]® Woven Ceramic Fabric - A woven fabric of  Alumina-
       Boria-Silica fibers, weighing 13 oz/sq. yd.  Maximum
       recommended temperature = 2100°F.  Supplier:  3M Co.,
       St. Paul, Minnesota.
                               31-2

-------
     • KE85 Felt -  A  self-supporting, thick, felt-like material made
       of refractory fibers held together with an organic binder.
       Maximum recommended temperature = 1650°F.  Supplier BWF KG,
       Offingen, West Germany.
     These  filter  media have  all  been independently  performance
tested in commercial  and pilot scale applications.  The Bekipor media
has been  used  in a substantial number  of  commercial applications.
These include a  cement  clinker  coolers,  electric  arc furnaces,  and
a fluidized  bed boiler.   The   Nextel media has  been  successfully
operated for 17,000 hours in a pilot  scale test of flyash filtration2.
The KE85  media also has  been pilot scale  tested on a  coal fired
boiler3  Studies  such as these have shown that these filter media can
be  utilized in  conventional baghouses and  perform similarly  to
conventional filter  fabrics;  however,  available information cannot
be used to directly compare  the  performance  of the filters.  Because
of this,  a test program was  carried  out in which the  four filter
media  described above were  operated  in parallel, under identical
conditions.   This  program  and  the  results achieved are described
below.
EXPERIMENTAL APPARATUS
       Flow System

       A nine-element apparatus was  used  for the testing described
here and  is  shown schematically by  Figure 1.   This  apparatus was
designed to simultaneously operate nine, I1 x I1  filter panels.  Each
filter module was assembled as shown in Figure 2.  The  filter modules
were then mounted on the common housing such that the filter planes
were vertical  as shown  in Figure  1.   The  positions of  the  four
filters to be evaluated are shown in Figure 3.
     The hot process gas was  produced by a  60,000 Btu/hr natural gas
burner.  Flyash  was  mixed with the hot gas  upstream  of  the filter
modules.  An Acrison feeder was  used to meter the ash, such that the
desired particulate  loading was achieved.   The  gas flow was evenly
distributed over the width of the  common  housing by means of vanes
and perforated  plates within  the  flow expansion  inlet.   The gas
subsequently made a 90°  turn  into the  individual filter modules, and
                               31-3

-------
passed through the filters.  Downstream orifice plates were used to
measure the flow passing through each filter module.  The flow rates
were adjusted by means of in-line ball valves.
     The nine individual  flows  were subsequently recombined within
a  common  manifold.    A fan was used  to draw  the gas through  the
system. Ambient  dilution air was  mixed with  the high temperature
process gas just upstream  of the  fan inlet in order to protect the
fan from excessive temperature.
     A  pulse  jet  type  cleaning  system  was utilized  to  allow
continuous  operation  over  long  periods.    Each  filter  module
contained, four, 1/4", compressed air nozzles, directed at the back
face  of the  filter  panel.   The  filter  modules were  cleaned  in
sequence by pressurizing the nozzles with a 200 msec pulse of 100 psi
air.   It  was  found that this yielded  an  instantaneous reverse air
flow rate  of  18  cfm per square foot of filter surface and a total
reverse air  volume of  .06 ft3  per  square  foot of  surface.   These
values are similar to those found for commercial pulse-jet filters.
     The  ash  used for  testing  was obtained  from the precipitator
hoppers of Public Service Electric and  Gas  Company's Mercer Station.
The ash was  generated from the combustion of  coal  with  average as
received properties:
                    6.53%          Moisture
                    5.45%          Ash
                    .84%           Sulfur
                 13,661            Btu/lb
     A mineral analysis of the  flyash is given in  Table 1.  The mass
mean diameter of the  ash was 7.8 microns and the standard deviation
was 3.43, values typical for PC fired boilers.

     Measurements Methods
          Temperatures, flow rates, particulate loading, and pressure
drops are parameters that define system performance,  and as such were
routinely  monitored.   Gas temperatures  were  measured by  means of
thermocouples at the gas burner  exit, within the common housing, and
within each filter module.  Gas  flow rates  were measured by means of
the orifice plates installed downstream of each filter module.  The
pressure drops across each filter  were measured by means of U-tube
manometers.
                                 31-4

-------
     Particulate loading was measured by the  insertion of an Alundum
thimble filter  downstream  of the orifice plate.   This pre-weighed
thimble was allowed to filter the full flow of one filter module for
several hours and was subsequently weighted.

       System Operation
       The system was brought up to temperature with the  ash feed and
filter  cleaning system  off.   Once  the operating  temperature was
achieved,  the ash feed and compressed air pulse cleaning  systems were
turned on.  The gas flow through each cell was generally maintained
at 4.5 ACFM,  which yields a  filter air cloth ratio of 4.5 ACFM/ft2 at
the typical temperature of  850°F.  Each filter was jet pulse cleaned
once every five minutes.
     Overnight,  the  ash feed and pulse  jet cleaning  system were
turned  off,  but the  gas burner  and  fan were  left on  to  keep the
system  hot.  The  entire system  was  turned  off on  weekends.   In
addition  to  recording  the  filter  pressure  drops,  the filtration
efficiencies  were periodically  measured by means  of  pre  weighed
alundum thimble type  filter.

       RESULTS
       The four filters  identified above were  operated intermittently
over  the  period  from 4/17/89 to  12/8/89.    During this period  a
cumulative time of 180  hours was  achieved.   The flyash loading was
held  at  approximately  2.3  gr./scf.    The   air  cloth  ratios were
generally held  at  4.5 ACFM/Sq. Ft.,  and the  gas temperature varied
between 600°F and 850°F. The flyash  specific resistance coefficient
was 1.82 N min/gm m,  a value which produced a pressure drop increase
of 2.0 inches water per hour at the conditions given above.
     The pressure drop versus time histories of each filter are shown
in  Figure  4.    The  vertical  lines  represent breaks in   system
operation, such as evenings and weekends.  The KE85 filter ruptured
due to  high pressure drop   (9 inches  water) after  seven  hours of
operation, and  no data is  shown  for this filter.   The structural
integrity of the remaining  three  filters was  good through the  entire
                               31-5

-------
program.    The  figure  shows   that   the  pressure  drops  differ
substantially, the Nextel filter often exceeding 8 inches of water,
with the Bekipor under 2.   The  positions  of these two filters were
switched  at  140  hours,   in order  to  determine whether  relative
position within the test apparatus would influence the results.  It
is seen that  the steady  state  filter  pressure drops were unchanged
as a result of the switch.
     Figure 5 shows how  the  pressure  drop of each filter varied as
a function of air cloth  ratio.   The curves can be described by the
equation AP^(ACR)X,  with 1.2 < x <1.6, which is typical for fabric
filtration.
     The results of filter  collection efficiency testing are given
in Table  2.   It is seen that  the metal mesh  filters (Bekipor and
Dynapore)   are much  more efficient  than   the  woven  ceramic fiber
(Nextel).   All of the  filters were operated at  an air/cloth ratio of
4.5  ACFM/SQ.  FT.,  a  cleaning  pulse  pressure of  100 psig,  and a
cleaning pulse  every five  minutes.    In  an effort to  improve the
Nextel efficiency,  tests  were performed on this filter using  60 psig
cleaning pressure,  and  no  pulse cleaning  at  all.    The resulting
outlet  particle  loadings   were  .021  Gr./SCF and   .004  Gr./SCF,
respectively, indicating that the act of  pulse jet cleaning causes
much of the ash penetration through the filter.

CONCLUSIONS
     The pressure drop of a  woven fiberglass fabric in a pulse jet
filter on a coal fired boiler would be expected to be  about 5 inches
water at an air cloth ratio  of  4 ACFM/FT2.   This was achieved by the
Nextel media.  The collection efficiency of an infrequently cleaned
woven fiberglass filter would be on the order of 99.9%. This  too was
achieved  by  the  Nextel  media  for  reduced cleaning.   It  can  be
concluded, then, that the performance  of the Nextel filter media was
very similar  to that  of  a  conventional woven  fiberglass,  a result
which  should  not be  unexpected,  since the two  media have  similar
physical characteristics.
     The pressure drop of the KE85 media could not be controlled at
a  steady  state  value by the  jet pulse  mechanism.    Cleaning was
partially effective because  rate of pressure drop increase was 1 inch
water per hour,  as compared  to  a 2 inch water per hour rate when the
                                 31-6

-------
jet pulse cleaning system is deactivated.  The KE85 media is a soft
shock absorbing  type of material,  and this  characteristic  is not
likely to be compatable with jet pulse cleaning.  Also  the KE85 media
had a relatively low permeability of 5  CFM/sq.  ft.  @ 1/2 inch water,
which tends to yield higher operational pressure drops.
     The  performance of  the  metal fiber  filters was  distinctly
superior to that of  the  ceramic  media,  and superior to performance
expected  from  conventional fabrics, either  woven or  felted.   The
pressure  drops  were less  than  half  of  the  pressure  drops  of
conventional  fabrics,  and  the  ash penetration was  an order  of
magnitude lower.  This improved performance by the  Bekipor media has
been cited by its manufacturer and demonstrated by other laboratory
scale tests.4
     In order to explain  the  difference in  performance  between metal
fiber filter  media and  conventional  media,  we  must  examine their
physical  differences.   The two characteristics  that  separate both
the Bekipor and Dynapore media from conventional filter fabrics are
the stiffness of the metal media  as compared to the flexibility of
fabrics and  the good electrical  conductivity  as  compared  to  the
dielectric nature of conventional fabrics.  Good jet pulse cleaning
is  favored by  a  flexible  filter  media,  and  hence   the  relative
rigidity of the metal filters should not be responsible for improved
performance.
     The good performance of the metal  filters may be the result of
their electrical conductivity.  It is known that electrostatic forces
are a major factor in filtration pressure  drop and efficiency5, and
good performance is generally obtained  when there is a difference in
static  charge between  the  particles  and  the  filter.   The metal
filters,  being  grounded  conductors, would carry  no  static charge,
while  the  flyash  particles  would  be  naturally charged.    This
electrostatic configuration enhances particle interception by fibers.
Once intercepted  by a metal  fiber,  the particles would discharge
through the fiber, which could result in improved jet pulse cleaning.
     In summary,  it has been seen that the Nextel  ceramic fiber media
has many  of  the physical characteristics  of fiberglass filtration
fabric and its  performance  is similar  to  that  of  fiberglass.   The
Bekipor and Dynapore metal fiber filters yield lower pressure drops
and higher efficiencies  than  conventional  filter fabrics,  possibly
resulting from their electrical conductivity.

                               31-7

-------
ACKNOWLEDG MENT

The  test  program  described  above  was  sponsored  by  the  U.S.

DDE/Pittsburgh Energy  Technology Center, through  a Helipump Corp.

subcontract.
REFERENCES
1.  W. Verplancke, "Progress in Hot Gas Filtration with Metal Fiber
      Media."  Filtration  &  Separations,  March/April 1982, pp 143-
145.

2.  G. F. Weber,  Private Communication.

3.  C. Weber, "Advances in Hot Gas Filtration Technology."
      Filtration & Separation, March/April 1988, pp. 100-103.

4.  H. C. Gurtler and R. de Bruyne, "The Economics of High
      Temperature Filtration."  Filtration & Separation, November/
      December 1977,  pp. 641-648.

5.  D. J. Helfritch,  "The Development of Electrostatic Filtration
      Technology."  Proceedings of the American Filtration Society
      Annual Meeting. March 1990.
                              31-8

-------
               Mineral
               Si O2
               A12 03
               Ti O2
               Fe203
               CaO
               MgO
               Na2O
               K20
               S03
               P205
               L.O.I.
               Undet.
                               Table  1

                        ASH MINERAL ANALYSIS
                                 Wt. Percent

                                    38.71
                                    26.61
                                     1.33
                                 15.13
                                  5.56
                                  1.68
                                   1.40
                                  2.21
                                  2.98
                                  0.34
                                  3.22
                                  0.87
Filter

No Filter
(Inlet
loading)
Bekipor
Nextel
Dynapore
             Table 2

       FILTER EFFICIENCIES

   Test #              Gr/SCF

1                  2.16
     2

     Ave

     1
     2
     3
     Ave

     1
     2
     3
     4
     5
     Ave
     6**
     7***
                      2
                      3
                      Ave
     2.46
2.30
     2.3

    <.0001
     .00048
     .00075
    <.00044

     .041
     .064
     .047
     .033
     .0731
     .052
     .021
     .004

,00039
     .00061
     .00057
     .00052
                   Efficiency*
99.99+
99.98
99.97
99.98

98.22
97.22
97.96
98.57
96.83
97.74
99.09
99.82
                                      99.98
                                           99.97
                                           99.98
                                           99.98
*Based on an average inlet loading of  2.3  Gr/SCF
**Low pressure cleaning
***No cleaning
                               31-9

-------
             AIR
                 GAS  BURNER
                                    ASH
                                  FEEDER
TEST
CELLS
             2"
CO
                            -4'-8'
                                                 I.D. FAN
                                             TT
                          DILUTION
                             AIR
                           EXCESS
                             ASH
     SIDE VIEW
        FRONT VIEW
      Figure 1.   High Temperature Filtration Test Stand
                            31-10

-------
                FILTER MEDIA
GAS
FLOW-
                          JET PULSE
                          NOZZLE
  Figure  2.   Filter Cell Construction
Not Used
Bekipor
Not Used
KE 85
Nextel
Not Used
Not Used
Dynapore
Not Used
Figure  3.   Filter Cell Identification
                     31-11

-------
g
10
W
I
H
s
8
M
g
09
CO
14
12
10
6
4
2
   Nextel
       Dynapore
                       i^iA
            20
                  40
                         60
                                80
                                   100
                                         120
                                                140
                                                           160
                                                              180
                           TIME ON STREAM,  HOURS
              Figure 4.  Filter Pressure Drops vs. Time on Stream

-------
          2345678
            AIR/CLOTH RATIO, ACFM/SQ.  FT.
10
Figure 5.  Filter Pressure Drops vs. Air Cloth Ratio
                       31-13

-------
                   PILOT-SCALE PERFORMANCE/DURABILITY EVALUATION
                OF 3M  COMPANY'S  HIGH-TEMPERATURE  NEXTEL  FILTER  BAGS
                            G.F.  Weber  and G.L.  Schelkoph
                      Energy and Environmental Research Center
                             University of North Dakota
                          P.O.  Box  8213,  University Station
                          Grand Forks, North Dakota  58202
ABSTRACT

A two  phase  program to  evaluate  the performance/durability of  Nextel  filter bags,
cost-shared by the  Electric  Power  Research  Institute  and  3M Company,  was performed.
The project  was  initiated in  December  1984 with the  design and construction  of a
pilot-scale  pulse-jet  baghouse filtering  a slipstream of  flue  gas from  a  stoker-
fired boiler at the  University of  North  Dakota steam  plant.  The results from Phase
I were previously reported.

Phase II was initiated in October 1985 with installation of nine  woven Nextel filter
bags.  The original intent of  Phase II was to operate  the baghouse slipstream system
for 6,000  hours.   As of May 15, 1989, the  Nextel bags  had  accumulated  16,877 hours
of  operating  time  and  3,441  cleaning cycles.   Baghouse operating conditions  were
nominally  atmospheric  pressure and  427°  to 538°C  (800°  to 1000°F).   Air-to-cloth
ratio was  nominally  1.8  m/min  (6.0  ft/min)  but ranged  from 0.5 to 2.4 m/min (1.5 to
7.9 ft/min).   Supplemental  ash  injection  was  used throughout Phase  II,  augmenting
fly ash from the stoker-fired  combustor.   Several ash  types were  used, including fly
ash from  a pilot-scale  AFBC  system, a  pilot-scale pc-fired system,  and  two full-
scale  pc-fired  utility  stations.    Baghouse  particulate  collection  efficiency,  as
determined by EPA Methods 5 and 8, ranged from 99.66%  to 99.99% depending on fly ash
mass loading and chemical/physical characteristics.

The bags/cages were removed  from the  baghouse  in June 1989.  Each bag was carefully
cleaned and tagged to identify the location from which it was removed  and the number
of  operating  hours experienced.  Photographs  were  taken  of  individual   bags  to
identify  ash  deposits  that  were  difficult to  remove  and those  that  were embedded
into the woven fabric.   Each bag was evaluated for fabric strength (tensile strength
and MIT Flex) by 3M Company and fabric samples from selected bags were characterized
using scanning electron  microscopy  analysis.   Overall, the  Nextel  filter bags were
found to be in excellent condition after 16,877 hours  of operation.


INTRODUCTION

The goal  of coal  combustion  technology development  programs  is to  improve system
efficiency,  thereby reducing   the  cost  of  electrical  power  and  maximizing   the
potential   of  fossil  energy  resources.    One  such   program  has  focused   on   the
development of pressurized fluidized  bed  combustion  (PFBC)  technology.  Particulate
control in a  PFBC system requires a  portion,  if not  all,  of  the particulate to be
captured at elevated temperature, -400°C or ~850°C (~750°F or -1560°F).  In  the case
                                        32-1

-------
of a  combined  cycle system, high-temperature  cyclones, operating  at about  850°C
(1560°F), are  used  to capture particulate matter  upstream  of  the gas   turbine to
prevent erosion.  In addition, a conventional ESP or baghouse is required downstream
of  the   heat   recovery   boiler  to  meet   EPA   particulate   emissions  standards.
Particulate control  for a PFBC turbocharged cycle can be accomplished using a single
device  operating  at  about  -400°C  (750°F)  downstream  of  the  steam  generator  and
upstream  of  the gas turbine-driven compressor.   In this case,  particulate control
can be accomplished with a high-temperature filter.

The Electric Power Research Institute (EPRI)  began evaluating filter options for hot
gas particulate control  for  PFBCs in 1978.  This  initial work  was completed by the
Westinghouse  Electric Corporation  and  reported  in  May  1981  (1).    Westinghouse
Electric  Corporation,  under  EPRI  sponsorship,  continued  evaluation  of  filtration
systems   for   high-temperature   and   high-pressure   (HTHP)   particulate   control
applications.  Evaluation of 3M Company's woven ceramic filter continued, along with
evaluation of rigid filter elements (2).

In December  1984  EPRI, through  the  Westinghouse  Electric Corporation, subcontracted
with the  Energy & Environmental  Research  Center (EERC) of the  University  of North
Dakota (UNO) to evaluate the performance/durability of 3M Company's high-temperature
Nextel  filter  bags  (3,4).   The  program  was  a cooperative  effort  between  EERC,
Westinghouse,  3M  Company,  and  Donaldson  Company.   Phase   I  involved  the design,
construction,  and  operation  of  a  baghouse  slipstream system  at  the UNO  steam
plant.   Baghouse design  permitted  installation of  nine woven  Nextel  filter bags.
Baghouse  operating conditions were  nominally atmospheric pressure  and 427°  to 510°C
(800°  to  950°F).   Air-to-cloth ratio was  nominally 0.9  m/min (3.0 ft/min).  Filter
cleaning  was  performed  using  a  on-line  pulse-jet  cleaning  system triggered  by
baghouse  differential pressure (AP).

Phase  I   included a  total of 1,022  hours of operation and 174  baghouse  cleaning
cycles. Baghouse  particulate  collection  efficiency, as  determined  from EPA  Method 5
sampling, ranged from 99.19% to 99.96%.  Scanning electron microscopy analysis of an
EPA  Method  5  outlet  filter  showed  few  dust  particles  present on the  filter,
indicating  that particulate  collection  efficiency may  have  been better  than  EPA
Method 5  data  indicated.

Close  inspection of  the  Nextel  filter bags  following  removal  from  the  baghouse
revealed  no  major flaws  or breaks  in the  fabric.   Creases  were observed in some of
the  filter  bags near the  bottom and  were a  result of  too  much longitudinal slack
left in the filter bags when they were installed.

In October  1985 the  subcontract  was modified  to  permit installation of a second set
of  nine  woven  Nextel  filters  to  be  evaluated  in a 6,000-hour  test.   Again,  the
program   was  a  cooperative   effort  involving   EERC,  Westinghouse,   EPRI,  and  3M
Company.   The approach was  similar to that used  in  Phase I with the exception of
minor  system  modifications   and  broader  ranges for  system  operating  conditions.
Subsequent  contracts with  EPRI  and  3M  Company were  implemented  in  September and
December  1987, respectively,  to  continue  Phase II  in  an attempt  to accumulate an
additional  10,000 hours of operating time.


DESCRIPTION OF  BAGHOUSE SLIPSTREAM SYSTEM AND PROCEDURES

The  baghouse slipstream  system  consists of eight  major components:   1) UNO steam
plant  boiler #7; 2)  particle knockout  device;  3) pneumatic  ash  injection system;
4) baghouse;  5) 3M Nextel  bags; 6) three water-cooled heat exchangers; 7) an  induced
draft   (ID)   fan;  and  8) system   instrumentation.   Individual   component  design
and operating   characteristics for  several  components are presented in the  following
                                        32-2

-------
paragraphs.    Figure  1  is  a  conceptual  illustration  of  the baghouse  slipstream
system.    Particulate  sampling  and  analytical   procedures   used  during  the  bag
durability test program are also presented.

UND Steam Plant Boiler

The  UND  steam  plant  boiler  (Boiler #7)  is  a  spreader-stoker  boiler designed  to
produce 33,000  kg/hr  (72,750  Ibs/hour)  of  steam at 180°C  (355°F) and  896  kPa (130
psig) while  firing a North Dakota  Lignite.   This boiler was  designed  and  built by
Zurn Industries,   Inc. in  1978.  Typical operation  of this boiler  results in  load
swings from  9,080  to  45,400 kg/hr  (20,000  to  100,000  Ibs/hour) of  steam  to support
the  University's heating  and cooling systems.   This  boiler typically operates from
October 1 to May 15 each year.

Particle Knockout  Device

A particle knockout device was installed upstream of  the baghouse to remove +147 um
(+100  mesh)  ash  and  uncombusted  carbon particles  from the  flue  gas  stream.
Preliminary  particulate  sampling at the  location in  boiler #7 from  which  the flue
gas  slipstream was to  be  removed  indicated  that 40% to  45%  of  the +147  ym (+100
mesh) particles were carbon on a mass basis.  Removal  of these particles upstream of
the  baghouse was  deemed  necessary  to  minimize the potential  for carbon  burnout in
the  baghouse.    Installation  of the  particle  knockout  device  reduced  the  carbon
content of  the fly ash entering the baghouse from  about  25% to 30%  down  to <15% by
weight.

Pneumatic Ash  Injection System

The  pneumatic  ash  injection system used to introduce  supplemental  ash into the flue
gas  slipstream consisted  of two  primary components,  the  volumetric  feeder and  the
air-operated vacuum pump.   Ash was  injected into  the flue gas slipstream immediately
downstream   of  the  particulate  knock-out  device.   An   Acrison  Inc.  Model  105
volumetric feeder  was used.

An air-operated  vacuum pump,  Model  TDHR 380M  manufactured by AIR-VAC Inc., was used
to disperse  fly ash particles  into  the  flue gas slipstream.  The air-operated vacuum
pump employed  compressed  air  entering  through  an annular orifice,  which resulted in
a straight-through vacuum passage  and  permitted  solids  to pass directly through the
pump with no directional  change or  reduction of vacuum flow.

Baghouse

The  baghouse for the  slipstream system was designed and built by Donaldson Company,
Inc.,  and  is  illustrated  in  Figure 2.   Materials  of  construction  are primarily
stainless  steel due  to  baghouse   operating  conditions.    Baghouse  design permits
operation  at temperatures of  up to 538°C (1000°F) and flue gas  flow rates of  19.2
m /min  (-680 acfm).   The baghouse is  a  single  compartment housing  nine  bags,  each
0.15 m (0.5  ft) in diameter by 2.4  m (8  ft) long.  Total bag filtration area  is  10.5
m  (113 ft ).

Access to the  bags and perforated  stainless steel  cages is  through a  0.70-  by  2.65-m
(2.3- by 8.7-ft) door  in  one wall  or by removing  the clean air plenum  at the  top of
the  baghouse.   Electric preheaters  are  located just above the  baghouse  ash  hopper to
prevent moisture condensation  during start-up and  nonoperational time periods.

Pulse-jet cleaning can be triggered as a function of baghouse differential pressure
(AP) or time.   During  Phase I  and  Phase  II of the  test program, filter bag cleaning
was  performed  as a function of baghouse AP.  The  baghouse  pulse-jet  cleaning  system
is operated  by  a programmable  controller which controls cleaning frequency  and pulse
duration.    Clocks and  counters in the   controller  display total system  operating


                                        32-3

-------
time  and  test  time  in  hours,  total  cleaning  cycles,  and  test cleaning  cycles.
Filter bag cleaning occurs when  the  programmable  controller opens solenoid-operated
valves between  the pulse-air  manifold  and the three  filter bag blow-pipes.   Each
blow-pipe contains  three  pulse-air  orifices,  one located over  each  individual bag.
Figure 3 illustrates  the  pulse-jet component  configuration  for  the  baghouse.  Three
filter bags are cleaned simultaneously with a short  delay  between each set of three
filter  bags  to  allow  air  pressure  to  recover  in the  pulse-air  manifold.    Air
pressure in the manifold  is  nominally 414 kPa (60 psig) but  has  ranged from 310 to
483 kPa (45 to 70 psig).  At a nominal  pulse-air manifold   pressure of  414 kPa (60
psig), a pulse-air  volume of  approximately 0.04  m  (1.27  scf) is discharged to each
set of three bags during a cleaning cycle.

3M Nextel Filter Bags

Filter  bags  used  in  Phase  II  were seamless  tubes.    The  Nextel  fabric  used  to
construct  the  seamless  tubes  was  a  5H  3/0  30x37.5  material  woven with  yarns,
consisting of  8-ym filaments.   Fabri?c  basis  3weight a/id permeability  were 410 g/m
(17.3 ounces/sq yd)  and 3.4 m /min-m  (11 ft /min-ft  ), respectively.   During both
phases, filter  bags were  mounted on rolled,  perforated, stainless  steel  cages (60%
open  area).  Fabric was carefully  drawn  over  the cages and the bags were clamped at
both  the  top  and   bottom.    Enough  slack  in  the  longitudinal  direction  was
deliberately left in  the bags to provide "snap" during filter bag cleaning and allow
for thermal growth  of the cages.

Instrumentation

Instrumentation  on the baghouse  slipstream  system  was limited  to monitoring  and
recording temperatures, static and differential pressures,  cooling water flow rates,
ID fan vibration, and ID fan amperage.  Thermocouples (TC's) were located at fifteen
points   in  the   baghouse  slipstream  system  and   all   thermocouple   values  were
automatically logged  on a time basis.

Three  static  and  two  differential   pressures  were  measured  continuously on  the
baghouse  slipstream  system.   Static pressures  were measured  at the   inlet  to the
baghouse slipstream system, the  clean-gas  side  of the  baghouse, and at the location
where  annubar  differential   pressure  was measured.    Differential   pressure  was
measured  across  the  baghouse  and  downstream  of  the baghouse with  an  annubar.  The
annubar  differential  and   static  pressures   and the   flue  gas   temperature down-
stream  of the baghouse were used to calculate and control  flue gas flow rates.

Particulate Sampling  Procedures

Three types  of  particulate  sampling were employed  during Phase II;  EPA  Method 5,
multicyclone,  and real-time  sampling with the TSI  APS-33  system.  EPA Method 5 was
the primary sampling  technique used  for determining baghouse  efficiency and required
sampling  at  both the baghouse  inlet and outlet.   Inlet  particulate sampling using
EPA  Method 5  sample  trains required  a  thirty-minute to  one-hour  sampling period.
Particulate sampling  at the  baghouse outlet  generally  consisted of a two- to  three-
hour  sampling period with  a  few extended to six hours.    Wet  S02/S03 measurements
 (EPA  Method  8)   were periodically  conducted  in conjunction with  the particulate
sampling.    Orsat  analysis  of  flue  gas  for  oxygen, carbon  dioxide,  and  carbon
monoxide  concentrations  was   routinely  performed  in  conjunction  with particulate
sampling.

Real-time  sampling with  the  TSI APS-33  sampling system was performed to  generate
emissions  data  before, during, and  immediately  after  a cleaning cycle.  The  APS  33
laser particle  sizer, manufactured  by TSI  Inc.,  can count and  size particles  in  the
0.5 to  15 ym range.   The primary  advantages  of  this system  are the high  resolution
and  short sampling time.   A more  detailed  description of  the TSI APS-33  sampling
system and its  capabilities can  be found in previous publications (5,6,7).


                                         32-4

-------
Analytical Procedures

Ash  samples  were  analyzed  for  loss  on  ignition,  particle size  distribution,  and
major mineral  oxides.   Loss  on  ignition  was  performed using ASTM  Method C618-72.
Particle  size  distributions  were  determined  by  Coulter  counter and  a  dry  sieve
technique.   Concentrations  of major mineral oxides  (Al,  Si, Na, Mg, Ca,  P,  K,  Fe,
Ti, and S) were determined by x-ray fluorescence.   Scanning electron microscopy and
microprobe analysis  were used to inspect/analyze  outlet filters from  EPA Method 5
sampling trains to determine the  presence or absence of ash particles and adsorption
of vapor phase gas constituents  (S02/S03) as a check on calculated efficiencies.


RESULTS AND DISCUSSION

Phase  II  was initiated  in  October 1985 with installation  and conditioning of nine
woven  Nextel  filter  bags.    The  baghouse was  initially  operated  at  temperatures
ranging from  427° to 482°C (800°  to  900°F)  and an air-to-cloth  ratio  of  0.9  m/min
(3.0 ft/min).  The  baghouse cleaning  cycle was  controlled  as a function of baghouse
AP, with  a  trigger point of  1.0 kPa  (4 inches  W.C.).   Flue gas  particulate loading
in the  slipstream system was  continuously supplemented by  injection of additional
fly  ash to  maintain a particulate  loading at the  baghouse  inlet of  >3.4 g/m  (>1.5
gr/scf) (8).

This  paper  presents  Phase  II operational results  for the  performance/ durability
evaluation of 3M  Company's high-temperature Nextel filter bags.  As of May 15,  1989,
a  total   of  16,877  hours of  operating  time  had  been accumulated, exceeding  the
16,000-hour  effort originally  planned.   Data  presented for  Phase  II  include  ash
analysis,  a  review  of  system  operating conditions,  and  results from particulate
sampling.

Ash Analysis

Ash  samples  collected for analysis during Phase  II  fall  into  three  categories:   1)
ash  collected  in  the hopper of  the particle knockout  device;  2) ash collected from
the  supplemental  ash  feeder;  and  3)   ash collected  in the baghouse  hopper.   Ash
collected  in  the  hopper of the  particle knockout  device  during  Phase II was coarse
and  contained  unburned   carbon   particles.   Samples  collected   from  the  particle
knockout  device   showed   that 50%   to  65% of  the  material  on a mass   basis  was
larger than  110  ym  (140 mesh).    LOI  data for  the particle-knockout-device samples
show that unburned carbon in the  ash amounted to about  5% to 9% on a mass basis.

Supplemental ash  injection  was used during most  of  Phase  II to  increase total mass
loading at the baghouse  inlet.   Samples  of ash were collected from the feeder system
periodically for  Coulter counter, LOI, and XRF  analyses.   The supplemental ash used
during  Phase  II  was  collected  from  the  pulse-jet  baghouse of  a  pilot-scale
atmospheric fluid-bed combustion  system  (AFBC), the reverse-gas  baghouse of a pilot-
scale pc-fired combustion system, and  the hoppers of particulate control devices on
full-scale pc-fired units.

Figure  4  presents  Coulter  counter data representing  fourteen  samples.   The data
indicate  that  ash particle  sizes ranged from a low of  1 to 2  ym to  a maximum of 80
vim.  Mass median  diameters (MMDs)  ranged  from about 3 to 40  ym.  The unburned carbon
content of the  Q^.I ranged from  about  1% to 5%  on  a mass basis.   X-ray  fluorescence
analysis  showed that  the  chemical  nature  of the ash, on a mass basis, covered a wide
range for oxide constituents such  as silica (20% to 60%), alumina (10%  to  25%), iron
(3% to 14%), calcium  (8%  to 35%),  and  sodium (0% to 5%).  The variability  of the  ash
material  used  for supplemental ash injection resulted  from the  firing  of  different
fuels  (lignite,   subbituminous,   and  bituminous),  bed materials  (silica  sand   and
limestone), and   percent  ash   recycle  in  the AFBC pilot-scale system, and  the  use of
                                         32-5

-------
ash material collected  in  the participate control devices  (baghouses  and ESP's) of
pilot- and full-scale pc-fired combustors.

Ash samples were also collected from  the  baghouse hopper for characterization using
Coulter counter,  LOI,  and  XRF analysis.   Coulter counter data  for  fifteen samples
are presented in Figure 5.   The data indicate that ash collected in the baghouse had
a particle size that ranged from a low of 1 to 2 ym to a maximum of 80 ym.  From dry
sieve  data,  particles  as  large  as 246  ym were  collected  in the baghouse hopper.
Since data  indicated that  ash particles  from supplemental   ash injection  did not
exceed 80 ym, the bulk of the ash  material collected in the baghouse  hopper having a
particle  size  of  >80  ym  must  have come  from  the stoker-fired  boiler.    LOI  data
showed that the baghouse hopper ash contained 0.5% to  5.6% unburned carbon on a mass
basis.

X-ray  fluorescence data showed  that  the  baghouse hopper  ash  was  similar  to  the
feeder  ash  except for  the quantities of  calcium and  sodium  oxide.   Baghouse  ash
contained more  calcium  and sodium  oxide  than was observed  in the feeder  ash  on a
given  day,  but remained within  the  ranges  stated above.   The higher calcium  and
sodium oxide content of the baghouse hopper ash was a  direct result of the ash being
generated in  the  stoker-fired boiler.   The fuel  fired  in  the  stoker  boiler was a
lignite  known  to  result  in  an alkaline  ash with  high  calcium oxide  and  moderate
sodium oxide content.

Figures  4  and  5 contain particle  size data for PFBC  ash from the Grimethorpe PFBC
facility  (9).   Figure  4 compares  particle  size data  for   ash  collected    in  the
tertiary  cyclone   at the   Grimethorpe facility  with  feeder ash data  from the  bag
durability  test.   Although the Grimethorpe data  falls  within  the  range of particle
size  data observed during  the  bag  durability test, it falls  near the low end of the
particle  size  range.   This  indicates that  most of  the  fly ash injected  into  the
baghouse slipstream  system at  the  UNO  steam plant was larger than the ash particles
entering  the filter  elements tested  at  Grimethorpe.    Figure  5 compares  primary
cyclone  ash   from the  Grimethorpe facility with  ash  collected  in   the  baghouse
hopper  during  the  bag  durability  test.   In this  case, most  of the ash collected in
the  hopper  of  the baghouse at the  UNO steam  plant  had  a smaller particle size than
the  ash collected in the  primary  cyclone at Grimethorpe.   In general,  the various
ash  samples  were  chemically  similar.   The most  significant  difference noted was the
high  loss-on-ignition value (22%)  for the Grimethorpe  primary cyclone ash.

Baghouse Slipstream System Operation

The  baghouse slipstream  system operated quite well during the Phase II evaluation of
the  Nextel  filter  bags.  With the  exception of a temporary vibration problem in the
ID  fan, all  system  shutdowns exceeding  a few  hours  were  a  result  of operational
problems with the  stoker-fired boiler.  Temporary  (two to four hours) shutdowns were
required periodically (once every six to eight weeks)  to clean ash deposits from the
boiler  probe  in  order  to maintain  control  over  flue  gas flow  rates.    Periodic
replacement  and  maintenance  of system  instrumentation  was  also required.   Table 1
briefly  summarizes Phase II operating data.

Figures  6,  7, and  8 present examples of baghouse temperature, differential pressure,
and  A/C ratio, respectively,  as  a function  of  run  hour.   Baghouse  temperature is
reported as  a me^n baghouse temperature.  This value was determined by averaging the
values  for  up  -,,_ seven  thermocouples located  in the  baghouse.    Fluctuations in
baghouse temperature during Phase  II were primarily due to boiler load swings.  More
dramatic changes  in  baghouse  temperature  were the result of adjustments to flue gas
flow  rate and temporary  system shutdowns for  probe cleaning or instrument repairs.

Baghouse  differential   pressure  data  is  presented as  a  function  of run  hour in
Figure  7.    The  figure   indicates  very clearly  the  time periods  during  which  the
cleaning  cycle  trigger   point  was   set   at  1   kPa  or  2 kPa  (4" or 8" W.C.).


                                         32-6

-------
Differential pressure spikes were  a  result  of  either substantially reduced flue gas
flow  rates,  temporary  system  shutdowns,  or  excursions  due to  an  improperly  set
cleaning cycle  trigger mechanism.   The Nextel  filter  bags experienced a  total  of
3,441 cleaning  cycles during  the  16,877 hours  of  operating time, resulting  in  an
average of  one  cleaning  cycle every 4.9 hours.   Actual  pulse  cycles  experienced  by
the bags totaled  5,827 due  to  an occasional  need for multiple pulse cycles to reach
the  lower  differential  pressure set-point  limit  and pulse  cycles  resulting  fro^
several  days of cold-flow testing.

Figure 8 presents A/C ratio data as a function of run hour.  Spikes toward  lower A/C
ratios resulted  from temporary  system  shutdowns.   Spikes  toward  higher A/C ratios
were a result of over-correction by the flow control valve.  For the most part, flue
gas flow rate through the system was stable.  Those periods of time during  which A/C
ratio  varied over   a  broad  range directly  resulted from boiler  load   swings  in
response to fall and spring weather conditions.

Flue  gas analysis during the Phase  II  filter  bag  durability  test  showed  oxygen and
carbon dioxide  levels  to  be 3% to 13% and  5% to 12%, respectively.  Carbon monoxide
was  detected  in the flue gas  at levels of  up to 1%.  Sulfur  dioxide values ranged
from  100 to 840 ppm.  Although S02 values were low (<400 ppm) during the first three
years of operation,  they  increased during the fourth year to values ranging from 600
to  840  ppm.   A corresponding  increase  in S03  was  also observed, 0.5-7  ppm to 18-30
ppm.

Baqhouse Efficiency

Particulate mass  loading  at the  baghouse inlet due to boiler fly ash was roughly 1.0
to  2.0  g/m     (0.4 to  0.9 gr/scf).    Therefore,    with   measured   mass  loadings
typically ranging from 3  to 10 g/m  (1.3 to 4.4 gr/scf)  the bulk of the particulate
entering   the  baghouse  was a  result  of supplemental  ash injection.   Variability
observed  for mass   loading  at  the  baghouse inlet  was  due to  variability  in  the
supplemental ash  injection  feed  rate and swings in boiler  load.   Values for overall
particulate  collection efficiency   ranged  from  99.66%   to 99.99%  with  most  values
>99.9%.  Those  values that  were  <99.9%  corresponded to sampling periods during which
supplemental ash  injection  was not used, resulting in low  inlet mass loading values.

EERC  personnel  have  speculated  that the particulate  collection efficiency  of the
Nextel filter bags  was actually better  than the EPA Method 5  data indicated.  This
speculation  was  based on the fact that the  outlet filters were consistently clean.
SEM  micrographs  of  the outlet filters  showed  very  few particles.   In an attempt to
prove  that  actual  mass  collection  efficiency  was  better  than  indicated,  an EPA
Method  5  sample  train  was modified  to make  use  of two  filter  holders/filters  in
series.  This modified sample train was used on  three separate occasions in February
and March 1989.

The objective of this approach to particulate sampling at  the baghouse outlet was to
determine  the percentage  of the  filter weight  gain that  could  be  attributed  to
adsorption  of  flue  gas  vapor phase  constituents.   Since  the  primary  filter would
trap  virtually  100%  of  the solid particles  entering the  sample  train,  the weight
gain  of the backup filter  would be  the result of surface adsorption of vapor phase
flue  gas constituents such  as S03.

The  modified EPM Method  5  sample  train was  first  used  on February  22,  1989.  The
observed weight gain for  the primary and backup  filter was  0.01618 and 0.01395 grams
(0.24947 and 0.21509 grains), respectively.   Particulate  collection efficiency was
calculated  based  on the weight  gain  of the primary filter and  found to be  99.96%.
When  the  weight  gain of  the  backup  filter was  subtracted  from  the  primary, the
overall mass  collection  efficiency increased from 99.96%  to >99.99%.  On two  other
occasions, March  1  and 15,  1989, similar results were observed.
                                         32-7

-------
SEM  microprobe   analysis  of   the   filters  was   performed   to   document  major
constituents.  Table  2 presents the results  of  the analysis along  with  data for a
blank  filter.    Results  of  the  analysis  show  that the  concentrations  of sulfur
species on the primary and backup filters are significant when  compared to the blank
filter  and  similar when  compared to  each  other.   Therefore,  the  data presented
support the  conclusion that  the overall mass  collection efficiency  of  the Nextel
fabric filters was better than EPA Method 5 data indicated if the weight gain due to
sulfur  (S02/S03)  species  is  ignored.   Similiar results were reported  for a fabric
filter study involving a pc-fired pilot-scale system (10).

Particle  size  distribution  at the baghouse  inlet was determined  using a five-stage
multicyclone  sample  train.    A  total   of   19  multicyclone sampling periods  were
completed.   In all  but one  case, roughly 80% to 95% of  the total  mass collected was
captured  in  the  first  cyclone.   The  one exception resulted in  69% of the mass being
collected in the first cyclone.  The D50 value for the first cyclone ranged from 6.1
to 8.4 ym.

Fifteen  sampling periods were  completed with the  TSI  APS 33  system from November
1985 through June  1989.   Figure  9 presents  mass emissions as a function of sampling
time  for  one data  set generated with  the  TSJ  APS 33  System.   The  data set shows
emissions to be  about 0.01 mg/m3  (4^4  x 10"   gr/scf)  with spikes  due  to cleaning
cycles approaching 1 mg/m  (4.4 x 10"  gr/scf).

Overall,  the APS data  indicated  that  baghouse particulate collection efficiency was
better  than  shown by  the EPA Method  5 data by  as much  as one  and  two orders of
magnitude.    EPA  Method  5  and  APS  sampling  were  not   performed  simultaneously,
therefore, a direct  comparison of the  data  is  not  possible.   Also,  the  TSI APS 33
system  only   measures particles   in the 0.5 to  15.0 ym   range.   Particles larger
than  15.0 ym and agglomerates of smaller particles  that may have contributed to the
mass  emissions measured by  EPA Method 5 would not have  been measured by the TSI APS
33 system.

Mass  emissions data generated with the  TSI  APS  33 system were  also used to evaluate
the effect of emissions spikes due to on-line cleaning.   Six different APS data sets
were  selected and  used   for  the evaluation.   An  overall  mass  emissions  rate was
determined   by   estimating  a   baseline   mass  emissions   rate  from  the   data  sets,
calculating  total  mass for  a  specific  time  internal, adding the  emissions spike to
the  total emissions  value,  and  recalculating an  average  mass  emission rate.  Based
on  t£is  approach^ mass emission^  were  determined to range from  0.00895 to 0.69132
mg/m   (3.9  x 10"   to  3.0 x  10"  gr/scf) for the  six data sets.    The two order of
magnitude range  of  the  data  resulted from  a  wide   range of   baghouse operating
conditions   and   the   use  of  three  different  ash types  for  supplemental  ash
injection.   The contribution of the emissions spike  to total  mass  emissions was
observed  to  range  from  15^o  to  55%.    Generally,  when  the  baseline mass emissions
value  was low (<0.01  mg/m  )  the percentage of total mass emissions contributed by
the cleaning cycle spike  was  high (>50%).

Condition of 3M  Nextel Filter Bags

During  the  Phase II effort,  three Nextel filter  bags were removed for physical and
chemical  characterization by 3M Company.   One  Nextel filter bag was  removed at
1,836,  a second at 4,433,  and  a third  at 8,616  hours.   Figure  10 indicates the
location  in  the  ^ghouse  from which  each of the three  bags were removed.  The  first
filter  bag   was  removed  from  position  "D"  to  permit  inspection  of  the filter bag
surface  for  fused ash  deposits  similar to  those  observed in  the  Phase  I test.  A
thorough  inspection of the  filter bag revealed  no deposits.  The second   filter bag
was  removed  from position "C" to  determine  if  any  moisture from the compressed air
system  was   entering  the  baghouse  during  pulse-jet   cleaning  cycles.     Fabric
discoloration  was  observed  on  the  upper   portion  of   the filter  bag  located in
position  "C" following completion of the Phase  I test and  was suspected to have  been

-------
caused by moisture.   No  discoloration  was  noted  during  inspection of the filter bag
removed after  4,433 hours of operation.   Selection of position  "E"  for  removal  of
the third filter bag at 8,616 hours was a random selection.

Each time a  filter bag was removed and  at  the conclusion  of  each operating period,
the  remaining  Nextel  filter  bags  were  visually  inspected.    Observations  made
include;  1)  the formation of a  few  folds  or creases in  the  bags,  although not  as
severe as those  observed  following  completion of  the Phase I  test, 2) the formation
of ash  deposits  in the top of the  baghouse  above  the filter  bags,  3) the formation
of ash  deposits between  filter bags in positions  "D,  E,  G,  and H", and 4) lightly
brushing of the fabric surface removed the dust cake present,  revealing a filter bag
surface that looked as new as it did the day it was installed.

Upon removal  from the baghouse,  each  bag  was gently brushed/vacuumed clean.   As a
result, EERC personnel observed two distinct layers of ash  on  the surface of each  of
the  nine  bags  installed  at  the  beginning  of the  Phase  II  operating period.   The
outer layer was  gray   in color  and thicker than  the inner  layer.  The reddish inner
layer was  a  thin, residual layer  of  ash resulting from precoating  the  bags with a
high-calcium  ash  material at the  beginning  of  Phase II.   Overall,  the  nine Nextel
filter bags removed from  the baghouse after completion of Phase II operation were  in
excellent condition with  no visible damage observed.


CONCLUSIONS

To  expand  the  performance data  concerning  3M  Nextel  filter bags,  a set  of  nine
filter  bags  was evaluated on the  baghouse  slipstream  system.   As of  May 15,  1989,
the  system  had operated  for  a  total  of  16,877  hours,  resulting  in  3,441  cleaning
cycles.    Conclusions resulting  from  the  Phase   II  durability  test include  the
following:

   1.     Operability  of the  baghouse  slipstream  system  during  Phase  II   was
          comparable to  that experienced during Phase I.  Minimal operator attention
          was  required  and  instability  in  the  system  was  primarily a result  of
          swings in boiler load.

   2.     Particulate  collection efficiency, as determined  by  EPA Method  5 sampling,
          ranged from  99.66% to 99.99%.   In some cases  collection efficiency values
          were  >99.99%   when  EPA  Method  5 filter weight  gain  due  to  S02/S03
          adsorption was  ignored.

   3.     Baghouse  emissions  measured  using the  TSI  APS   33  system  indicate  that
          particulate  collection efficiency was at  least as good as the EPA Method 5
          data indicated.  Limited  data evaluating the effect  of emission spikes due
          to on-line cleaning, indicated that 15% to 55% of the total mass emissions
          observed  during an  operational  period  resulted  from  a  cleaning  cycle
          spike.

   4.     Operational  data and  visual   inspection  of  the  Nextel  filter  bags  both
          during and  after completion  of the Phase II  test showed no major  flaws or
          breaks  in the   fabric.   The use of perforated stainless steel  cages (60%
          open  area),  rather than wire  cages,  most  likely  contributed  to fabric
          durability.    Also,  infrequent bag cleaning  and moderate  pulse  pressure
          minimized the  stress experienced by the Nextel filters.

   5.     Although  creases were  again  observed  in  some  of the  filter  bags during
          Phase  II,  they were  fewer and  less pronounced than those observed at  the
          conclusion of  the Phase  I test.
                                        32-9

-------
   6.      Each of the nine bags  installed  at  the  beginning  of Phase II was found to
          have two  distinct   layers  of ash.   The  thin, residual  inner  layer was
          reddish in color and resulted from precoating the  bags with a high-calcium
          ash material prior  to  initiating operation in October  1985.   The purpose
          of the precoat was  to protect the surface of the Nextel  fabric from carbon
          particle  burnout.    But  the precoat  may  have  also  contributed  to the
          particulate collection efficiency and fabric durability observed.


ACKNOWLEDGMENTS

The  Energy   and  Environmental  Research  Center  (EERC)  would  like  to express its
appreciation  to  the  Electric Power  Research Institute  and 3M  Company  for  their
support of  the  program.   Special recognition  is  due Mr. Steven  G.  Drenker and Dr.
Owen J. Tassicker, EPRI; and  Mr.  Lloyd R.  White,  Mr. Don O'Brian,  and Mr.  Stephen M.
Sanocki, 3M  Company  for  their technical  support and  guidance.   Recognition is also
due the Westinghouse Electric Corporation; specifically, Mr.  T.E.  Lippert  for his
technical  assistance.   Donaldson Company  is recognized  for  its  assistance with the
design  and   construction  of   the baghouse  and baghouse control  system.    Special
recognition  is also  due  Mr.  LeRoy Sondrol for making  the UNO  steam plant available
for this  program,  and  Mr.  Ray  Tozer Jr.  and the  UND steam  plant  personnel for
providing assistance with the day-to-day monitoring of system operation.


REFERENCES

 1.   Lippert, T.E.  and  Ciliberti,  D.F.,   "Evaluation of Ceramic Fiber Filters for
      Hot Gas Cleanup  in Pressurized  Fluidized-Bed Combustion Power  Plants."  EPRI
      CS-1846, Topical Report May 1981.

 2.   Lippert,   T.E.  and  Ciliberti,  D.F.,   "Gas   Cleaning  Technology   for  High
      Temperature, High  Pressure Gas  Streams."  EPRI CS-3197, Annual  Report August
      1983.

 3.   Lippert,  T.E., Ciliberti,  D.F-  and Drenker,  S.G.,   "Hot   Gas   Cleaning for
      Pressurized  Fluidized-Bed  Combustion,"  j_n  Proceedings  of  the EPRI  Second
      Biennial Pressurized Fluidized-Bed Combustion Power Plants Utility Conference,
      Milwaukee, Wisconsin,  June 18-20, 1986.

 4.   White,  L.R., Forester,  R.J., O'Brien, D.L.  and Schmitt, G.A., "Ceramic Fabrics
      for Filtration at  High  Temperatures,  550°-1600°F,"  In  Proceedings of the EPRI
      Second  Biennial   Pressurized  Fluidized-Bed  Combustion  Power Plants  Utility
      Conference, Milwaukee,  Wisconsin, June 18-20, 1986.

 5.   TSI Incorporated,  APS 33 Aerodynamic Particle  Sizer  Instruction Manual,  p. 2-
      4, 1983.

 6.   Miller,   S.J.,   and  Laudal  D.L.    "Real-Time   Measurement  of  Respirable
      Particulate Emissions From a  Fabric Filter,"  16th Annual Meeting of the Fine
      Particle Society, Miami Beach, Florida,  April 22-26, 1985.

 7.   Chen,  B.T.,  Cheng Y.S.,  and  Yeh  H.C.,  "Performance of  a  TSI Aerodynamic
      Particle  Sizer,"  Aerosol  Science  and  Technology, Vol.  4,  No.  1,  pp. 89-97,


 8.   Weber,  G.F.,   "Performance/Durability   Evaluation  of   3M  Company's   High-
      Temerature  Nextel  Filter  Bags," Final  Project  Report  submitted  to  EPRI  under
      EPRI Contract  1336-16,  February  1990.  (Pending Publication).
                                        32-10

-------
 9.   Yung, Shui-Chow,  "Properties of  Ashes Resulting  from Pressurized  Fluidized-Bed
      Combustion,"  Prepared  for  Electric  Power  Research  Institute,  Contract  No.
      RP1336-5,  November 15,  1986.

10.   Miller,   S.J.,      "Flue   Gas   Conditioning   For   Fabric   Filter   Performance
      Improvement,"   Final  Project  Report  for work  performed under DOE  contract No.
      DE-AC22-880C 88866, Grand Forks, ND, December  1989.
         System Inlet
                                 Fly-ash
                                 Feeder
                                       Vacuum Transducer
                                 Knockout
                                 Hopper
                                  Ash Collection
                                    System
                                                Inlet        ~ PU|SB Air
                                              Sample Port 1     Q Reservoir
                        Spreader
                        Stoker
                        Boiler
Knife Valve TJ
» 	 ^

T
Heat
xchanger


TC.
7

TC.

1_
agnouse ^-"^^
y-Pass I




TC
"•
TC

TC


w
•^K_

=TN
DP f
TC
6

TC
0
{ Expansion Joint
Heal
Exchanger
Outlet
Sample Port 2
Heat Exchanger
\ \^ SP DP
Strip Heaters
_j 	 ^ Ash Collection
^ System
                                                                             ! I D Fan Damper Valve
                          Figure 1.   Baghouse  Slipstream System.
                       Pneumatic Ash

                      Conveying System
                                                      Pulse System

                                                  Compressed Air Manifold
                                                        9-3M Nextel Bags On Rolled

                                                       Perforated Stainless Steel Cages
                                                     	Hinged Access Doo
                      Figure  2.   High-Temperature Pulse-Jet Baghouse.
                                              32-11

-------
                                                                     Expansion Joint
                                                                     I D Fan Damper Valve
                                                                    - TC 12
                                                      Ash Collection
                                                        System
       Figure  1.   Baghouse  Slipstream  System.
                                       Pulse System
                                   Compressed Air Manifold
 Pneumatic Ash

Conveying System
                                           -3M Nextel Bags On Rolled
                                        Perforated Stainless Steel Cages
                                         Hinged Access Door
 Figure  2.   High-Temperature Pulse-Jet Baghouse.
                             32-12

-------
      Figure 3.   Baghouse  Pulse-Jet Component Configuration.
    99 9
    99 8
    99 5
     99
 c   8C
 o
 jE   70
 i   60

 O   40
 5   30
 o
 ""   20
    0 5
    0 2
             	Gnmethorpe Tertiary Cyclone
      01     02  030405    10
                                 2   345
                            Particle Diameter 1pm)
Figure  4.   Coulter  Counter Data  for Ash Samples Collected
            From the Supplemental  Ash Feeder.
                                 32-13

-------
     99 9
     99 8
     99 5
      99
      95
      90

      80
      70
      60
      50
      to
      30
      20

      10
       5

       2
       1
      0 5
      0 2
	Gnmethorpe Primary Cyclone
        01     020304050810      2   345   8 10      20  30  40 50   80 100
                                Particle Diameter (^m)

       Figure 5.  Coulter Counter Data for Ash  Samples
                   Collected From the  Baghouse Hopper.
 600
 500 -
 400 -
 300 -
 200 -
 100 -
    0.0
                                                                       -1,000
                                                                       -800   2
                                                                       -600
                                                                       -400
                                                                       -200
                              2.0
                          Run  Hour (Thousands)
Figure  6.  Baghouse Mean Temperature  Versus  Run Hour  for Phase II.
                                   32-14

-------
      0.0
      0.0

                                                       4.0
                            Run Hour (Thousands)
Figure 7.   Baghouse Differential  Pressure Versus Run Hour  for Phase II.
                            Run Hour (Thousands)
  Figure 8.  Baghouse  Air-to-Cloth Ratio  Versus Run Hour for  Phase II.
                                   32-15

-------
   10 -
E
en
E
en
O 10 -'-E
en
en
en
D
  10
        December  15,  1!

— • —  Total  Mass

— o —  Respirable Mass


AIR/CLOTH  =  1.3 m/min

BH TEMP. = 390 C

PC ASH FROM  EERC  PILOT COMB.
                                               E-10
      0            100           200           300
             Test Time  (minutes)
  Figure 9.  Emissions Data Generated Using the TSI
            APS 33 System During Phase  II.
                                                      en
                                                      c
                                                      o
                                                      en
                                                      C
                                                      O
                                                      en
                                                      en
                                      UJ
                                      en
                                      en
                                      o
       HOT GAS
         INLET
                              HOT GAS
                              OUTLET

A
16,877 Mrs
15,041 Hrs
D
1,836Hrs
G
16,877Hrs

B
16,877Hrs
8,261 Hrs
E
8,616Hrs
H
16, 877 Hrs
12,444Hrs
C
4,433Hrs

F
16,877 Hrs

16,877 Hrs
 Figure 10.  Perspective of Individual Bag Location
            for  the Phase II  Durability Test.
                          32-16

-------
                                      TABLE 1

                         SUMMARY OF PHASE II  OPERATING DATA
         Bag  Material
         Filter Temperature
         Filter Face Velocity
         Operating A P
         Operating Hours Completed
         Filter Cleaning Cycles
         Filter Pulse Cycles
         Inlet Dust Loading
         Filter Performance
           Outlet Dust Loading
           Collection Efficiency
                             Nextel  5H  3/0  30x37.5
                             440°C (260 to  540°C)
                             1.8 m/min  (0.5 to  2.4  m/min)
                               1.4 kPa  (0.6 to  1.7  kPa)
                             16,877
                              3,441
                              5,827
                              5.7 g/m3  (3 to 10 g/m3)


                              0.0022 g/m3  (0.0002 to  0.006  g/m3)
                             99.95%  (99.66  to 99.99%)
                                      TABLE 2

                  SEM MICROPROBE ANALYSIS OF DUST LOADING FILTERS
                     Na,0
Percent Concentration By Weighta
A1203          Si02          CaO
Blank Filter
                                                            SO,
21.0 ± 0.1     0.9 ± 0.1    73.0 ± 0.1     5.0 ± 0.1    0.00 ± 0.0
Primary Filter5   19.4 ± 0.2     1.4 ± 0.7    72.2 ± 1.6     5.6 ± 0.1     1.4 ± 0.6
Backup Filterb    20.1 ± 0.7     1.8 ± 0.2    72.0 ± 0.2     5.3 ± 0.3     0.8 ± 0.1

a  Reported values are averages of two analyses.
5  Filters used during sampling at outlet of baghouse slipstream system.
                                        32-17

-------
          PARTICLE CONTROL IN ADVANCED COAL-BASED POWER GENERATION SYSTEMS
                          S. J. Bossart  and  C. V. Nakaishi
                             U. S. Department of Energy
                         Morgantown Energy Technology Center
                                    P.  0. Box 880
                        Morgantown,  West Virginia  26507-0880
ABSTRACT

The Morgantown Energy Technology Center of the U.S. Department of Energy is actively
sponsoring a variety of research projects aimed at developing improved electric-
power generation systems that use coal.  The most promising of the advanced coal-
based power generation systems are pressurized fluidized-bed combustion combined-
cycle,  integrated gasification combined-cycle, and direct coal-fueled turbine.  An
essential feature of each of these systems is control of particles at high-temper-
ature and high-pressure (HTHP).  Particle control is needed in these systems to meet
environmental regulations and to protect major system components such as the gas
turbine.  Incorporating HTHP particle control technologies into these advanced power
generation systems can increase plant efficiency, lower plant capital costs, lower
the cost of electricity, reduce wastewater treatment requirements, decrease gas tur-
bine maintenance requirements, and eliminate the need for post-turbine particle con-
trol to meet New Source Performance Standards for particle emissions.  The ceramic
cross-flow filter, ceramic candle filter, screenless granular-bed filter,  acoustic
agglomeration,  and nested fiber filter are being developed as HTHP particle control
technologies for advanced power generation systems.  Each of these technologies has
successfully demonstrated the ability to reduce particle loading and particle size
to levels acceptable for meeting environmental regulations and protecting major
system components.  In general,  the remaining technical issues require engineering
solutions to enhance the performance reliability and economic advantages of HTHP
particle control technologies.  If the technical issues can be resolved, one or more
of these technologies can realize its commercial potential for incorporation into
advanced coal-based power generation systems.
                                        33-1

-------
          PARTICLE  CONTROL IN ADVANCED COAL-BASED POWER GENERATION SYSTEMS
INTRODUCTION

The Morgantown Energy Technology Center of the U.S. Department of Energy is sup-
porting research to develop advanced power generation systems that will utilize the
large coal reserves available in the U.S.  Some of the more promising of the ad-
vanced coal-based power generation systems are depicted in Figure 1:  pressurized
fluidized-bed combustion (PFBC)  combined-cycle, integrated gasification combined-
cycle (IGCC),  and direct coal-fueled turbine (DCFT).   A key aspect in each of these
systems is control of particles  at high-temperature and high-pressure (HTHP).   Par-
ticle control technologies  are incorporated into advanced power generation systems
at operating conditions that match the temperature and pressure of the coal conver-
sion process (i.e., gasification or combustion) and the power generation block
(i.e.,  gas turbine).  The careful matching of operating conditions eliminates re-
guirements for expensive heat-recovery equipment and the avoids thermal efficiency
losses associated with fuel gas  quenching.  As a consequence of improved thermal
efficiency and simplified HTHP particle-control systems, the capital costs and cost
of electricity can be lower for  power generation systems that include HTHP particle-
control technologies.

HTHP particle-control technologies can reduce particle size and particle loading to
acceptable levels for meeting New Source Performance Standards (NSPS) for particle
emissions and to below tolerance limits for a gas turbine.  These technologies elim-
inate the need for large post-turbine particle control devices to meet NSPS.  The
elimination of post-turbine particle control technology is especially important for
IGCC applications, since the volumetric flow rate is increased roughly twenty-fold
downstream of the gas turbine.

Typical process conditions  and particle control goals for each of the advanced power
generation systems are shown in  Table 1.  The development of particle control tech-
nologies has focused on removing particles at pressures greater than 100 psig and
temperatures greater than 1,000  °F.   Particle loading and particle size  in the coal-
derived gas at the inlet of HTHP particle-control systems is substantially higher
than the particle control requirements to meet environmental regulations and to pro-
tect a gas turbine.  Although there is some disparity regarding particle tolerance
limits for gas turbines among manufacturers,  Figure 2 is a typical particle toler-
ance graph showing the relationship between particle loading and particle size.

The ceramic cross-flow filter, ceramic candle filter, screenless granular-bed fil-
ter,  acoustic agglomeration,  and nested fiber filter are the current leading candi-
dates for particle control  in the HTHP environment of advanced power-generation
systems.
                                       33-2

-------
CERAMIC CROSS-FLOW FILTER

The Westinghouse ceramic cross-flow  filter  element, mounting  arrangement,  and  system
assembly is shown in Figure  3.  The  ceramic cross-flow  filter element  is  constructed
of multiple layers of thin,  porous ceramic  plates that  contain ribs  to form gas  flow
channels.  Consecutive layers of the  ceramic plates are  oriented  such  that  the chan-
nels of alternating plates are at an  angle  of  90 degrees  ("cross-flow") to  each
other.  The current size of  a ceramic cross-flow filter  element is 12-in.  x 12-in. x
4-in.  Both sides of the 4-in. channels  are exposed to  the particle-laden coal gas.
One end of the 12-in. channels is sealed while the other  end  of the  long  channel is
mounted to the clean gas plenum.  The particle-laden  coal gas flows  through the
"roof and floor" of the porous ceramic plates  that comprise the short,  "dirty side"
channels.  Particles form a  dust cake on the ceramic  plate and the gas  flows through
the porous plates to the long, "clean side"  channels  and  finally  to the clean gas
plenum.  The dust cake on the "dirty  side"  channels is periodically  removed by
applying a reverse pulse of  high pressure gas.  The "clean side"  of multiple filter
elements is connected to a clean gas  plenum to form a filter  module.   A commercial-
size filter module may contain up to  40  filter elements on a  single  gas plenum.
Several gas plenums are supported on  a tubesheet in a single  pressure  vessel.  An
important aspect of the ceramic cross-flow  filter system  is its large  filtration
surface area-to-volume ratio.  A single  filter element has a  filtration surface area
of about 7 ft2  within a  volume of only 0.33  ft3.

A 15-element ceramic cross-flow filter system was tested at the New York University
(NYU) subpilot-scale PFBC facility.   Nominal test conditions  were temperatures of
1,300 to 1,500 °F,  pressures of 120 psia, and inlet particle  loadings of 250 to
1,060 ppmw.  For the first 50-hour period,  the ceramic cross-flow filter  system was
operated at a nominal filtration velocity of 5.2 ft/min with  a baseline pressure
drop of 35 in.  of water.  The particle loading at the outlet  of the cross-flow
filter system ranged from 3  to 9 ppmw.   As  shown in Figure 4, the outlet particle
loading and size distribution were sufficient to meet NSPS and particle tolerance
limits for gas  turbines.

For the next test,  the ceramic cross-flow filter system was operated at a nominal
filtration velocity of 10 ft/min with a  baseline pressure drop of 78 in. of water.
For the first 11 hours of testing, the particle loading at the outlet  of the ceramic
cross-flow filter was less than 14 ppmw, but increased to 103 ppmw during the next
19 hours.  Failure of solenoid valves  on the pulse system and failure  of a  sorbent
feed valve occurred during this test.  A post-test inspection of the ceramic cross-
flow filter system revealed  that the  major  cause of excessive particle penetration
was the failed dust seals that were installed between the filter body  and the
mounting flange.

The performance of the ceramic cross-flow filter under gasification conditions is
being evaluated at the Texaco subpilot-scale, entrained-bed gasification  facility in
Montebello,  California.   The test program is investigating the effects of reducing
gas and coal char on the filter material.  A ceramic  cross-flow filter system con-
sisting of four elements has accumulated approximately 250 hours of testing on the
Texaco gasifier.   The first  50 hours  of  testing were  performed in the  oxygen-blown
mode,  while the remaining tests were  conducted in the air-blown mode.  Typical oper-
ating conditions  were temperatures of  1,250  to 1,300  °F, pressures of  365 psia,
inlet particle  loading of 250 to 22,500  ppmw, and outlet dust  loading  of  5  to 114
ppmw.   The excessive particle loading  at the outlet of the cross-flow  filter system
was caused primarily by sealing failures similar to those experienced  during testing
at the NYU-PFBC.   The pressure drop across the cross-flow filter  system was about
three-fold higher than that  experienced during previous tests  under PFBC  conditions.
The higher pressure drop was probably  caused by smaller particles penetrating the
ceramic filter.
                                        33-3

-------
Two ceramic cross-flow filters have completed approximately 1,000 hours  of  a  planned
2,000-hour test under simulated PFBC conditions at the Westinghouse  Science & Tech-
nology Center.  A 2,000-hour test series using two filter elements has also been  in-
itiated under simulated gasification conditions.  These long-term tests  will  address
the durability and reliability of the components and materials used  in construction
of a ceramic cross-flow filter system.

There are several remaining technical issues and uncertainties that  must be resolved
before the ceramic cross-flow filter system can achieve commercial status.  The rel-
atively short-term test programs have not addressed long-term durability and  reli-
ability for many cross-flow filter system components.  Improved materials,  fabri-
cation methods, and mounting designs are needed to avoid problems such as chemical
degradation, thermal fatigue,  delamination, and dust seal failures.  Recent test
results have shown that mounting ceramic cross-flow filters in compression  prevents
catastrophic delamination of filter elements.  Previous test programs have  evaluated
performance of relatively small-scale cross-flow filter systems.  These tests  have
not addressed the cleanability of commercial-scale filter modules that may  contain
up to 40 elements.  Individual ceramic cross-flow filter elements can vary  widely in
terms of strength, permeability, and filtration characteristics.  More consistent
fabrication methods and a quality inspection program need to be implemented to re-
duce the variation in properties from one filter element to another.


CERAMIC CANDLE FILTER

A ceramic candle filter and system assembly are shown in Figure 5.   The external
surface of the candle is composed of a fine porous layer of ceramic  fiber,  clay, and
fine silicon carbide.  This fine porous layer provides a surface for forming  a par-
ticle cake layer and prevents particles from penetrating and blinding the filter
element.  A coarser,  inner layer of silicon carbide structurally supports the  fine
porous layer.  Typical ceramic candle filters are 3.3 to 5.0 ft long with an  inter-
nal diameter of 1.2 to 1.6 in. and an external diameter of 2.0 to 2.4 in. Multiple
candle filters are suspended from a tubesheet across the filter pressure vessel
through a number of counter-bored holes.  Gas flows from the outside to the inside
of the ceramic candle filter,  and the particles form a cake layer on the external
sur-face.  The clean gas is exhausted from above the tubesheet.  The particle  cake
on the external surface of the candle filter is periodically removed by applying a
reverse pulse of high-pressure gas.

A ceramic candle-filter system was tested under gasification conditions at the KRW
subpilot-scale fluidized-bed gasification facility.   Candle filters  3.3 ft and 5 ft
long were tested at the KRW fluidized-bed gasifier.   The test series involved  13
startup-shutdown cycles, 653 hours of exposure to coal gas,  and 2,253 pulse cleaning
cycles.  Nominal test conditions were temperatures of 800 to 1,150 °F,  pressures of
145 to 245 psia, filtration velocities of 1.5 to 5.2 ft/min, and inlet particle
loadings between 200 and 1,000 ppmw.   Particle loadings at the outlet of the  ceramic
candle filter system were 1 to 5 ppmw.   These ceramic candle filters are currently
undergoing evaluation to measure changes in strength and filtration properties after
exposure to the coal gas.

Despite the encouraging filtration performance of the ceramic candle-filter system,
numerous filter failures have occurred during operation,  or filter elements have
shown a significant loss in strength after the tests.  Failures have included  ther-
mal shock of the filter, possible chemical attack of the clay binder, mechanical
failures caused by a large "slug" of particles striking the filter elements,  dust
seal leakage, and improper operation of solenoid valves on the pulse cleaning
system.
                                       33-4

-------
In 1990, a ceramic patch-test unit will be installed at an operating  coal-based fac-
ility to evaluate the filtration properties of materials used  in constructing cer-
amic candle filters and ceramic cross-flow filters.  The effects of filtration velo-
city, particle loading, particle size distribution on pressure drop,  and filtration
characteristics will be studied during these tests.  In 1990,  Solar Turbines and
Allison Gas Turbines will evaluate ceramic candle-filter systems for  DCFT applica-
tions.  In 1990, Solar will also evaluate the potential of many ceramic materials by
exposing coupon samples in a DCFT.

The remaining technical issues with the ceramic candle filter  are similar to those
described for the ceramic cross-flow filter.  The relatively short-term test pro-
grams have not addressed the durability and reliability of many of the system com-
ponents in the candle filter.  Long-term tests are needed to determine the mechan-
ical integrity and chemical stability of the tubesheet, pulse  cleaning system, and
candle filters.  These tests will establish the long-term filtration  and pressure
drop characteristics of the candle filter system.

Improved ceramic materials, fabrication methods, and mounting designs are needed to
avoid problems such as chemical degradation, thermal fatigue, bridging, and dust
seal failures.  Alternative commercial-scale designs, such as a multiple-tiered
arrangement, need to be investigated to optimize the configuration of the ceramic
candle-filter system.  More consistent fabrication methods and a quality control and
assurance program need to be implemented to reduce the variation in properties from
one filter element to another.
SCREENLESS GRANULAR-BED FILTER

The Combustion Power Company, screenless granular-bed filter  (GBF) system is shown
in Figure 6.  The particle-laden coal gas enters the 5-ft diameter filter element
through the inner annulus of a central pipe submerged in filter media  (typically
0.08 or 0.12 in. diameter alumina spheres); the media move continuously downward
toward the cone section of the filter vessel.  Particles are removed by the filter
media as the coal-derived gas turns and flows upward through the filter media.   The
particle-laden media are withdrawn at the bottom of the filter element and are
transported pneumatically in a lift-pipe to a de-entrainment vessel where the filter
media and particles are separated.  The cleaned filter media flow by gravity to a
media reservoir located above the filter vessel.  The media flow to the main filter
vessel through eight distribution pipes and the outer annulus of the central pipe.
The particles leaving the de-entrainrnent vessel are removed in a pressurized bag-
house after the lift-pipe transport gas is cooled to 400 °F.   The lift-pipe trans-
port gas is re-pressurized in a boost blower before reuse to convey particle-laden
media up the lift-pipe.  In a commercial-scale system,  several filter elements will
be contained in a single pressure vessel.  A single media recirculation system will
be used to supply media to all filter elements.

The GBF was successfully tested for 164 hours during three test periods at the NYU-
PFBC facility.   A media size of 0.08 in. diameter was used during the first two test
periods while the 0.12 in. diameter media were used during the third test.  The fil-
tration velocity during the first two tests was 30 to 36 ft/min while the third test
was operated at a filtration velocity of 59 ft/min.  Tests were conducted at gas
temperatures between 1,550 and 1,600 °F,  pressures of 105 to 135 psia,  and inlet
particle loadings between 360 and 1,580 ppmw.  Outlet particle loadings were consis-
tently less than 20 ppmw and as low as 1 ppmw.

The benefit of  using 0.12 in. media during the third test was to permit the PFBC
exhaust gas flow through the filter bed to be increased by about 64 %.  Test results
indicate that the larger media were as effective in removing particles as the smal-
                                        33-5

-------
ler media.  The filtration velocity through the filter bed during the second  and
third tests was about 30 to 35 % of the minimum fluidization velocity.  The use of
larger media can result in substantial savings in capital costs for a commercial GBF
system because fewer filter elements will be required to process the same quantity
of PFBC exhaust gas.  Post-test inspection of the GBF system at the conclusion of
the NYU-PFBC test program revealed significant wear in the refractory-lined sections
of the lift pipe and the media injection-valve area.

Despite the successful testing performed, several technical problems and issues must
be resolved before a moving granular-bed filter system is ready to enter the market-
place as a viable particle control technology.  The major technical issue is  compat-
ibility of moving granular-bed filter technology to gasification, PFBC, and DCFT
systems because of weight, size,  and cost considerations.  More durable, inexpensive
liners are being evaluated for the lift-pipe, since a key issue for commercial-
ization of the GBF system is durability of the media circulation system.  An auto-
matic control system needs to be developed for the GBF system.  The control system
may require the development of specialized instrumentation,  including a media flow
monitor.  Key process variables such as media size,  shape,  and filtration velocity
require further optimization.

In 1990, there are plans to reduce the size,  weight, and cost of the moving granu-
lar-bed filter system by developing alternative design concepts.  Proposed solutions
for concerns with selected components will be evaluated in component test facil-
ities.  Once the technical issues are resolved for each GBF component, an integrated
GBF system will be tested at a coal-fired operating facility.


ACOUSTIC AGGLOMERATION

The concept of acoustic agglomeration is shown in Figure 7.   Acoustic agglomeration
is a pretreatment process for increasing the average particle size in a coal-derived
gas so that the particles can be efficiently collected in conventional cyclone sep-
arators.  An air siren or pulse combustor produces high-intensity acoustic waves
that generate an oscillating gas velocity as the wave travels through the particle-
laden coal-derived gas.  Small particles tend to move with the gas,  but large parti-
cles move slower than the gas.  The vibrating small particles collide with each
other and with larger particles to form agglomerates.  The effectiveness of agglom-
eration is dependent on the initial particle loading, initial particle size distri-
bution, sound intensity, sound frequency, and residence time.  The agglomerated
particles have been shown to be sufficiently robust to withstand the high shear
stresses encountered in separator cyclones.

Pennsylvania State University conducted several HTHP acoustic agglomeration tests on
re-entrained PFBC flyash using an acoustic 2000 W siren as the sound source.  Nom-
inal test conditions were temperatures of 800 to 1,500 °F,  pressures of 125 to 150
psia, inlet particle loadings  of 950 to 4,800 ppmw,  residence times of 3.5 to 4 s,
sound pressure levels of 152 to 157 dB, and sound frequencies of 820 to 910 Hz.  In
general, more than 30 % of the particle mass was transferred from the smaller par-
ticle sizes (< 5 (im) to the larger particle sizes (> 10 |0m) because of  acoustic
agglomeration.

Acoustic agglomeration upstream of conventional cyclones offers a potentially low-
cost and reliable particle control system for PFBC applications.  Little hardware is
required,  and no parts are exposed to the hostile HTHP environment when a pulsing
combustor is used as the sound source.  The major technical challenge is developing
a durable and cost-effective sound source.   Pulsing combustors are being developed
for possible use as energy-efficient,  cost-effective sound sources.
                                       33-6

-------
In 1989, an acoustic agglomeration  system will be tested  for  200 hours by  Solar
Turbines using a simulated PFBC exhaust gas.  This test will  demonstrate,  for the
first time, the concept of an acoustic agglomeration  system consisting of  a sound
source, agglomeration chamber, and  downstream cyclones.   In 1990, Manufacturing &
Technology Conversion International will evaluate acoustic agglomeration to enhance
particle and sulfur control in DCFT systems.


NESTED FIBER FILTER

The Battelle nested fiber-filter concept is depicted  in Figure  8 .  The nested fiber
filter consists of interlocking needle-shaped fibers.  The fibers can be made from
different alloys suited for particular applications.  The needle-shaped fibers have
a length-to-diameter ratio of about 100:1.  Filtration action occurs when  particles
form dendritic or chain-like structures on the nested fibers.   The major advantages
of the nested fiber filter includes high filtration velocities, high particle col-
lection efficiencies, low pressure  drop, and compact  equipment  size.  Battelle Col-
umbus Division conducted a series of HTHP tests using re-entrained AFBC and PFBC
flyash as feed material to a nested fiber filter.  Nominal test conditions were tem-
peratures of 740 to 1,220 °F,  pressures of 30 to 80 psia,  and filtration velocities
of 100 to 300 ft/min.  The particle loading at the outlet of  the filter was less
than 5 ppmw with a relatively low pressure drop of only 8 to  30 in. of water.   At
these low outlet particle loadings, the nested fiber  filter can meet NSPS  for par-
ticle emissions and particle tolerance limits for gas turbines.

The major technical obstacle facing the nested fiber  filter is  cleaning the filter
following excessive particle accumulation.  Rotation, mechanical vibration, acoustic
waves,  and pulse-jet flow are being evaluated as possible methods for cleaning the
nested fiber filter.


DEMONSTRATION TEST PROGRAM AT THE OHIO POWER COMPANY  TIDD-PFBC FACILITY

The Department of Energy and the American Electric Power Service Corporation (AEPSC)
plan to evaluate integrated engineering designs of selected advanced particle-con-
trol technologies using a l/7th flow slipstream at the 70-MWe Ohio Power Company
Tidd-PFBC facility.  Currently, the candidate technologies for installation and
testing at the Tidd-PFBC are the ceramic cross-flow filter, ceramic candle filter,
screenless granular-bed filter, and the ceramic tube  filter.  One or two of these
technologies will be tested over a 2-year period from January 1992 to January 1994,
coinciding with planned operation of the Tidd-PFBC under the  Clean Coal Technology
Demonstration Program.  Table 2 provides some of the  expected operating conditions
for these tests.  The results from the large-scale test program at the Tidd-PFBC
facility may be used to establish a design basis for  installing and operating a
selected,  advanced particle-control technology at the 330-MWe AEPSC Philip Sporn
plant.   The Philip Sporn plant is scheduled to begin  its test program in January
1996.   The hot gas cleanup tests at the Tidd-PFBC facility represent the first op-
portunity for the developers of HTHP particle-control technologies to evaluate inte-
grated  engineering designs of commercial-scale modules on an  operating advanced,
power-generation system.


CONCLUSIONS

The ceramic cross-flow filter,  ceramic candle filter, screenless granular-bed
filter,  acoustic agglomeration,  and nested fiber filter are promising particle
control concepts for potential application to advanced coal-based power generation
systems.   Each of these technologies has successfully demonstrated the ability to
                                       33-7

-------
reduce particle loading and particle size to levels acceptable for meeting environ-
mental regulations and protecting major system components.  The remaining  technical
issues require engineering solutions to enhance the performance reliability and  eco-
nomic advantages of HTHP particle-control technologies.  If the technical  issues^can
be resolved,  one or more of these technologies can realize its commercial  potential
for incorporation into advanced coal-based power generation systems.


BIBLIOGRAPHY

R.C. Bedick.   "High-Temperature Contaminant Control for Coal-Based Gas  Turbine
Systems," Presented at International Specialty Conference on Combustion and the
Environment,  March 8-10, 1989, Seattle, Washington.  Sponsored by the Air  and Waste
Management Association.

R.C. Bedick and V.P. Kothari, Editors.  Gas Stream Cleanup Papers From  DOE/METC
Sponsored Contractors Review Meetings in 1988.  October 1988, Springfield,  Virginia:
National Technical Information Service, DOE/METC-89/6099, NTIS/DE89000901.

S.J. Bossart.  "Advanced Particle Control Technologies for Pressurized  Fluidized-Bed
Combustion Applications," Presented at 51st Annual Meeting of American  Power
Conference, April 24-26, 1989, Chicago, Illinois.  Sponsored by the  Illinois
Institute of Technology.

P. Cherish and J.D. Holmgren. "Hot Gas Cleanup Process Demonstration of Ceramic
Filters for IGCC Power Generation," Presented at 1988 Seminar on Fluidized-Bed
Combustion Technology for Utility Applications, May 4, 1988, Palo Alto,  California.
Sponsored by the Electric Power Research Institute.

R.J. Dellefield and R.C. Bedick. "Evaluation of Three High-Temperature  Particle
Control Devices for Coal Gasification," European Federation of Chemical Engineering
Publication Series No. 52, Gas Cleaning at High Temperatures.  Sponsored by the
Institution of Chemical Engineers, Geo. E. Davis Building, 165-171 Railway Terrace,
Rugby, Warks, England, CV21 3HQ.

G.H. Koopma and G. Reethof.  Evaluation of Acoustic Agqlomerators for High-Pressure
High-Temperature Particle Control, Task 13 Final Technical Report. KOH  Systems,
Inc., for U.S. Department of Energy under Contract DE-AC21-85MC22012, March 1989
 (in press).

T.L. Lippert, G.B. Smeltzer, and J.H. Meyer.  Performance Evaluation of a  Cross-Flow
Filter on a Subpilot-Scale Pressurized Fluid-Bed Coal Combustor, Final  Report.
Westinghouse Electric Corporation, Science and Technology  Center, for  U.S.
Department of Energy under Contract No. DE-AC21-85MC22136, December  1989 (in press).

Morgantown Energy Technology Center.  Gas Stream Cleanup Technology  Status Report.
October 1987, Springfield, Virginia:  National Technical Information Service,
DOE/METC-87/0255, NTIS/DE87006493.

Morgantown Energy Technology Center.  Hot Gas Cleanup for Electric Power Generating
Systems.  May 1986, Springfield, Virginia:  National Technical Information Service,
DOE/METC-86/6038, NTIS/DE86006607.

G. Reethof.  Acoustic Agglomeration of Power Plant Fly Ash, Final Report.   December
1985, Pennsylvania State University, Springfield, Virginia:  National Technical
Information Service, DOE/PC/60270-T3, NTIS/DE86005469.

-------
G. Reethof, G.H. Koopman, and T.P. Dorchak.  "Acoustic Agglomeration for Particulate
Control at High-Temperature and High Pressure - Some Recent Results," Presented at
ASME Winter Meeting - Session on High Intensity Acoustic Applications in Coal Fired
Power Plants, December 10-15, 1989, San Francisco, California.  Sponsored by the
American Society of Mechanical Engineers.

M.C. Williams and R.C. Bedick.  Gas Stream Cleanup Technology Status Report. October
1988, Springfield, Virginia:  National Technical  Information Service,  DOE/METC-
89/0263, NTIS/DE89000925.

K.B. Wilson and J. Haas.  Performance Analysis of a Screenless  (Counter-Current
Granular Bed Filter on a Subpilot-Scale PFBC, Final Report.  Combustion Power
Company, for U.S. Department of Energy under Contract No. DE-AC21-84MC21335,
November 1989  (in press) .
                                        33-9

-------
CO
CO
                                             Steam Turbine
                  Integrated Gasification Combined Cycle


! Hot
:- Sulfur
: V Removal ;
k

;.. Participate
; Control :
Second St
Combus
                                           Steam Turbine

                        Direct Coal Fueled Turbine
      Pressurized Fluidized-Bed Combustion
Advanced Pressurized Fluidized-Bed Combustion
                                     Figure 1.   Advanced Coal-Based Power Generation System

-------
        10,000
 -
    ro
   '(D
JD _CO
O ^

E
CD
         1,000
   u,      100

   Q.
   a
            10
                      Projected
                       Turbine
                      Tolerance
                  1            5   10               100
                                    Particle Size, /um

                 Figure 2.  Particle Tolerance for Gas Turbines
1,000
                                   33-11

-------
Dirty Gas
  Inlet

  Flange
                    Sealed Face
  View Shown
Rotated 90° CCW
                                                                            Blowback
                                                                            Manifold
                                -Plenum
                                                                             Dusty
                                                                           Gas Inlet   ^
Cleaned Gas
    Exit
     A   Plenum
    cifc
                                                                                  Collected Dust Outlet
                                                           Venturi
                                                            Pulse
                                                                                                                       Filter
                                                                                                                      Element
                                                                                                           Filter Module
                                            Figure  3.   Ceramic Cross-Flow Filter

-------
                    g
                    E
                    o.
                    Q.
                    73
                    to
                    O
                    ro
                    Q.
                         10,000  _
1,000  =
                    J2     100   =
                    D

                    3
                    _o
                    CO
                    c
                           10   =
                           1.0   =
                           0.1
                                  Inlet Particle Loading
                                  /  (after cyclone)
                                    Outlet Particle Loading

                                          A  Sample 1
                                          o  Sample 2
                                                        10
                                                    Particle size, j
                      Figure 4.   Performance of Ceramic Cross-Flow Filter
                                at  New York University  PFBC Facility
Gas Outlet
Tube Sheet
Tangential
 Gas Inlet
Pressure Vessel


  Filter Elements
Collector Sector
                                                     Fine-Grained
                           Typical Boundary
                           Shape Between
                           Fine and Coarse
                           Material
                                                     Filter Surface Layer
                                                     End Plug o) Coarse -
                                                     Material, Fired In
                                                                                             Mating Flange tor
                                                                                             Suspension in
                                                                                             Tube-Sheet

                                                                                             Typical Limit of
                                                                                             Filter Surface Layer
                                                                   Coarse-Grained
                              Figure  5.   Ceramic  Candle-Filter  System
                                                   33-13

-------
                                     "Dusty" Gas
                    Disengagement
                    Vessel

                                                                      Water-Cooled
                                                                      Heat Exchanger
                 Figure 6.   Screenless Granular-Bed Filter
                             DIRTY GAS
         - SOUND PRESSURE   (AGGLOMERATED
              WAVE           PARTICLES)
                                                                        CYCLONE
                                                                  (PARTICLE SEPARATION)
            HIGH INTENSITY       '
            SOUND SOURCE
 INLET PARTICLE   AGGLOMERATION
SIZE DISTRIBUTION     PROCESS
PARTICLE SIZE DISTRIBUTION
 SHIFTED TOWARD LARGER
       PARTICLES
            AGGLOMERATION CHAMBER
               (PRESSURE VESSEL)
                                                                   PARTICLE REMOVAL
                                                                       DISPOSAL
                     Figure 7.   Acoustic  Agglomeration
                                     33-14

-------
                                        Dirty Gas IN
                                       Clean Gas Out
       Collected Particles
       After Regeneration


Figure 8.  Nested Fiber-Filter System
                33-15

-------
         Table 1.   Particle Control Conditions and Cleanup Goals
Inlet Particle
Loading (ppmw)1

Outlet Particle
Loading (ppmw)2

Outlet Size
                        PFBC
                      IGCC
  1,000-20,000
 200-3,000
                                                           DCFT
Temperature (°F)
Pressure (psig)
1,500-1,700
100-240
1,000-1,800
120-1,500
1,800-2,250
120-500
 Undefined
Lower than NSPS3   Lower  than NSPS3  Lower than NSPS3
     15-30            120-300            15-30
 No particles
   > 5-12 Om
No particles
  > 5-12 (Urn
No particles
  > 5-12 (am
   1  Possible  ranges  at  inlet  of  advanced particle  control filter.
   2  At  outlet  of  the advanced particle  control  filter.
   3  New Source Performance  Standards  for particulate  emissions
     from coal-fired power plants is 0.03 Ibs/MMBtu.
           Table  2.   Expected Operating Conditions for Hot Gas
                  Cleanup  Testing at  Tidd-PFBC Facility
         Gas  Flow:   100,700  Ibs/hr  (l/7th  of  Total  Flow at Tidd)

                     Operating Temperature:  1,550°F

                     Operating Pressure:   165  psia

                        Gas  Density:   0.22 lb/ft3

                    Volumetric Gas Flow:  7,715 ACFM1

         Expected  Particle Loading  at  Inlet of  Advanced Particle
                     Control  Device:   500-2,500 ppmw

                       Total  Test Period:   7000 hr


       1  Actual cubic  feet per minute
                                 33-16

-------
                  PERFORMANCE OF A  HOT GAS CLEAN UP  SYSTEM ON A
                        PRESSURIZED  FLUIDIZED BED COMBUSTOR
                                    J.  Andries
                                    J.  Bernard
                     Laboratory for Thermal  Power Engineering
            Department of Mechanical  Engineering  and Marine Technology
                          Delft University of Technology
                  P.O. Box  5037,  2600 GA Delft,  The Netherlands
                                    B.  Scarlett
                         Laboratory of Particle Technology
                        Department of  Chemical  Engineering
                          Delft University  of  Technology
                              Delft, The  Netherlands
                                   B.  Pitchumani
                          Chemical Engineering Department
                     I.I.T.  Hauz Khas, New  Delhi  110016 India
ABSTRACT

A sub-pilot  scale,  coal burning, pressurized  fluidized  bed combustor at the
Delft University  of Technology with a maximum thermal  capacity of 1.6 Mw  has
been used  to study the performance  of  a hot  gas clean  up  system.
The 0.5 m  diameter combustor has been  operated at pressures up  to  9 bar and
fluidization velocities up  to 1.2 m/s.
The hot gas  clean up system  consists of two  cyclones in series  followed by a
fixed granular bed filter which cleans  a  controllable  part of the exhaust  gas
stream.  Both cyclones  are equipped with an  ash removal  system which  enables  a
continuous measurement  of the amount of captured particulates.
The particulate loading and size distribution  in  the  outlet of  the  second
cyclone have been determined using a high  temperature,  high pressure sampling
probe operated at near-isokinetic conditions.
The performance of the  cyclones has been  determined using  the cyclone  catch
material and the  samples  obtained with  the  sampling probe. The  size
distributions have  been determined  using a  Coulter  Multisizer  instrument and the
particulate  loadings have been determined gravimetrically.
The results  have  been  analyzed to investigate the  effect of operating  conditions
of the  combustor  on the  dust concentration  and the particle size distribution  in
the outlet of  the second cyclone. Using  statistical regression analysis a
relation has been obtained between the  combustor-operating parameters
bed temperature,   bed pressure,  expanded bedheight,  fluidization  velocity   and
the particulate concentration in the outlet of the  second  cyclone.
                                       34-1

-------
                  PERFORMANCE OF A  HOT  GAS CLEAN UP  SYSTEM ON A
                        PRESSURIZED  FLUIDIZED BED COMBUSTOR
INTRODUCTION

Research on pressurized  fluldized bed combustion at  the  Delft University of
Technology dates back  to 1974 when investigations  on the applicability of  the
pressurized fluidized bed  combustor  for the firing of  low-quality residues in
marine gasturbines were  initiated.
The experimental test  facility was originally designed for  the study of  two-
phase flow and  heat  transfer phenomena. The characteristics  and the behaviour  of
the combustor were experimentally investigated and a theoretical  description of
the fluidization and heat  transfer phenomena was developed.  A full description
of the experimental  and  theoretical  results is  given in  [10].
The aim  of  the  second research project  was  to contribute to  a better
understanding of the  transient behaviour of a coal-fired FBC,  with main  emphasis
on the pressurized version.  A comprehensive theoretical  model  was developed
which included  gas flow  dynamics, coal  combustion, attrition,  elutriation and
bed heat-up/cool-down. A linearized  version of  this  model  was used for the
comparison with experimental  results.  In the experiments Pseudo  Random Binary
Sequence (PRBS) pertubations  in coal feed rate,  air  feed rate, position  of  the
outlet valve  and  oil  feed rate have  been  used to obtain impulse-  and step
responses for bed  temperature and flue  gas  composition.  A  full description  of
the experimental and  theoretical  results is given  in [12].
The third  research project was aimed at refining the existing model  using  the
techniques which were  developed in the  previous  research project.  Refinements
were made  in  the  modeling of  char combustion,  volatiles  combustion and
elutriation.   Furthermore the  water-steam side  and the  remaining  components  of an
FBC base energy conversion unit have been modeled  and  an integrated  model  of a
complete FBC  based energy  conversion system has  been developed.  The  integrated
model was  used  to  design a control  strategy for FBC-based energy  conversion
systems.  Results are  given in  [13].
Experimental work on  hot gas clean up  started  in January 1987. The  current
project,  which  is jointly  financed by  the Dutch Ministry of Economic Affairs  (as
part of  the  National  Coal Research Program  NOK),  the European Community  (as  part
of the EC  R&D program in  the field  of non-nuclear energy),  the  Delft University
of Technology,  OUT,  and  the  Dutch research  organization  TNO, will run  until June
1990 and is  jointly  executed by  the  OUT laboratories for Thermal  Power
Engineering and Particle Technology  and the TNO  Centre for Polymeric Materials.
A  detailed description of  the project  is  given  in [1,2,3,4,5,6,7,8].
                                       34-2

-------
The final  aim  of the project  is  to test an  electrostatically  enhanced moving
granular bed filter  in the outlet  of  a pressurized fluidized  bed  combustor.
The purpose of the work described  in  this  paper is to  assess  the influence  of
combustor  operating  parameters  like bed temperature, bed  pressure,  fluidization
velocity and expanded bed height on the  particulate concentration in  the  outlet
of the  second  cyclone.


TEST FACILITY

The combustor

Since the  beginning  of the research program  on pressurized fluidized  bed
combustion there  have been 3 versions  of  the combustor which  are  described in
[4,10,12,13].   The  latest  version of the  combustor,  which  was  commissioned  in
1984, is shown in figure  1,  the  main  design data  are  given in table  1.
Compressed air produced by two diesel  driven,  high pressure compressors is
supplied to the combustor by a valve  operating above  the  critical pressure drop
and enters the bed through a nozzle  type air distributor,  after having been
preheated  in the  annulus between the  pressure  vessel  and  the  heat resistant
inner vessel.  The fluidization velocity is controlled  by  a second valve
discharging to the stack.  The heat exchanger and pressure vessel cooling  system
are part of  a  closed circuit  filled with demineralized water.  Heat  release is
adjusted by varying  the bed content and  hence  the cooling  surface immersed in
the bed.
Coal and gas can be  supplied  to  the  combustor respectively via  an  inbed and  an
underbed feeding  system.  The solids feed pipe  extendeds vertically to  0.2  m
above the  air  distributor plate.
By means of  three solid feed  systems,  each using a separate screw  feeder,  coal,
inert bed  material and sorbent material  are  dumped through a  common chute  in
the combustor.   By a pressure lock  system  the pressurized  storage bunkers  are
periodically filled  from  atmospheric  ones.
A nozzle type  air distributor is used.  In  every nozzle an atomizer  is mounted
which can  supply  natural gas or  oil  to the  fluidization  air.
The freeboard  region is heavily  insulated  and  free of  internals to  enable
unobstructed experiments  on freeboard  phenomena.


The hot  gas  clean up system

The test rig is equipped with two  high temperature cyclones of  the  standard
Lapple high efficiency  type.  The dimensions  of  the cyclones are based on
standard design rules.  The cyclones,  made  of a heat resistant metal alloy,  are
installed  in two  identical pressure vessels. The  space between the  outside  wall
of the  cyclones  and the pressure vessel  wall is filled with heat  insulating
material (kao  wool type material)  The diameter  of the first  and  the  second
cyclone is 0.3 m, resp. 0.2 m.  The collected fly  ash  is  automatically removed
through lock hopper  systems and  the amount is  measured every  10  seconds.  A
controllable  part  of the fly ash,  collected  by the first  cyclone, can be
reinjected into the  combustor.
A 0.4 m  diameter  fixed granular  bed filter,  which can  clean a variable part
(10   75 %)  of the  exhaust gas  stream, is being  used to  study  the  influence of
several parameters  (filter  velocity,  filter  material  and  fly ash composition) on
the collection processes.
The installation  including  the high temperature  cyclones  and  the granular  bed
filter is  shown in figure  2.
                                        34-3

-------
Operation and  control

The installation  is  automatically controlled by  means  of a microprocessor-based
control system which is integrated with  the  data acquisition system.  All
measurement values are  stored every 10  seconds  for further analysis.  Flue gas
can be  sampled at several  locations of  the installation using  2  sets of  gas
analysis instruments  (02,  C02, CO,  N0x,  S02) .
During normal  operation bed temperature, pressure,  fluidization velocity  and  bed
content are controlled  by an on-line  control  system.
The test facility is normally operated  on  day shift only,  in  general 6  to  8
hours  including  the  start-up period.  Figure  3  shows a typical  registration of a
test run during  which  3 different  pressure levels were progressively  adjusted.
While  the  operating  temperature  of the  fluidized bed can be reached already  30
minutes after  start-up, the total  system needs  about 2 hours before achieving
full stability  (except  desulphurization in the  case of sorbent addition).  This
can be  clearly noted from  the 02-  and NO^  concentrations.
The composition  of the  coal used  is  given  in table  2.

Sampling

The sampling  arrangement  is shown  schematically  in figure  4.  The  probe  consists
of a  sampling tube surrounded by  a concentric pipe through which  water is
circulated to  cool the  sample stream  (figure  5). The sample stream is passed
through a  sampling cyclone which collects  the  larger particles. The  small
particles,  which  pass  the sampling cyclone,  are collected  in a pre-weighed,
glass-fiber reinforced,  thimble  filter in  the cyclone  exhaust.  The  gas  stream
then passes through  a heat  exchanger  and a silica  gel drier before it enters a
gas meter. A  rotameter  downstream  of  the gas meter is used  to  adjust the
sampling velocity with  reference to the  gas  velocity in the exhaust stream from
the second cyclone.  Condensation in the  sampling cyclone and the  thimble filter
is prevented  by  insulating the  sampling  system  externally.
After  a  steady state of the  combustor has  been  reached, the flow rate of the
sampling stream  is adjusted to  establish isokinetic sampling  conditions.
Sampling is carried  out for 30  minutes.  After the  sampling has  been finished,
the pre-weighed  thimble filter and the  collection pot  of the sampling cyclone
are cooled and weighed. The dust  concentration  in the hot  gas  stream is
calculated from  the  total mass  increase  and  the gas volume which  flowed  through
the sampling  system.


EXPERIMENTAL  RESULTS

The dust concentration  values,  measured  in the  outlet  of the second cyclone,  are
presented  in  table 3.
Using  statistical regression techniques the following  empirical  relationship has
been found between the  combustor operating parameters    expanded bedheight  H
(m),  bedtemperature  T   (°C), bed pressure P (bara),  ,fluidization velocity V
(m/s)    and  the particulate concentration  C  (mg/Nm3)  in  the outlet  of the second
cyclone:

           C  = -7275.2 +  60194.7*H   23.27*T +  284.4*H2    0.0585*T2 +

               75.1*H*T     2078.6*P  +  15.4*P2  + 644.1*H*P +  1.75*T*P +

           +  50934.5*V    8751.6*V2    3993.5*H*V    37.7*T*V    71.99*P*V         (1)

Some size  distributions determined with  a  Coulter Multisizer  are  shown  in
figure  6 .
                                        34-4

-------
DISCUSSION

The dust  concentration In the outlet  of  the second cyclone  is  determined by the
coal feed  rate,  the specific elutriation  rate  and the collection  efficiency  of
the cyclones.  Each  of these three parameters  is influenced by  the combustor
parameters H,T,P  and  V.
Equation  (1)  is  used  to calculate the  dust concentration in  the  outlet of the
second cyclone as  a function of  each  of  the four combustor  operating parameters
(H,T,P and V), keeping the other  three constant.  The results are  shown in
figures 7,8,9  and 10.  From these  figures  a number of observations can be made:

    •    increasing  the fluidization velocity  initially  increases   the  outlet
         concentration.  At higher  gas  flow rates the increasing cyclone
         efficiencies  offset the  increase  in specific elutriation  rate.
    •    increasing  the bed temperature causes  a decrease in  outlet
         concentration.  The improved combustion  rate  decreases the specific
         elutriation rate.
    •    increasing  the pressure  causes a  decreasing outlet concentration. A
         higher combustion rate decreases  the  specific elutriation rate.
    •    increasing  the bed height causes  initially a decrease  in  the outlet
         concentration.  The increasing coal  feed rate at deeper beds,  due to  the
         increased in-bed heat exchanger  surface,  offsets the decreasing  specific
         elutriation rate caused  by  the  longer residence time of  the coal
         particles in the deeper  bed.


FUTURE WORK

Work is  going on to substantiate  the  qualitative  statements  made  in  the  previous
section:

    •    simultaneous  concentration-  and size  distribution measurements at three
         positions in the PFBC  installation (outlet combustor, outlet  first
         cyclone  and outlet second cyclone)  to determine the  mass-  and grade
         efficiency  of the cyclones.
    •    measurement of the coal  feed  rate and  the specific  elutriation rate  of
         the  combustor at different  combustor  operating  conditions.

The experimental  values  will  be  used  to  validate  theoretical models used  to
predict the  inlet concentration of  the final  filter.
A moving  granular bed filter using  electrostatic  enhancement (particle charging
and bed polarization)  will be installed  in April  1990.  Experiments will  continue
during the second half of 1990 to  assess  the properties of  the filter system.
The experimental  results  will be used  to  validate theoretical models,  which  will
be used  to  optimize the hot  gas  clean up system.
Besides work  on  particulate emissions, gaseous  emissions will be  studied:


    •   measurement of gaseous nitrogen  components.
    •   modeling  the  behaviour of nitrogen compounds in the  combustion system.
    •    testing  the effectiveness of  a number of NOX abatement techniques  (air
         staging,  additive injection) and  their  influence on  other harmful
        components  (gases and particulates).
                                        34-5

-------
ACKNOWLEDGMENTS

This research  is  supported by the  Dutch  Ministry of Economic Affairs  (PEO
contracts 20.36-0110.10  and 20.36-0110.11).  the  European  Community (contract
EN3F-0028-NL(GDF),  TNO  and Stork Boilers.

REFERENCES

1. J  Andries,  "The Delft  experimental  pressurized fluidized bed  test  facility",
   Paper presented at the  second biennial PFBC power plants  utility conference,
   June 18   20, 1986,  Milwaukee, Wn, USA.
2  J.  Andries,  "Hot gas  cleanup  research on the  Delft  pressurised fluidised bed
   combustion test facility", Paper  presented  at the 9th  international conference
   on fluidized bed combustion,  May 3    7,  1987.  Boston,  Ma, USA.
3  J  Andries,  "Hot gas  clean up research at  the Delft pressurised fluidised  bed
   combustion test facility", Paper  presented  at the International  conference
   Advanced Coal Power  Plant Technology,  December 2    4,  1987,  Dusseldorf,  FRG.
4  J.  Andries et al.,  "Closed loop controlled  integrated  hot gas  clean up",
   Periodic Report June  1986   November 1987,   Laboratory  for Thermal  Power
   Engineering,  Delft University  of  Technology,  1987.
5  J.  Andries,  "The Delft  pressurised fluidised bed combustion  test facility",
   Paper presented at the  EPRI  1988 seminar on Fluidized  bed combustion
   technology for utility  applications,  May  3     5,  1988,  Palo Alto, CA,  USA
6  J.  Andries,  "Closed  loop controlled  integrated hot  gas  cleanup",  Paper
   presented at the EC  contractor's meeting  'Flue gas  treatment,  fly  ash
   properties,  re-use and  disposal',  June 16,  1988,  Brussels, Belgium.
7  J.  Andries et al.,  "Closed loop controlled  integrated  hot gas  clean up",
   Periodic report December 1987    May 1988, Laboratory for  Thermal Power
   Engineering,  Delft University  of Technology,  1988
8  J  Andries et al.,  "Closed loop controlled  integrated  hot gas  clean up",
   Periodic report June  1988   December 1988,   Laboratory  for Thermal  Power
   Engineering,  Delft University  of  Technology,  1989
9  J.G. Bernard, J. Andries and  B.  Scarlett,  "Cyclone  research  for application
   at high  temperatures  and pressures",  Paper  presented at the  1. European
   Symposium Separation  of  Particles  from gases,  April 19    21, 1989,  Nurnberg,
   West Germany
10 G.  Boelens,  "Turbulent  fluidization  and heat  transfer  in  a pressurized
   fluidized bed combustor",  Ph.D.  Thesis.  Delft: Laboratory  for  Thermal  Power
   Engineering,  Delft University  of Technology,  WTHD 178,  March 1986,  Delft.
IIP  de  Haan,  P.J.  Nederveen,  P.J. Droppert,   K.E.D. Wapenaar and B.   Scarlett,
   "Electrical  characterisation of  filter  bed  material  and fly  ash at  high
   temperature",  Paper presented  at  the  third   international  conference  on
   electrostatic precipitation,  October  25    29,  1987,  Abano,  Italy
12R.J.M. Kool,  "Dynamic modeling and identification of a  coal-fired  pressurized
   fluidized bed combustor",  Ph.D.  Thesis.  Delft: Laboratory  for  Thermal  Power
   Engineering,  Delft University  of Technology,  WTHD 175,   September 1985, Delft.
13J.M.P. van der  Looij, "Dynamic modeling and control of coal fired  fluidized
   bed boilers", Ph.D.  Thesis.  Delft:  Laboratory for Thermal  Power Engineering,
   Delft University of  Technology,  November  1988.
14C.A.P. Zevenhoven, J. Andries  and B.  Scarlett, "Hot gas clean up using
   electrostatically enhanced granular bed filters",  Paper presented at  the 1.
   European Symposium Separation  of Particles  from gases,  April 19   21,  1989,
   Nurnberg, West Germany.
15 J.  Andries,  D.  Boersma  and K.R.G. Hein,  "Research into particulate  removal
   and N0x  emission control of  pressurized fluidized bed  combustion",  Paper
   presented at  the  Symposium on Low-grade fules, June 12-16,  1989, Helsinki,
   Finland
                                        34-6

-------
                        Table  1
                    MAIN  DESIGN DATA

            Bed  diameter
            Static  bed height
            Freeboard  height
            Operating  pressure
            Fluidization velocity
            Air  distributor
            Heat  exchanger
                   0.49
                   0.3
                   2.6
                   0.1
                   0.8
                          1.5  m
                          3.8  m
                          1.0  MPa
                          1.8  m/s
                   nozzle plate
                   water cooled
                   vertical pipes
            Maximum  thermal  capacity 1.6 MW

                        Table  2

            COAL COMPOSITION (AS RECEIVED)

        Proximate analysis  (raw)      Elemental analysis  (raw)
      moisture
      ash
      volatiles

      LHV
 2.0
 11
31.6
     1,
30.0 MJ/kg
carbon
hydrogen
nitrogen
oxygen
sulfur
76.3 %
4.7 %
1.4 %
7.2 %
0.7 %
                        Table  3
DUST CONCENTRATION  IN  THE OUTLET OF THE  SECOND  CYCLONE
expanded
bedheight
(m)
0.81
0.81
0.81
0.81
0.81
0.69
0.69
0.74
0.74
0.55
0.55
0.55
0.58
0.68
0.72
0.72
0.72
0.72
0.67
0.67
0.71
0.71
0.71
bed
temperature
<• C)
846
853
850
853
850
850
850
850
850
851
864
831
841
740
830
831
833
827
853
849
828
830
829
bed
f luidization
pressure velocity
(bara) (m/s)
4.0
4.0
5.5
5.5
7.0
7.0
7.0
7.0
7.0
3.0
3.0
3.0
3.5
5.0
5.0
5.0
6.1
6.0
5.0
5.0
5.0
6.0
6.0
0.
0.
0.
0.
0.
0.
0.
1.
1.
0,
0.
0,
0.
0.
1
1
1
1
0
0
1
1
1
8
8
,8
.8
.8
,8
,8
,0
.0
.8
.8
.8
.9
.8
.0
.0
.0
.0
.8
.8
.0
.0
.0
                                                       dust
                                                       concentration
                                                       (mg/Nm3)
                                                       334
                                                       347
                                                       363
                                                       402
                                                       370
                                                       282
                                                       326
                                                       278
                                                       280
                                                       820
                                                       1059
                                                       695
                                                       787
                                                       636
                                                       490
                                                       460
                                                       314
                                                       429
                                                       345
                                                       350
                                                       546
                                                       340
                                                       390
                          34-7

-------
                                             FLUE CAS (TO CASCLEANINC)
                                            .INSULATED EXPANSION PART
                                            -PORT TOR FLUE CAS
                                             SAMPLING APPARATUS
                                            -CERAMIC WOOL INSULATION
                                            •FREEBOARD THERMOCOUPLES
                                            -FREEBOARD
                                            .COAL/DOLOMITE SUPPLY
     FURHACEDIAHETER 0,5 H

     EEDHEICHT  (MAX.)  2 M

     FREEBOARDHEICHT 3 M
     PRESSURE 1-10 BAR

     THERMALPOWER" 1 ,6  HW
                                            .AIR SUPPLY
                                            -HATER HEAT EXCHANGER
                                             MAX.  2,4  H
.FLUIDIZED BED

^DISTRIBUTOR ?LATE WITH AIR,OIL,
 NATURAL CAS NOIZLES,
 AND  BED ASH OFF-TAKE
Figure  1.  The  Delft  pressurized  fluidized bed  combustor
                             34-8

-------
                                                                 STORAGE
                                                                  VESSEL
Figure 2.  The  Delft hot  gas  clean up  test  facility
                        34-9

-------
                   ].rcit.irc   (0 - 10 b.r.)  	 Uti! {0 - 1000 C)
                   I.eH IK-UU (0 - ^ u>)         Vfl  (Q - ; in/,) |
                Lime [hours]
                   o;  (o - 10 ».)
                  ' CO  (0 - '/OOj.
               Lime  [ 11 ours]
                       d (0 - 100 ti/J,J  	 cj-c! (0 - J  X,/),J
                         (0 - < lf/h)    - —_.- crcl [0 - 10 L'f/h)
Figure  3.   Typical  test  run
                34-10

-------
                                                               9.    10.  11.
flow
               1- Measuring point
               2. Insulation
               3. Cooling pipe
               4. Valve
               5. PrecolUctor
               6. Absolute  fil'.cr
               7. Cooler
               0. Drier
               9. Gas mctei-
10. Needle val-, e
11. Pump
FT. Flo;,- nieler
TT. TherniocouDle
P. Pilot lube "
Cl. G.TS ?.11?.lVais
               Figure  4.   Sampling  arrangement

                 Figure  5.  Sampling  probe
                               34-11

-------
                                  200
                             sire [micrometer]
100

 90

 eo

 70

 60
 50


 30

 20  -

 10

  0
             20
                                60        £0
                             sire [micromol-fr]
                             s;:« [microrrvM-rr]
                                                 300
                                                                •'.00
                                                  100
                                                           120
                                                 Cyclone 2 exit
                                                 Cone. 378 mg/m3
                                                 Pressure D bcr
                                                 Temperature  853 C
                                                 April •>.•;. 19S9
    Figure  6.  Fly-ash  size  distributions
                           34-12

-------
           700
     O
     O
            200
        Bed ht. 0.6 m   Temp. 550 C


         •  4 bcr         +  5 bor
                                  0.85
                                                     0.9
                                                                       0.95
                                    VELOCITY [ m/s 1

                                      o  6 b c r
                                                         7 bcr
Figure  7.  Effect  of fluidization  velocity  on outlet dust  concentration
      o
      0
      O
                                         T	1	1	T

               320   832    S34   835   838   840   842


     ted press. 4 bcr bed height 0.8 m     TEMPERATURE [ C ]


             •  O.°m/s         +  O.SSm/s          o
                                                    544
                                                          346   848   S50
                                               o   0.9n/s




Figure  8.  Effect  of bed  temperature on  outlet  dust  concentration
                                     34-13

-------
                                               0.9 m/s
Figure  9 .  Effect of  bed pressure  on outlet  dust concentration
       750
 o
 z
 o
 o
       5CO
     ei  O.S m/s  temp. S50 C
                            0.6
                                BED HEIGHT [ m ]

                              5 tor         o   6 bar
 Figure  10.  Effect  of  bed height  on outlet  dust concentration
                               34-14

-------
       ELECTRIFIED GRANULAR FILTER FOR HIGH TEMPERATURE GAS FILTRATION
                                 P.H. de Haan
                      TNO Centra for Polymeric Materials
                          P.O. Box 71, 2600 AB Delft
                               The Netherlands
                              M.L.G. van Gasselt
                                L.M. Rappoldt
                           TNO Division for Society
                       P.O. Box 342, 7300 AH Apeldoorn
                               The Netherlands
ABSTRACT

The results of an experimental study on an electrified granular filter  usable
at  high  temperatures are presented. This filter combines the best properties
of an Electrostatic Precipitator and a common ceramic filter, viz.  it  has  a
low  pressure  drop  and  a high efficiency. The system consists of a layer of
coarse granules with a grid electrode embedded in the layer which generates  a
strong  electric  field.  The particles to be captured from the gas stream are
charged with a corona device upstream the filter layer.  Experimental  results
on  penetration  and  dynamic  pressure drop have been obtained for a layer of
loose and of sintered granules. It has been  shown  that  the  filter  can  be
regenerated  by applying a  high pressure air pulse. The filter offers various
promising applications.
                                   35-1

-------
       ELECTRIFIED GRANULAR FILTER FOR HIGH TEMPERATURE GAS FILTRATION
1. INTRODUCTION

Hot  gas  filtration  plays  a key role in the development of several advanced
power generation processes, such as Pressurized Fluidized Bed Combustion, Coal
Gasification Combined Cycle, and many other industrial processes.
Possibilities for cleaning gas flows at a high  temperature  (and  at  a  high
pressure)  are,  for  instance,  electrostatic  precipitators,  cyclones, metal
fiber fabrics, ceramic fiber fabrics, and ceramic candle filters.

A  promising  alternative  is  the use of granular bed filters. Because of the
free choice of the granular material such filters can be  used  under  extreme
conditions.
Three types of  granular  bed  filters  are  described  in  literature.  Fixed
granular  bed filters consist of a layer of granular material confined between
grids. The filter layer may be positioned either horizontally  or  vertically.
The upstream grid often has an open structure, e.g. a louvre panel. The filter
is cleaned by blowing off the contaminated top  layer  of  the  filter  [1,2].
Sometimes,  the  granular material is periodically removed from the filter bed
and externally washed [3]. The performance of such filters can be improved  by
charging the dust particles in the contaminated gas stream by corona.
Moving granular bed filters are found in a gridless  version  [4],  and  in  a
version  in  which  a  layer of granular material moves between vertical grids
[5]. The contaminated granules are continuously extracted from the filter  and
externally  washed.  High  voltage electric fields can be used to increase the
filter performance and to stabilize the flow of the granules.  Moreover,  also
corona  precharging  of  the  dust is applied to enhance the filter efficiency
still further.
In  the  case  of  fluidized granular bed filters a layer of granular material
lies on a gas distribution plate. Violent bubbling of  the  fluidized  bed  is
prevented by the application of an electric or magnetic field [6].
Also the fundamental aspects of gas filtration with granular bed  filters  has
been investigated and reported in literature [7-12].
                                     35-2

-------
In this paper a particular form of a fixed electrified granular  filter   (EGF)
will  be  described. A relatively simple and direct concept has been developed
for a filter system suitable for high temperature and extreme conditions  [13].
The  system  consists  of  a layer of coarse granules. The granules are either
confined between two grids,  or  are  fused  together  by  sintering.  A  grid
electrode  within  the  filter  layer  generates  a strong electric field. The
coarseness of the granules and the application of electrostatics result   in  a
filter  concept  with low pressure drop and high efficiency even for submicron
particles.
The  filtering  performance  of  the  concept has been investigated at ambient
conditions. Experimental results were obtained from  penetration  and  dynamic
pressure  drop  measurements.  It will be shown that, after a certain pressure
drop has been reached, the filter can be regenerated. The investigations  have
revealed  that  the  application  of  a pulsed flow of air is an efficient aad
practical way of removing the dust deposited on the filter.
Besides   the  experimental results, also the further development aims  will be
discussed, as well as some potential applications.
 2.  FILTER  DEVICE

 The  filter device  comprises  a  charging  system,  e.g.  a  corona charger,  and  an
 electrified  granular   filter  downstream of  the charging  system. The  filter is
 formed  by  a layer  of granules  of  non-conducting material  and  a  high voltage
 grid for   generating  an electric field across  the  granules. The set-up of the
 test equipment  for the device  is  shown  in Figure 1,  which shows a wire  corona
 charger and a flat filter plate.
 In  first instance  the  filter layer was  composed  of  loose granules  confined
 between two heavy  supporting grids. The grids were  electrically grounded  and  a
 high voltage grid  halfway the  filter layer provided for  the   strong  electric
 field.   In  the   experiments   performed,  granular  material  with  different
 diameters  was used. The thickness of the filter layer was about 2.5 cm, and an
 electric field  strength of about  6 kV/cm was used.
                                     35-3

-------
I 1
i E' \
! U
1 E « /
/
tl „

" 1
                   (a)
                                                    0,2m
                    -U
                              0,3m
                     -n
                   (b)
                                           FTh-A
                                                     0,3m
       Figure  1. Electrified Granular Filter with corona charger in
       top view (a) and aide view (b). A) filter frame, B)  grounded
       grids,  C)  high  voltage  grid,  D)  grounded electrodes for
       corona charger, E) corona wires, F) fly ash collector.
Secondly,  a  filter was constructed of alumina granules (average diameter 1.5
mm) which were sintered together. The filter layer  was  3  cm  thick  and  an
electric  field of about 5 kV/cm was used. Two simple open metallic grids were
used to provide for the grounding of the filter.
An  improved  concept  of this filter device also makes the metallic grounding
grids redundant. A top layer of  (semi)conducting ceramic granules   (e.g.  SiC)
is fused together with the filter layer. Since the layer only has to provide a
ground for an electrostatic potential, the demands on the conductivity of  the
layer  are only modest. In this way a self-supporting, all-ceramic  electrified
                                   35-4

-------
granular  filter  system  is   obtained,   suitable   for   high   temperature
applications.

The advantages of this kind of  filter  are  the  following.  Because  of  the
presence  of  an  electric  field the filter efficiency is very high, also for
submicron particles. Due to the large size of the granules the filter layer is
highly  porous.  Therefore, the pressure drop of the filter is relatively low.
As a. consequence, the superficial gas velocity through the filter can be high,
typically around 0.2 m/s, which is high in comparison with a fabric filter.
Due to the presence of the electric field a large part (on a mass base) of the
incoming  dust  is  deposited in the outermost layer of the filter. Therefore,
there are good possibilities for cleaning the filter after a certain  pressure
drop  has  been  reached. In the case of the filter system with loose granules
several techniques for cleaning  the  filter  have  been  investigated;  these
methods  comprised  e.g.  reverse  gas  pressure  pulses, reverse air flow (at
relatively low velocity) optionally accompanied by additional  means  such  as
vibration or mechanical  shocks.
The best results were obtained with the pressure pulse method. This method  is
similar  to the one used in pulse jet fabric filters. The maximum overpressure
behind the filter plate  (see also Figure 1) was 0.1 bar,  the  pulse  duration
was  about  0.5 s. The maximum velocity of the cleaning gas through the filter
amounted to about 10 m/s. Because of the high porosity of the granular filter,
the  cleaning  pulse  does  not  lose  much of its energy in the filter layer.
Therefore, most of the energy can be dissipated  in  the  layer  of  deposited
dust, which causes the pressure drop across the contaminated filter.

In conclusion, an electrified granular filter, with either loose  or  sintered
granules, might show a relatively low pressure drop, a high filter efficiency,
and a good cleanability.
The rnnin advantage of sintering the granules of the filter layer in comparison
with loose granules, is  the disregarding of  the  heavy,  metallic  supporting
grids.  Therefore,  the  self-supporting  electrified granular  filter can very
well be used at high temperatures  (up to 900  °C). Moreover,   the  all   ceramic
filter can be used in highly corrosive environments.
                                    35-5

-------
3.  EXPERIMENTAL RESULTS

First,  the  influence of the granular shape and size on the filter efficiency
and pressure drop will be reported. Besides the initial pressure drop also the
pressure  drop  increase  versus  the  amount  of  deposited  fly  ash  is  of
Importance.
The  experiments  were  performed  on  a  2.5  cm  thick filter plate of loose
granules. The filter plate area was 0.2x0.3 m2. The filter was tested with air
at  room  temperature  (40  m3/hr,  i.e.  0.2  m/s  gas velocity). The gas was
contaminated with redispersed fly ash collected from  the  bag  filter  of  an
Atmospheric  Fluidized  Bed  Combustor.   The  bulk density of the fly ash was
around 370 kg/m3 and the specific density 2300 kg/m3. The carbon  content  was
about  211; as a consequence the resistivity of the fly ash is quite low  viz.
about 103 12n. The concentration of fly ash in the gas stream amounted to about
3 g/m3.
The penetration of fly ash through  the  filter  was  measured  with  a  Laser
Aerosol Spectrometer (PMS CSASP100-HTHP).  which counts the number of particles
within a specified sample volume in the size range from 0.45 up to 12 /jm. From
the particle size distribution the mass concentration behind the filter can be
obtained. Taking into account the capture efficiency of  the  corona  charger,
the penetration of the electrified granular filter can be determined.
The filter was challenged with the flow of contaminated gas until  a  pressure
drop across the filter of about 1200 to 1500 Pa was reached.
Four different granular materials have been used:
1. fine sand, with a granule diameter 0.7-1.4 mm;
2. coarse sand, with a granule diameter 1.4-2.0 mm;
3. glass beads, monodisperse, with a diameter of 3 mm;
4. granite flakes, with dimensions 1x2x4 mm3.

The results of the filter penetration and the pressure drop are shown in Table
1.  The filter with the coarse sand was tested at a lower  gas  velocity.  For
the fine sand the range of the results of several experiments La shown.
The table gives the gas velocity, U,  the  initial  pressure  drop,  Ap0,  the
average  pressure  drop increase per amount of deposited fly ash at a pressure
drop of 1500 Pa, Ap/M, and (an estimation of) the filter penetration P .
                                                                      m
                                     35-6

-------
                                   Table 1
                  INFLUENCE OF THE KIND OF GRANULAR MATERIAL
                 ON THE PRESSURE DROP AND FILTER PENETRATION.
Material U
fm/31
fine sand
coarse sand
glass beads
granite flakes
0
0
0
0
.2
.1
.2
.2
Apo
[Pal
150-260
30
30
40
Ap/M P
fPa m2/gl rZT
0.3-0.8
0
0
0
.04
.15
.33
<0.01
<0.01
0.04-0
0.03-0


.6
.06
An  electrified  granular  filter  with  coarse  sand  was loaded with fly ash
repeatedly. After a pressure drop of about 1500 Pa was reached, the filter was
cleaned by a reverse air pulse. The filter was challenged with redispersed fly
ash from an atmospheric fluidized bed combustor.
Due  to  the  incomplete   cleaning  of  the filter some fly ash remains on the
upstream side of the   filter,  and  hence  the  open  filter  area  decreases.
Therefore,  the pressure drop at the beginning of a filtration cycle increases
and also the rate of increase of the pressure drop versus  filter  loading  is
enhanced every cycle.
The initial pressure drop, i.e. the pressure drop after  the  cleaning  pulse,
and  the  average   rate of increase of the pressure drop during the filtration
cycle are shown in  Figure  2. The first three data points were obtained with  a
gas  velocity  of U-0.1 m/s, the other data points with a velocity of 0.2 m/s.
The most important  result  is the stabilization of both  the  initial  pressure
drop  and  the  rate   of   pressure  drop  increase.   That means,  that a stable
operation of the filter will be possible.
Regarding  the  operation  of the filter the following should be mentioned: The
filter penetration  is  invariably low,  below 0.01Z on  a masa base.  There is  no
significant  penetration   of  fly   ash into the  filter bed. Only  the  outermost
layer of sand granules has been  contaminated  with   fly  ash.   It  should  be
mentioned that another experiment has  revealed  that  the precharging  of  the fly
ash particles with  a corona  charger is of great  importance.  Without  charging
the fly ash penetrates to  at least  halfway  the  filter. Therefore, charging  the
fly ash particles is essential  for  a  good cleanability of  the  filter.
                                    35-7

-------
                 0   5   10   15   20   25   30   35   4O   45
                                          30   35   *0   45
       Figure 2. a) Pressure drop across the filter  bed  after  the
       cleaning  pulse,  Apo,  and  b) average rate of pressure drop
       increase during loading with fly ash, , as  a  function
       of  the number of cycles. During the first three cycles U-0.1
       m/s, thereafter U-0.2 m/3. The corona voltage  is  -12.5  kV,
       and the grid voltage -7.5 kV.
About the cleaning pulse the following is noted: A cleaning pulse of about 0.1
bar  overpressure and a maximum flow rate of 10 m/s during a 0.5 3 corresponds
to a peak power of about 10 to 30 kW per m2 of filter area, and a total amount
of  dissipated  energy of 3 to 8 kJ per m2. It is interesting to note that all
other cleaning methods investigated showed about  the  same  total  amount  of
energy  dissipation.  However,  the  cleaning action lasted much longer (up to
about 30 s), and the maximum peak power was much lower. Obviously, a high peak
power  is  more  important  for an efficient cleaning than the total amount of
dissipated energy.

-------
The  same kind of experiment has been performed with a 3 cm thick filter plate
(area 0.2x0.3 m2) made of alumina granules (average diameter 1.5 mm)  sintered
together.  Halfway  the  filter plate a wire grid has been imbedded to provide
for the electric field. Two grids with a large open area on both sides of  the
filter plate provide for the electric ground.
The pressure drop of the new and clean filter plate amounts to about 90 Pa  at
a  gas  velocity of 0.2 m/s. First the filter plate has been contaminated with
redispersed fly ash by subsequently loading  and  cleaning  the  filter  at  a
relatively  low  pressure  drop  of  150 Pa. Subsequently, the filter has been
loaded with redispersed fly ash until a pressure drop of 1500 Pa  is  reached.
Then, the filter has been cleaned by a reverse air pressure pulse. The results
of the last two experiments are summarized in Table 2.
                              Table 2
               RESULTS OF THE LOADING OF THE CERAMIC
                    ELECTRIFIED GRANULAR FILTER.
       Initial pressure drop               Apo     • 110  Pa
       Initial rate of increase            (Ap/M)i - 0.02 Pa m2/g
       Final rate of increase              (Ap/M)2 =» 0.15 Pa m2/g
       Deposited dust at Ap-1500 Pa        M       - 20   kg/m2
       Pressure drop after cleaning pulse  Apo"    - 130  Pa
       Rate of increase after cleaning       - 0.12 Pa m2/g
       Filter penetration on a mass base   P       - <0.05Z
Note that the initial  pressure  drop  of  the  sintered  granular  filter  is
comparable  to the one with  the  loose granules. Moreover, the rate of increase
in the pressure drop during  the  first cycle is about a factor of 10 lower than
with  the  fine sand filter. This is probably due  to the fact that in the case
of the loose granules the heavy  grid with a low free area contributes  to  the
blocking of the filter by the deposited dust.
So, the results with the sintered filter bed are at least  as  good  as  those
with the filter with loose granules.
                                      35-9

-------
4 .  OUTLOOK

In this section some potential applications of the Electrified Granular  Filter
will be summarized and, when appropriate, elaborated.
In  the  framework of  the European Community for Coal  and  Steel  a  proposal has
been submitted for testing a sintered filter plate at  a  temperature  of 500 °C.
The  investigations  are  incorporated in the development  of  Coal  Gasification
Processes. The size of the filter plate  to be tested is  still  rather  modest,
viz. 0.2x0.3m2.
Another application of the EGF will be the cleaning of flue gas  from a  waste
incinerator. It is assumed that a part of the pollutive  components of the flue
gas is formed by condensation and subsequent chemical  reactions  on  the  dust
particles  in  the  gas   (e.g. dioxines). Removing the dust particles from the
flue gas at a high temperature might therefore reduce  the  emission of  gaseous
pollutants, which would be of great importance.
Moreover, a suitable design of the corona precharging  system might enhance the
beneficial  effects  of   the  system.  The  employment of  a so-called positive
streamer corona [14] might lead to the oxidation of SOz  and NO,  and,  besides,
to favorable chemical  reactions of some  organic components.
A third way in which the  EGF might be used with advantage  is  in  the  free board
of an Atmospheric Fluidized Bed Combustor. A schematic diagram of  this concept
is shown in Figure 3.
On  top  of  the free  board, eventually  above a superheater a series of  filter
elements is installed. Upstream of the filter elements a  charging   system  is
situated, consisting of a series of parallel corona wires.
When the pressure drop across the  filter  has  increased  above  a  specified
value, high pressure air  pulses applied  to the backside  of the filter elements
remove the deposited filter cake back into the free board.  As   a  result  the
fly-ash concentration  in  the free board  will increase. For this  reason,  a fly-
ash extraction system  is  provided which  incorporates  a  fan  and   a  cyclone
separator.  The  gas   is  extracted  above  the  heated  surface. The flue gas,
partially cleaned by the  cyclone, is returned to the free  board  again  below
the  heated  surface.  When  5Z  of  the flue gas flow  is circulated into the
cleaning system, the fly  ash concentration in the free board  will   amount  to
about  30-50  g/m3,  which  equals  the  concentration  present  in  convential
retiring systems.
                                   35-10

-------
Before  this option of the EGF can be incorporated in commercial fluidized bed

boilers, additional experimental research is needed.   High  temperature,  high

dust  concentration  experiments  are  essential  for  assessing the practical

feasibility of the concept.

                                                                   10
       Figure  3. Electrified Granular Filter in the free board of an
       Atmospheric  Fluidized  Bed  Combustor.  1)  air  supply,  2)
       fluidized bed,  3)  free  board,  4)  superheater,  5)  corona
       wires,  6)  filter  elements,  7) cyclone, 8) ash removal, 9)
       heat exchangers, 10) to stack.
                                    35-11

-------
5.  CONCLUSIONS

The Electrified Granular  Filter  comprises  self-supporting,  ceramic  filter
elements,  with  a  high  voltage  wire  grid  within  the granular layer. The
grounding grids needed for the generation of the strong electric field, may be
composed  of  conducting  ceramic  granules  sintered together with the highly
insulating granules.
Obviously.  the  concept  shows  quite  promising  characteristics. The filter
efficiency of the system is high. In addition,  because of the  use  of  coarse
granules  the  pressure  loss  across  the  filter  is  low;  alternatively, a
relatively high gas velocity can be  allowed.  Due  to  the  presence  of  the
electric  field the dust particles are deposited in the outermost layer of the
filter. As a consequence, the filter cake  can  easily  be  removed,  e.g.  by
applying  high  pressure  air  pulses. The cleanability of the filter has been
confirmed by experimental investigations at room temperature.
Basically, the filter concept combines the best properties of an Electrostatic
Precipitator and the usual ceramic filter; it shows a low pressure drop and  a
high efficiency.
Future research should be aimed at testing filter elements (first on  a  small
scale,  later  on  a  larger  scale)  under practical conditions, such as high
temperature and high dust loadings.
ACKNOWLEDGMENTS

This work has been performed under contract with the  Netherlands  Agency  for
Energy  and  the  Environment  NOVEM,  within  the framework of NOK, the Dutch
National Programme on Coal Technology and is financed by the Dutch Government.
                                    35-12

-------
REFERENCES


 1.  Patent nr. US 4067704, Granular Bed Filter. Ducon Co. Inc.

 2.  Patent nr. US 3926537. Electrostatic Filtration Panelbed. A.M. Squires

 3.  Patent nr. GB 2070973, Moving Bed Ga3 Filter, Rexnord Inc.

 4.  J.L.  Guillory.  "Progress  in High Temperature Moving Bed Granular Filter
     Development at CPC", In Proceedings 2nd Ann. Contractors Meeting on  Gas
     Stream Glean Up. Morgantown WV. USA, 1982

 5.  Patent nr. GB 2090773, Electrostatically Augmented Granular  Bed  Filter
     for High  Temperature Particulate Removal. GE Co.

 6.  Patent  nr.  US  4038049,  Electrofluidized  Beds  for   Collection   of
     Particulates, MIT

 7.  K.  Zahedi, J.R.  Melcher,  "Electrofluidized Beds  in  the  Filtration  of
     Submicron Aerosol", JAPCA, 26_,  1976, p. 345

 3.  P.W.  Dietz,   "Electrostatic  Filtration  of  Inertialess  Particles  by
     Granular  Beds",  J.  Aerosol Sci. 12, 1981, p. 27

 9.  T.W. Kalinowski, D. Leith, "Aerosol Filtration  by  a  Cocurrent  Moving
     Granular  Bed;  Penetration Mechanisms", Environ. Int. §_, 1981, p. 379

 10.  G.M. Colver,  "Bubble Control  in Gas Fluidized Beds with Applied Electric
     Fields",  Power  Technol.  17,  1977, p. 9

 11.  C.  Gutfinger,  G.I.  Tardos, "Theoretical and  Experimental  Investigations
     on  Granular  Bed  Dust Filters",  Atmos. Environ.  13, 1979, p. 853

 12.  M.  Shapiro,  Aerosol Filtration  by  Electrostatically  Enhanced  Granular
     Bed Filters,  PhD Thesis,  Israel Institute of Technology, Haifa  (Israel),
     1984

 13.  Patent nr.  EP  88201469,  Filter  Device, TNO

 14.  A.  Mizuno,  J.S.  Clements, R.H.  Davis,   "A  Method  for   the  Removal  of
     Sulfur   Dioxide  from   Exhaust  Gas Utilizing Pulsed  Streamer Corona  for
     Electron Energization",  IEEE Trans, on Ind.  Appl.  LA-22,  1986,  p.  516
                                     35-13

-------
                    NESTED FIBER FILTER FOR PARTICULATE CONTROL
                       Robert D. Litt and H. Nicholas Conkle
                                      Battelle
                                 Memorial  Institute
                                  505 King Avenue
                              Columbus,  OH  43201-2693
ABSTRACT

The Nested Fiber Filter system is based on the concept of a nest of needle-like
fibers made from an alloy suited for the application.  Filter action results from
particulate matter forming dendritic, or chain-like structures on the surface of the
fibers.

Baseline parametric tests were conducted to determine the technical feasibility of
the system.  The major test parameters included:  face velocity (0.5-1.5 m/s),
temperature (20-850 C),  pressure (1-6 atm),  and particulate loading (1000-10,000
ppm).   Particulate capture levels over 99.9 percent with 5 micron particles at 200
mm water pressure drop were achieved.  DOE supported developing the concept of
nested fibers for use as a high-temperature filter for integration with advanced
power systems.   The technology is also being evaluated for conventional particulate
control  applications and special environmental control requirements.

The major advantages of the Nested Fiber Filter system included high throughput,
high  efficiency, low pressure drop, and compact equipment size.  Reduced capital and
operating costs result from these technical  advantages.
                                        36-1

-------
                    NESTED FIBER FILTER  FOR PARTICULATE CONTROL
INTRODUCTION

Battelle is developing the Nested Fiber Filter (NFF)  as a high throughput
particulate filter with the potential  to be significantly smaller and less
expensive,  as low as 50 percent,  than  conventional  bag filters or electrostatic
precipitators.  The Nested-Fiber Filter (NFF)  system is based on the concept of a
nest of needle-like fibers made from an alloy  suited for the application.  Filter
action results from particulate matter forming dendritic, or chain-like,  structures
on the surface of the fibers as illustrated in Figure 1.  The fibers form an
interlocking nest with sufficient structural  integrity to maintain high voidage and
low pressure drop over a wide range of operating conditions.

Baseline experiments were conducted to determine the technical feasability of the
system.  Particulate capture levels over 99.9  percent with 5 micron particles at 230
mm (9.3 in.) water pressure drop were  achieved.   A system for implementing the
concept of nested fibers for use as a  high-temperature filter was desired for
integration with advanced power systems.  The  general process concept is illustrated
in Figure 2.

The development  approach has focused on the high-temperature, high-pressure  (HTHP)
applications  because of DOE's interest and funding.  Most of  the work and results to
be presented  are based on the HTHP testing but are generally  applicable  to other
applications.  Battelle has evaluated other applications which will also be
di scussed.
                                        36-2

-------
TECHNICAL DISCUSSION

Experimental Facility

A NFF test module was fabricated from 6-inch (159 mm) diameter, stainless steel
pipe.  A schematic diagram of the simplified test facility is shown in Figure 3.  An
existing air supply, flyash feeder, instrumentation and auxiliary equipment were
integrated to operate over the following range of conditions:

     •    Temperature:  Ambient to 900 C (1650 F)
     •    Pressure:  1 to 6 atm absolute (0 to 70 psig)
     •    Face Velocity:  0.5 to 1.5 m/s (100 to 300 ft/min)
     •    Bed Depth:  100 to 356 mm (4 to 14 inches)
     •    Particulate Loading:  1000 to 10,000 ppm.

Flyash was proportionately fed into the air stream, heated and introduced below the
NFF test module in a mixing zone at the bottom of the vessel.  Pressure drop was
continuously monitored across the filter.  Inlet and outlet sample points were used
to determine particulate loading (by EPA Method 5) and particle size distribution
(by Andersen Impactor®)  periodically during the testing.   The weight of flyash  fed
and that collected on the NFF were measured to establish a material balance that
typically agreed within 95 percent.  Cooling by natural convection was provided at
the NFF outlet so that an ultimate filter (glass filter paper with 0.02 /im openings)
could be used to accurately determine particulate capture/loss.

Tests were conducted for varying time periods depending on the established test
parameters.  Tests were stopped when the pressure drop across the NFF  reached
500 mm H20 (20  in.  W.C.)  including  the perforated support plate.   This was generally
equivalent to filling 30 percent of the available void volume within the  NFF module.
Increased particulate loadings were observed at high filter loadings  and  pressure
drop.
Experimental Results

Approximately 100 tests have been conducted to evaluate particulate  capture  as  a
function of expected operating conditions.  The term collection  efficiency  is  used
as the percent of mass collected in the NFF.  Data are presented  as  particulate loss
                                        36-3

-------
to provide greater sensitivity; particulate loss equals 100 percent minus collection
efficiency.

The test results show that face velocity is the most significant filter parameter.
Bed depth is also important, but less significant than face velocity.  The remaining
parameters (particulate loading, particle size, fiber length and diameter, void
filled, temperature and pressure) produce only minor effects.  The tests  show that a
180- to 250-mm bed of 13-mm-long fibers operated at 0.5 to 1 m/s face velocity with
< 44-/jm flyash result in collection efficiencies ranging from 99.0 to 99.9 percent
with 200- to 300-mm-water pressure drop.  Size analysis of the flyash leaving the
NFF indicates that all material was smaller than 3 /*m in diameter.  This, combined
with the high-collection efficiency, indicates that the NFF output ranges from 5 to
30 ppm and can thus meet both the New Source Performance Standards and turbine
particle-size/loading specifications.

A series of high-temperature tests evaluated the NFF performance up to 1570 F.
These tests confirm the expectation that outstanding particulate capture  is
maintained at high face velocity and low-pressure drop.  Table 1 summarizes results
from tests at high temperature and high pressure.

Flyash from the Grimethorpe PFBC was used to test collection efficiency with a
material known to have a small average particle size (4.6 /im).  The AFBC  ash used
for most of the tests had a mean size of 6.2 /im and a broader size distribution.
The NFF collection efficiency was excellent with both flyash particulates.
 Face Velocity.  The accumulated data indicate that when face velocity is at or below
 0.5 m/s  (100 ft/min), particulate losses on the order of 0.1 to 0.5 percent were
 achieved over a range of conditions.  Increasing the velocity to 1 m/s resulted in a
 modest increase in losses from 0.5 to 1.0 percent.  Particulate losses increase to
 unacceptable levels above 1.5 m/s as shown on Figure 4.  Even at its lower value,
 the NFF  face velocity is significantly greater than most other HTHP particulate
 control  devices, see below.
                                        36-4

-------
                                          Face Velocity (m/s)  for 1°
                    Device                     Particulate Loss
          Nested Fiber Filter                    0.5 - 1
          Ceramic Cross — Flow Filter             0.03-0.09(1)
          Granular Bed Filter                    0.5 -0.6(2)
          Woven Ceramic Bag Filter               0.02-0.05(3)
          Ceramic Candle Filter                  0.03-0.06(4)
          Electrostatic Precipitator             0.6 -1.2(5)
The higher face velocity possible with a NFF participate removal device means that
less filter volume is required.  This translates into smaller equipment requirements
and lower capital costs.
Bed Depth.  The second most important variable is bed depth.  Figure 5 shows
representative data.  A depth of 180- to 250-mm of 13-mm fibers appears adequate to
provide sufficient collection surface to meet NSPS and/or turbine specifications.
Such shallow bed depths (compared with 600 mm for granular bed filters) mean that a
very compact filter and low pressure drop are possible.
Temperature.  The test data indicate that elevated temperature does not adversely
affect collection efficiency over the 285 to 936 K (54 to 1226 F) range tested.
Figure 6 presents a subset of the data.  This result was anticipated based upon the
offsetting influences of gas density and viscosity.(6-7)   At  temperatures  above  1144
K (1600 F),  the flyash begins to soften and adhere to the perforated support plate.
An alternative gas entry arrangement avoided problems with the fiber support plate
and provided excellent collection efficiency consistent with lower temperatures.
Pressure.  Elevated pressure, over the range of  1 to 6 atmospheres absolute,  shows
no statistically significant effect on capture efficiency.   Figure 7  presents  a
subset of the data showing the effect of pressure on particulate  loss.   Pressure
drop, however, increases with pressure due to the increased  gas density.   While  the
pressure range was limited by the experimental facility, the limited  data  are
consistent with the expectation that higher pressures will not  affect collection
efficiency.
                                        36-5

-------
Cleaning/Regeneration.  Three techniques have been tested for cleaning/ regenerating
the NFF:  rotation, vibration, and pulse-jet flow.  Vibration is the most effective
cleaning technique for the present test modules.  Cleaning is being further
evaluated to establish an integrated system for scaleup to pilot plant testing.
ECONOMIC EVALUATION

A preliminary economic evaluation of the NFF was made based on the above
experimental data.  The evaluation considered both a high-temperature, high-pressure
application, (pressurized fluidized bed combustor) and a conventional coal-fired
boiler with a baghouse.
Pressurized Fluidized Bed Combustor

Three NFF modules were designed to match the three combustors and three turbines in
a 660 MUe PFBC reference pi ant. ^  A stage of primary cyclones between the
combustor and NFF removed the largest particle sizes and reduced the inlet  loading
to 2150 ppm.  The NFF face velocity was set at 1 m/s (200 ft/min) and the bed depth
was 254 mm  (10 inches).  These conditions provide 99.5 percent collection efficiency
which meets the gas turbine requirement and NSPS.  The filter pressure drop was
estimated at 250 mm H20.   A pressurized containment vessel  (13.5 feet diameter x 40
feet  long)  housed the NFF beds.  This is less than half the volume required for  a
ceramic cross flow filter to meet the same performance.^

Capital cost of the NFF was estimated to be less than $20 million for a module.
This  is roughly half the projected cost for a ceramic cross-flow filter  ($35
million) and less than 40 percent of the baseline plant cost for 3 stages of
cyclones plus a downstream baghouse ($52 million).(1)
 Conventional  Coal-Fired  Boiler

 An  NFF  conceptual  design was prepared for a 500 MWe pulverized-coal-fired  boiler.
 Four modules  were  used with off-line, in-situ cleaning.  The design was  based  on
 1 m/s  (200 ft/min) face  velocity and 254 mm (10 inches) bed depth  to  provide 99.5
 percent collection efficiency at approximately 250 mm  H20 pressure drop.   This
                                        36-6

-------
conceptual design was compared to an actual baghouse on an operating 500 MWe boiler
for which data were available.  The NFF volume was less than 10 percent that of the
baghouse with an air/cloth ratio of 2.5.

The projected capital cost of the NFF would be less than half the cost of the
baghouse.  Variable operating costs are expected to be comparable.
FUTURE PLANS

DOE continues to support the NFF development for high-temperature, high-pressure
applications.  Current efforts involve evaluation of alternative cleaning
techniques, scaleup to a pilot plant and evaluation of engineering and economic
merits.  This work should conclude in early 1991 with subsequent large-scale testing
or demonstration.

Battelle plans to commercialize the technology with an industrial manufacturing
partner to capitalize on both future and immediate markets.  Conventional
particulate control applications represent an immediate market for which the NFF
presents significant opportunities to reduce the capital  cost relative to state-of-
the-art equipment.

Battelle is also investigating other applications of the NFF for submicron particle
capture, integrated emissions control (especially NOX)  and  catalytic  reactions.
Submicron particle capture is a difficult problem for many industries which are
faced with increasing environmental, health and safety regulations.  The NFF may
offer an advanced particulate control device which collects the submicron particles
primarily through diffusion.

The high surface area to volume ratio of the fibers makes the NFF a good candidate
for surface specific catalytic reactions.  NOX reduction  from combustion flue gases
is one such application where the fibers are the catalyst.  There are several
differences from state-of-the-art selective catalytic reduction but one  advantage is
the ability to operate in a particulate laden gas stream and regenerate/clean the
fibers to maintain the catalyst activity.
                                        36-7

-------
CONCLUSIONS


The NFF offers significant potential to reduce equipment volume and capital cost
compared to conventional particulate control equipment.  The high throughput, high
voidage and compact size provide outstanding particulate collection efficiency at
reasonable pressure drops.  The NFF can be designed for high-temperature or
corrosive environments by selecting suitable materials of construction.


The small-scale testing to date has been very encouraging and plans for larger-scale
testing are in progress.  This NFF offers high potential for new technology for
particulate control in some very challenging applications as well as conventional
applications.
ACKNOWLEDGEMENTS


The authors wish to acknowledge the U.S. Department of Energy for their interest and

support of the NFF technology.  Tom Dorchak, the DOE Technical Project Officer, has

been especially helpful in guiding the early research efforts.  The authors also

thank the many additional Battelle staff who have contributed to this work.
REFERENCES


  1.   Rubow, L. N., et al, "Technical and Economic Evaluation of Ten High-
      Temperature, High-Pressure Particulate Cleanup Systems for Pressurized
      Fluidized Bed Combustion," Report prepared by Gilbert/Commonwealth for the
      U.S. Department of  Energy, Morgantown Energy Technology Center, Contract No.
      DE-AM21-82MC19196,  DOE/MC/19196-1654 (July, 1984) p 4-25.

  2.   Ibid, p 4-43.

  3.   Ibid, p 4-11.

  4.   "Gas Cleaning Technology for High-Temperature, High-Pressure Gas Stream," EPRI
      CS4859 (October, 1986).

  5.   Ibid, p 6-30 through 6-37-

  6.   First, Melvin, W.,  "Gas Cleaning Systems for the High Temperature, High
      Pressure Fluidized  Bed Combustor," Journal of the Air Pollution Control
      Association, 35:1286-1297  (Dec., 1985^	

-------
"Development of a Novel Nested-Fiber Filter Concept for High-Temperature and
High-Pressure Physical Cleanup, Design Alternatives Assessment Report (Task
1)," prepared by Battelle Columbus Division for U.S. Department of Energy,
Morgantown Energy Technology Research Center, Contract No. DE-AC21-86MC23251
(March 2, 1987).
                                   36-9

-------
CO
01
                 Cleaned Fiber
                        Initial Capture of Particles
                                      Development of Branching
                                      Chain  Structures (Dendrites)
                                                                 Well Developed

                                                                 Dendritic Structures
                            Figure 1.  Formation and Growth of Dendrites  on An  Individual  Fiber

-------
Fiber Filtration Concept
  Collected Particles
  After Regeneration
                             Fiber Bed
                                       Dirty Gas  IN
                                      Clean Gas Out
   Fi gure
Basic Elements of the Battelle Regeneratively
   Cleaned,  Nested Fiber-Filter System Concept
                      36-11

-------
CO
CTl
                                                                                                  TO
                                                                                                        :=:  10-Finned Tube
                                                                                                        : i:    Ai r-to-Ai r
                                                                                                  : = :    :i:    Heat Exchanger
                                                                                                  : i :    :!: :    Tubes
Vent
 t
                                                                                                  m    !!;                    ©     Y
                                                                                                  :E:    ;E;                     T  n
                                                                                  [Electronic Scale
                                        Figure 3.   High  Temperature,  High  Pressure NFF Facility

-------
              0.5
CO
01
              0.4-
(0
§   0.3

O
«J
O   0.2
              0.1
              0.0
         Average Loading- 7,500 ppm
         Bed Depth - 254 mm
         Fiber Length - 13 mm
         Pressure - 2.1 to  5.8 Atmospheres
                                                      o
                                                     63
                                                             D  304 K
                                                             O  646 to 675 K
                                                                 64
                       0.2     0.4    0.6    0.8     1      1.2    1.4     1.6
                                        Face Velocity, m/s
                                                                 1.8
                            Figure 4.  Effect of Face Velocity on Particulate Loss

-------
OJ
en
            (0
            CO
            O
            _J

            
-------
              0.5
OJ
CD
              0.4 H
CO
S  0.3-
_l
5,000 ppm
               Bed Depth - 254 mm
               Fiber  Length - 13 mm
               Pressure - 2 to 6 ATM-ABS

                     D  1 m/s Velocity
                     O  0.5 m/s Velocity
                         0.2
0.4
                               0.6
0.8
1.2
1.4
                                  Temperature, °F (Thousands)
1.6
                   Figure 6.  Effect of Temperature on Particulate Loss as A Function of Velocity

-------
CO
en
             0.8-
                  Loading- >5,000 ppm

                  Bed Depth - 254 mm

                  Fibers - 13 mm
          0)
          (0
          o


          CD
0.6
             0.4-
         cd
         Q.
             0.2-
                                            n  0.5 m/s Velocity

                                            O  1.0 m/s Velocity
                                            43 o
          39
                     41
                                  40
                                     u
                                             42 °  45
                                         n D 46
                                     234

                                       Pressure, atm-abs

                             Figure 7.  Effect of Pressure on Particulate Loss
                                                             49 n  56
                                                            48n
                                                              47 n
                                                                   6

-------
                           Table 1

           HIGH-TEMPERATURE AND  HIGH-PRESSURE  TESTS



Test No.                        62       63       64        70
Flyash                        AFBC     AFBC     AFBC      PFBC
Outlet Temperature, °F          756      745      746      1217
Inlet Temperature, °F         1100     1100     1100      1570
Pressure,  atm                   5.5      2.1      2.2       2.0
Face Velocity, m/s              0.5      1.1      1.5       1.1
Pressure Drop, mm H20           200      750      750       625
Collection Efficiency, %      99.96    99.96    99.92    99.93
Outlet Loading, ppm             2.7      2.6      4.4      3.7
                          36-17

-------
            PARTICULATE EMISSIONS FROM PREPARED FUEL (RDF)
                     MUNICIPAL WASTE INCINERATORS
                                R. M. HARTMAN
                        RESOURCE RECOVERY SYSTEMS
             ASEA BROWN BOVERI/COMBUSTION ENGINEERING, INC.
                            7 WATERSIDE CROSSING
                         WINDSOR, CONNECTICUT 06095
ABSTRACT

Municipal waste to energy facilities have three distinct sources of particulate
emissions to the environment: boiler stack emissions, fugitive dust emissions and
emissions from waste processing facilities.

The paper will review each of these sources of particulate emissions at an RDF facility
(Mid-Connecticut) in Hartford, CT. Boiler stack emissions at that facility are controlled
by a lime spray dryer absorber and fabric filter baghouse.  Fugitive dust emissions are
controlled by appropriate equipment design and operating procedures. Emissions
from RDF processing are controlled by enclosure of equipment and routing of air
containing high dust loads to cyclones and baghouses.

The paper provides particulate, mercury and dioxin and furan emission rates and
removal efficiencies for RDF boiler stack emissions. It also discusses the association
between combustion and flue gas  cleaning conditions and particulate, mercury and
dioxin/furan boiler stack emissions at the Mid-Connecticut facility.  The paper also
reviews specific sources of fugitive dust emissions and means for controlling fugitive
dust at an RDF facility. Finally, the paper presents particulate data from the waste
processing portion at an RDF facility.
ACKNOWLEDGEMENTS

I would like to thank the many scientists from Environment Canada, U.S. EPA, and
contractors such as Alliance Technologies Corporation and MacLaren Plansearch.
In particular, I am grateful for the help of Mr. Abe Finkelstein from Environment
Canada, Mr. James Kilgroe and Dr. Theodore Byrna from U.S. EPA, Mr. Edward Peduto
from Alliance Technologies and Mr. Glenn Gross from Connecticut Resources
Recovery Authority in the review of this paper.
                                     37-1

-------
                 PARTICULATE EMISSIONS FROM PREPARED FUEL (RDF)
                          MUNICIPAL WASTE INCINERATORS
INTRODUCTION

Municipal  waste  incinerators,  or municipal waste  combustors (MWC),  have  three
distinct sources  of particulate  emissions  to the  environment.   One  source is
particulate emissions  which may  be  produced  at  the waste  receiving  or  waste
processing  area.  Another is fugitive dust  from miscellaneous  equipment such as
lime  storage  silos,   ash  conveyors  and  ash  loadout  equipment.   The  most
significant  source  of  particulate   emissions,  however,  comes  from  combusting
refuse in the  boilers  and ultimately emitted  from  the  boiler stacks.   Each of
these sources  of particulate emissions  will  be reviewed  using  the  Asea  Brown
Boveri/Combustion Engineering's  municipal waste cotnbustor as the  reference type.

RDF FACILITY DESCRIPTION

One example  of a modern RDF facility  is the Mid-Connecticut  Resource Recovery
facility in  Hartford,  Connecticut.   Most of  the  information presented  in this
paper is based on this facility.

The  Mid-Connecticut  facility is  owned  by  the Connecticut Resources  Recovery
Authority.   The  facility contains  a waste  processing  plant  and  an  RDF  power
plant.   The power  plant,  which is operated  by  Asea  Brown  Boveri  Resource
Recovery Systems (ABB/RRS),  contains three ABB/CE  boilers, each consisting of an
RDF spreader stoker, a  natural  circulation  water-wall boiler,  a  superheater, an
economizer, and a tubular combustion air preheater.   The boilers  are designed to
burn  coal  or  RDF.   The  design  steam flow  from each boiler is  231,000 Ibs/hr.
when burning RDF with an average Higher Heating Value of 5690 BTU/lb.

The fuel burning portion for each boiler  consists of an RDF injection system, a
traveling grate stoker, and  a combustion air system  (Fig. 1).  RDF is
                                       37-2

-------
pneumatically injected through 4 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
over-fire air (TOFA)  system  and a wall  system.   The  tangential  system consists
of  four tangential  overfire  air  windbox   assemblies  located  in the  furnace
corners.  Each TOFA  assembly contains three elevations of nozzles  which can be
manually adjusted in the  horizontal  plane.   The  wall  system  contains  one row of
overfire air ports on the front wall and two rows on the rear wall.

The ABB/Combustion  Engineering  designed  flue gas cleaning system  consists  of a
lime-based  spray  dryer  absorber  (scrubber)  followed  by a  reverse air  fabric
filter baghouse. The  scrubber is  capable of controlling the  temperature  at the
baghouse  inlet  and  the  sulfur  dioxide (S0?)  concentration at  the  baghouse
outlet.   The  baghouse  inlet temperature   is  controlled  by  the  lime  slurry
flowrate.   The   SCL   removal  rate   is  controlled   by  adjusting   the   lime
concentration in the lime slurry  feed.   Each baghouse  has  12  compartments,  each
with 168 Teflon-coated glass  fiber bags.

BOILER PARTICULATE EMISSIONS

RDF facilities  generally produce a  greater mass concentration of  uncontrolled
boiler particulate  emissions (that  amount of particulate existing  prior  to the
air pollution control  devices)  than mass  burn  or rotary kiln combustors.   The
primary reason for this  is that RDF boilers  incur some suspension burning of the
fuel,  while  other  types  of  MWC's push  the  fuel  onto a grate  in a  large  mass
where it is burned.  In  suspension  burning,  the  light  ash  formed  by the rapidly
combusted heterogeneous  fuel  tends  to fly  with  the flue gas  stream  instead of
remaining on the grate.  This permits the .units to operate at a lower excess air
level  than most mass burn units.

Studies  at  the  Mid-Connecticut RDF  facility have demonstrated that  boiler fly
ash (not including economizer ash)  represents  25-32%  by  weight of the total ash
produced by the  facility.  Most mass  burn  facilities with  a  scrubber baghouse
produce boiler  fly  ash that  constitutes  approximately 15-20% by  weight of the
total  ash produced.  Method  5 particulate tests  of  the Mid-Connecticut flue gas
stream leaving  the  boiler,  but before the  air pollution control  devices,   show
uncontrolled particulate  loading  around 4.0 gms/Nm  at 12%  C09   (Approximately
                                        37-3

-------
2.1 gr/dscf).   Most  modern mass  burn  facilities have  uncontrolled particulate
emissions  around  2.8  gms/Nm   (approximately  1.5  gr/dscf).    With  use   of  a
scrubber  baghouse,   these  uncontrolled  particulate  emissions  can readily  be
reduced  by  99.90%.   Use  of  an  electrostatic  precipitator  designed   for  high
efficiency   particulate   removal   can   achieve   overall   particulate  removal
efficiencies very similar to baghouses.

Extensive emission  studies,  funded by  U.S.  EPA  and  Environment Canada  at the
Mid-Connecticut  facility,  were  conducted in January  and February  1989.   Those
studies included 13 different tests on one of the three identical boilers, under
conditions  of  changing  steam  load,   good   and   intentionally poor  combustion
conditions (using CO as  a  criterion  of good or  poor  combustion  conditions) and
changing  temperature   and  lime  addition   rate  to   the  scrubber/baghouse.
Table l' ' shows  the preliminary compiled results  of these 13  different  tests
for particulate  and various  metals  at  the  scrubber  inlet (SDI)  and baghouse
outlet  (FFO).  From examining this table, it is  possible to  conclude:

    1.    The  concentration  of  uncontrolled  particulates  is  reasonably
          uniform and does not appear  to be  affected by changing steam load
          or boiler combustion conditions.

    2.    The  particulate   removal  efficiency  of  the  baghouse  is   very
          consistent,  staying  between  99.7  and  99.9%  removal  efficiency
          under  loads  varying  from  85  to  112%  of normal   full  load  and
          changing  combustion  conditions.   A  slight  drop  in  particulate
          removal  efficiency  to  99.5% was  measured  when  steam flow was
          reduced to 70% of full steam load.

    3-    All  of the  metals  tested  appeared to attain  approximately the
          same removal  efficiency as  achieved on  particulates,  also showing
          little effect  of change in  load or combustion  conditions on the
          concentrations  of  uncontrolled  metals  produced  or on  removal
          efficiency.

Examination  of  dioxin  and  furan emissions at  the  Mid-Connecticut  facility
supports  the  hypothesis  that  these  uncontrolled emissions become  attached  to
particulates  and may  be  removed with  the  same  removal  efficiencies as  the
particulates themselves  (see preliminary  results  in Table 2).   These dioxin and
                                        37-4

-------
furan  removal  efficiencies' '  are  somewhat higher  than  attained  at  mass burn
facilities with  scrubber baghouses  which are  reported  to  average around  93-97%
removal  efficiencies.    One   possible   explanation   for  the  greater  removal
efficiency at  RDF facilities  is  the higher mass  concentration  of uncontrolled
particulates relative to mass  burn  facilities.  These uncontrolled particulates
may  offer convenient  nuclei  onto  which  gaseous  pollutants  can condense  as
temperatures are  reduced  below their vaporization point.  There may also be some
active  chemical  absorption of  these pollutants,  since a  small  fraction  (i.e.
2-3%) of the  fly ash contains unburned carbon which  acts  like activated  carbon
and chemically absorbs many gaseous pollutants.

While particulates can act  like a sponge  or filter to help remove some toxic air
pollutants, they can also  do the  opposite.   It  has  been  shown  in  laboratory
experiments by Karasek^  '  and Hagenmaier' ',  that fly  ash  laced  with  carbon  13
labeled chlorophenols and chlorobenzenes, when held at temperatures in the  range
of approximately  450-670°F  can form dioxin  and furan molecules.  This  data  helps
explain why many  of  the  older facilities  (i.e. Niagara Falls,  Albany,  etc.) with
hot ESPs have higher concentrations of dioxin leaving the ESP than entering.

Particulate emissions can help form certain toxic organic compounds and they can
help remove such  compounds.  What the Mid-Connecticut preliminary results appear
to demonstrate  is that  by reducing the  uncontrolled  particulates compared  to
older  RDF  units  (i.e.  from  4.0  to  2.1  gr/dscf)  and  by maintaining  better
combustion  conditions   in  the  boiler   as  a  result   of  improved  air  and  fuel
distribution and  mixing, the  uncontrolled  dioxin  concentrations  going  into  the
scrubber can be  reduced  from  approximately  2,000 ng/Nm   for PCDD/PCDF  to around
             o
400-500 ng/Nm .  Although the uncontrolled dioxin concentration appears somewhat
high compared to modern  mass burn facilities, it is this writer's belief that by
reducing the  flue gas  temperature  to  below  300°F,  additional benefits may  be
gained  by  utilizing the properties  exhibited by  the  higher  concentrations  of
particulates in  an  RDF  system, i.e. greater surface area  available  for  nuclei
condensation,  and to some degree their increased absorption potential.  Hense,  a
modern RDF facility may  achieve even greater removal  efficiencies with PCDD/PCDF
concentrations as low,  or lower  than  mass  burn  systems  in  spite   of  having
slightly higher PCDD/PCDF concentrations  at the  scrubber inlet.
                                        37-5

-------
The  above  preliminary  findings  show  if participate  emissions  are  controlled
below .015 gr/dscf at 7% CL, with the use of scrubber baghouse technology, there
is excellent control  of all other metals.

In analyzing the Mid-Connecticut data, one  might  surmise  that the improved flue
gas removal efficiency  for dioxin  and heavy metals  at  Mid-Connecticut  is based
on simply  transferring  such  pollutants  onto and into the  fly ash.   The results
of  the   state   of  Connecticut's  EP   toxicity   tests  of  the   ash   at  the
Mid-Connecticut RDF facility  and  at  the  nearby Bristol, CT  mass  burn facility,
both burning  refuse  from  essentially the  same waste stream,  consistently show
slightly   lower   values   for   lead,   cadmium,    arsenic,   and   mercury   at
Mid-Connecticut.   Heavy  metals  should logically  be somewhat  lower  in  RDF  fly
ash,  since the front-end waste processing system removes a significant amount of
these metals prior to combustion.

The relative  differences  of dioxins  in  RDF and mass burning ash is harder  to
demonstrate because there is little published data on dioxin-ash concentrations.
The test  data  at  Mid-Connecticut  during good  combustion conditions  indicates
total PCDD + PCDF concentrates in fly ash ranges from 74 ng/g (0.074 ppm) to 213
ng/g (0.213 ppm).   From the  few  available  data points for  mass burn  fly ash,
their total concentrations appear to be about the  same.   Bottom ash  has much
lower  dioxin   concentrations  than   fly  ash   and   was  hardly  detectable  at
Mid-Connecticut (<0.1 ng/g).  Thus,  the  combination of  two  parts of  bottom ash
to one  part of fly  ash,  which is  roughly  in  proportion  to what  is produced,
means  that the average  PCDD +   PCDF  concentration  in   the  combined  ash  is
approximately 50 ng/g, or less than .05  ppm.

FACILITY FUGITIVE DUST EMISSIONS

At a  typical  modern  RDF  facility, or  any  other municipal  waste combustor,  a
second source of particulate emissions is fugitive dust  emissions from:

    1.     ash  storage/load out buildings
    2.     1ime storage areas
    3.     conveyors transporting  ash or  process residue  to
          storage/loadout buildings
    4.     debris that falls on plant roadways from transport of refuse or ash
          to or from the plant.
                                        37-6

-------
Most MWC's  handle  bottom ash wet,  since this ash typically  drops  off the grate
into a water quench tank.  Thus, unless  bottom  ash  is held in storage and has a
chance to dewater, bottom  ash  does  not  contribute  to fugitive dusting.   Fly ash
however, is typically collected  in  a  dry form and,  since it  has the consistency
of fine  face  powder,  can  easily become  a fugitive  dusting  problem.   Thus,  all
conveyor transport of fly  ash  is done in enclosed  systems.   Also,  at some point
prior to  mixing fly ash  with  bottom ash,  the  fly  ash  is wetted to  about  25%
moisture.   The  fly  ash  and  bottom  ash  are  then  typically combined  prior  to
loadout in  ash  trucks.   To reduce  fugitive dusting,  the  combined  ash  should  be
directly loaded into trucks  or handled  in an enclosed  building.   That  building
should have misting  sprays available over the drop  chutes to minimize  fugitive
dusting should  the combined ash become  dry  and start  causing  fugitive dusting
while being loaded into trucks.

All  storage of  dry reagents  used  at   a  facility,  such  as  hydrated  lime  or
limestone  used  in  scrubbers,  should  be conveyed  pneumatically   from  delivery
trucks or  railroad tank  cars and put into fully enclosed  silos with  a  baghouse
to filter air displaced as the silo is filled.

A  daily  cleanup program  should  be instituted  around MWC facility  roadways  to
pick up debris  that has  fallen from trucks and,  when necessary, to wash or hose
down  the   roadways.   By   following   these   fugitive  dust   control   measures,
particulate emissions  from these sources  should be  too  low to measure  and  no
different from one type of MWC to another.

WASTE PROCESSING FACILITY  PARTICULATE EMISSIONS

RDF  facilities   process  municipal  waste   to   make a  prepared  fuel  with  a
significant  portion  of  the  non-combustibles removed.   Hence,  the  processing
system becomes an added source of particulate emissions.  In order to understand
this added  source  of particulate emission,  a short  description of the workings
of one type of waste processing facility is in order.

In  the  ABB/CE  waste processing  facility,  refuse  is  delivered  by packer  or
transfer  trucks.   After  weighing,   the  waste  is   discharged  onto  the tipping
floor within the enclosed  waste  processing  building.  The refuse   is then  pushed
onto apron conveyors which convey the refuse  to the  processing equipment.
                                        37-7

-------
The  first  piece  of  processing  equipment  in  the  process  line  is  the primary
shredder.  Prior  to  the refuse  entering  the primary shredder,  it  is  inspected
again,  by  a  plant  operator,  for  inappropriate,  dangerous  or  potentially
hazardous items.

Primary shredders are typically  low  horsepower flail  mills that  break open bags
and coarse shred material to a 6 to 12 inch size.  Each shredder will handle 100
tons  per  hour,  and  is housed  in  a reinforced concrete room,  with  a "rip-open"
canvas roof,  to vent any explosions.

The  waste next  passes   a  magnetic  separation  system,  which removes  magnetic
ferrous materials.  From there,  the  waste  enters a  two-stage rotary trommel,  of
ABB/C-E  design,  where  most  of  the non-combustible  material  is   removed  for
disposal  and  where   some   initial  materials  sizing  takes  place.   The  sized
material  stream  from the  first  trommel,  which   contains  mostly  combustible
material less than 4" as well  as a small  quantity of sand, glass, grit and other
non-combustibles,  is   passed   through   a  second   trommel,   where  additional
non-combustible materials are removed.  The properly sized combustible materials
from  this  separation  are  conveyed  to  the  RDF storage  room.  Meanwhile,  the
larger  combustible materials  that did not pass  through  the holes  of  the  first
rotary  trommel  are conveyed  to the secondary  shredder, where  they  are shredded
to the  appropriate fuel  size.  This  shredded  and sized fuel  is also conveyed  to
the RDF storage room.

Each  RDF process line  has a  particulate  collection  and control   system.   The
primary  shredder  in  each line has a  dust  control  system  comprised  of  a fabric
filter  baghouse and exhaust fan with a stack which vents  through the roof.

Each  secondary  shredder has  a  control   system consisting  of a   cyclone  and
baghouse.   The  cyclone  receives  dust-laden  air   purged   from  the  secondary
shredder, trommels and ferrous metal  air classifier.   Air from the  cyclone then
passes  through  a  dedicated  baghouse,  from  which  clean  air is discharged through
an exhaust fan  and stack to the atmosphere.

Method  5 particulate emission  tests carried out  on  each  of the  two primary
shredder vent  stacks  at Mid-Connecticut showed  particulate emissions averaging
             (4)
0.003gr/DSCFv   while the two secondary shredder vents averaged 0.0016 gr/DSCF.
                                        37-8

-------
CONCLUSION

The preliminary  test results presented  from the Mid  Connecticut  facility show
that RDF  facilities with  scrubber/baghouse  technology can achieve  some  of the
highest  removal  efficiencies  for  particulates,  trace metals,  and  dioxin  and
furans shown  for  any type of municipal waste  combustor.   This  data shows, that
even with intentionally poor combustion conditions and varying steam load, stack
emissions for particulates and these other pollutants  are extremely low.

The other  two sources  of particulate emissions,  fugitive  dust and  vents from
waste  processing  equipment,  at  the  Mid Connecticut  facility  show  negligible
particulate emissions.
                                        37-9

-------
REFERENCES
1.    Unpublished data  from  a  study  of Environmental  Characterization  of RDF
     Combustion Technology   Mid-Connecticut Facility, Hartford, CT.  This  study
     was sponsored  by  Environment  Canada and  U.S.  EPA and the  whole  report is
     expected to be published in the summer of 1990.

2.    Model   Studies  of   Polychlorinated  Dibenzo-p-Dioxins   Formation  during
     Municipal  Refuse  Incineration.   Report,   F. W.  Karasek and L.  C.  Dickson,
     University of Waterloo, Waterloo Ontario, 1987.

3.    H. Hagenmaier, M.  Kraft, H.  Brunner and R. Haag.  "Catalytic Effects of Fly
     Ash from  Waste  Incineration  Facilities  on the  formation  and  Decomposition
     of  Polychlorinated   Dibenzo-p-Dioxin  and  Polychlorinated  Dibenzofurans."
     Environmental  Science and Technology,  Vo. 21,  No. 11, 1987.

4.    Air   Emissions   Compliance   Test   Results   and  Related   Data    at   the
     Mid-Connecticut  Resource  Recovery  Facility   Waste  Processing  Facility.
     Report, TRC Environmental  Consultants,  April  22-26,  1988.
                                       37-10

-------
                                                         TABLE 1

                                 MID-CONNECTICUT UNCONTROLLED AND CONTROLLED EMISSIONS*,,,
                              AND REMOVAL EFFICIENCIES FOR PARTICULATE AND TRACE METALSU;
                                      GOOD CONDITIONS/
                                         STEAM LOAD
POOR CONDITIONS/
   STEAM LOAD

STEAM LOAD %
(100% = 231,000
Ibs/hr)
PARTICULATE
SDI
FFO
% Removal Eff.
ANTIMONY
SDI
FFO
% Removal Eff.
ARSENIC
SDI
FFO
% Removal Eff.
CADMIUM
SDI
FFO
% Removal Eff.
CHROMIUM
SDI
FFO
% Removal Eff.
LOW
70%

3.45 X 10,
6.2 X 10J
99.5

113
100.0

205
100.0

573
100.0

1050
22.0
97.9
INTERIM
85%

4.98 X 105
4.9 X 10J
99.9

120
100.0

240
100.0

584
100.0

983
12.4
98.7
NORMAL
100%

4.2 X lo!j
5.1 X 10J
99.7

135
100.0

211
100.0

694
100.0

984
16.9
98.3
HIGH
112%

3.4 X lo!?
4.0 X 10J
99.8.

173
100.0

247
100.0

562
100.0

745
7.6
99.0
INTERM
87%

4.46 X lo!j
3.9 X 10J
99.9

122
100.0

230
100.0

527
100.0

623
14.8
97.6
NORMAL
100%

4.05 X lol?
5.8 X 10
99.7

60
100.0

186
100.0

552
100.0

539
8.9
98.3
HIGH
110%

3.30 X ID,
2.7 X 10J
99.9

51
100.0

194
100.0

437
100.0

353
7.6
97.8
* CONCENTRATION ug/SnT 0 12% C02

-------
                       TABLE 1 (CONT.)

   MID-CONNECTICUT UNCONTROLLED AND CONTROLLED EMISSIONS*,,,
AND REMOVAL EFFICIENCIES FOR PARTICULATE AND TRACE METALS^1'
        GOOD CONDITIONS/
           STEAM LOAD
POOR CONDITIONS/
    STEAM LOAD

COPPER
SDI
FFO
% Removal Eff.
LEAD
SDI
FFO
% Removal Eff.
MERCURY
SDI
FFO
% Removal Eff.
NICKEL
SDI
FFO
% Removal Eff.
ZINC
SDI
FFO
% Removal Eff.
* CONCENTRATION uc
LOW

2010
100.0

10826
67
99.4

723
12.2
98.3

3381
32.6
99.0

48270
100.0
3/Sm3 @ 12% CO,
INTERIM

1992
100.0

8714
44
99.5

722
7.5
99.0

1417
4.4
99.7

43992
18
100.0
NORMAL

2531
100.0

5164
45
99.1

650
12.0
98.2

805
26.0
96.8

44338
100.0
HIGH

1112
100.0

4036
47
99.8

558
3.2
99.4

523
3.5
99.3

34660
100.0
INTERM

1429
100.0

14286
47
99.7

634
6.8
98.9

2030
2.2
99.9

31169
100.0
NORMAL

1531
100.0

10211
32
99.7

594
14.1
97.6

503
4.9
99.0

35563
41
99.9
HIGH

1264
100.0

7229
37
99.5

583
11.5
98.0

257
3.8
98.5

31029
100.0

-------
                                                  TABLE 2

                          MID-CONNECTICUT UNCONTROLLED AND CONTROLLED EMISSIONS
                            OF PCDD/PCDF AND RESULTANT REMOVAL EFFICIENCIES1'
                                    GOOD CONDITIONS/
                                       STEAM LOAD
POOR CONDITIONS/
   STEAM LOAD

PCDD
SDI
FFO
% Removal Eff.
PCDF
SDI
FFO
% Removal Eff.
LOW

109
0.060
99.9

404
0.145
100.0
INTERIM

228
0.130
99.9

579
0.112
100.0
NORMAL

125
0.333
99.7

591
0.385
99.9
HIGH

67
0.067
99.9

215
0.075
100.0
INTERM

580
0.371
99.9

1281
1.124
99.9
NORMAL

196
0.365
99.8

732
0.336
100.0
HIGH

317
0.346
99.9

885
0.162
100.0
CONCENTRATION ug/Sm  @ 12%

-------
             CONDENSIBLE  EMISSIONS  FROM MUNICIPAL WASTE  INCINERATORS
                        Ashok  S.  Damle  and  David  S.  Ensor
                           Research  Triangle  Institute
                                  P.  0.  Box  12194
                        Research  Triangle Park,   NC  27709
                                   Norman Plaks
                 Air  and Energy Engineering Research  Laboratory
                      U.  S.  Environmental Protection  Agency
                         Research Triangle Park,  NC  27711
ABSTRACT

Incineration of solid waste is an effective method for reducing the volume of waste
requiring ultimate disposal and for recovering possible energy value of the waste.
For successful utilization of incineration technology, resulting air pollution
problems must be adequately controlled.  This paper analyzes in detail emissions of
semivolatile materials, specifically emissions of dioxins,  furans,  and trace metal
species in the incineration process.  Available municipal incinerator emission
sampling data are reviewed, and the measured concentrations are compared with the
saturation  concentrations of the respective species.  Possible options for
controlling these emissions are discussed.
                                         38-1

-------
           CONDENSIBLE EMISSIONS FROM MUNICIPAL WASTE INCINERATORS
1.0  INTRODUCTION

Incineration of solid waste is becoming an attractive waste disposal option due to
inherent liabilities and increasing costs associated with long-term storage or land
disposal.  It is an effective method of reducing the volume of waste requiring
ultimate land disposal and, where the waste has significant heating value, its
combustion can be used for cogeneration of steam and energy.  Such waste combustion,
however, may produce certain air pollution problems which must be dealt with for
successful utilization of the incineration technology.  For example, trace
quantities of toxic organics such as dioxins and furans may be formed by incomplete
combustion, or noncombustible trace metallic species present in the waste may be
emitted by volatilization or along with the fly ash.

High temperature processes such as incineration and combustion often generate
materials which may be in the vapor phase at the process conditions prior to their
emission, but may condense downstream of the process before or after being released
into the atmosphere.  Such condensible vapor emissions are not efficiently
controlled by conventional particulate control devices, and are not measured in an
EPA Method 5 particulate sampling train.  The condensation of vapors in the
atmosphere generates new particulate matter, especially in the submicron size range,
and/or  alters existing material in the particulate phase again predominantly in the
submicron size range (Damle et al., 1987).  The size distribution of the particulate
matter  significantly influences its effects on opacity, its inhalation
characteristics, and its potential health effects.  The submicron size fraction in
the range of 0.1-1.0 /im, diameter, has the greatest effect on obscuration of light
as well  its potential for inhalation.  The condensed emissions may also contain a
number  of toxic chemicals and elements, such as dioxins, furans, and mercury, which
increases their potential health effects.

This paper analyzes emissions data collected in several municipal incinerator
sampling studies and investigates possible options to control emissions of toxic
semivolatile material into the atmosphere.  During data collection the emphasis has
usually been placed on chlorinated dibenzo-p-dioxins and dibenzofurans, and metal
emissions because of their known toxic effects.  This paper will thus focus upon
these same chemical species as in the measurement studies; however, the conclusions
regarding  formation and capture may be applicable to the other condensibles as well.

For this study of condensibles the emphasis was on emissions from municipal waste
incinerators.  This was because municipal waste incinerators are a significant and
very important source of these emissions.  However, there are other important
industrial sources of condensible emissions.  The conclusions reached in this study
on approaches for improving control of condensible emissions would likely be useful
for application to other industrial sources.


2.0  CONDENSIBLE SPECIES

A condensible species may be defined as that which has a significant vapor pressure
at normal air pollution control device operating temperatures at the emission point,
                                         38-2

-------
but a sufficiently low vapor pressure at ambient conditions so that saturation might
occur.  Such a species thus may escape the air pollution control device, but may
condense to form a particulate phase upon release into the environment.  A number of
volatile organics, such as acetone which is used as a solvent, would not be
considered as condensibles by this criterion, because such compounds might not
actually condense upon release due to relatively high vapor pressures at ambient
conditions.

Among the organic species, polychlorinated dibenzo-p-dioxins and dibenzofurans
(PCDDs and PCDFs) have caught public attention due to their known toxic effects,
their ubiquitous presence in combustion and incineration emissions, and their
anthropogenic sources.  Of special interest and concern are the tetrachlorodibenzo-
p-dioxins (TCDDs).  These compounds are solids at room temperature, and have very
high boiling points.  The most toxic of these compounds, 2,3,7,8-TCDD, has a melting
point of 305°C, a boiling point of 450°C, and very low vapor pressures at ambient
conditions.   There is an analogous furan, 2,3 , 7,8-tetrachloro dibenzofuran (2,3,7,8-
TCDF).  There are a total of 75 dioxin and 135 furan isomers (Oakland, 1988).  These
compounds meet the above mentioned criterion for being condensible emissions.  The
emissions of metal species such as, Cd, Cr, Zn, Se, and Hg are quite often
associated with-vapor phase emissions and also meet the criterion for condensible
emissions.
3.0  SATURATED VAPOR CONCENTRATIONS
3.1  PCDD and PCDF Compounds

Recently vapor pressures of several PCDD and PCDF compounds were measured and
predicted from theory by Rordorf and coworkers  (Rordorf, 1985,1986; Rordorf et al.,
1986).  The physicochemical properties of 2,3,7,8-TCDD were independently measured
by Schroy et al., (1985) which agree within an  order of magnitude with those of
Rordorf et al.,  (1986).  The data for some of the PCDF compounds as given by Rordorf
(1986) are reproduced in Figure la, data for some of the PCDD compounds as given by
Rordorf et al.,  (1986) are shown in Figure Ib,  and data for 2,3,7,8-TCDD as given by
Schroy et al., (1985) are shown in Figure Ic.   The measured vapor pressures and
saturation concentrations of 2,3,7,8-TCDD by Schroy et al., (1985) at 30.1°C is
4.68E-7 Pa (-60  ng/m3) ,  and at 71°C is 1.59E-4 Pa  (=20,400 ng/m3) .  The measured
vapor pressures  and saturation concentrations for PCDF compounds are much higher
than those for PCDD compounds for a given temperature as seen from Figures la and
Ib.
3.2  Metals

The saturated vapor pressures of various metal  species are shown  in Figure  2  for the
temperature range 0 to 300°C.  The volatility of a metal depends  upon  the specific
chemical compounds of that metal that are present; e.g., for certain metals the
oxides or chlorides are more volatile than  are  the pure metals  themselves.  Metal
emission measurement studies usually determine  the concentration  of a  metal without
regard to the specific compounds that are present.  This makes  it difficult to
determine the saturation concentration  of the metals.  Alkaline metals such as Na
and K have very high elemental vapor pressures, but are usually detected  in the
particulate phase indicating that these elements are present in the form  of
compounds with very low volatilities.   Figure 2 indicates that, in general, Hg, As,
Se, Cd, and Zn elements and their compounds have the highest volatility among the


                                        38-3

-------
various metallic species.  Chlorides of elements such as Al, Fe, Sb, Bi, and Ti are
highly volatile; whereas, the elements themselves have very little volatility; thus
their volatilization in a process may depend upon formation of  the respective
volatile chlorides.
4.0  AVAILABLE DATABASE

Due to the public concern about emissions of chlorinated organic compounds and
metals from waste incinerators, several sampling studies have  recently been
undertaken to measure these emissions.  A number of incinerators in western Europe
were sampled earlier in this decade, whereas several incinerators in the U. S. and
Canada have been sampled in the last 3-4 years.


4.1  PCDD and PCDF Emissions

After the first reported detection by Olie et al. in 1977 of PCDD and PCDF compounds
in municipal incinerator emissions, several sampling studies in various European
countries such as Sweden, Norway, Italy, and Austria confirmed the presence of these
compounds in the emissions from municipal incinerators  (Scheidl et al., 1985;
Cavallaro et al.,  1980, 1982; Marklund et al.,  1986;  Gizzi et al., 1982;
Ballschmiter et al., 1985; and Olie et al., 1982.)  The total PCDD and PCDF
emissions measured in these studies show high variability in the  emissions
depending upon the type of incinerator and refuse feed, and operating conditions --
primarily the combustion temperature.  Even for a given incinerator, up to 2 orders
of magnitude variation was found due to differences in operating conditions.  In
general, high moisture content in the feed and low combustion temperature resulted
in high PCDD and PCDF emissions; e.g., Gizzi et al., 1982.

The average values of TCDD and TCDF emissions reported  in the European incinerator
studies are tabulated in Table 1.  The reported average emissions of TCDD were from
about 5 to 1,000 ng/m3,  with 2,3,7,8-TCDD being 2 to 10% of the total TCDD
emissions.  The concurrently measured average TCDF concentrations were significantly
higher  than the TCDD concentrations, reflecting their higher rate of formation or
volatility.  The average TCDF concentrations ranged from about 75 to 2000 ng/m3.

Table 2 summarizes reported measurements of TCDD and TCDF in the municipal
incinerator emission sampling studies that were conducted in the U. S.  These
studies often included sampling before and after particulate control equipment to
determine the capture efficiency of the particulate control equipment.  The measured
TCDD concentrations were significantly lower than those observed in the European
studies and the average ranged from about 5 to 150 ng/m3   The proportion of the
average 2,3,7,8-TCDD isomer in the total average TCDD was somewhat higher  than in
the European studies, ranging from about 2 to 40% of the total TCDD compounds.  The
average concurrent concentrations of TCDF compounds were again higher than the TCDD
concentrations and ranged from about 13 to 450 ng/m3.

As seen from Tables 1 and 2 the maximum reported concentration of total TCDD
compounds is about 1,000 ng/m3   Figures la to  Ic suggest that the saturated vapor
concentrations for dioxins and furans are much higher than the maximum reported
dioxin  and furan concentrations in the incinerator emissions that have been  studied.
Thus, all dioxin and furan emissions might be expected  to be in the vapor  phase.
However, during all of these sampling studies a  large fraction of the dioxins  and
furans were found in the particulate phase rather than  in the vapor phase.   Also,
the proportion of total dioxins and furans in the particulate phase was  found to be
                                         38-4

-------
dependent on the sampling location.  A typical sampling train for volatile organics
includes a "front half" section which collects the in-stack particulate phase and
any adsorbed organics associated with those particulates,  and a "back half" section
which collects the organics present in the vapor phase by condensation and
absorption.  In the PCDD and PCDF sampling studies at the inlet of a particulate
control device, where the particulate concentration in the gas phase is very high,
almost 70 to 95% of the dioxins and furans were collected in the particulate phase;
whereas, at the outlet of the particulate control device,  70 to 90% of the dioxins
and furans were collected in the condensate and impingers constituting the back half
of the sampling train.  The breakdown of the collected dioxins and furans in various
sampling studies is shown in Table 3.  The substantially greater proportion of the
dioxins collected in the front half of the sampling train at the inlet locations
clearly suggests the possibility of strong adsorption of dioxin on the particulate
phase collected on the filter during sampling.  The high proportion of the dioxins
and furans collected in the back half of the sampling train at the particulate
control device outlet locations also strongly indicates that the dioxins and furans
are present in the flue gas in the vapor form (as expected by saturated vapor
pressures) but are transferred to the particulate phase during flue gas sampling at
inlet locations, which is often carried out for several hours to collect enough
organics for analysis.

The most common particulate control device used on municipal incinerators is an
electrostatic precipitator  (ESP)   The data from both the inlet and the outlet of an
ESP collected at three incinerator facilities (Peekskill,  NY, Radian,  1988;  North
Andover, MA, Anderson et al., 1988; and Pinellas County, FL, Entropy,  1987)  indicate
very poor efficiency of an  ESP, operating at temperatures above the acid gas dew
point, in collecting dioxin and furan emissions.  In fact at the Signal RESCO
facility in Pinellas County, FL, an increase in both the TCDD and TCDF concentration
was observed with the gas passage through the ESP; the Signal RESCO facility at
North Andover, MA, showed an increase in TCDF.  Such increases can be attributed to
two possibilities: 1) formation of additional dioxins at the precipitator
conditions; and/or 2) incomplete recovery of the dioxins adsorbed on the particulate
phase (during sample collection) which becomes very important for the sampling at
the ESP inlet conditions due to very high loading of the particulates.  However, it
is clear from these data that particulate  collection at high temperature will not
provide significant reduction in the dioxin and furan emissions.

At three municipal incinerator facilities (Marion County,  OR, Anderson et al. , 1987,
Zurlinden et al., 1986; Millbury, MA, Entropy, 1988; and the Flakt pilot plant at
the Quebec City Incinerator, Flakt, 1986), spray dryers have been installed
primarily for acid gas removal where alkaline lime slurry is contacted with the flue
gas in a spray dryer.  The  lime slurry absorbs acid  gases such as HC1, and S02.
The slurry is dried in the  spray dryer, and the resulting product and unreacted
reagent particles are collected in a particulate control device.  The gas phase is
also cooled and humidified  adiabatically in the spray dryer.  These spray
dryer/particulate collection  combinations have been shown to be quite efficient  in
collecting dioxin and furan emissions as seen by the low outlet concentration levels
of less than 15  ng/m3.   Among the facilities,  the incinerator at Marion County uses
a baghouse for particulate  collection; whereas, the incinerator at Millbury,  MA,
uses an ESP.  The spray dryer/baghouse combination produced much lower outlet dioxin
and furan concentrations, compared to the spray dryer/ESP combination  (a  factor of 5
for TCDD emissions and a factor of 15 for TCDF emissions).  The Quebec City
incinerator, operated with  a spray dryer and baghouse pilot unit for which  only PCDD
and PCDF were reported, also had low levels of emissions at  the outlet.   Since  the
observed dioxin and furan concentrations at the spray dryer  inlets were well  below
the saturation concentrations, condensation of dioxins may be ruled out as  a
                                        38-5

-------
possible collection mechanism.  The likely mechanism may be postulated to be
adsorption on the particulate phase which, like condensation, is also improved by
the lower temperature resulting from gas cooling in the spray dryer.  The greater
efficiency of the baghouse, compared with the ESP for dioxin and furan collection,
may be explained by the adsorption that results during the intimate contact of the
collected particulates on the filter, with the vapor bearing gas passing through it.
In an ESP the gas flows over the particulate cake instead of through it.


4.2  Metal Emissions

Most of the incinerator sampling studies referenced in Table 2 also measured the
emissions of various volatile elements from incineration.  The  measured
concentrations for the elements As, Cd, Cr,  Hg, Pb,  Sb,  Se, and Zn  are summarized
in Table 4.  Of these elements, Cr, Pb, and Zn may be considered relatively less
volatile than the other species.  The analytical techniques determine the
concentration of a metallic species regardless of its chemical form.  Thus from the
data, it is not possible to compare the saturated vapor concentrations for these
metal species with the measured concentrations.  Most of the measurement studies
referenced in Table 4 found all the metals,  except Hg, predominantly (> 95%) in the
particulate phase in spite of high volatility of some of their chemical compounds.
Elemental forms of As, Cd, and Se as well as the chlorides of Sb and Zn have high
enough volatility for some of the observed concentrations to be completely in the
vapor form.  Unlike the organics,  the inorganic metal compounds do not have a
significant tendency for adsorption on particulates.  Thus the high proportion of
the  total metallic species found in the particulate phase in these sampling studies
suggests that at stack conditions the metals were primarily present in their less
volatile chemical forms.  For the installations summarized in Table 4, Hg was always
collected, using EPA Method 101, predominantly in the condensate and impingers of
the  sampling train, indicating that the Hg was in the vapor phase at stack
conditions.

The  metals, which are predominantly in particulate form at stack conditions, may
have been present in vapor form at incinerator temperatures, which are in excess of
700°C.  The vapors condense as the gases cool downstream of the incineration process
in passing through the waste heat boiler and other heat recovery devices.
Condensation concentrates the volatile species in the fine submicron fraction of the
particulate phase.  Enrichment in the fine fraction of the particulate phase, by
condensation of the volatile species, has been confirmed and documented in detail
for  fly ash from coal combustion (Damle et al., 1982)   Similar enrichment of the
fine particulates by condensation of the volatile metals may also be expected in the
case of the incinerator particulates.

The  submicron particulate fraction is collected less efficiently than the bulk or
total mass in a particulate control device such as an ESP.  Because the volatile
species are likely to be concentrated in the fine fraction of the particulates,
their collection in an ESP may also be expected to be less efficient than the
overall particulate collection efficiency of an ESP   This trend is seen very
clearly in the detailed sampling studies at the North Andover RESCO facility
(Anderson et al., 1988)   In this  incineration facility, although Sb, As, Cd, Cr,
and  Se were concentrated predominantly in the particulate phase at the ESP  inlet,
the  collection efficiencies of  the ESP with respect to these metals were 78.1, 98.9,
95.3, 81.7, and 89.9%,  respectively (Table 5), which were consistently lower than
the  overall particulate mass collection efficiency of the ESP of about 99.5%.  In
another sampling study at a RESCO facility in Millbury, MA  (Entropy, 1988),  the  ESP
collection efficiencies, with respect  to the Cd and Cr emissions, were  94.6 and
                                         38-6

-------
97.8%, respectively, which were lower than the overall particulate  collection
efficiency of 99.9%.  These observations are consistent with the supposition that
the metallic species are predominantly in the fine particulate fraction, which are
collected less efficiently than are the large particles.  At stack conditions,  the
metal species, with the exception of Hg, are predominantly present in the
particulate phase, with the majority present in the fine fraction.  Efficient fine
particulate collection would thus be essential for control of such emission of
semlvolatile metallic species.


5.0  CONTROL APPROACHES FOR CONDENSIBLE EMISSIONS

     As discussed in Section 4.1, and as seen from Tables 1-2 and Figures la-lc,  the
measured concentrations of PCDD and PCDF compounds are well below their saturation
limits at normal stack temperatures.  Therefore, it will not be possible to reduce
their concentration in the flue gas by simple gas cooling and condensation followed
by removal as particles.  The same also applies for emissions of Hg,  which also has
a very high vapor pressure even at ambient conditions.  Other metallic species  are
already present in the particulate phase at stack conditions, as indicated in the
sampling studies.  Thus simple gas cooling is not likely to be effective in
controlling the emissions of semivolatile materials of concern.


5.1  PCDD and PCDF Compounds

The collection of dioxins and furans with the particulate phase under unsaturated
gas-phase conditions during sampling indicates strong adsorption on the particulate
matter.  This has obvious control implications.  Intimate contact of the flue gas
with its particulate phase, as in the dust cake of a fabric filter,  may be expected
to provide good collection of the dioxins and furans.  Lower temperature favors
stronger adsorption and is thus likely to improve collection by the adsorption
mechanism.

The spray dryer and fabric filter pilot plant on an incinerator flue gas slipstream
at Quebec City (Flakt, 1986) indicated very high (>99%) removal efficiencies of PCDD
and PCDF compounds even when the system was operated with an outlet temperature of
200°C.  Two modes were used in these studies: 1) humidification and cooling of gas
with separate injection of dry lime sorbent; and 2) injection of lime sorbent in the
form of a slurry.  The dry product and unreacted lime particles were subsequently
collected in a fabric filter.  The adsorption of PCDD and PCDF compounds at the
lower cooled gas temperature is likely to be the dominant mechanism for the observed
collection.  However, it is not clear whether the alkaline reagents such as lime or
caustic react with or enhance the PCDD and PCDF collection.  In another spray
dryer/fabric filter pilot plant study (Nielsen et al., 1985), the dioxin removal
efficiency was found to depend on the spray dryer outlet temperature.  A lower
temperature of 110°C was required in this study for achieving greater than 97%
dioxin collection efficiency (Brna, 1988.)

Reduction of PCDD and PCDF compounds to low levels by the spray dryer/fabric filter
combination with lime reagent was also observed in the incineration facility at
Marion County, OR (Zurlinden et al., 1986)   The lime spray  dryer/ESP combination,
however, appears to be less efficient than the fabric filter combination as seen
from data collected at the Mlllbury, MA, RESCO facility (Entropy, 1988).  As
discussed in Section 4.1, this trend may be explained by the intimacy of the contact
between the gas and the particulate cake on the filter fabric.  In general, contact
of the flue gas with particulates (fly ash or lime) at a low temperature appears  to
                                        38-7

-------
be important for PCDD and PCDF removal.  Better removal may be expected with a more
intimate contact and a lower temperature.

Adsorption upon particulate matter, either before or after particle collection,
appears to be the dominant mechanism for capture of PCDD and PCDF compounds.  It  is
well known that adsorption is a surface effect.  In a free stream, smaller particles
will adsorb relatively more material on a mass basis than will larger particles
because of their relatively higher cumulative surface area.  Filtering a gas through
a sorbent/fly ash filter also enhances adsorption because of the relatively higher
surface areas which the gas stream experiences.  It is further known that adsorption
increases with decreasing temperatures.  Control by adsorption requires temperature
reduction and intimate gas-particle contact to enhance the adsorption followed by
efficient fine particle collection.

Spray drying of droplets of an alkaline slurry, such as slaked lime, followed by
fabric filtration appears to meet much of the above criteria.  The temperature is
reduced by evaporation of the slurry droplets.  Intimate contact between the gas
stream and the particle matter occurs in the particulate filter cake.  PCDD and PCDF
collections in the high 90% range have been reported.  The majority of current
municipal incinerators have ESPs as the gas cleaning device.  Retrofit with a spray
dryer/fabric filter would require replacement of the existing ESP.  The spray dryer
also collects acid gases present in the stream.

The data from the RESCO spray dryer/ESP combination at Millbury, MA (Entropy 1988)
indicate that the collection of adsorbed PCDD and PCDF is less than that of a spray
dryer/fabric filter.  One possibility is that there isn't enough intimate contact
between the particle matter and the gas.  Unlike a fabric filter, in which the gas
passes through the filter cake, the gas in an ESP passes over the particle matter
which collects upon the grounded plates. In an ESP the fine particles are generally
collected less efficiently than are the coarser particles.  As the efficiency of  the
ESP is increased, more and more of the fine particles are collected.  Unfortunately
the limited data on spray dryer/ESP systems did not indicate the size distribution1
of the dioxin and furan bearing particles, and whether or not increasing the
efficiency of the ESP would result in improved collection of the dioxins and furans.
For current municipal incinerators the retrofit of a spray dryer to an existing ESP
may be possible, which would likely be less costly than a total new installation  of
a spray dryer/fabric filter system.

A variation of the spray dryer/ESP system is the E-SOX technology (Sparks et al.,
1986).  In the E-SOX process the inlet section of the ESP is removed and replaced
with an array of spray drying nozzles for injection of an alkaline slurry, thus
effectively making over the inlet of the ESP into a spray dryer.  The remaining ESP
sections are upgraded by use of the multi-stage ESP concept to collect not only the
original particulate matter, but also the  dried slurry particles.  The evaporation
of the slurry droplets cools the gas stream in a manner similar to the spray dryer.
Like the conventional spray dryer, E-SOX would also collect acid gases that would be
present.


5.2  Metals

The spray dryer/fabric filter pilot plant (Flakt, 1986) reported greater than 90%
collection efficiency for Hg at a spray dryer outlet temperature less than 140°C.
For a spray dryer outlet temperature of 200°C, no Hg collection was observed.  This
indicates that cooling is essential for efficient Hg collection.  Since the  inlet Hg
concentrations are well below the saturation concentration at 140°C, the temperature
                                        38-8

-------
dependency of the collection efficiency cannot be explained by condensation.   Thus,
like for PCDD and PCDF, adsorption on the solids may be important.  It is again not
clear whether the alkaline reagents such as lime or caustic react and/or enhance the
collection of Hg.

In another pilot plant study (Moeller et al.,  1983), Hg removal was  reported to be
75-85% with the lime spray dryer/fabric filter system and only 35-45% with the lime
spray dryer/ESP system.  In both cases the  spray dryer outlet temperature was 140°C.
These data further indicate the importance  of contact between the gas and solid
phases, again suggesting that there is an adsorption mechanism.  In the two full
scale incinerator facilities, using the lime spray dryer (Marion County, OR,  and
Millbury. MA), Hg concentrations at the outlet were quite high (280 and 720 mg/m3,
respectively) although the spray dryer outlet temperature was less than 140°C.  In
both cases the Hg concentration at the spray dryer inlet was not measured.  It is
interesting to note that at other incinerator facilities shown in Table 4 the Hg
concentrations at the ESP outlet ranged from 150 to 720 mg/m3   Thus,  in full  scale
installations the lime spray dryer did not  appear to be effective in reducing Hg
concentrations, although it was shown to be effective in pilot scale studies.   The
reasons for this discrepancy are not clear.

All other metallic species besides Hg are present predominantly in the particulate
phase at stack conditions.  As discussed in Section 4.2, these semivolatile species
may be expected to be concentrated in the fine submicron fraction of the particulate
matter.  Efficient collection of these species is thus directly dependent upon the
efficient collection of the fine submicron  particulate material in the particulate
control equipment used.  The spray dryer/fabric filter pilot plant system at the
Quebec City incinerator indicated almost complete removal of Zn, Cd, Pb, Cr,  Ni, As,
and Sb.  Since fabric filters are generally efficient for fine submicron particulate
collection, these reported efficiencies may be explained by efficient collection in
the fabric filter.  Except for Hg, it is thus not clear how much of a role the lime
spray dryer played In the collection of these metallic species.  The gas cooling
would certainly be helpful in improving collection of some of the material present
in the vapor fraction.  For example, the data at the resource recovery  facility in
Millbury, MA  (Entropy, 1988) (which uses a  spray dryer/ESP combination) show lower
concentrations of Se" and Sb when compared with the data at the North Andover RESCO
facility (Anderson et al., 1988) which uses an ESP only

The heavy metals, except for Hg in the gaseous state, generally are concentrated in
the finer particle fraction -- the size range that is collected least efficiently in
an ESP.  Recent ESP technology developments have shown that the collection
efficiency for the fine particulates can be enhanced significantly  (Rinard et  al.,
1986; Durham et al., 1986).  This new technology, called the multi-stage  ESP,
operates by separating the charging and collection functions thereby allowing  each
to be optimized.  Retrofitting of this technology to existing  ESPs would  be expected
to improve their fine particle collection thereby decreasing the  emission of  the
heavy metals and any PCDD and PCDF compounds that may be adsorbed upon  them.
Unfortunately, unless the gas stream is cooled, condensation and  adsorption of the
volatile species will be limited.


6.0  CONCLUSIONS

The flue gas from municipal incinerators often contains toxic  semivolatile organics
such as dibenzo-p-dioxins and dibenzofurans and semivolatile metallic  species  such
as Hg, Cd, and As.  Available incinerator sampling data indicate  that  measured
concentrations of dioxins and furans as well as Hg are well below their saturation
                                         38-9

-------
limits at stack conditions.  Hg is thus predominantly collected in the back half of
a sampling train.  Therefore it is not feasible to reduce the concentration of these
substances by simple gas stream cooling and condensation.  The dioxins and furans,
however, are often collected in the front half of a sampling train, indicating
strong adsorption in the particulate phase.  All other metallic species are mostly
present in the particulate phase at stack conditions and are likely to be
concentrated in the fine submicron fraction of the particulates.
A lime spray dryer/particulate control system, often used for acid gas control,
appears to be effective in controlling dioxins and furans as well as Hg emissions in
pilot scale studies.  Cooling of the flue gas below 140°C as well as providing an
intimate contact of the gas with solids are two important parameters affecting
dioxin/furan removal.  Thus a spray dryer/fabric filtration system, in which the gas
contacts the dust cake on the filtration medium, is expected to be more efficient
than a spray dryer/ESP system operated at the same temperatures.   Full scale units
using a lime spray dryer were also found to be effective in dioxin/furan removal;
however, these units indicated poor Hg collection in contrast to pilot scale
results.  The cooling of flue gas with subsequent adsorption of dioxins/furans may
also be achievable in configurations other than a spray dryer, such as in the ESP by
use of the E-SOX process.

Control of submicron particles is necessary for efficient removal of semivolatile
metallic species enriched in the fine fraction of the particulates.  Since mo.st
existing municipal incinerators use ESPs,  improvement in their performance in the
typical low efficiency size range of 0.1   1.0 urn in diameter, would be expected to
improve the removal of the semivolatile metallic species.  A potential retrofit
approach for existing incinerators, using ESPs, is the E-SOX process to provide
cooling of the gas stream which would enhance the adsorption of dioxins/furans.


ACKNOWLEDGEMENT

The authors gratefully acknowledge the cooperation of Mr. Michael G. Johnston of the
the Environmental Protection Agency Office of Air Quality Programs and Standards for
providing several of the EPA reports referenced in this work.


7.0  REFERENCES

Anderson, C. L., W. P. Gergen, J. W. Mayhew, and P. O'Hara, "Emission Test Report,
CDD/CDF, Metals, HC1, S02 and Particulate  Testing:  Marion County  Solid Waste  to
Energy  Facility, Inc. Ogden Martin Systems of Marion, OR,"  Vol.  1, U. S. EPA-EMB
Report  No. 86-MIN-3, DCN No. 87-222-124-06-16, September 1987

Anderson, C. L., T. S. White, and M. A. Vancil, "Summary Report for CDD/CDF, Metals
and Particulate Uncontrolled and Controlled Emissions: Signal Environmental Systems,
Inc. North Andover RESCO, North Andover, MA," U.S.  EPA-EMB Report No. 86-MIN-02A,
DCN No. 87-222-124-15-03, March  1988.

Ballschmiter, K., W. Kramer, H. Magg, W. Schafer, and W. Zoller, "Distribution of
Polychlorodibenzodioxin and Furan Emissions Between Particulates, Flue Gas
Condensate, and  Impinger Absorption in Stack Gas Sampling," Chemosphere, 14, No.
6/7, 851-854 (1985)

Brna, T. G., "State-of-the-Art Flue Gas Cleaning Technologies for MSW Combustion,"
AIChE Spring National Meeting, New Orleans, LA, March 6-10, 1988.
                                        38-10

-------
Cavallaro, A., G. Bandi, G. Invernizzi, L. Luciani, E. Mongini, and A, Gorni,
"Sampling, Occurrence, and Evaluation of PCDDs and PCDFs from  Incinerated Solid
Waste," Chemosphere, 9, 611-621  (1980)

Cavallaro, A., L. Luciani, G. Ceroni, I. Rocchi, G. Invernizzi, and A. Gorni,
"Summary of Results of PCDDs Analysis from Incinerator Effluents," Chemosphere  11
No. 9, 859-868 (1982).
Damle, A. S., D.  S. Ensor, and L. E. Sparks, "Options for Controlling Condensation
Aerosols to Meet Opacity Standards," J. of Air Pollution Control Association, 37,
No. 8, 925-933 (1987)

Damle, A. S., D.  S. Ensor, and M. B. Ranade, "Combustion Aerosol Formation
Mechanisms: A Review," Aerosol Science and Technology, 1, 119-133 (1982)

Dumarey, R.,  R. Heindryckx, and  R. Dams, "Determination of Mercury Emissions from a
Municipal Incinerator," Environmental Science and Technology.  15, No. 2, 206-209.
(1981)

Durham, M., G. Rinard, D. Rugg,  T. Carney. J. Armstrong, and L. Sparks,  "Field Study
of Multi-Stage Electrostatic Precipitators," In Proceedings: Fifth Symposium on the
Transfer and Utilization of Particulate Control Technology, Volume 2, EPA-600/9-86-
008b  (NTIS PB 86-167 160),1986.

Entropy Environmentalists, Inc.,  "Stationary Source Sampling Report,  Signal RESCO
Pinellas County Resource Recovery Facility, St. Petersburg, FL," Volume I,  EEI Ref.
No. 5286-B, March 1987

Entropy Environmentalists, Inc.,  "Emission Test Report, Municipal Waste Combustion
Study, Wheelabrator Resource Recovery Facility, Millbury, MA," ESED Project No.
86/19a, U. S. EPA-EMB Report No.  88-MIN-07, April 1988.

Environment Canada, "The National Incinerator Testing and Evaluation Program:  Two
Stage Combustion (Prince Edward  Island)," Report No. EPS 3/UP/1, September 1985.

Flakt Canada, "Quebec City Mass  Burning Incinerator, Flakt Pilot Plant Test Results,
Wet/Dry Scrubber,"  Environment Canada Report No.EPS 3/UP/2, 1986.

Gizzi, F. , R. Reginato, E. Benfenati, and R. Fanelli, "Polychlorinated Dibenzo-p-
dioxins  (PCDD) and  Polychlorinated Dibenzofurans (PCDF) in Emissions from an Urban
Incinarator,  Chemosphere, 11, No. 6, 577-583 (1982)

Greenberg, R. R., W. H. Zoller,  and  G. E. Gordon, "Atmospheric Emissions of Elements
on Particles from the Parkway Sewage-Sludge Incinerator," Environmental Science and
Technology, 15, No. 1,  64-70, (1981).

Lavalin Inc., "National Incinerator  Testing and Evaluation Program, Mass Burning
Incinerator Technology, Quebec City," Volume II, Main Report,  Prepared  for
Environmental Protection Service, Environment Canada, September 1987

Marklund, S., L.  0. Kjeller, M.  Hansson, M. Tysklind, C. Rappe, C. Ryan, H.  Collazo,
and R. Dougherty, "Determination of  PCDDs and PCDFs in Incineration  Samples  and
Pyrolytic Products," Chapter 6 in Chlorinated Dioxins and Dibenzofurans in
Perspective,  edited by  C. Rappe,  G.  Choudhary. and L. H. Keith, Lewis Publishers,
Inc., 1986.

Moeller, J. T., C. Jorgensen, and B. Fallenkamp, "Dry Scrubbing of Toxic Incinerator
                                        38-11

-------
Flue Gas by Spray Absorption," ENVITEC 83, Duesseldorf,  West Germany, February
21-24, 1983.

Nielsen, K. K.,  J. T. Moeller, and S. Rasmussen, "Reduction of Dioxins and Furans by
Spray Dryer Absorption from Incinerator Flue gas," Dioxin 85, Bayreuth, West
Germany, September 16-19, 1985.

Oakland, D., "Dioxins, Sources, Combustion Theories and Effects, Filtration and
Separation, January/February 1988.

Olie, K.,  J. W.  A. Lustenhouwer,  and 0. Hutzinger, "Polychlorinated Dibenzo-p-
dioxins and Related Compounds in Incinerator Effluents," Pergamon Series on
Environmental Science, 5, "Chlorinated Dioxins Related Compounds," 227-244 (1982).

Olie, K.,  P. L.  Vermeulen, 0. Hutzinger,  "Chlorodibenzo-p-dioxins and
Chlorodibenzofurans are Trace Components  of Fly Ash and Flue Gas of some
Municipal Incinerators in the Netherlands," Chemosphere, 6, 455-459 (1977).

Radian Corporation, "Results from the Analysis of MSW Incinerator Testing at
Peekskill, New York," Volume I, Draft Report, Prepared for New York State Energy
Research and Development Authority, DCN No. 88-233-012-18, March 1988.

Rinard G., M. Durham, D. Rugg, and L. Sparks, "Optimizing the Collector Sections of
Multi-Stage Electrostatic Precipitators," In Proceedings: Fifth Symposium on the
Transfer and Utilization of Particulate Control Technology, Volume 2, EPA-600/9-86-
008b  (NTIS PB 86-167  160), 1986.
Rordorf, B. F.,  "Thermodynamic and Thermal Properties of Polychlorinated Compounds:
The Vapor Pressure and Flow Tube Kinetics of Ten Dibenzo-Para-Dioxins," Chemosphere,
14, 885-892 (1985).

Rordorf, B. F ,  "Thermal Properties of Dioxins, Furans and Related Compounds,"
Chemosphere,  15,  1325-1332 (1986).

Rordorf, B. F.,  L. P. Sarna, and G. R. B. Webster, "Vapor Pressure Determination for
Several Polychlorodioxins by Two Gas Saturation Methods," Chemosphere, 15, 2073-2076
(1986) .

Scheidl, K.,  R.  P. Kuna, and F. Wurst,  "Chlorinated Dioxins and Furans in Emissions
from  Municipal Incineration," Chemosphere, 14, No. 6/7,  913-917 (1985).

Schroy, J. M., F. D.  Hileman, and S. C. Cheng,  "Physical/Chemical Properties of
2,3,7,8-TCDD," Chemosphere,  14, 877-880 (1985).

Smith,  R. M., P.  W. O'Keefe, D. R. Hilker, K. M. Aldous, L. Wilson, R. Donnelly, R.
Kerr,  and A.  Columbus,  "Sampling, Analytical Method, and Results for  Chlorinated
Dibenzo-p-dioxins and Chlorinated Dibenzofurans from Incinerator Stack Effluent and
Contaminated  Building Indoor Air Samples," Chapter 7 in Chlorinated Dioxins and
Dibenzofurans in  Perspective, edited by C. Rappe, G. Choudhary, and L. H. Keith,
Lewis Publishers, Inc.,  1986.

Sparks, L. E., G. H.  Ramsey, R. E. Valentine, and N. Plaks,  "Results  of Pilot Scale
Tests of E-SOX,"  In Proceedings: Sixth  Symposium on the Transfer and  Utilization of
Particulate Control Technology, Volume  1, EPA-600/9-86-031a  (NTIS PB  87-147617),
1986.

Tiernan, T. 0.,  M. L. Taylor, J  H. Garrett, G. F. VanNess, J  G. Solch, D. A. Deis,
                                        38-12

-------
and D. J. Wagel, "Chlorodibenzodioxins,  Chlorodibenzofurans,  and Related Compounds
in the Effluents from Combustion Process," Chemosphere, 12, No. 4/5, 595-606 (1983)

Zurlinden R. A., H. P. V. D. Fange, and J. L. Hahn, "Environmental Test Report" by
Ogden Projects, Inc., Report No. 108, Prepared for Ogden Martin Systems of Marion,
Inc., December 5, 1986.
                                        38-13

-------
                      TREATMENT OF FLUE GAS AND RESIDUES
               FROM MUNICIPAL AND INDUSTRIAL WASTE INCINERATORS
                              G. Mayer-Schwinning
                                Lurgi G.rn.b.H.
                                 Lurgi Allee 5
                               D 6 Frankfurt 50
ABSTRACT
The strict emission limits stipulated by the 1986 legislation of the Federal
Republic of Germany can largely be complied with by means of the dry, quasi-dry
and wet standard processes.

The new bill now under discussion, which contains even tougher requirements,
will call for reconsidering not only the designs of recently built plants but
also the concepts of plants still in the design stage.

Moreover, the refuse disposal legislation according to which residues from
refuse incinerator plants have to be utilized may have a considerable impact on
waste gas cleaning technologies. This contribution therefore deals with further
flue gas cleaning and introduces a new process for thermal treatment of
precipitator residues.
                                      39-1

-------
1.   INTRODUCTION
At present the most comprehensive measures for reducing particulate and gaseous
emissions are being taken in the field of municipal and industrial waste incin-
eration.

The legal requirements have been repeatedly tightened in the past few years, so
that even recently built waste gas cleaning plants frequently have to be up-
graded after short periods of operation.

In the Federal Republic of Germany, the design, construction and operation of
industrial plants is subject to the Federal Air and Noise Abatement Act as
amended in 1986. This requires the installation of waste gas cleaning systems
that correspond to the latest state of the art. The emission limits to be met
are stipulated in a general administrative regulation   the Technical Instruc-
tions on Air Pollution Control (TI-Air)   as amended in 1986 (see table 1).

As far as municipal and industrial waste is concerned, this regulation has been
repeatedly amended. The emission limits shown in table 1, which for the first
time cover dioxin/furan and NO , also tally with those sipulated in its latest
draft amendment.

Thus gaseous mercury and dioxin/furan become the principal items of considera-
tion in the entire collection process. Since the latter are mainly  tied up in
the dust, a highly efficient dust collection process ensuring residual values
of < 2   5 mg/m3  will have to be another of the main points of consideration.

The limit emissions stipulated in TI-Air 86 can largely be met with the noxious
gas treatment systems at present available on the market. However, certain cir-
cumstances, such as
     1.    the requirement that the emission limits to be met according to the
          licencing procedure have to be clearly below those stipulated by the
          TI-Air 1986
     2.    the requirement of handling extremely high raw gas values  (e.g.
          gaseous mercury in the case of industrial waste incineration)
                                      39-2

-------
    3.    the requirement to meet  special NO  limits
    A.    the requirement to guarantee that dioxin values in the clean
          gas will be  low
    A.    the operator's requirement to meet the emission limits stipulat-
          ed as reference values in the 17th administrative regulation to
          the Federal  Air Quality  and Noise Abatement Act of 1989
may call for the employment of more efficient waste gas cleaning systems and
even the application of special residue treatment processes.

Very low limits for special components which have to be met in several European
countries  (e.g. the Netherlands or Austria) may have a considerable influence
on the process concept when supplying flue gas cleaning systems to said
countries.

2.   BASIC FLUE GAS CLEANING MODULES
2.1  Spray Absorption
Spray absorption was developed on the basis of spray drying. It is a long
established technology, the first patents having been filed by Metallgesell-
schaft in the early twenties and utilized by Lurgi Warme as soon as 1922.

The spray absorption technology has been frequently applied in recent years for
cleaning flue gases from waste incinerators and power stations.

In the spray absorption process, the absorption agent is fed onto a fast
rotating disk located in the spray absorber. Under the influence of the
centrifugal force, the absorption agent is forced to the outer rim of the disk
where it is atomized by the high shearing forces. (Fig. 1)

Both, electrostatic precipitators and bag filters, are used for dust collection
downstream of the absorber. Bag filters have the advantage of adsorbing
residual gaseous pollutants onto the particulate matter deposited on the bag
surface.  However, the lower temperature limit of the flue gas must not be
underrun since the crystal water added onto the CaCl- may lead to fabric
clogging. Moreover, special measures have to be taken to prevent the
                                      39-3

-------
temperature from dropping below the dew point during start-up and shut-down
procedures. Due to the reduced danger of deposit formation, precipitators, on
the other hand, permit a further reduction of the flue gas temperature and thus
a further increase in the adsorption efficiency in the spray reactor.

Depending on the stoichiometric factors, the following clean gas values can be
reached on application of this process:

     HC1       10 to 30 mg/m3
     SO        20 to 50 mg/m3
                            3
     dust      10 to 20 mg/m

Problems may arise in connection with the separation of gaseous mercury as
collection efficiencies cannot be increased beyond 30 to 50% unless special
additives are used.

2.2  Wet Waste Gas Cleaning
Wet processes for flue gas cleaning are applied whenever the following has to
be taken into account:

          minimization of the volume and weight of the residue, since
          stoichiometric factors of approx. 1 can be achieved,
          possibility of discharging chlorides to the receiver,
          very high content of gaseous heavy metals in the raw gas,
          stringent requirements as to the heavy metal content in
          the flue gas (< 0.1 mg/rn ),
          existence of vital plant sections, such as dust collector,
          waste water treatment plant, stack etc.

Wet processes for waste gas cleaning may involve waste water or not. Waste
water can be avoided by subjecting the water discharged to an adequate
treatment  (neutralization, heavy metal precipitation) before conducting it
to an evaporation or crystallization plant.

The vital part of wet gas cleaning systems are scrubbers for gas cooling,
                                      39-4

-------
solids separation and adsorption of gaseous pollutants. While in the past
scrubbers were primarily used for dust collection,  they are nowadays employed
as fine cleaning stage in waste incineration plants for the removal of S0?,
HCL, HF and gaseous heavy metals.

The radial flow scrubber is a device that has proven very successful in
practice. It is used in a diversity of designs depending on the application
and the prevailing operating conditions. It may be combined with a separate
cooling tower or a venturi scrubber or equipped with a venturi scrubbing stage
at the scrubber inlet. Fig. 2 and 3 show two basic designs. The first is
characterized by reduced space requirements which make it particularly suitable
for retrofitting. The second design, although requiring more space, offers the
advantage of a reduced pressure drop and of a clear separation of the
individual circuits. Scrubbers may be followed by a wet precipitator as
secondary cleaning stage for the removal of particulate matter, aerosols and
heavy metals. Wet  precipitators permit high collection efficiencies to be
achieved at relatively low utility cost.

3.   FURTHER WASTE GAS CLEANING
3.1  Wet Flue Gas Cleaning
The concept for further cleaning of flue gases from industrial waste
incinerators provides for an indirect condensation stage to follow a dry type
precipitator and a double stage scrubber. In this indirect condensation stage
the waste gas is cooled down to less than the saturation temperature, i.e. to
approx. 50°C. The fine mists formed offer a large surface for the collection of
residual gaseous and particulate flue gas constituents. The mists formed are
collected in the subsequent wet precipitator ( see Fig. 4).

Apart from aerosols, the processes described above particularly permit gaseous
mercury to be efficiently removed. The following collection efficiencies were
demonstrably reached in the scrubbing, condensation and wet precipitator stages
[2, 3]:
     HC1                      > 99.9%
     SO                       > 90%
     Hg (gaseous and solid)   86   97.7
     dust                     95   98 %
                                      39-5

-------
3.2  Wet Flue Gas Cleaning with Dry Residue Discharge
With Spray Dryer.   In a few important aspects, the advantages of wet gas  clean-
ing may be linked to those of dry gas cleaning. This process variant works
according to the principle shown in Fig. 5. The flue gases of the incinerator
pass directly to the spray absorber where they are allowed to come into
intimate contact with atomized hydrate of lime suspension for extensive
pollutant removal. HC1, HF and SO- react with the hydrate of lime forming  a dry
particulate residue which is discharged together with the fly ash. The
particulate matter is then collected either in a bag filter or an electrostatic
precipitator.

The next step is to pass the flue gas to a two stage scrubber and to treat it
first with an acid and then with an alkaline solution to reduce the remaining
contents of gaseous pollutants, particulate matter and the HgCl,-, to the
values stipulated. If a high proportion of aerosols has to be reckoned with, an
electrostatic precipitator may be added downstream of the scrubber.

To prevent the recirculated scrubbing fluid from being concentrated in the
radial flow scrubber, part of the fluid is continuously withdrawn from the
scrubber circuit, neutralized and passed to the absorber for evaporation of the
water content.

Apart from the normal dry reaction products, such as CaSO., CaSO,-,, CaCl,,,
relatively small quantities of Na_SO, are obtained in the absorber.

The plants operating according to this process principle have several
advantages, the most important being:

          the wet gas cleaning stage with its highly efficient removal of
          gaseous heavy metals, HC1, HF and SO-
          the absence of a waste water treatment system
          the low stoichiometric factors   in spite of the low pollutant
          concentration
          the possibility of shifting the separation process to the spray
          absorber and/or scrubber which permits greater freedom  in the
          utilization of absorbents.
                                        39-6

-------
A plant consisting of 2 strands was built and commissioned for the Coburg-
Neuses power station, which is based on refuse combustion (fig. 5). The results
achieved up to now show that the aim of avoiding the production of waste water
has been reached and that the garanteed emission limits indicated in Fig. 6 are
met.


With Evaporation Crystallizer.   Another way of avoiding the production of
waste water on application of the wet gas cleaning process is the use of an
evaporation crystallizer. In this case the flue gases are treated in the way
described above.


After waste water treatment the clean effluent is passed to the multi-stage
evaporation crystallizer for evaporation of the water and separation of the
solids from the mother liquor to obtain chlorides (NaCl, KCl) as well as
sulfates and sulfites (Na_SO,, Na_SO.,) that permit industrial utilization.


A.   TREATMENT OF RESIDUES
The requirement to avoid or utilize waste forms a basic element of the
environmntal control legislation.


The residues, such as precipitator/ filter dust or reaction products
have to be reduced to minimum volume and weight, and reutilized, i.e. they have

to be recycled to serve some useful purpose.


The residues are treated in one of the following ways:


          Scrubbing of the fly ash, addition of binding agents ( such as
          cement), compacting. A plant which is to operate according
          to this process is being built by Lurgi for the Turgi refuse
          incineration plant in Switzerland.

          Alkaline scrubbing. Chloride burdens can easily be removed,
          heavy metals dissolve only to a minor degree. The residue can be
          solidified by means of small quantities of binding agents. The
          elution properties of heavy metals are considerably improved.

          Thermal treatment of the residue with a view to achieving a product
          which can be marketed as raw material or intermediate product for the
                                       39-7

-------
          building industry. The drawback is that this market will probably
          decrease in the future and that more substitute products will
          certainly appear on the market.

As far as the lastmentioned way is concerned, the vitrification of residues
seems to be particularly promising. Lurgi and Sorg have developed a process
(the SOLUR process) [4] which aims at the following:

          treatment of fly ashes and of fly ashes plus reaction products
          fixation of heavy metals
          dioxin/furan destruction
          volume and weight reduction
          minimum waste gas volume
          fixation of the acid noxious gases during the vitrification
          process
The vitrification plant consists of a melting unit which is followed by a dust
collection system and, if necessary, by a mercury removal system.   The
precipitator/filter residues are subjected to electrical melting by immersing
the electrodes into the melting chamber. The energy required for melting
amounts to approx. 1 MWh per tonne of residue.

The final product obtained on application of the SOLUR  process is glass
which can be used as construction material in the building industry and in
civil works (substitute for gravel, concrete additive).

Heavy metals are irrevocably fixed in the glass matrix. So the only product
that remains   it can either be further processed or dumped   is a concentrate
of soluble salts which represents 10 - 13% of the original quantity of residues
( see Fig. 7).

A process developed in accordance with these criteria is available for
application in full scale plants.
                                      39-8

-------
5.   CONCLUSION
Stringent legal requirements aiming at a reduction of the emission of waste
gases from waste incineration plants led to complex waste gas cleaning
installations with combined systems for particulate and gaseous pollutant
removal. The low emission limits for dioxin and gaseous heavy metals invariably
call for the dust emission limits of the waste gases to be reduced down to
< 2   5 mg/m3 (NTP).

However, it has not yet been ascertained whether the measures described above
will really permit the miminum emission limits to be met. So-called "0-emission
limits" are being discussed everywhere at the moment which would mean that a
final cleaning stage using rotary hearth coke and/or activated carbon would
have to be used.

REFERENCES
 1.  Pohle, R.; Arndt: Primar- und sekundarseitige Ma3nahmen zur
     Emissionsminderung bei der therm. Abfallverwertung in Niirnberg. Mull und
     Abfall 4.86
 2.  K. von Beckerath; P. Luxenberg: Weitergehende Rauchgasreinigung.
     Entsorgungspraxis 10.89, p. 492-499
 3.  J.-D. Herbell: Kondensationsstufe und Napelektrofilter
     Betriebsergebnisse zur weitergehenden Rauchgasreinigung   GVC Conf.
     Baden-Baden, 11.89
 4.  G. Mayer-Schwinning; H. Pieper et al.:Vitrification Process for the
     Immobilization of Residues Produced on Cleaning the Waste Gases from
     Refuse Incineration Plants   Int. Recycling Congress, Berlin 89
                                      39-9

-------
MSW  I
Emissions
                                     TI - AIR  Draft:  Enviro.
                                       1986    Quality Contr.Act
                                               24 hours/average'
Solid Particulate
Matter
Hydrogen
Chloride
( inorganic )
Hydrogen
Fluoride
( inorganic )
Sulfur Dioxide
Heavy Metalls
Solid and Vapor
NO
X
PCDD/PCDF
TE
mg/m3 (N,d)
gr/dscf
mg/m ( S , d )
ppmv
mg/m ( S , d )
ppmv
mg/m ( S , d )
ppmv
mg/m ( S , d )
mg/m ( S , d )
ppm
ng/ni (N,d)
30
0,012
50
31
2
2
100
35
Hg+Cd+Tl
0,2
500
240
10
0,004
10
6
1
1
50
18
Cd 0,1
Hg 0,1
100
48
0,1
  dry gas basis,  11 %  0,
                               39-10

-------
Flue Gas Cleaning System
Spray Absorption
Gas Inlet
                           Atomizing
                           System
                             Gas Outlet
                    Residue
Intimate mixing of the
atomized absorbent and
the pollutant-laden flue
gas

Sufficient residence time
to permit physico-chemical
absorption

Dry and pourable residue
         -e  1
                              39-11

-------
1  Saturator venturi/lst scrubbing stage
2  Centrifugal droplet separator
3  Radial flow scrubber/2nd scrubbing stage
4  Lamellar droplet separator
             Saturator venturi with
              radial flow scrubber
              (space saving design)
                    39-12

-------
1 Saturator venturi/ 1st scrubbing stage
2 Lamellar droplet separator
3 Radial flow scrubber/2nd scrubbing stage
4 Lamellar droplet separator
               Saturator venturi with
                radial flow scrubber
                (separate arrangement)
                   39-13

-------
©
                                 ©
(D © ©
— 	
i-'-u
[
-
               ©

1   Dry precipltator
2   Radial flow scrubber
3   Gas cooler
4   Wet precipitator
5   Flue gas reheating system
                           6    Induced draught fan
                           7    Stack
                           0    Acid scrubbing stage
                           9    Alkaline scrubbing stage
                           10   Water recooling system
   4  Flue gas  cleaning  in  hazardous
         waste  incineration  plant
                    39-14

-------
Flue Gas Cleaning System
Refuse - Fired degeneration Plant
                     39-15

-------
Balances for
Refuse-Fired Cogeneration Plant
                                           Flue Gas
Refuse
11 t/h
Feed Water
130"C
Co
Refuse Firing
System and
Waste Heat
Boiler
I
mbustion Slag
Air
Flue Gas
60000 m NTP/h
Steam 24,7 t/h
400 "C,
40 bar
>
, 200 "C
CaO
83kg/h
Flue Gas
Cleaning
System


— (([) 6950 kW
2 Lines
                         4,5 bar
Pollutants
(Basis: 1 1 Vol % O2)
HCI
HF
S0?
Staub
En

15
0
35
3
lission Values
Vn, dry
mg/m3 NTP
3 mg/m3 NTP
mg/m3 NTP
mg/m3 NTP
Pollutants
(Basis: 1 1 Vol % O?)
Heavy Metals (ace TA-Luft)
Class 1 (Hg solid and gaseous)
Class 2
Class 3
Emission Values
Vn, dry
0,2 mg/m3 NTP
0,3 mg/m3 NTP
0,5 mg/m3 NTP
                     39-16

-------
Mass and Volume Changes on Melting

Fly ash
Additive
Losses on melting
Valuable materials %
Glass
Glass product
Valuable materials %
'rod. to be dumped %
Inert Gases %
RIP fly ash
reaction product
Mass/kg
750
250
200
800


107
11
14
Vol/m3
1,25
0,17


0,33
0,57
26V45*
~4

RIP fly ash
Mass/kg
1000

150
850


85
13
2
Vol/m3
1,7



0,35
0,61
21*736*
«4

* Glass mass,  * Glass product
                      39-17

-------