EPRI
Electric Power
Research Institute
Topics:
Particulates
Electrostatic precipitators
Fabric filters
High-temperature filtration
Sulfur dioxide
Gaseous wastes
EPRI GS-7050
Volume 1
Project 1129-18
Proceedings
November 1990
Proceedings: Eighth Particulate
Control Symposium
Volume 1:
Electrostatic Precipitators
Prepared by
Electric Power Research Institute
Palo Alto, California
-------
REPORT SUMMARY
SUBJECT Fossil plant air quality control
TOPICS Participates
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, particulate 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 Participate Control Symposium
Volume 1: Electrostatic Precipitators
GS-7050, Volume 1
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.
ii 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
Session 1A, New Controls for Precipitators I
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.C. Landham, M.B. Tucker,
and W. Piulle
Experimental Evaluation of Improved Design of ESPs 6-1
B. Bellagamba, G. Dinelli, and E. Riboldi
-------
Section Page
Delaying Sodium Depletion in Electrostatic Precipitators 7~1
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 5 A, Fly Ash/ESP Studies
Scott Thomas, Chairman
The Effects of Fireside Process Conditions on Electrostatic -^
Precipitator Performance in the Electric Utility Industry
H.J. Hall
-------
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 7 A, 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 Page
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. Killer
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 33-1
M. Vakili and A.W. Ferguson
VI 1 1
-------
Section Page
Session 11 A, Innovative Pollution Control Technologies
Ralph Altman, Chairman
Improved Carbon Participate 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,
participate 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
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 -1
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. Cushing, T.J. Heller, R.F. Altaian, T.J. Boyd,
M.A. Friedman, and R.L. Chang
xi
-------
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 I
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
xm
-------
Section
DuPont's Engineering Fibers for Hot Gas Filtration Case ^
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
D.J. 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
-------
ADVANCED MICROPROCESSOR TECHNOLOGY
FOR ELECTROSTATIC PRECIPITATOR HIGH VOLTAGE CONTROL SYSTEMS
Edwin H. Weaver, P.E.
Frank A. Gallo
Belco Technologies Corporation
7 Entin Road
Parsippany, New Jersey 07054
ABSTRACT
The use of advanced microprocessors to monitor, interpret and instruct electrostatic
precipitator high voltage control systems has lead to major advancements in the
operating philosophies of these systems. The Merlin II microprocessor control
regulates the amount of power applied to the precipitator by monitoring several
hundred data points during each half cycle of operation and using this information
to maintain the desired precipitator performance. The maximum power level without
inducing a spark, the precipitator reaction to a spark, and the presence of back
corona are monitored, evaluated and acted upon. The various operating philosophies
which can be utilized by plant operators are discussed along with their potential
impact on performance. Finally, the performance of the Merlin II is examined through
an operational case history.
1-1
-------
ADVANCED MICROPROCESSOR TECHNOLOGY
FOR ELECTROSTATIC PRECIPITATOR HIGH VOLTAGE CONTROL SYSTEMS
INTRODUCTION
Effective control of the power applied to an electrostatic precipitator is essential
in optimizing its performance. Since the power level at which the electrostatic
precipitator's performance will be optimized is constantly changing, it is important
that the high voltage control system quickly and accurately adjust power levels so
that optimum power is maintained.
The development of a sophisticated microprocessor-based control system provides a
tool with the necessary capacity to analyze the operating conditions of an
electrostatic precipitator. Once the operating conditions have been analyzed, an
intelligent decision can be made regarding the power level in the electrostatic
precipitator at which performance will be optimized. The control is then able to
energize the electrostatic precipitator to the selected power level.
Special functions and integration to centralized monitoring data recording and
reporting systems are readily achievable. Intermittent energization, or power
modulated precipitation can be utilized. Data is reported to a central monitoring
unit (CMU) which make possible the reproduction of secondary voltage and current
waveforms on the cathode ray tube (CRT) display of the CMU.
The functions that a microprocessor must perform to ensure that the high voltage
control operates in an optimum manner will now be reviewed.
CONTROL FUNCTIONS
The primary control function is to optimize the collection efficiency of the
electrostatic precipitator. This is achieved by operating at the maximum sustainable
power level. This power level comprises the voltage and current levels that are
slightly below the level which will cause a breakdown of the electric field in the
form of a spark. If the conditions at which sparking occurs were constant, the
1-2
-------
control of power levels to the precipitator would be quite simple. However, the
maximum power level achievable depends upon a set of constantly changing variables.
These variables include the size distribution and chemical composition of the dust,
the amount of dust in the gas stream, flue gas temperature, gaseous constituents in
the flue gas, the amount of dust on the discharge electrodes, and the amount of dust
deposited on the collecting plates. Because of the large number of variables and
their propensity towards rapid change, the variations in the maximum power level
achievable can be large and can occur very quickly.
Power Optimization
Microprocessor based control systems must be capable of analyzing and responding to
these changing conditions in real time so as to ensure that maximum electrostatic
precipitator collection efficiency is maintained. In order to achieve this, the
control must effectively perform several functions. First, the control must
constantly check to determine the maximum power level that is achievable. This is
accomplished by comparing operating information to historical data and other
preprogrammed control performance data. If a spark does occur, the process of
extinguishing the spark and returning to the optimum power level must be as rapid
as possible. Also, the spark must be identified as self extinguishing or non-self
extinguishing. If the spark is self extinguishing, the power level can be returned
close to the pre-spark power level in the next half cycle of operation. Should the
spark not self extinguish, the control must turn off power for one half cycle in
order to extinguish the spark. In the next half cycle of operation, the alternating
current (AC) side of the transformer-rectifier will have the same polarity as the
half cycle in which the spark occurred and energization to a high power level would
saturate the transformer and induce another spark. This condition is effectively
avoided by control design so that re-energization occurs in the following half cycle
of operation. The power level that the system is energized to will then be as close
to the pre-spark power level as possible. A simplified control logic for power
optimization is shown in Figure #1.
Back Corona Detection
Another essential control function is back corona detection and prevention. Back
corona occurs when collecting high resistivity dusts. As the high resistivity dust
accumulates on the collecting plate, the normal flow of negative ions through the
dust layer to the collecting plate is slowed. If the condition becomes severe, the
gas in the dust layer will ionize and break down, sending positive ions out towards
1-3
-------
the discharge electrode. This condition is extremely detrimental to electrostatic
precipitator performance.
The secondary voltage waveform will indicate when a back corona condition occurs.
The voltage at the onset of conduction, that is the point where the positive slope
of the voltage waveform begins, will decrease in each ensuing half cycle of
operation. This is due to the fact that as back corona continues to increase in
intensity the equivalent resistance in the RC constant changes, the voltage decays
more rapidly with time, and conduction begins sooner in each ensuing half cycle of
operation. By monitoring the secondary voltage and current waveforms, it is possible
to detect the presence of back corona. Power levels can then be adjusted so that
the operation of the system is at its optimum point, just below the onset of back
corona. Waveforms that are typical of back corona are shown in Figure #2.
Intermittent Energization
A final control function to be examined is intermittent energization (IE), or power
modulated precipitation (PMP). IE can increase precipitator efficiency while
decreasing power consumption in cases of high resistivity dust, and can substantially
reduce power consumption with little impact on precipitator efficiency in cases of
moderate dust resistivity.
To appreciate why precipitator power consumption and performance are affected by IE,
it is necessary to understand the basic principles of IE. Simply, IE modifies the
energization of the precipitator by normally energizing the precipitator for one or
more half cycles and then not energizing the precipitator for a series of half
cycles. Figure #3 shows typical oscilloscope tracings of secondary voltage and
current for normal energization and for a one in five (one half cycle on, four half
cycles off) mode of IE. Two items to note are the peaks of the voltage and current
and the manner in which secondary voltage dissipates until the precipitator is re-
energized. In typical cases of high resistivity dust, the peak voltages that can
be obtained with IE are much higher than in cases of conventional energization. This
is due to the fact that the maximum voltage is normally limited to the value at which
back corona is induced. By not energizing the precipitator for several half cycles
in an IE mode of operation, the ions produced during energization have more time to
dissipate through the dust layer. Thus, when the precipitator is again energized
it can be at a much higher voltage level without causing back corona The average
voltage is increased and thus the precipitator efficiency is increased m cases
1-4
-------
of moderate resistivity, the voltage peaks are also increased slightly. However,
the voltage dissipates to a noticeably lower level than during conventional
energization. The result is very little change in the average voltage levels, and
thus the collection efficiency of the precipitator remains unchanged.
MICROPROCESSOR OPERATION
With the advent of economical 16 bit microprocessor circuits, it has become feasible
to re-focus the high voltage control design approach by transposing more of the
control and integration functions from circuitry to program instructions. This
approach, in addition to reducing the number of components and improving reliability,
can support a wide range of control philosophies. The primary benefit of this type
of control is the ability to modify any control operation through the software. The
hardware becomes a true servant of the operation and is not solely designed for any
specific function. A second benefit of this type of design is the ability to
reconstruct operating waveforms from the memory of individual controls which then
can be transmitted to a remote display, allowing for oscilloscope waveforms to be
available and recorded during any operating period without additional test equipment.
These waveforms also provide detailed information which allows the microprocessor
to fully analyze the operation of the system. The advent of integrated circuitry
with the necessary speed and power has permitted this method of operation to be truly
realized.
Design considerations to support this mode of operation and monitoring are primarily
concerned with data acquisition without integration or modification, sufficient speed
of operation to allow all data to be timely refreshed, program benchmarks at
fundamental power frequencies and scaled functions thereof, and algorithms developed
solely to make full use of dynamic waveform availability.
Circuitry Considerations
The signals which must be monitored and used for control include the following:
1. Primary voltage, linked through a potential transformer and
presented to an analog to digital converter as a rectified but
unfiltered signal.
2. Primary current, linked through a current transformer and
presented to an analog to digital converter as a rectified but
unfiltered signal.
1-5
-------
3. Secondary voltage, derived from a voltage divider and presented
to an analog to digital converter as an unfiltered signal.
4. Secondary current, derived from applying the voltage across a
resistor to an analog to digital converter as an unfiltered
signal.
5. Reactor voltage, linked through a potential transformer and
presented to an analog to digital converter as a rectified but
unfiltered signal.
These signals are level translated through circuitry with a band width of at least
one megahertz. The rectification of the AC signals is primarily to expand the range
of the analog to digital converter as the zero crossing of these signals should
remain constant. Figure #4 illustrates the monitoring points for these signals.
Circuitry is incorporated to notify the program when different timing benchmarks
have transpired; the most common of these benchmarks being the zero crossing of the
input power line. This is coordinated with a higher frequency benchmark to
represent the maximum sampling rate with which to decipher the input waveforms.
This frequency is high enough to allow a maximum of 1 degree of silicon controlled
rectifier (SCR) angle per increment of timing. This benchmark is used for SCR
output and also to digitize the input signals into frequency points during each half
cycle of line operation.
Metering
The signals in memory are used for local and/or remote metering. Each half cycle's
number of readings for each point must be averaged and combined with the average of
other half cycles to arrive at a smooth value to be used for operator interface.
Normally the total number of half cycles to be averaged is operator set allowing for
personal preference. The primary voltage and primary current are coordinated with
the operating phase angle of the SCR's to arrive at an RMS value for each level.
Such a system of metering and data retrieval separates average voltage or current
readings for each half cycle. This allows the control to separate the displayed
currents in each rectifier leg even though the transformer-rectifier has no
provisions for independent metering of each leg. Only one secondary current
metering wire is required. This is illustrated in Figure #5. These readings are
sensitive to SCR unbalance or an unsymmetrical power source, allowing for better
annunciation of possible problems. The metering will display peak voltages and
1-6
-------
currents for any half cycle when in the intermittent mode of operation, giving a
wealth of information about actual activity during the application of power and
about the voltage decay when power is removed or the time between energizing pulses.
System Control
The digitizing of the phase angle and zero crossing allows for the SCR's output and
the metering input to be placed in phase with the operation by delaying the zero
crossing in the program based on the size of the reactor and the delay it presents.
The inductance of the linear reactor causes a phase lag in the voltage applied to
the transformer rectifier which also limits the available control range of the
SCR's. Proper framing of the firing angle to the SCR's prevents this from
occurring. Digitizing the phase angle provides for all secondary information to be
properly stored in memory and recalled for the appropriate half cycle.
Having peak and valley information available for the secondary high voltage allows
the control to operate around an acceptable ratio of change between the peak and
valley to further detect the presence of back corona. This is evaluated against
different time frames for more accurate detection of back corona activity.
Spark rate limits have proven to reduce control capabilities and cause severe
degradation of applied power simply to satisfy a preset spark limit. A more
reliable system is to weigh the spark rate, but also keep close records of different
operating currents prior to spark initiation and use these for short range goals in
the never-ending search for greater secondary voltage. This spark current limit,
sometimes called a floating current limit, makes the control's operation truly
independent of gas conditions.
CASE HISTORY
The Merlin II microprocessor high voltage control system was installed on an
electrostatic precipitator collecting flyash generated from a pulverized coal fired
boiler firing low sulfur western coal with a moderate flyash resistivity. The
electrostatic precipitator is a weighted wire design.
Because this particular installation has several boilers, all with identical
electrostatic precipitators, it was possible to compare the performance of the
microprocessor-based controls against a previous generation of analog controls.
Figure #6 illustrates the average stack opacity for the unit with the new
1-7
-------
microprocessor controls and with the old analog control system. This data was
accumulated and averaged over several months of operation.
The opacity with the new control system was improved significantly and averaged from
60 to 75 percent of the original system stack opacity, or an opacity reduction of
between 25 to 40 percent. The impact of this opacity reduction on stack emissions
can be estimated by the formula shown in Figure #7. Assuming that all values except
opacity and stack emission concentrations remain unchanged, relative emission values
for various stack opacities can be obtained. The theoretical emission reduction,
utilizing the microprocessor-based control system, ranged from 30 to 40 percent from
the high load end to low load end of the system.
As can be seen from this case history, compliance with stack opacity and particulate
emission standards can be realized with state of the art high voltage control
systems.
-------
Figure #1
Precipitator Power Optimization
-*.
Increase ESP
Power
No
Spark
Occurs
I
Yes
Spark
Self Extinguished
I
No
Extinguish
Spark
I
Adjust Floating
Current Limit
Re-apply
Power
Yes
1-9
-------
Figure #2
Back Corona
Voltage and Current Waveforms
V
0
L
T
A
G
E
TIME
C
U
R
R
E
N
T
TIME
1-10
-------
Figure #3
Normal and Intermittent Energization
Normal Energization
V
0
L
T
A
G
E
TIME
Secondary Voltage
C
U
R
R
E
N
T
TIME
Secondary Current
Intermittent Energization (one in five half cycles)
V
0
L
T
A
G
E
TIME
Secondary Voltage
TIME
Secondary Current
1-11
-------
Figure #4
Control Monitoring Points
Vp = Primary Voltage
Vr = Reactor Voltage
Vs = Secondary Voltage
Ip = Primary Current
Is = Secondary Current
1-12
-------
Figure #5
Separation of Current Readings
r\
Waveform Trace of Secondary Current Fee^back
r
1-13
-------
Figure #6
Merlin II Control Performance
40 -
S
T
A 30
C
K
0
P
A
C
I
T
Y
20 --
10 --
Analog Controls
Microprocessor Controls
100
120 140 160
Boiler Load
180
1-14
-------
Figure #7
Opacity/Emissions Relationship
-0.7 wl
kp
op = 1 - e
op = stack opacity (%)
w = dust concentration (gr/acf)
1 = stack diameter (feet)
k = constant based on particle size and color
p = particle density (gm/cm3)
1-15
-------
CASE STUDY OF A HOT SIDE PRECIPITATOR USING
VOLTAGE LIMIT, CURRENT LIMIT,
PULSE BLOCKING, AND PULSE BLOCKING WITH BACKGROUND POWER.
Elliott M. Drysdale
FORRY, Inc.
Cleveland, OH
David Wakefield
Wakefield Assoc., Inc.
Houston, TX
John Wester
Central Power and Light
Corpus Christi, TX
ABSTRACT
This is a case study of the Central Power and Light (CPL), Coleto
Creek Power Station, Joy-Western hot side electrostatic precipitator
(ESP). The precipitator operates at 5% opacity with low resistivity
ash characteristics.
Data was collected, including opacity, as all 36 Automatic Voltage
Controls were simultaneously modified. Secondary voltage (K'volts),
secondary current (milliamps), Pulse Blocking, and Pulse Blocking
with "BACKGROUND POWER" were used to reduce the power and improve the
performance. All the Automatic Voltage Controls were modified
uniformly. Operating conditions were kept normal with continued
rapping and soot blowing and one (1) Automatic Voltage Control turned
off for maintenance.
Test results show that there are energy savings without increase in
opacity. Pulse blocking was made tolerant of operating changes in
flue gas and ash conditions with "BACKGROUND POWER" and specifically
operate with low resistivity ash and on the outlet fields.
Preliminary test results suggest even better results can be obtained
when the Automatic Voltage Controls are matched with the precipitator
performance contour. This would include Pulse Blocking on the inlet
fields while Pulse Blocking with "BACKGROUND POWER" on the outlet
fields, intermixing Pulse Blocking with conventional type limits such
as current limit, and programming individual Automatic Voltage
Controls for best TOTAL precipitator performance.
2-1
-------
CASE STUDY OF A HOT SIDE PRECIPITATOR USING
VOLTAGE LIMIT, CURRENT LIMIT,
PULSE BLOCKING, AND PULSE BLOCKING WITH BACKGROUND POWER.
PURPOSE
Central Power and Light, Coleto Creek Power Station operates a hot-
side precipitator with opacity in the 5% range. Forry, Inc.
introduced several control methods to evaluate which is most
efficient in reducing opacity while using minimum energy. This
includes limiting Transformer Rectifier (T/R) Set power using
secondary current limit, secondary voltage limit, Pulse Blocking, and
Pulse Blocking with Background Power.
PROCEDURE
For the purpose of this test, all normal operating conditions were
allowed to continue. This includes soot blowing, rapping, and one
Transformer/Rectifier (T/R) Set turned off.
Each Automatic Voltage Control (AVC) has 10 sets of program
parameters which consists of the following adjustments:
1. Spark Rate
2. Set Back
3. Secondary Current Limit
4. Secondary Voltage Limit
5. Pulse Blocking
6. Background Power
These ten (10) programs were preloaded into all of the AVCs and then
selected by the Central Control Unit (CCU). For this test, all T/R
Sets use the same program parameters. In actual use, each AVC has
ten (10) site specific programs that allowed custom installation
while maintaining the 10 step philosophy.
For this evaluation the following methods were used:
1. Limit all T/R Sets secondary current in ten (10) steps, from
180 ma. to 1800 ma.
2. Limit all T/R Sets secondary voltage in ten (10) steps, from 15
KV to 40 KV-
3. Limit all T/R Sets with Pulse Blocking in ten (10) steps of
ratio from 1:0 to 1:9.
4. Repeat step three (3) (Pulse Blocking) with Background Power of
10% and 20%.
For each test, ten (10) programs were loaded into each AVC This
process took one (1) minute for all thirty-six (36) AVCs. *Then each
program was manually selected and allowed to run for ten (10)
minutes, or until stable, before proceeding to the next set of
program parameters. During each step, all AVC readings and opacitv
were automatically data logged every minute onto a 20 megabyte hard
2-2
-------
disk. The data was in the LOTUS 1-2-3 format so that LOTUS 1-2-3 was
used to compile and graphically display the results for the report.
After all the tests were run, the original CPL/Coleto Creek program
parameters were reloaded into all the AVCs in which CPL personnel had
experimentally optimized. The ten (10) programs were then run for
comparison.
Normal operation of the T/R Set controls the power into the T/R Set
by phase firing anti-paralleled Silicon Controlled Rectifiers (SCR).
The current and the voltage are maintained every half cycle (see
Figure 1). By measuring the secondary current and secondary voltage,
one or both may be used as limits to the phase firing of the SCR's,
which limits the power.
o.
E
OFF CYCLES:
BACKGROUND-
AMPS:
VOLTS:
MILLIAMPS:
K'VOLTS:
0
0
193
244
1320
21.9
I I
:.33 Milliseconds per Division
I I I I I I I
I I
SECONDARY CURRENT k VOLTAGE WAVEFORMS
Without Pulse Blocking
Figure 1
2-3
-------
Pulse Blocking uses the phase firing of the SCR except for
Sit are completely turned "off" or "blocked". The cycles
firing are intermixed with "off" cycles (see Figure 2).
OFF CYCLES:
BACKGROUND:
AMPS:
VOLTS:
MILLIAMPS:
K'VOLTS:
00
23
55
90
15.2
iiiir~
.33 Milliseconds per Division
I I I I I
SECONDARY CURRENT k VOLTAGE WAVEFORMS
Pulse Blocking with 0% Background Power
Figure 2
Pulse Blocking takes advantage of the relationship of the voltage and
capacitance from the wire to the surface of the fly ash on the plate
and across the fly ash itself. The wire is pulsed with voltages with
less than 50% duty cycle so that the wire is at full voltage but the
ash layer is never allowed to charge. The more cycles "off", the
lower the fly ash voltage, and the greater the relative emitting
electrode voltage.
2-4
-------
Background Power is used with Pulse Blocking. The "off" cycles are
set at a percentage of the "on" cycles (see Figure 3).
OFF CYCLES: 4
% BACKGROUND: 10
AMPS: 65
VOLTS: 163
MILLIAMPS: 320
K'VOLTS: 18.6
A A A A A A A A
I I I
\ \ \
8.33 Milliseconds per Division
I I I I I I I I
SECONDARY CURRENT & VOLTAGE WAVEFORMS
Pulse Blocking with 10% Background Power
Figure 3
Since the Background Power can be adjusted independently of the
"on:off" ratio, it gives another dimension of control.
2-5
-------
The usefulness of Background Power is explained with the analogy of
powering a light bulb with Pulse Blocking. The Background Power
keeps the filament hot during "off" cycles. The result is that the
light bulb is not allowed to operate with a cold filament, thus the
bulb lasts longer, produces more light and uses less energy. In the
precipitator, Background Power keeps the corona at a low level during
the "off" cycles (see Figure 4).
3.33 Milliseconds per Division
I I I I I I I I L
PULSE BLOCKING ALONE
18.6 KV, 320 mA
PULSE BLOCKING WITH 20% BACKGROUND POWER
NO PULSE BLOCKING, 21.9 KV, 1320 mA
SECONDARY VOLTAGE WAVEFORMS
Comparison
Figure 4
The advantages are:
1. The minimum voltage across the fly ash layer is set
independently of the Pulse Blocking ratio.
2. Corona is always generated and does not allow fly ash to
collapse back onto the wire during the "off" cycle and foul the
wire. It also reduces re-entrainment due to loss of voltage
across ash layer.
3. Eliminates Pulse Blocking ratio changes due to change in fly
ash characteristics when rapping.
4. Reduces inrush current to recharge precipitators on the first
"on" cycle.
2-6
-------
RESULTS
Using the ideal algorithm results in a rapid drop in power usage with
a declining or constant opacity. Eventually, the opacity will
increases with reduced power. However, the sharper the knee of the
graph curve, the greater the possibility of energy savings and
opacity reduction.
Current
All 36 (35) AVC secondary current limits were reduced in ten (10)
equal steps from 1800 ma. down to 120 ma. (see Figure 5). Note the
dramatic reduction in power (Megawatt) used compared to the relative
small change in opacity.
1.8 —
1.7 —
1.6 —
1.5 —
LU 1,4 —
cr>
5 1.3 —
=> 1.2 —
*~ 1.1 —
3: 1.0 —
o 0.9 —
LJJ
^ 0.8 —
< 0.7 —
g 0.6 —
a: 0.5 —
•- 0.4 —
0.3 —
0.2 —
0.1 —
PROGRAM SMA LIMIT
09 (MA)
Q8 9 1800
8 1600
7 1400
6 1200
71 5 1000
T>6 4 800
1 3 600
1 2 400
I 1 200
| 0 120
O5
4O
\
\
O3
A
2 >v
^v
<3— — °
1 °
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
0 1 2 3 4 5 6 7 8 9 10 1 1 12 13 14 15 16 17 18
PERCENT MEASURED OPACITY
OPACITY vs. WATTAGE
Secondary Current Limit
Figure 5
2-7
-------
Voltage Limit
All 36 (35) AVC secondary voltage limit was reduced from 48 KV to
18 4 KV (see Figure 6). Note that the power savings compared to
opacity is apparent, but not as good as the current limit method.
CD
•<
CO
0
1—
cc
1.8 -
1.7 -
1.6 -
1.5 -
1.4 .
1.3 .
1.2 •
1.1 .
1.0 .
0.9 .
0.8 .
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
PROGRAM
9
8
7
6
5
4
3
2
1
0
SKVLIM1T
(KV)
48
26.4
24
224
20.8
18.4
T n i i i M i IM i i i II i
1 2 3 4 5 6 7 8 9101112131415161718
PERCENT MEASURED OPACITY
OPACITY vs. WATTAGE
Secondary Voltage Limit
Figure 6
Each program reduces the secondary voltage by a different value to
operate on or about the onset of corona. The test was limited to
program "4" due to high opacity
2-8
-------
Pulse Blocking
All 36 (35) AVC Pulse Blocking ratios were set at 1:1 and increased
to 1:3 (see Figure 7). The test was terminated due to the excessive
opacity above 20%. This may be explained by:
1. Low resistivity and low voltage across the ash layer allowed
re-entrainment, especially on the outlet fields.
2. Rapping caused re-entrainment.
CD
*t
to
1.8 —
1.7 —
1.6 —
1.5 —
1.4 —
1.3 —
1.2 •
1.1 •
1.0 •
0.9 —
0.8 —
0.7 —
0.6 —
0.5 —
0.4 —
0.3 —
0.2 •
0.1 —I
0 •
PROGRAM
9
8
7
6
PB RATIO
(ON:OFF)
1:0
1:1
1:2
1:3
8
01234
I I I I
5 6 7 8
9 10 1
I I I I I I I
12 13 14 15 16 17 18
PERCENT MEASURED OPACITY
OPACITY vs. WATTAGE
Pulse Blocking with 0% Background Power
Figure 7
2-9
-------
Background Power
All 36 (35) AVC repeated the Pulse Blocking ratio, but with 10% and
20% Background Power (see Figure 8 and 9). Note that the test was
able to go up to the ratio 1:9 without excess opacity. Also note the
reduction of energy used before the increase of opacity.
1.8 —
1 . / —~"
1.6 —
1.5 —
CO
-< 1.3 —
CO
=> 1.2 —
i —
i- 1.1 —
-t
3: 1.0 —
- 0.4 —
0.3 —
0.2 —
0.1 — I
0 —
PROGRAM PB RATIO
(ON:OFF)
9 1:0
8 1:1
7 1:2
6 1:3
5 1:4
4 1:5
3 1:6
2 1:7
1 1:8
0 1:9
0
2
i i i i i i i i i i g
1 1 i 1 1 i I
0 1 2 3 4 5 6 7 8 9 1 0 1 1 12 1 3 14 1 5 16 17 1 8
PERCENT MEASURED OPACITY
OPACITY vs. WATTAGE
Pulse Blocking with 10% Background Power
Figure 8
2-10
-------
1.8 —
1.7 —
1.6 —
1.5 —
1.4 —
1.3 —
1.2 —
1.1 •
1.0 -
0.9 —
0.8 —
0.7 •
0.6 —
0.5 —
0,4 —
0.3 —
0.2 •
0.1 —
0 •
PROGRAM
9
8
7
6
5
4
3
2
1
0
PB RATIO
(ON:OFF)
1:0
1:1
1:2
1:3
1:4
1:5
1:6
1:7
1:8
1:9
I I I I I I I i I I I I I I I I I I
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
PERCENT MEASURED OPACITY
OPACITY vs. WATTAGE
Pulse Blocking with 20% Background Power
Figure 9
2-11
-------
Plant personnel developed a algorithm of limiting power into the
precipitator by using the Central Control Unit (CCU) and IBM graphic
screen to evaluate the performance. The results were graphically
displayed as bar charts and they determined the best method was to
keep the inlet and outlet field at full power and reduce the
secondary current limit of the in between fields. Because the
current limit was used, corona was maintained in all fields (see
Figure 10).
•<
CD
-------
CONCLUSIONS
1. Reducing secondary current limit will reduce energy usage with
a small change in opacity.
2. Using secondary voltage limit is not stable enough to achieve
the same results as secondary current limit.
3. Pulse Blocking alone does save energy, but with an increase in
opacity.
4. Background Power works with Pulse Blocking to lower opacity
with the low resistivity ash on the outlet fields, and is
unaffected by normal operation disturbances such as rapping.
5. All the previous tests were to make identical changes to all of
the T/R Sets. The CPL/Coleto Creek test demonstrated that
power and opacity can both be reduced. The basic advantage of
ten (10) sets of parameters is that each AVC may be optimized
for its location in the gas flow. Additionally.- secondary
current limit, Pulse Blocking, and Pulse Blocking with
Background Power may be intermixed with location, boiler load,
type of coal and opacity level.
6. By using opacity as a control feedback signal, one (1) of ten
(10) sets of parameters may be selected. Each set (profile)
optimizes each AVC for that opacity and location in the gas
flow.
7- In actual use, the CPL/Coleto Creek precipitator uses energy
proportional to the boiler load and the amount of fly ash that
is collected. In the automatic mode it saves up to 90% energy
while the opacity is kept constant.
2-13
-------
Table 1
PROGRAM
7
6
5
4
3
2
1
0
DATA TEST RESULTS
Secondary Current Limit
MILLIAMP LIMIT KWATT
1800
1600
1400
1200
1000
800
600
400
200
120
1684
1779
1314
1344
944
858
559
442
183
136
OPACITY
2.4
2.4
2.7
2.4
3.1
3.0
3.9
4.4
7.7
10.0
PROGRAM
9
8
7
6
5
4
Secondary Voltage Limit
K'VOLT LIMIT KWATT
48
26.4
24
22.4
20.8
18.4
1782
1677
1146
950
502
332
OPACITY
2.58
3
2.96
3.56
7.49
16.14
PROGRAM
9
8
7
6
Pulse Blocking
PULSE BLOCKING KWATT
1:0
1:1
1:2
1:3
1726
146
193
90
OPACITY
2.5
12.1
11.9
16.3
Pulse Blocking with 10% Background Power
PROGRAM PULSE BLOCKING KWATT OPACITY
9
8
7
6
5
4
3
2
1:0
1:1
1:2
1:3
1:4
1:5
1:6
1:7
1:8
1:9
1726
628
486
405
238
229
191
197
177
175
2.5
4.9
5.3
5.8
7.2
7.6
8.9
9.2
9
9.6
2-14
-------
1696
893
566
386
274
28
202
204
163
172
2.
3.
3.
5.
6.
7.
8.
8.
9.
11
49
43
77
47
52
44
4
67
52
.32
Pulse Blocking with 20% Back ground Power
PROGRAM PULSE BLOCKING KWATT OPACITY
9 1:0
8 1:1
7 1:2
6 1:3
5 1:4
4 1:5
3 1:6
2 1:7
1 1:8
0 1:9
CPL/Coleto Creek Method
PROGRAM KWATT OPACITY
9 1726 2.5
8 1189 2.77
7 1017 2.45
6 770 2.98
5 556 2.93
4 462 4.52
3 502 4.96
2 245 5.44
1 296 7.23
0 312 6.47
2-15
-------
AN EVALUATION OF THE ENERGY SAVINGS AND ELECTRICAL WAVEFORMS
FROM THE INTERMITTENT ENERGIZATION OF ELECTROSTATIC PRECIPITATORS
AT COAL FIRED STOKER UTILITY BOILERS
Peter Gelfand, P.E.
P. Gelfand Associates
60 Clover Hill Road
Trumbull, CT 06611
J. A. Alden
D. J. McKay
C. M. Richardson
New York State Electric & Gas Corporation
4500 Vestal Parkway East
P. 0. Box 3607
Binghamton, NY 13902-3607
ABSTRACT
The electrical power consumption of full scale utility precipitators was
evaluated with AC cycle blocking, or intermittent energization. Tests were
conducted on stoker fired coal boilers at the Jennison and Hickling Stations
of New York State Electric & Gas Corporation. The effect of intermittent
energization was to decrease power consumption by more than 70% without
significantly increasing opacity or particulate emission rate.
When current limits were appropriately set, intermittent energization did not
add significantly to electrical equipment stress in either steady state or
transient (sparking) operation.
3-1
-------
AN EVALUATION OF THE ENERGY SAVINGS AND ELECTRICAL WAVEFORMS
FROM THE INTERMITTENT ENERGIZATION OF ELECTROSTATIC PRECIPITATORS
AT COAL FIRED STOKER UTILITY BOILERS
INTRODUCTION
New York State Electric & Gas Corporation (NYSEG) was interested in evaluating
Intermittent Energization (I.E.) as a possible method of decreasing the
operating cost of their electrostatic precipitators (ESP). While I.E. had
been successfully applied to pulverized coal utility precipitators (1, 2),
uncertainties existed as to the effectiveness of the method for stoker fired
plants. Particularly, whether enough power could be saved to justify the
cost; and if I.E. had any long term detrimental effect on the electrical
equipment.
I.E., or AC blocking, can be described as an electronic method of interrupting
the flow of energy to the ESP at appropriate times and relying on the
precipitator's capacitance to sustain voltage and, thereby, dust collection.
Figure lisa typical schematic of an ESP power supply. During normal
operation, the silicon-controlled rectifiers' (SCRs) adjust the ESP voltage
and current by applying a voltage to the transformer-rectifier (TR) primary
winding for a portion of the time of each half cycle of line voltage. In I.E.
operation, the blocking action is achieved by using the SCR phase control to
turn off the power entirely for a preset number of half cycles. During the
time when the I.E. control allows power to flow, phase control is used to
adjust the output to spark-limit.
I.E. increases the peak voltage applied to the ESP while lowering the average
voltage and the power consumption. The amount of charge that a dust particle
receives is proportional to the peak voltage.
The evaluation program was conducted at two sites: energy measurements,
opacity and mass emission testing were made at Jennison Station; while an
assessment of energy savings and electrical stresses was performed at Hickling
Station.
TEST SITES
Jennison Station located in Bainbridge, NY and Hickling Station located in
Corning, NY are similar in design. Each plant has a total of four traveling
grate stoker boilers, manufactured by Combustion Engineering, that are
utilized to provide steam for two turbine generator units. Boilers 1 and 2
provide steam for unit 1, and boilers 3 and 4 for unit 2.
3-2
-------
Table 1 indicates the design parameters for the two plants. All boilers are
fired with bituminous coal. Average coal and ash analysis during 1988 is
shown in Table 2.
Flue gases from the boilers pass through a tubular air heater, multicyclone
mechanical collectors manufactured by Western Precipitation and an
electrostatic precipitator designed by Environmental Elements Corporation.
Mechanical dust collector and ESP design parameters are the same for both
stations and are tabulated in Table 3.
Fly ash resistivity is estimated to range between 4 x 109 to 1011 ohm-cm at 300
to 380°F.
I.E. RESULTS AT JENNISON
Opacity vs. I.E. Ratio
At the end of March 1987, eight intermittent timers (Redkoh Industries, model
#RK 1231) were installed on TR control cabinets Nos. 1-8 on the number 1
precipitator (Boiler 1). Unit 1 has two precipitators each with eight
electrical fields in the direction of gas flow. The gas flow from each
precipitator exhausts into a common stack. A single in-stack transmissometer
monitors the opacity.
In April 1987, preliminary testing was performed to determine the effect of
I.E. on precipitator performance. This was done by monitoring the opacity
changes as varying I.E. ratios were applied to the ESP's eight power supplies.
The generator was at full load.
Since the opacity measured is that of the combined gases from both ESPs, it
can be used to determine whether the I.E. precipitator performance improves or
degrades compared to normal powering. I.E. ratio applied during testing
ranged from normal, 2/2 to 2/80.
Figure 2 plots I.E. vs. opacity. The opacity remains unchanged to I.E. ratios
of 2/14 to 2/20. Further decreases in I.E. duty cycle significantly increased
opacity.
Energy Savings
The precipitator was operated with I.E. at 2/20 until February 1988 when the
energy measurement tests were performed. Because of the harmonic distortion
and waveform crest factors generated by the precipitator and I.E. operation,
care must be taken to apply an instrument that can accurately measure energy
and power (3). In order to compare the power differences required to maintain
energization of the precipitator, individual watt/watthour transducers and
current transformers (CTs) were installed in each of eight cabinets. The
watt/watthour transducers were Scientific Columbus Exceltronics and the CTs
were Square D split-core type. Prior to installation, the watt/watthour
transducers were calibration tested and found within the published standard of
±0.2% of reading accuracy. The transducer should be unaffected by influences
3-3
-------
of power factor, temperature, light loads, overloads, positioning, external
magnetic fields, relative humidity, or frequency. The CTs were made with
split cores to eliminate possible saturation effects caused by unbalanced
current pulses. The transducer pulse outputs were recorded with an Acurex
datalogger programmed to count the pulses.
A test was conducted at a 28mw load over an eight day period. The power mode
was changed every eight hours to distribute the I.E. operation over different
time of day shifts and over different days. Each cabinet's controls were
optimized just prior to starting the test. Boiler and precipitator operation
over the test period was considered typical. The average stack opacity was 2-
4% over the testing period and did not display any noticeable change with I.E.
operation. The results are summarized in Table 4.
The total energy savings was 2,759,457 watthours (28.745 kw) or 88.61%. The
economic value of the savings based on 2.62tf/kwhr (fuel cost savings) is
$18.10/day or $6,335 (350 days) per ESP per year. On a unit basis, that is
with both ESPs using I.E., the savings would double, equalling $12,670.
ELECTRICAL TESTING AT HICKLING
Hickling Station was selected for electrical testing. The boilers are similar
and the precipitators are identical to those at Jennison Station. Testing was
conducted to determine if discontinuous power, I.E., would induce voltage or
current in a component of the precipitator power supply that could adversely
effect its life, particularly the high voltage transformer rectifier.
Measurements were taken at the inlet and outlet fields of precipitator number
4.
The tests were conducted with conventional and intermittent power during both
steady state and transient (sparking) operation. Inlet and outlet fields were
selected because they represent the extremes of electrical conditions present
in the collector. The inlet field treats gasses directly from the boiler and
contains the highest dust concentration, while the outlet field treats the
smallest dust concentrations. The influence of the dust concentration on
electrical operation can be seen in Figure 3 , the corona discharge
characteristic of the inlet and outlet field. The charging of the dust
results in a space charge effect that shifts the voltage and current
relationship. The inlet field has higher voltages and lower current while the
outlet field operates at lower voltages but very high currents.
The plant selected an I.E. ratio that provided good operation, 2/14. We used
this as the I.E. ratio for testing and varied electrical levels to observe
steady state and sparking operation. Similarly, we used normal conventional
power, 2/2, to observe and compare electrical set operation. A diagram of the
test equipment connections is shown in Figure 1. Power measurements were
taken using a Scientific Columbus transducer. This is the same equipment used
at Jennison Station to measure the I.E. energy savings. Also, the power was
calculated directly from the measured waveforms using the Tektronics 2430/4041
oscilloscope measurement system thus providing a check of the transducer.
The waveforms generated in the power supply circuit have complex non-
sinusoidal shapes because of the corona discharge load presented by the ESP
3-4
-------
and this can make analysis difficult. Figures 4a, 4b, 4c, 4d, 4e, 4f, and 4g
are typical circuit waveforms measured on normal operation, 2/2. In addition,
there are two situations which must be carefully evaluated for both
conventional and I.E. operation: first, non-sparking or steady state; and
second, when sparking occurs. To aid in the analysis, two simple circuit
models of the power supply and ESP were developed, a steady state and a
sparking network.
Steady State Operation
Figure 5a represents the circuit model used when power flows from the supply
to the precipitator (the charging period) and Figure 5b represents the circuit
applicable during the time interval between charges (the precipitator
discharge). The major circuit elements and their parameters are referred to
the secondary side of the transformer. The current limiting reactor impedance
is referred to the secondary side of the transformer by multiplying the
inductance by the turns ratio squared and similarly, the primary voltage is
referred to the secondary by multiplying by the turns ratio. Since anti-
parallel SCRs switch the primary voltage, a phase angle a is applied to this
voltage. Circuit 5a applies when the voltage from the transformer exceeds the
voltage at the precipitator so that charging current can flow from the supply
to the precipitator and voltage on the collector can be maintained or
increased. The current through the inductor LI, when rectified will be the
same as the transformer-rectifier precipitator current, lesp. In theory, if
we multiplied the inductor current by the turns ratio, this would be the
current that flows in the primary, Ipri. The voltage at the capacitor and
resistor is the precipitator high voltage, Vesp. No current flows in the
inductor during the discharge period. The discharge circuit 5b then applies.
The equation for the precipitator voltage (charging interval):
Vesp = Vmax ^'/(c^V) [COS(a)(COS(wt)-COS(Wlt)-w SIN(a) *
(SIN(wt)/w-SIN(Ult)/Wl)] + V(0+)COS(Ult)
The equation for the inductor current:
I = Vmax/((Wl2-^2) L) (COS(a) (-« SIN(wt) + W1SIN (^ t)))
(wSIN(«) (COS(wt)-COS(Ult)) + (Vesp- V(0+))/R + V(0+)SIN(Wlt)/LWl
Where:
w = 27rf = 27r60 = 377
(1/LC)5
a = phase conduction angle of the SCR's
L = 43 henries, (transformed value, actual inductance is 2.43
millihenries)
3-5
-------
C = Precipitator capacitance, not considering corona, is about 30 E-09
farads for 10,000 ft2 of collecting surface at 9 inch duct spacing.
The capacitance increases as corona forms and higher voltage levels
apply.
R = Precipitator resistance is approximately 750,000 ohms, or discharge
time constant of 0.02 seconds.
Vmax = RMS transformer secondary voltage * 1.414 (maximum ESP voltage)
V(0+) = The voltage at the start of the charging period (minimum ESP voltage)
The corona capacitance and resistance values can be estimated from:
C = (q/delv) * (tdis/(tdis+delt))
Where: q is the change in capacitor charge =
(2/?r) * precipitator peak current pulse * delt
delv is the change in ESP voltage = Vmax- V(0+)
tdis is the capacitor and resistor discharge time constant =
(time of the discharge -delt)/-Ln((VO+)/Vmax)
delt is the time duration of the charging current pulse
The models apply to both conventional and I.E. operation. A computer was used
to solve these equations and the results are compared in Table 5.
Because of the non-linear nature of both the precipitator capacitance and
resistance, these models should be viewed as indicating approximate values
rather than exact results.
The conduction time for each half cycle is short, 2-2.8 milliseconds out of a
possible 8.33 milliseconds. The short conduction time is due in part to the
impedance of the current limiting reactor, about 25%. Many power supplies use
a larger impedance of 40-50% and this can sometimes result in better ESP
operation by providing a smoothing of the high voltage waveform and longer
conduction times before sparkover.
By using the charging model we can also estimate the peak voltage and current
that results if Vmin is held constant and only the conduction phase angle (on-
time) is varied. This is plotted in Figures 6a and 6b. Note how sharply both
voltage and current change even for small increments in angle. In theory, the
peak voltage generated could exceed the maximum voltage of the transformer's
secondary winding, but, in practice there will be added capacitance due to
corona space charge effects and increased resistive losses as well as sparking
within the precipitator that will intervene, limiting the voltage before
higher levels are reached. If we now hold the conduction angle constant and
vary the minimum voltage, we can see the effect on the peak voltage and
current in Figures 7a and 7b. The lower the minimum voltage, the greater the
peak voltage and current. With conventional powering, the minimum voltage is
normally about 50-70% of the peak voltage, but the minimum voltage will become
3-6
-------
very low immediately after a spark or arc. If the automatic control were to
ignore the sparking and keep the conduction angle the same or increase it, a
great surge of current would be applied in the next half cycle. This could
cause intense arcs and is why manual control, that adjusts power by setting a
constant conduction angle, should only be used well below the point of
sparking. If you disregard sparking and increase the control past this level,
large currents will occur that could damage the equipment and reduce
precipitator performance.
This same effect, lower minimum voltage and a higher peak, is used to
advantage by I.E. operation. Since power is applied for a short interval, one
cycle, a longer discharge time occurs between charging cycles. The Vmin is
lower than that found in continuous powering. When the I.E. charging cycle is
applied, the peak precipitator voltage reaches a higher level.
Figure 8, taken on the outlet field during 2/14 operation, indicates this
effect (-6kv minimum and a -48kv peak). To ensure the peak current on I.E.,
operation is limited to the capacity of the power supply components
(transformer-rectifier and current limiting reactor), the current limit
control is adjusted to limit the maximum conduction angle that can occur.
This will keep the peak primary current within equipment ratings.
The average TR current output is rated at 750 milliamperes. This rating is
based on a resistive load, so that in terms of peak current:
Peak current = (n/2) * (750 milliamperes) = 1.18 ampere
The rectifier bridge in the TR consists of two doublet assemblies of 1N5407
diodes. These have an average forward current rating of 3 amperes or 4.71
ampere peak. The inverse voltage rating of the rectifier assembly is about
lOOkv.
Figure 9 plots the outlet field current under I.E. operation and indicates
that the peak of the precipitator current during the first half cycle reaches
about 1.18 amperes. At the same time, the primary current waveform, Figure
10, indicated some slight distortion on the trailing edge of the first half
cycle of primary current. The primary current is rated at 120 ampere rms or
170 peak ampere. The maximum primary peak current in the first cycle was 174
ampere. We had noted this same primary current distortion in an earlier
measurement with conventional powering but on manual control, Figures 10 and
lOa. Increased trailing edge distortion is also present with higher peak
currents, Figure 11. It appears the current limiting reactor begins
saturating when the peak current is above 170 amperes. Saturation can
increase internal heating losses and reduce the impedance of the reactor and
may lead to premature failure.
By adjusting the maximum phase angle, the primary first pulse and the
precipitator current, 162 amps peak primary current and 0.86 amps precipitator
current (Figures 12 and 13) were brought within the equipment ratings. Based
on our testing, and with this limiting adjustment made, steady state ESP
operation with I.E. was found to add no additional stresses to electrical
equipment.
3-7
-------
An asymmetry in primary current was observed when using both conventional and
I.E. powering. This uneven triggering is probably due to an out of balance
condition in the automatic voltage control's trigger circuit. Care should be
taken when applying I.E. to control systems with magnetic type of SCR trigger
networks. Consideration should be given to replacing them with modern solid
state designs.
Transient Operation (Sparking)
High Voltage Circuit: To help discuss the operation of the power supply under
sparking, the charging model was modified by adding inductor 12, as shown in
Figure 14. L2 is a choke coil that is inside of the transformer-rectifier and
in series with the rectifier bridge output. When a spark occurs, the
precipitators's voltage rapidly collapses and places the difference in voltage
between the transformer secondary winding and the precipitator across the
choke coil. The inductance of the coil is about 0.1 henry. This limits the
rate of change of current through the rectifiers, thereby isolating and
protecting them. The rectifiers should have a minimum half cycle rating of
200 amps and in the microsecond range they can carry several times their half
cycle rating. The rectifier's overload current rating (I2t) is 165, and the
rate of rise on sparking is typically 50 microseconds. The worst case maximum
current at maximum secondary voltage can be calculated as shown:
Delta I = (delta time)(delta voltage)/inductance
Delta I = (50E-6 seconds)(70,000V peak)/.lH = 35 amps peak
therefore I2t .06125 amp2.sec
The spark shown in Figure 15 was taken on the inlet field under conventional
powering. The rise time is 11.9 usec and the peak is 6.4 amps. The spark
shown in Figure 16 was taken on the inlet field under a I.E. ratio of 2/14.
The rise time is 35.05 usec and the peak is 5 amps. The voltage transients on
sparking are normal. Figure 17 shows a spark on the inlet with normal
powering, and Figure 18 during I.E. operation. The voltage and current
transients were found to be about the same whether using I.E. or conventional
powering. We conclude that sparking on I.E. does not additionally stress the
high voltage rectifier or the secondary winding of the transformer.
Low Voltage Circuit: Figures 19 and 20 plot the primary waveforms during
conventional operation, 2/2, of the inlet field. Note the extreme asymmetry
in the primary current prior to the sparking event. Figures 21, 22, and 23
show the primary voltage and current waveforms during I.E. operation. The
maximum primary current that the transformer is rated for is determined from
the impedance of the transformer.
Maximum primary current = Primary current rating/Impedance
= 120A/0.061(.2ohms) = 1967.2 ampere
The current limiting reactor provides the circuit impedance to keep the
primary current low when a spark or arc occurs. The reactor's impedance is
calculated:
XL WL = (377)0.00243 henries = 0.916 Ohms
-------
The primary circuit surge - 480 volts/1.116 ohms = 430 amperes. The maximum
surge current observed was about 180 amps, well within the equipment ratings.
No abnormal voltage transients were observed. The primary waveforms under
sparking conditions were normal during both conventional and I.E. powering.
Power Measurement
The Scientific Columbus transducer was connected so as to measure the total
power used by the electrical set. This was achieved by energizing the
transducer's potential input directly from the line voltage and the current
input from a current transformer measuring the line current.
The average power measurements in automatic ESP operation are indicated below:
Inlet Field Outlet Field
Normal (2/2) 1776W 9952W
I.E. (2/14) 480W 1248W
Percent Reduction 73% 87%
MASS EMISSIONS TESTING AT JENNISON STATION
By the end of 1988, I.E. timers had been installed on all the Jennison Units.
In December 1988, source emissions testing was conducted. The first six
electrical fields were adjusted to an I.E. ratio of 2/20, the last two fields
to 2/10. A total of twelve test runs were performed to determine emissions of
particulate matter when continuously and intermittently energizing the ESPs.
Tables 6 and 7 summarize these results.
The average stack opacity was 4 to 5 percent during the tests.
The emission increase of unit 1 was larger than expected due to problems
encountered with the software controlling the ESP rapper sequence. The
rappers were not operating properly during these tests. Even though the
emissions were higher, I.E. operation easily achieved compliance with the
state regulations.
These tests indicated a need to readjust the rapping pace and intensity so as
to avoid increased reentrainment losses. The precipitator's lower average
current density may be reducing the forces holding the dust layers on the
collecting plates, making it easier to dislodge and reentrain dust.
CONCLUSIONS
1. I.E. operation reduced the energy used by the ESPs by greater than 70%
saving an estimated $25,000 per year in electric generation costs at
Jennison Station.
3-9
-------
2. The effect of I.E. on average opacity did not indicate any significant
performance change. Readjustment of the rapping intensity and pace is
necessary to avoid increases in particulate emissions.
3. Electrical operation on I.E. did not adversely affect the power supply
components provided that the current limit control was adjusted to limit
peak primary and precipitator currents.
REFERENCES
1. J. L. Dubard, E. C. Landham, Jr., W. Piulle, J. Riley and P. Gelfand.
"Evaluation of ESP Intermittent Energization." Third International
Conference on Electrostatics, Abano, Italy, October 1987.
2. E. C. Landham, Jr., J. L. Dubard, W. Piulle, and L. F. Rettenmaier.
"Pilot-Scale Evaluation of ESP Intermittent Energization". Sixth
Symposium on the Transfer and Utilization of Particulate Control
Technology, New Orleans, LA, February 1986.
3. P. Gelfand, E. C. Landham, Jr., and L. F. Rettenmaier. "Electrostatic
Precipitator Power Measurements." Ibid.
3-10
-------
me
Vind
480V
60HZ
10
o o 6 — 1| — o-
ine
1
line fpWR\_
— n
I t I I KWiaruKMtrs
CLR i dllt j
/ i r1" —
« rori VPn
Ipri J |
— "" J
c
; r» '
rm
< — •
i RECTIFIER
>
lespl 1 H.V. _
(MA) DIVIDER 1:
* | SCOPE
-45 KV avg
0.750A DC
Vesp
CHOKE /
PARAMETER
Vline, Vscr,
Vind, Vpri
Ipri
lesp
Vesp
MEASURED BY:
DIGITAL METER
FLUKE 6080
CLAMP ON
6080
6080
& METER
12K
WAVEFORM PROBES
TEKTRONIX MODEL:
A6902B, VOLTAGE ISOLATOR
MODEL A503/6J03
A503/6302
A6902, VOLTAGE ISOLATOR
OSCILLOSCOPE/CONTROLLER/PRINTER-TEK.2430/4041/HP THINK JET PRINTER
ESP
Q
^
Et
>-
[—
O
<
Q.
O
^ •
s
Puffing
I.E O,03
%
•V •
• j\
*^>^^ • •
^i^~L s
t •
•
0 0.05 0,1 0,15 0.2 0.25
I.E Ratio
• Opacity/I.E
ratios
— Curve Fit
Figure 1. Schematic Diagram Precipitator
Power Supply and Test Equipment Connections
Figure 2. Opacity vs I.E. ratio
Normalized Opacity vs Duty Cycle
-------
c
o>
z>
O
300
200-
o
'5.
D_
12
40
16 20 24 28 32 36
ESP Voltage (KV)
• Inlet Field <• Outlet Field
Figure 3. Volatcje-Current
Characteristics
Inlet and Outlet Fields
4 6 8 10 12 14 16 18
TIME
MILLISECOND
Figure 4a. Input line
(2/2)
60
40
20
0
-20
-40
-60
T 1 1 T
T T
J I
24 6 8 10 12 14 16 18
TIME
MILLISECOND
Figure 4b. Primary Current
(2/2)
600
400
g 200
o
> -200
-400
-600
j I
2 4 6 8 10 12 14 16 18
TIME
MILLISECOND
Figure 4c. Voltage across
inductor (2/2)
TIME
MILLISECONDS
Figure 4d. Voltage across
SCR (2/2)
8 10 12 14 16 18
TIME
MILLISECONDS
Figure 4e. Voltage across
Primary (2/2)
-------
ce
an
a
300
200
100
0
-
~
Hill
i-376 Ma :
MM
-
1 1
1
;
,,,,i,,,;
j
1 :
1'
avg
53
1
1 1 1 1
tiii
-
value- ~
Ma
2 4 6 8 10 12 14 16 18
">
¥
VOLTAG
TIME
MILLISECONDS
Figure 4f, Current in
ESP (2/2)
0
-20
-40
-60
-Rn
i i i i
t.
-
j
•*"-T_^ _*-Ul**%t«n
IT-*— ,„..,„ ^"-UTMV*:
_;
(III-
- -36 KV „
l 1
IN i
-
!N-%,HM • '"H
-44.8 KV
1 1
-
i i
246
8 10 12 14 16 18
TIME
MILLISECONDS
Figure 4g. Voltage on
ESP (2/2)
PRECIPITATOR MODELING CIRCUITS
LI-43H
Vesp
CHARGING MODEL * 5A
Vesp
t^.uo
uF/10,000 FT
ohms
DISCHARGING MODEL *5B
Figure 5
234
Conduction time (ms)
234
Conduction time (ms)
0 5 10 15 20 25 30 35 40 45
ESP Voltage Minimum (KV)
Figure 6a, ESP Peak Voltage vs
SCR Conduction Angle
Voltage Minimum 34.4 KV
Figure 6b. Current vs
SCR Conduction Angle
Voltage Minimum 34.4 KV
Figure 7a. ESP Peak Voltage vs
ESP Minimum Voltage (KV)-
Conduction Time 2.6 ms
-------
o
1.6
1
0.6
0.4
0.2
0
0 5 10 15 20 25 30 35 40 45
ESP Voltage Minimum (KV)
Figure 7b. ESP Peak Current
vs ESP minimum voltage (KV)
Conduction Time 2.6 (ms)
o
>
0
-50
-100
-150
"^
-:
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1.
-6kv 2
~ \rV '
\- rr-48kv -
1 1 1 1 :
: 1 I I I
L
^ ll44 ""I1
Illl
2 4 6 8 10 12 14 16 18
(/)
CURRENT AMI
o In o In
;
1.18A ~
jVj
1111 =
; 1 1 1 1
r 1.1A
illlLttJI Illl Illllmi
i/\ :
: 1 I 1 I
2 4 6 8 10 12 14 16 18
TIME
MILLISECONDS
Figure 8. Voltage on ESP-
outlet field (2/14)
TIME
MILLISECONDS
Figure 9. Current in ESP-
outlet field (2/14)
2U
-------
-------
10 12
TIME
MILLISECONDS
Figure 18. Voltage on ESP-
inlet field (2/14) with spark
20 25 30
TIME
MILLISECONDS
Figure 19. Primary current and voltage-
inlet field (2/2) with spark
10 12 14 16 18
TIME
MILLISECONDS
Figure 20. Primary current and inductor
voltage- inlet field (2/2) with spark
CO
01
o
20U
150
100
50
0
-50
-100
-150
-200
_
Al
1
\
^ -
I I 1 1 -
; 1 1 1 1
-
r
_ f
' SPARK
1 1 1 1
2 4 6 8 10 12 14 16 18
TIME
MILLISECONDS
Figure 21. Priamary current-
inlet field (2/14) with spark
8 10 12
TIME
MILLISECONDS
14 16
Figure 22, Voltage across primary-
inlet field (2/14) with spark
5 10 15 20 25 30 35 40 45 50
TIME
MILLISECONDS
Figure 23. Voltage across
SCR- inlet field (2/14)
-------
Steam Temperature
Steam Pressure
Steam/Hour
Generator Output
Hlckling Sta
900'F 960*F 850'F 900'F
875 PSIG 875 PSIG 670 PSIG 670 PSIG
175,000 Ibs 220,000 Ibs 200,000 Ibs 200,000 Ibs
37,000 KM 49.000 KM 33,000 KW 38,000 KW
infcal Collector:
Type
Number Tubes
Tube Diameter
irostatic PreclpHator:
Design Gas Flow (ACFM)
Specific Collection Area
(ft'/ 1000 ftVmtn.)
Gas Temperature 'F
Number of Fields
Number of T-R Sets
Guarantee ESP Efficiency
Guarantee ESP Outlet
(gr/scf)
9 VG 1Z
208
9"
135,000
530
350-380
8
8
99.5
0.015
9 VG [2
208
9"
150,000
544
300-350
8
8
99 5
0,015
Ming
Total Carbon
Hydrogen
Nitrogen
Oxygen
Moisture
Sulfur
Ash
Fuel (as received)
S10,
Al,0j
FeA
710,
CaO
HgO
Na,0
K,0
PA
SO,
Trace/Other
Loss on Ignition
Fisher
66.35%
3.73%
1.08%
3.35%
4.96X
0.85X
19.66%
11504 Btu/l
Ash -Minerals
56,97%
30.16%
6 96X
1.80%
0.1 IX
0.60%
0.21%
Z.09%
0.05%
0.73%
0.32%
79.29%
Antrini
64 6?'.
3 70%
1-10%
3 39%
4 60%
1 31%
21.24%
11293 Btu/l
57 08%
23 08%
8 99%
I.62T.
0.06%
0-65%
0 19%
2 39%
0. 10%
0 65S
0 19%
77 74%
JEMISON NO. 1 PRECIPITATOR
POKER SUMHARY • I.E. 'Off/I.E. "On-
Cabinet No. 1
Cabinet No. 2
Cabinet No. 3
Cabinet No. 4
Cabinet No. 5
Cabinet No. o
Cabinet No. *
Cabinet No. 8
8 DAY TOTALS
0-0800
117725
14862
123954
15947
119511
7566
84858
9777
117747
15744
178436
16292
231352
24408
112836
14448
1086419
119044
QBQO-1600 1600-2400 Dally Totals
104847 112766 33S338
16169 14709 45740
119066 120084 363104
16844 14267 47058
118454
7549
83339
9135
113211
15951
143080
17580
237132
23080
108576
12984
1027705
119292
116851
7373
79530
10590
110205
15948
123848
17120
230052
23252
106588
12996
999924
116255
(88. 6%)
354816
22488
247727
29502
341163
47643
445364
50992
698536
70740
32BOOO
4Q428
3114048
354591
Off
On
Off
On
Off
On
Off
On
Off
On
Off
On
Off
On
Off
On
Off
On
TABLE 5
Electrical
field
Inlet
Inlet
Inlet
Inlet
Electrical
Outlet
Outlet
Outlet
Outlet
ESP during
Hode
2/2
2/14
2/14
2/14
Hode
2/2
2/2
2/14
2/14
the charging
Heas. Cal.
34.4 34.4
23 2 23.2
23.2 23.2
22.4 22.4
V»1n(kv)
Heas. Cal .
22 22
IB 18
6 6
2 2
Interval .
Heas.
46.4
48.8
SO 4
55.2
Vpeakl
56
48
48
40
CaL
44.4
44.7
44.7
53.5
[Kv)
ClL
52.9
47
47.8
42.6
Heas , Cal .
0.500 0.43
0.960 0.75
0 860
0 860
0.75
0.81
Ipptr/1nd(Amps Pk)
Heas . Cal .
1.36 0.97
1.16
1.12
.75
0.86
0.93
0 70
2.6
2 6
2.5
2 8
onlmsi
2 8
2.5
2 3
2.0
SUMMARY OF RESULTS
JENHISON UNIT 1 PAR7ICULATE EMISSION TESTS
SUMMARY OF RESULTS
JENNISON UNIT 2 PART1CULATE EMISSION TESTS
Heat Input
Test (10* BTU/Hrl
(Cont. Energized)
1 490
2
3
AVG
(I.E.)
2
3
AVG.
490
490
490
490
490
490
490
fGr/dscfl
0.0034
0
0
0
0
0
0.
0.
.0107
.0027
.0056
.0159
.0204
0130
0164
ILbs/hrl
3.57
11.6
2.76
5.98
13.9
18.1
11.2
14.4
0 009
0.027
0.007
0.014
0.042
0.049
0.031
0.041
Percent of
Standard '
3
10
3
5
16
19
12
16
Heat Input
(Cont. Energized}
1 470
Z 491
3 470
AVG. 477
(I.E.)
1 491
2 491
3 470
AVG. 484
(Gr/dscfl
0.0091
0.0047
0.0043
0.0060
0.0097
0.0117
0.0120
0.0111
ILbs/hrl i
8.90
4.52
4.04
5.83
9.44
11. S
11.7
10.9
Lbs/10' BTU1
0.020
0.012
0.011
0.014
0.025
0.029
0.029
0.028
Percent
Standar
8
5
4
6
10
12
12
11
* 0.26 1bi/lO*BTU bisid on heat Input
• 0.25 lbi/10*8TU based on heat input
3-17
-------
INTERMITTENT ENERGIZATION OPTIMIZATION
ON PSI GIBSON STATION - UNIT #1 PRECIPITATOR
S. Szczecinski
J. Lantz
M. Neundorfer
Neundorfer, Inc.
4590 Hamann Parkway
Willoghby, Ohio 44094
R. Pepmeier
Public Service Indiana
Gibson Generating Station
Owensville, Indiana 47665
ABSTRACT
Experiments were performed with Gibson Station's Unit #1 precipitator to
optimize the use of intermittent energization for performance enhancement and
energy savings. Six different combinations of intermittent energization duty
cycles were evaluated with multiple trials in most cases.
The results indicate that significant energy savings can indeed be obtained
with no deterioration in opacity or even with improvement in opacity. The
recommended IE duty cycle of one half-cycle on/two half-cycles off in the
first four fields will potentially save nearly 3,000,000 kW hours of
electricity per year, equivalent to an energy cost savings of around $50,000
per year.
4-1
-------
INTERMITTENT ENERGIZATION OPTIMIZATION
ON PSI GIBSON STATION - UNIT #1 PRECIPITATOR
INTRODUCTION
This paper discusses the findings and recommendations resulting from the
intermittent energization optimization effort performed on Unit #1 of Gibson
Station, Public Service Indiana. The on-site work was performed during April,
1988.
BACKGROUND
Back Corona
High resistivity dust accumulation on collection plates of electrostatic
precipitators can seriously diminish collection efficiency. An electric field
is generated across the dust layer on the collecting plates when corona
current is present in the precipitator. When this field exceeds the breakdown
level, back corona occurs. Back corona is a primary cause for reduced
collection efficiency.
In some cases, when the dust layer is thick, uneven, or when electrode
alignment is poor, the back corona may be more concentrated in small areas.
This can cause sparking at low power levels. In more uniform layers of high
resistivity dust, however, more stable back corona is likely to occur. In
this case, the breakdown is distributed more uniformly and sparking occurs
less frequently, if at all (1).
Back corona causes large quantities of positive ions to be emitted from the
collecting plates. This barrage of ions results in increased currents but
reduced collection voltage. The positive ions cancel the charge on
electronegative gas ions and dust. This effect causes the current to increase
and reduces the collection efficiency.
Intermittent Energization
The equivalent circuit representation of the dust layer is a parallel
combination of a resistor and capacitor. The value of the resistance is
determined by the resistivity of the dust and the thickness of the dust layer.
The dielectric constant of the dust determines the capacitance (2).
4-2
-------
The capacitance of the dust layer requires a charge time on the order of
hundreds of milliseconds before the layer breakdown and back corona occurs.
Back corona can be minimized if the precipitator can be energized for a time
period less than the charge time of the dust layer, then deenergized for a
period to allow the dust to discharge.
The intermittent energization (IE) feature of the voltage control energizes
the transformer-rectifier set for an integral number of half-cycles of the
line, followed by an even number of half-cycles off. This is illustrated in
Figure 1.
The notation for describing the duty cycle of the IE is as a ratio of the
number of half-cycles of on-time divided by the sum of the on-time and off-
time half-cycles. For example, one half-cycle of on-time followed by four
half-cycles of off-time would be represented as a duty cycle of 1/5.
IE has an implicit energy savings associated with the cycles of
deenergization. The half-cycles of off-time result in a power savings that
usually exceeds the increased power level in the half-cycles of on-time.
Public Service Indiana, Gibson Station Unit #1
The Gibson Unit #1 generating facility is a Foster Wheeler, front fired, dry
bottom boiler rated at 635 Megawatts. The boiler is fueled by Midwestern
bituminous coal containing approximately 2% sulfur by weight.
The effluent from the boiler is treated for particulate abatement by two five
field Joy Western precipitators. These precipitators treat a total of
2,282,000 cubic feet of flue gas at 288 degrees F. Expected efficiency for
the precipitator is 99% at these conditions. Each precipitator has one
chamber. The total collecting surface is 583,200 square feet, duct spacing is
nine inches and the SCA is 255.
The precipitator is equipped with a total of 24 Neundorfer MVC II
microprocessor-based voltage controls.
Because opacities of Unit #1 were typically well below the emission
requirements and the precipitator power was quite high there was a strong
desire on the part of PSI to evaluate the performance enhancement and energy
saving capabilities of IE on this precipitator.
OBJECTIVES
The objective of this study was to evaluate intermittent energization at
different IE settings to find the optimum conditions with respect to opacity
reduction and energy savings.
SCOPE
The scope of the investigation included the following:
4-3
-------
1. Obtained readings of true precipitator power (in watts rather than
volt-amps) and opacity under continuous full-wave operation of the
precipitator.
2. Obtained voltage current curves for over half of the controls under
normal operation to get an idea of the flyash resistivity
characteristics, which are an indication of which IE ratios might
be most reasonable to try.
3. Intermittent energization was evaluated at ratios of 1/3 and 1/5 in
six combinations. In most cases, two or more experiments were
performed for each combination to increase the significance of the
observations.
The boiler was base loaded during the experiments and the boiler's computer
was set up to print out averages of opacity, air heater exit temperatures and
boiler load every 6 minutes. Power was logged manually for each of the
voltage controls via their internal true-power readouts (watts).
The boiler load was held constant at 95% of rated capacity, and air heater
exit temperature averaged 280 degrees Fahrenheit.
The experiments were synchronized with the printout of 6 minute averages. In
this way the average of opacity readings could be compared with a
representative reading of precipitator power during the period. Experiments
typically lasted at least 30 minutes, providing at least five 6-minute opacity
averages.
The experiments were typically interspersed with equivalent no-IE trials.
Except in a few cases at least one no-IE trial was time-adjacent to each IE
trial. In this way power and opacity with and without IE could be compared
under nearly equivalent conditions.
RESULTS
Voltage-Current Curves
A typical V-I curve is shown in Figure 2. Separate plots of the minimum and
maximum voltage are shown. The difference between the maximum and minimum
voltages at any given current represents the ripple in the waveform. Figure 2
differs from the usual V-I curve, which indicates average voltage rather than
the minimum and maximum.
The plot of the minimum instantaneous secondary voltage is particularly
valuable in revealing the existence of back corona. In these tests many of
the voltage minima stayed either almost constant with increasing current or
bent back slightly. Such curves are indicative of some back corona although
certainly not severe back corona.
Based on this information it was anticipated that intermittent energization
ratios between 1/3 and 1/5 would be successful.
4-4
-------
Intermittent Energization Trials
Table 1 describes the trials that were performed during the investigation.
Each IE trial was immediately preceded or followed by an interval of
continuous full-wave operation as a baseline trial. This allowed each trial
to be associated with a consistent reference for comparison. Table 1 shows
the opacity and power level averages for each type of trial with respect to
its reference trial.
All opacities are the averages of the multiple 6 minute averages. Each table
entry represents the average power for each experiment type (that is, averaged
over all the experiments of that type).
The data for each of these six experiments was averaged and plotted in Figures
3 and 4. Figure 3 correlates the observed opacity with the precipitator
power. Opacity improves as power decreases down to about 2/3 of the full-
power level. Additional reduction in power causes increased opacity.
Figure 4 shows how saving power with IE alters the opacity relative to
baseline (full wave) operation. Each point shows the percentage of power
savings with respect to the change in opacity from the full-power level.
Positive values indicate higher opacity and negative values indicate lower
opacity. The point at the origin shows that full-power operation results in
zero power saved and zero effect on opacity.
It is apparent that the performance improvement and energy savings with IE are
significant. The biggest energy savings were obtained with a duty cycle of
1/3 in all fields (Trial 6). However, this was obtained at the cost of
significant increases in opacity. By contrast, this same setting in only the
first four of the five fields (Trial 2), caused a slight decrease in opacity
on the average while still providing very beneficial energy savings.
The remaining combinations tried were less optimal. Other combinations could
be tried, but the ones tried were selected to suit the resistivity of the ash.
Ratios of 2/x and 4/x were omitted because these are in most cases found to be
less effective, and may also cause asymmetry in precipitator half-wave
currents.
It was interesting to note that intermittent energization significantly
reduced instantaneous opacity spikes. This may partly explain the overall
improved opacities in some cases. However, less dramatic correlations were
seen on the instantaneous opacity charts during some experiments.
CONCLUSIONS
The following are concluded, based on these evaluations:
1. It has been demonstrated that for the Gibson Station Unit #1
precipitator significant improvements in performance are possible
with energy savings as well.
4-5
-------
2. The most optimal performance was obtained with a ratio of 1/3 (one
half-cycle on/two half-cycles off) for all fields except the last
field (Trial 2).
3. At least for this particular precipitator it appears undesirable to
use intermittent energization in the last field.
4. Intermittent energization was observed to reduce opacity spiking.
RECOMMENDATIONS
In order to improve collection efficiency, it was recommended that Gibson
Station operate Unit #1 with intermittent energization in the first four
fields at a ratio of 1/3 (one on/two off). The experiments indicate a power
savings in the neighborhood of 330kW. On an annual basis this equates to
2,890,800 kW hours of energy. At a unit power cost of 1.7 - 1.8 cents per kW
hour, this suggests a savings of approximately $49,000 - $52,000 per year with
a reduction in opacity.
DISCUSSION
Ideally, the selection of an appropriate IE ratio for a particular
precipitator will include more experiments under a wide variety of conditions.
Variations in fuel, boiler load, and other parameters can significantly affect
the precipitator operation. In addition, longer trials will provide more
reliable information.
4-6
-------
REFERENCES
1. K. J. McLean and L. E. Sparks. "Some Electrical Characteristics of Back
Corona." In Proceedings of Sixth Symposium on the Transfer and
Utilization of Participate Control Technology, vol. 2, 1986.
2. Y. Matsumoto, S. Sugiura, T. Ando, N. Teramura. "Development of Mitsubishi
Intermittent Energization System (MIE) for Electrostatic Precipitators for
Coal Fired Boilers." Technical Review, Mitsubishi Heavy Industries, Ltd.,
October, 1982
3. E. C. Landham, Or., J. L. DuBard, W. E. Piulle, and L. F. Rettenmaier.
"Pilot-Scale Evaluation of ESP Intermittent Energization." In Proceedings
of Sixth Symposium on the Transfer and Utilization of Particulate Control
Technology, vol. 2, 1986.
4-7
-------
CO
SECONDARY
CURRENT
A
HMf
A
4 HALF CYCLES _
A
A
A
TIME
TIME
SECONDARYi
VOLTAGE W
Figure 1. Typical current and voltage wave forms for IE 1/5 duty cycle
-------
10
-p.
1C
10OO
8OO
6OO
fc 400
3
CJ
200
2O
Minimum voltage
3O
40
2O
1OOO
8OO
6OO
4OO
2OO
4O
50
Volts
-------
12
11
1O
1 9
h
a
^ a 8
0 0
7
6
5
O 2OO 4OO 6OO SOO 1OOO
\
c\
Trial 6\,
\
\ Trial 5
\ "
\
\ ~
\
~ O\ Baseline "~
Trial 4 \ Trial 3 Q
\ / 1
\ Trial 1 /
0 X^ 0 ^ 1
Trial 2 ^^_ _^--
~" —
1,1,1,1.
12
11
10
9
a
7
6
5
2OO 4OO 600 8OO
Precipitator Power, KW
Figure 3. Opacity vs. total precipitator power
1000
-------
0)
a
-1
-2
-3
Baseline
\
\
\
10
20
3O
40
50
Trials /O Trial 6
O /
60
Trial 4
Trial 1
O
/ Trial3
O
Trial 2
O
10
20
3O
4O
5O
Percent power savings
Figure 4. Change in opacity versus power savings
-i
-2
-3
<5O
-------
TABLE 1 RESULTS OF IE EXPERIMENTS
TRIAL NUMBER I FIELDS (FLD)/DUTY CYCLE(D.C.)
1 FLD:1,2,3,D.C.1/3
2 FLD:1, 2,3,40.0.1/3
3 FLD:1 D.C.1/5 FLD'2,3,4 D.C.1/3
4 FLO: 1, 2 D.C.1/5FLD:3,4 D.C.1/3
5 FLD:3, 4,5 D.C.1/3
6 FLD:1,2,3,4,5D.C.1/3
I AYE. OPACITY
6.40%
6,50;?
7.20%
7,90%
10,20%
11, 40%,
CHANGE
-1,00%
-1.50%
-0.50%
1,60%
3,40%
3.70%
POWER SAVED
24%
37%
37%
45%
45%
cr^q?
Oo*
-------
FULL SCALE DEMONSTRATION OF INTERMITTENT ENERGIZATION
ON A 500 MW HOT-SIDE ELECTROSTATIC PRECIPITATOR
Wall is A. Harrison
Robert P. Gehri
Southern Company Services
800 Shades Creek Parkway
Birmingham, Alabama 35209
E. C. Landham, Jr.
Southern Research Institute
2000 Ninth Avenue South
Birmingham, Alabama 35255
Morris B. Tuck
Mississippi Power Company
P.O. Box 950
Escatawpa Mississippi 39552
Walter Piulle
Electric Power Research Institute
3412 Hill view Avenue
Palo Alto, California 94303
ABSTRACT
This paper describes a full scale test program performed by the Electric Power
Research Institute, Southern Company Services, Southern Research Institute, and
Mississippi Power Company on the benefits of intermittent energization (IE) when
applied to the 500 MW hot-side electrostatic precipitators on unit 1 at Plant
Victor J. Daniel. IE is a technology developed to improve electrostatic
precipitator operation and performance by selectively blocking portions of the
alternating current input sine wave. Methods for IE optimization, potential
problem areas, and results are presented.
5-1
-------
FULL SCALE DEMONSTRATION OF INTERMITTENT ENERGIZATION
ON A 500 MW HOT-SIDE ELECTROSTATIC PRECIPITATOR
INTRODUCTION
The Electric Power Research Institute (EPRI), Southern Company Services (SCS),
and Mississippi Power Company (MPC) recently completed an investigation of the
benefits of intermittent energization (IE) on the 500 MW hotside electrostatic
precipitators on unit 1 at Plant Victor J. Daniel. This program is part of a
larger program funded by EPRI to evaluate the usefulness of the IE technology
for the American utility industry. IE is a technology developed to improve
electrostatic precipitator operation and performance by modification of the
voltage and current waveforms used to energize electrostatic precipitators.
Previous research has demonstrated that IE can reduce transformer-rectifier
power consumption, and under back corona conditions, improve precipitator
collection efficiency.
The objectives of the intermittent energization test program were to:
1. Determine the power savings possible using IE on a hotside
precipitator without negatively impacting precipitator performance.
2. Determine if IE could significantly improve hotside precipitator
performance under back corona conditions.
The test program for the evaluation was divided into three phases. Phase one of
the test program was an evaluation of the first generation of IE firmware
supplied by Environecs. During this phase the IE firmware was loaded into the
microprocessor voltage controllers, and varying IE ratios were tested to
determine their effect on precipitator performance. This test phase was
terminated due to problems with the IE firmware which resulted in numerous
silicon controlled rectifier (SCR) failures. Phase two was an evaluation of the
second generation of IE firmware developed by Environecs which included scaling
factors and other modifications to eliminate the SCR failure problem. The major
portion of the test program described in this report was conducted with the
second generation IE firmware. Precipitator optimization, performance
evaluations with and without sodium conditioning, emissions testing, and power
measurements were documented with the phase two firmware. Phase three was an
5-2
-------
evaluation of a third generation of IE firmware developed by Environecs that
eliminated the scaling factors identified as a possible performance limiting
factor in phase two.
Plant Daniel is a two unit coal-fired electric generating station located near
Escatawpa, Mississippi. Both units are 500 MW systems utilizing Westinghouse
turbine/generators and combustion engineering steam generators. These boilers
are controlled circulation drum units arranged for tangential pulverized coal,
burner tilt firing, and balanced draft operation. The boiler design operation
is 2,400 psi with 1,000 F/1,000 F superheat and reheat. Approximately 200 tons
of coal per hour is burned at full load with a hourly heat input between 5,200
and 5,500 x 105 Btu. Exit gas flow is 2,400,000 ACFM. The units are equipped
with piggy-back Joy-Western hotside electrostatic precipitators with a design
SCA (specific collection area sq. ft./lOO ACF) of 325 and a design collection
efficiency of 99.25%. The weighted wire designed precipitators have 9 inch
plate-to-plate spacing, and each plate is 35 feet high. Each precipitator box
is divided into four chambers with six electrical fields in the direction of gas
flow. The twenty four 1,800 ma transformer-rectifiers (TR) energize two
chambers in each field.
Plant Daniel unit 1 was selected for the IE investigation because MPC and SCS
have been involved in a program for a number of years to improve the performance
and reliability of the hotside precipitators. The addition of state-of-the-art
microprocessor based automatic voltage and rapper controllers, a diligent
operation and maintenance program, and sodium addition for hotside resistivity
modification have greatly improved the performance record of these units.
BACKGROUND
During normal operation, the high-voltage waveform used to power the
precipitator fields is full-wave rectified direct current (DC). Conventional
automatic voltage controllers (AVC) maintain maximum TR output current and
voltage by continually seeking the point where sparking occurs in the field.
The AVC's adjust TR primary input power, and therefore the precipitator
secondary voltage and current, by proportioning the percent of "on" time during
each half-cycle on the alternating current (AC) line. This process of adjusting
the percent of output during each half-cycle is termed silicon controlled
rectifier (SCR) phase control.
To produce intermittent energization, the voltage controller utilizes the same
SCRs to alternately block and pass integral numbers of half-cycles to the
primary side of the TR. During the time when the IE control allows half-cycles
to pass, normal phase control maintains the peak currents and voltage at
spark-limit. The process of power control by blocking and passing whole
half-cycles is termed half-cycle control. In such a system the duty cycle
established by the integral half-cycle control is used to adjust the
precipitator power input.
5-3
-------
In several other publications on intermittent energization the notation for the
duty cycle was:
Half Cycles ON
IE Ratio = Half Cycles ON + Half Cycles OFF
However, for this program and paper, the notation for duty cycle is:
Full Cycles ON
IE Ratio = Full Cycles OFF
Thus an IE Ratio of 1/3 for this program would correspond to an IE Ratio
of 2/8 in some of the earlier literature on work performed by EPRI.
The advantages claimed for IE include reduction in power consumption and also an
improvement in precipitator performance for high resistivity ashes that produce
stable back corona. The improvement in performance has been attributed to the
elimination of back corona. IE eliminates or reduces back corona by reducing
the precipitator field strengths to less than the amount necessary to produce
the voltage drop across the high resistivity dust layer. The reduction in field
strength is achieved by energizing the ESP for one full cycle followed by
several cycles off. The Environecs P370 voltage controllers were set up to
energize two half-cycles and the number of de-energized half-cycles were always
an even number. The even number of de-energized half-cycles prevented
polarization and potential damage to the TR set.
PHASE ONE TEST PROGRAM
During late 1986 and early 1987, SCS R&EA, EPRI, and SRI attempted to
demonstrate IE on Unit 1 at Plant Daniel. These early attempts (Phase 1) were
not successful. A lack of understanding of how IE should be implemented and
rapper control problems interfered. Later in 1987, the IE demonstration program
was attempted again; however, when various test sections of the precipitator
were placed in IE and allowed to run for extended periods of time, numerous
SCR's failed. The SCR's failed because their current ratings were being
exceeded. An investigation by Environecs into the SCR failure problem revealed
that the IE firmware was allowing the TR sets to occasionally turn on at full
conduction angle following a spark. This condition occurred because the
controller was incorrectly sensing the current limit in the IE mode. The test
program was again terminated while Environecs modified the IE firmware to
eliminate the SCR failure problem.
PHASE TWO TEST PROGRAM
The initial testing of IE resulted in blown SCR's as discussed previously;
therefore, a second generation of the firmware was prepared by Environecs to
avoid the problem. This second generation firmware included the scaling factors
shown below.
5-4
-------
Scaling Factors vs. Cycles Off
Cycles Off 0 1 2 3 4 5 6
Scale Factor 1.0 1.5 1.9 2.7 3.7 4.7 6.0
The operating current limit in the IE mode would be Imax divided
by the scale factor for the selected number of cycles off.
The scaling factors were developed by Environecs to limit the maximum current
through the SCR's and TR sets while in the IE mode. The scaling factors were
required because of the method used by Environecs to measure the current limit.
The current limit measuring technique in the first and second generation
firmware was to measure the current limit for some number of half cycles of the
primary AC current and then calculate an average value. This method of
determining current limit will work well in the normal energization mode, but in
the IE mode some number of half cycles were missing from the averaging period.
These missing half cycles caused the average value calculated by the AVC to be
much lower than it actually was. Therefore, in the IE mode, the AVC would
compare the calculated value to the set current limit, determine that the
calculated value was below the setpoint and increase the current to the SCR's.
The normal current limit for the TR sets installed at Plant Daniel is 262 amps.
This value is the maximum recommended by the manufacturer and normally should
not be exceeded. With the second generation firmware the scaling factors were
effectively reducing the current limits to safe values, but the values were well
below the recommended maximums. For instance, with an IE ratio of 1/5, the
current limit allowed by the scaling factor was only 56 amps (262 divided by 4.7
or 56). This lower current would, of course, prevent problems with the SCR's,
but it would not provide sufficient input power in the IE mode to obtain good
ESP performance.
Therefore, a method was sought to allow the IE mode current limit to increase to
a value more representative of the normal 262 amp setting. In conversations
with the AVC manufacturer and plant personnel, it was determined that the
current limit setpoint could be increased in the microprocessor to a maximum
value of 511 amps.
The current limit was changed to 511 amps on one of the AVC's to determine the
benefit and to assess any potential problems with this mode of operation. The
combination of the existing scaling factors and the higher current limit of 511
amps now gave an effective current limit of 109 amps with an IE ratio of 1/5
(511 divided by 4.7 or 109). The operating current limit in the IE mode, was
much improved although still not optimum. The measured peak secondary voltages
were correspondingly increased, and no problems were found with this mode of
operation. Therefore, all of the AVC's current limits for IE operation were
subsequently changed to 511 amps.
IE OPTIMIZATION
An optimization program was then undertaken to determine the best IE ratios for
each precipitator field. Calibrated high voltage dividers were installed in the
1-2 chamber of the lower ESP. Based upon previous experience with this ESP, the
performance of these fields in the lower 1-2 chamber was deemed to be
representative of the performance of the other fields in the remaining three
5-5
-------
chambers. The voltage dividers were connected to a digital oscilloscope in the
precipitator AVC control room. A digital display of the 6-minute average
opacity and a chart recording of the instantaneous opacity were also located in
the AVC control room. With this arrangement, it was possible to monitor the
peak secondary voltage and wave-form of a particular TR set while changes in the
IE ratios were made. In addition, the response of the ESP to changes in the IE
ratios could also be seen immediately.
One important observation from the IE program was that a significant short term
opacity excursion occurred immediately after the initiation of IE operation.
The opacity excursion was due to the electrical conditions in the ESP being
significantly disturbed when IE was implemented. Often several hours were
required for the ESP to reach stable operating conditions following IE
initiation. Once the ESP had regained stable operation, additional changes in
the IE modes did not cause significant upsets or opacity excursions.
With the unit operating at stable conditions, the secondary peak voltages of
each field were monitored as the IE ratios were changed. The secondary voltage
test sequence was begun at the outlet field and progressed to the inlet field so
that testing done in one field would not influence testing done on the next
field. As expected, there was a relationship between the number of cycles off,
the ESP field, and the peak secondary voltages as shown in Table 1.
The data in Table 1 were used to determine the best IE ratio for each field of
the ESP. The underlined values represent the maximum peak voltages obtained and
except for the "E" field, the IE test ratio chosen for that field. The "E"
field was operated in the 1/4 ratio, as no impact on opacity could be observed
between 1/3 and 1/4. The 1/4 ratio increased the amount of power saved, and as
there was only a 0.2 kilovolt difference in the peak voltage, the 1/4 ratio was
used. The outlet field of the ESP was left in the normal mode of operation
because large rapping spikes occurred when the outlet fields were placed in the
IE mode. These rapping spikes would have increased the six minute average
opacity beyond acceptable limits.
There was a consistent two percent increase in opacity when the ESP was placed
in the IE mode. However, the emission testing revealed no change in outlet mass
loading. There was probably a change in the particle size distribution leaving
the ESP which would account for the increased opacity. Particle size
distribution testing was not performed during the test program, as previous
pilot scale testing had shown no change in the particle size distribution during
IE operation.
POWER MEASUREMENTS
The pilot scale testing by SRI had identified the difficulty involved in making
accurate voltage and current measurements of the TR set primary. The problem
was due to the phase angle control characteristics of a SCR controller, which
caused a distortion in the normally sinusoidal AC wave-form that made it hard to
measure accurately. Typical power line monitoring instrumentation requires an
undistorted waveform to make accurate measurements. It was therefore decided to
make the power measurements at a location ahead of the automatic voltage
controllers where the AC waveform was still sinusoidal. The precipitators at
Plant Daniel are powered by four 4160/480 volt transformers Two of the
transformers feed only the upper ESP and two feed only the lower ESP Four
Dranetz Model 808 power line analyzers were connected to these transformers to
-------
Table 1
PLANT DANIEL I.E. PROJECT
I.E. PEAK VOLTAGES
Section 1-2 A 1-2 B 1-2 C 1-2 D 1-2 E 1-2 F
Ratio
1/0
1/1
1/2
1/3
1/4
1/5
1/6
36.6 kv
42.4 kv
44.1 kv
43.2 kv
43.1 kv
41.9 kv
40.2 kv
38.6 kv
42.4 kv
44.8 kv
43.6 kv
41.0 kv
39.0 kv
37.7 kv
35.3 kv
40.3 kv
42.0 kv
43.0 kv
40.8 kv
39.2 kv
37.7 kv
35.7 kv
39.4 kv
40.4 kv
40.5 kv
40.3 kv
39.3 kv
39.2 kv
34.8 kv
36.9 kv
37.6 kv
38.1 kv
37.9 kv
37.8 kv
37.7 kv
33.6 kv
35.6 kv
36.8 kv
36.8 kv
36.3 kv
35.4 kv
34.7 kv
-------
monitor the input power to the entire ESP during the test program. The power
line analyzers provided a measurement every five minutes of the power used by
the precipitators. The analyzers operated continuously during the test program
providing accurate measurements of the power input to the ESP during the various
modes of operation. With the ESP in the normal mode and the boiler at full
load, the precipitator consumed 880 KW of electrical power. Implementation of
IE with the optimized ratios reduced power consumption to 596 KW, a savings of
284 KW or 32.3%, with no loss in ESP performance.
BACK CORONA TEST
One of the principle objectives of the IE test program was to document the
performance of IE in alleviating performance problems associated with back
corona in hotside precipitators. As mentioned previously, the earlier work at
Arapahoe had shown a significant improvement in precipitator performance for IE
under back corona conditions. The Plant Daniel tests were designed to see if IE
could improve the performance of the hotside precipitators enough under degraded
back corona conditions to eliminate the need for sodium conditioning. Following
the IE optimization phase, Southern Research Institute and Sanders Engineering
conducted several tests to measure the outlet emissions of the unit with both
normal and intermittent energization. These tests were conducted while the
fuel was being sodium treated. SCS R&EA monitored unit performance and measured
the precipitator input power. The results of the tests were to be compared to
the results of a more comprehensive test program that was to be conducted with
the precipitators in a degraded, back corona condition.
The outlet emissions tests were completed on September 1, 1988 and the sodium
conditioning of the unit 1 precipitator was halted on September 4. Over the
next several weeks the unit opacity and precipitator power consumption were
monitored in order to document the degradation in performance. By the end of
September, the unit performance had degraded, at low load and during load
changes, to the point that the operations staff believed that the unit would
soon be out of compliance. Even though the precipitator performance was
degraded at low load and during load rise, the performance at stable full load
operation was still acceptable; however, because of the plant's concern for
operating out of compliance, it was decided to test intermittent energization to
see if the unit performance could be improved.
The precipitator was placed in IE on September 26, and the performance was
monitored for the next several days. Intermittent energization did not help the
precipitator performance at low load or during load rise in this partially
degraded state. Based on these observations, the decision was made to resume
sodium treatment and concentrate the program on documenting the power savings
and performance of intermittent energization with sodium treated fuel. Sodium
treatment was resumed on the evening of September 28, and the emissions testing
began on October 3.
RESULTS AND DISCUSSION
During the emissions testing program, coal samples and hopper flyash samples
were collected for each test. The coal samples were blended together, and a
daily composite sample was then analyzed. The hopper flyash samples were
collected from a front, middle, and outlet hopper and then blended into a
composite sample for each test condition. The composite flyash sample was then
-------
analyzed for each day of testing. The coal analyses are shown in Table 2. The
flyash analyses are shown in Table 3. The data show very consistent coal and
ash properties for the test period with the only possible exception being some
variability in the ash content.
The emissions testing program was conducted from October 3-6. Two days of
testing were conducted under the both IE and normal energization modes. The
Unit 1 boiler and turbine conditions were held as constant as possible at full
load for all testing periods. Southern Research Institute and Sanders
Engineering were responsible for the inlet and outlet mass testing using EPA
Reference Method 17. The precipitator efficiency results are presented in
Table 4. Even though the precipitator collection efficiency was higher and the
outlet mass loadings were lower for the IE tests, the outlet opacity for the
unit was approximately 2% higher for IE than for normal energization. It is
believed that there was a change in the size dependent collection efficiency
of the ESP when in the IE mode of operation that accounted for the slight
opacity increase; however, no particle size data were collected at either the
inlet or outlet test locations. It should be noted that the performance of the
precipitator in both cases was well within both state and federal compliance
standards.
The most dramatic result of the test program was the amount of station service
that was saved by using intermittent energization. The Dranetz power meter data
were collected for each test, and the results were then averaged for each series
of tests. The average full load station service for the Unit 1 precipitator in
the normal energization mode of operation was 880 KW. The average full load
station service for the Unit 1 precipitator in the intermittent energization
mode of operation was 596 KW. This is a 284 KW or 32% reduction in station
service for the Unit 1 precipitator. The dollar savings that this reduction in
station service represents depends on a number of factors such as unit load
factor and cost of replacement generation; however, the electrical power
consumed by the precipitator T-R sets represents a significant portion of the
total cost of precipitator operation.
PHASE THREE TEST PROGRAM
Following the completion of the emissions test program, it was obvious that a
better way of implementing IE in the Environecs P370 AVC was needed. The
technique of changing the current limit to 511 amps for the IE mode of operation
was cumbersome, and presented a potential problem if the current limit was not
changed back to the TR set value of 262 amps for normal mode operation.
Discussions with the manufacturer's representative and others at Environecs were
initiated to with the idea of improving the P370 firmware. Environecs
immediately began to investigate the possibility of rewriting the firmware, and
within a short time supplied a third generation of the firmware for evaluation.
Initially the third generation firmware was loaded in one AVC for testing. The
new firmware was found to operate as desired, and STI Environecs was asked to
provide copies for the remaining AVC's. The additional copies of the new
firmware were received and installed, and an evaluation program began in the
summer of 1989.
It was anticipated that the increased peak voltage, for the longer off times,
would allow the ESP to operate with off times longer than those used during the
test phase when the emissions tests were performed. The use of longer off times
was expected to result in additional power savings. However, ESP performance
5-9
-------
Table 2
CHEMICAL COMPOSITION OF COAL SAMPLES
WT %, As Received
Moisture
Carbon
Hydrogen
Nitrogen
Chlorine
Sulfur
Ash
Oxygen
Volatile
Rxed Carbon
Btu/lb
10/03/88
7.71
72.76
4.47
1.51
0.13
0.75
6.81
5.86
34.12
51.36
12,822
10/04/88
7.46
72.01
4.40
1.51
0.12
0.70
7.40
6.40
33.85
51.29
12,803
10/05/88
7.65
71.00
4.32
1.50
0.12
0.71
8.63
6.07
33.39
50.33
12,544
10/06/88
6.79
72.79
4.50
1.56
0.12
0.75
7.42
6.07
34.30
51.49
12,862
Table 3
CHEMICAL COMPOSITION OF ESP HOPPER ASH SAMPLES
Lithium Oxide
Sodium Oxide
Potassium Oxide
Magnesia
Lime
Ferric Oxide
Alumina
Silica
Tltania
Phos. Pentoxide
Sulfur Trioxide
Loss On Ignition
Upper ESP
Normal
0.05
1.40
1.80
0.89
1.30
5.60
30.50
55.50
1.20
0.15
0.09
7.30
Lower ESP
Normal
0.04
1.40
1.80
0.87
1.30
6.10
29.70
56.10
1.20
0.18
0.10
9.90
Upper ESP
IE Test
0.05
1.40
1.70
0.87
1.30
5.60
30.30
56.00
1.20
0.17
0.08
5.60
Lower ESP
IE Test
0.04
1.30
1.70
0.85
1.20
5.60
30.20
56.80
1.20
0.16
0.09
6.80
5-10
-------
Table 4
PLANT DANIEL I.E. PROJECT
Inlet Mass Outlet Mass ESP Efficiency
Normal Operation 2.88 gr/SCF 0.0197 gr/SCF 99.31
Intermittent Energization 2.96 gr/SCF 0.0170 gr/SCF 99.42
°7
/o
/o
-------
began to deteriorate quickly when longer off times were selected. The original
IE ratios of 1/2, 1/2, 1/3, 1/3, 1/4 and 1/0 (A-F fields) remained the best
choice for precipitator performance and power savings. The third generation
firmware did provide slightly higher peak voltages for the 1/3 and 1/4 ratios,
but significant improvements in ESP opacity were not observed. Therefore, no
additional emissions testing was performed. The third generation firmware
eliminated the scaling factors, that have been discussed previously, and also
improved the user interface for implementing the IE mode from the AVC keyboard.
CONCLUSIONS
The following conclusions can be drawn from the IE test program:
• Intermittent Energization reduced the power input to the ESP
by 30-35 %.
• Outlet mass emissions from the precipitator and collection
efficiency were equivalent for IE compared to normal
operation.
• Opacity was approximately 2% higher in IE than in normal
energization.
e Outlet field rapping spikes were higher in IE mode than in
normal energization.
• IE did not improve ESP performance under back corona
conditions.
• The ESP was more sensitive to load rise conditions while in
the IE mode.
• IE works satisfactorily on large hotside precipitators
and will probably work equally well on large coldside
precipitators.
• IE ratios should be chosen carefully by evaluating peak
secondary voltages and not by simple guesswork.
• Rapping system programs may need to be modified in order to
work effectively with IE.
• Significant reductions in precipitator station service loads
are possible with IE, without increasing outlet emissions.
The Electric Power Research Institute and Southern Company Services, Inc.
believe that the demonstration program conducted at Plant Daniel proves that
Intermittent Energization is a commercial technology that is ready for
implementation in the electric utility industry. IE is not a technology that
can be retrofitted to all precipitators, but within certain design constraints,
it is a technology that may enhance precipitator performance and will save
significant amounts of station service.
5-12
-------
REFERENCES
1. Electrostatic Precipitator Power Measurements, Peter Gelfand, E. C.
Landham, Jr., Louis F. Rettenmaier, February 1986, Sixth EPA/EPRI
Symposium on the Transfer and Utilization of Particulate Control
Technology, New Orleans, LA.
2. Pilot-Scale Evaluation of ESP Intermittent Energization, E. C.
Landham, Jr., J. L. DuBard, Walter Piulle, Louis F. Rettenmaier,
February 1986, Sixth EPA/EPRI Symposium on the Transfer and
Utilization of Particulate Control Technology, New Orleans, LA.
3. Evaluation of Intermittent Energization On Mississippi Power Company
Plant Daniel Unit 1, E. C. Landham, Southern Research Institute,
March 1989.
5-13
-------
EXPERIMENTAL EVALUATION OF IMPROVED DESIGN OF ESPs
B.Bellagamba, G.Dinelli, E.Riboldi
ENEL-Thermal and Nuclear Research Center
Via A.Pisano, 120
56100 Pisa-Italy
ABSTRACT
The influence of the electrodic structure (emitting electrode
shape and collection plate spacing) on the collection efficiency
of an electrostatic precipitator was investigated with a pilot
precipitator at a nominal gas flow rate of 12-000 Nm3/h. The
pilot ESP was specifically designed by Enel and installed
slipstream of the flue gas duct of a 35 MVe coal boiler and it
was equipped with conventional and narrow pulse power supplies.
The following ESP configurations were tested: collection plate
spacing from 250 mm to 500 mm; emitting electrodes made with
round wires, serrated strips and helical wires.
In total 12 test campaigns were performed each consisting of 8
efficiency measurements according to ASME standard. The total
operation of the pilot ESP, including the start-up warming, was
over 3-000 hours.
The experimental results have evidenced a significant advantage
of the following configuration among the others that were tested:
collection plate spacing equal to 400 mm, emitting electrode made
with round wires having 5 mm diameter. In accordance with such
results it has been decided to further prove the optimum
configuration on an industrial ESP at a nominal gas flow of
175-000 Nm3/h and to retrofit with the narrow pulse energization
the ESP of a 240 MWe coal unit.
6-1
-------
INTRODUCTION
Although the electrostatic precipitation process has been applied
for years for removing suspended particulate present in the flue
gases from industrial processes and a lot of study has been
carried out in order to optimize the collection efficiency, there
are still several design and operation factors, that need to be
investigated, that may influence the actual performance of an
electrostatic precipitator especially when the solid particulates
are originated from the combustion of coals of different sources.
Within the scope of a research programme aimed at improving the
performance of electrostatic precipitators for thermal power
plants, a pilot unit of 12'000 Nm /h has been designed by ENEL so
that it can be easily displaced and put in slipstream of the gas
ducts of thermoelectric units.
The pilot electrostatic precipitator offers the possibility of
carrying out experiments with real flue gases without influencing
the normal operation of the thermoelectric unit.
The main objectives of the experiments performed with the pilot
ESP are:
1. to asses the best electrode geometry under conventional
and impulse power energization;
2. to verify the most effective rapping system and operation
as a function of the fly ash characteristics;
3. to investigate the efficiency of gas conditioning
systems.
Previous experimental results have been reported concerning
either the wide spacing and electrode geometry of ESP under
conventional energization [1], [2], [3] or the application of the
impulse energization to an industrial ESP with standard 300 mm
spacing of the collection plates [4].
The tests carried out with collection plates spacing set at 250,
300, 400 and 500 mm and with different types of emitting
electrodes (round wires with a diameter of 5 mm and 8 mm of
serrated strips and helical electrodes) either with conventional
6-2
-------
or impulse power energization are reported.
In fact during preliminary tests carried out with different plate
spacings it was noted that the electrodes geometry has a very
important influence on the collection efficiency. Therefore a
correct electrodes geometry was designed for each spacing to
ensure the highest performance of the pilot ESP both with
conventional and impulse power supply
The impulse power energization system consists in superimposing a
short pulse voltage, of about 70 fisec, on a d.c. base voltage.
The frequency of the impulse voltage may be varied from 20 to
300 Hz. Basically it can be assumed that the impulse peak voltage
provide the electric charge to the particles while the d.c. base
voltage, wich is maintained slightly higher than the corona onset
voltage, guarantees the electric field necessary for the particle
collection.
TEST FACILITY
PILOT ESP
The pilot ESP, shown in Fig.l, has two electric fields in
series, each one is 4980 mm long and 3670 mm high. The collection
plate spacing can be set at: 250-300 mm (4 parallel channels),
400 mm (3 parallel channels) and 500-600 mm (2 parallel
channels) The design and nominal operating parametres are
summarized in Table 1.
The pilot ESP has been placed slipstream of the gas duct of a 35
MWe boiler at Marghera (Venice) coal power station. For each
electric field of the precipitator the collection plates are
subdivided in three mechanically indipendent sections that are
rapped from the top by means of pneumatically driven hammers.
Each sectionalized collection plate of an inner channel is
weighed by means of loading cells at which the plates are
suspended. A microprocessor takes care of the rapping system
giving the possibility of assigning the desired rapping sequence
and frequency.
The fly ash, coming for each electric field from five hoppers,
is collected during the test in separate containers and weighted
6-3
-------
by means of loading cells (two for each container).
The fly ash at the outlet of the pilot ESP is collected in a
fabric filter and weighted separately.
The amounts of the ash collected in the pilot ESP and in the
fabric filter are elaborated by a microprocessor to control the
correct withdrawal of the fly ash..
The gas flow rate is regulated from 12'000 to 22'000 m3/h by
means of a fan placed at the outlet of the pilot ESP while the
depression in the pilot is mantained at about -50 mm H20 by means
of a fan placed at the inlet.
POWER SUPPLIES
The pilot ESP is equipped with conventional and impulse power
supplies
The narrow impulse power supply sets consist of:
- a h.v. generation system for the d.c. base voltage;
- a h.v. generation system for the impulse voltage, obtained
directly at high voltage without using a step-up
transformer;
- an automatic control for voltage and current levels
delivered to the electric fields of the ESP.
The main features of the conventional and the impulse power
supplies are shown in Table 2.
TESTING PROCEDURES
The tests were carried out with the thermoelectric plant running
conditions as steady as possible; a South Africa (AMCOAL) coal
was used during all tests concerning 300, 400 and 500 mm spacing;
South Africa (SHELL) coal was used for the 250 mm spacing tests
as well as the 300 mm spacing tests with different types of
emitting electrodes.
Each test campaign was preceded by a pre-heating of the pilot
ESP, to reach the nominal thermal operating conditions, by means
of an indirect hot air generator. Before starting with the tests
the pilot ESP was kept running for at least 90 hours.
During each test campaign the pilot remained in continues
operation. The isokinetic sampling of the gas at the inlet and
6-4
-------
outlet of the pilot was carried out during a complete rapping
cycle (144 minutes)
For the 300 mm collection plate spacing (4 parallel channels),
the rapping system (30 hammers) was programmed as follows (see
figure below):
- every 16 minutes for collection plates 1A;
- every 24 minutes for collection plates IB;
- every 36 minutes for collection plates 1C;
- every 48 minutes for collection plates 2A;
- every 72 minutes for collection plates 2B;
- every 144 minutes for collection plates 2C.
FIELD 1 FIELD 2
! o 0 0 | 0 0 0
2 0 0 0 | 0 0 0
3 0 0 0 | 0 0 0
4 o 0 0 | 0 0 0
5 0 0 0 | 0 0 0
1A IB 1C 2A 2B 2C
collecting plates 0 hammers
Before each testing cycle the power supplies were finely tuned in
order to obtain at the outlet of the pilot ESP the lowest
opacimetric signal.During this stage the following events were
noted:
1. with the collection plates spacing set at 300 mm and
conventional power energization a number of sparks per
minute higher than 10 and lower than 3-4 determined a
drop of the average voltage applied to the emitting
electrodes and consequently increased the outlet
opacimetric signal;
2. with collection spacing of 400 and 500 mm and
conventional energization the number of sparks per minute
had less influence on the outlet opacity of the flue gas;
the best results were anyhow obtained with a sparking
rate ranging from 3 to 5 per minute;
6-5
-------
3. with the impulse power energization the best performance
was obtained by setting the d.c. base voltage at values
close to the corona onset voltage ensuring about 3 sparks
per minute.
During the tests measurements and analysis of the following
quantities were made:
- gas velocity, temperature and pressure at the inlet and
outlet of pilot ESP;
- gas flow rate;
- oxigen content in the gas at the inlet and outlet;
- opacity of the gas at the inlet and outlet;
- in situ ash resistivity;
- isokinetic particle sampling at the inlet and outlet;
- voltage/current characteristics of each electric field;
- weight of the ash collected with each sub-sections of the
collection plates;
- amount of the ash withdrawal from the ten hoppers;
- chemical and particle size analysis of the ash, at the
inlet and outlet and as collected in the hoppers.
EXPERIMENTAL RESULTS
COLLECTION PLATE WIDE SPACING
The results of the tests, summarized in Table 3, clearly show the
improvement of the ESP performance when changing from
conventional to impulse power energization for all the electrodes
configurations.
The best results have been, anyhow, obtained with a collection
plates distance of 400 mm and round wire emitting electrodes of
5 mm diameter, either when using conventional or impulse power
energizat ion.
It must be noted that with the impulse power supply the best
operating conditions were obtained with an impulse repetition
rate of about 30 Hz for the first electric field and 15 Hz for
the second electric field.
From the voltage/current characteristics, shown in Fig. 2 and
Fig. 3, it can be noted the gradual increase of the onset corona
voltage and the spaikover voltage with the increase in distance
6-6
-------
between the collection plates from 250 mm to 500 mm; it can also
be noted that backcorona occurs when conventional power supply is
used but never occurs when the impulse power supply is used.
EMITTING ELECTRODES SHAPE
Significant differences in collection efficiency were obtained by
using different types of emitting electrodes, mainly with impulse
power supply. Considering the results reported in Table 3 and the
voltage/current characteristics of Fig. 4 and Fig. 5 it is
evident that the best performance is obtained with the 5 mm round
emitting electrodes. A round emitting electrode of 8 mm diameter
has also been tested with the collection plates spaced 400 mm.
The results show a certain identity of values with the 5 mm
electrode when using impulse power supply but a worse behaviour
when using conventional power supply.
SPECIFIC PERFORMANCE PARAMETERS
The fly ash penetration (I-"*?) as a function of the particle
diameter is shown in Fig. 6 and Fig. 7. It is possible to notice
that the best conditions are obtained under the impulse power
energization and with the 400 mm collection plates spacing
The amount of fly ash collected in the ten hoppers of the pilot
ESP is plotted in Fig. 8 and Fig. 9. From these figures it is
possible to estimate the efficiency of the first field for
different operating conditions; it results:
- at 300 mm spacing, an efficiency of 83.9% with
conventional power supply and 90.5% with impulse power
supply;
- at 400 mm an efficiency of 85.6% with conventional power
suppply and 93.3 with impulse power supply.
CONCLUSIONS
The main design parameters influencing the ESP electrical
energization have been investigated at a pilot unit equipped with
both conventional and narrow pulse power supplies.
By changing the electrode configuration, i.e.: the emitting
6-7
-------
electrode shape and the spacing between the collection plates, it
has been possible to define those design conditions that optimize
the ESP collection efficiency
It has been experimentally verified that the round wire emitting
electrodes, with outer diameter enlarged to 5 mm, by providing a
higher onset corona voltage, ensure a more effective charging and
collection of the dust. Such benefit is enphasized by increasing
the spacing between the collection plates up to 400 mm.
The narrow pulse energization, as compared with the conventional
one, always improves the collection efficiency since it allows
much higher peak voltages for a more effective particle charging
and a strongly reduced and more uniform current density at the
collection plates thus avoiding the incurrence of the back-
corona. This event has been verified with fly ash of different
chemical composition and physical properties with the electrical
resistivity varying from 10 to 10 Ohm-cm.
It is belived that the improved ESP design may become very
promising for those plant retrofitting projects that need the
upgrading of the already installed precipitators as well as for
those new plants designed to burn coals of different sources
generating fly ash of varying captability since the narrow pulse
energization, in particular when combined with the enlarged plate
spacing, ensures a high flexibility in the optimal operation of
plant.
Following the very promising results of the pilot ESP
experimental programme two retrofit projects are at present under
development to upgrade with the narrow pulse energization the
performance of the electrostatic precipitators of two coal units;
in addition the mechanical structure of a precipitator will also
be modified to enlarge the collection plate spacing from 300 mm
to 400 mm.
ACKNOWLEDGEMENTS
The authors acknowledge with thanks the personnel of Marghera
power station for the active a^uistance and collaboration
supplied during the tests.
6-8
-------
REFERENCES
1. K.Darby, "Plate Spacing Effect on Precipitation
Performance" , Proc . of the Second Int. Conf . on
Electrostatic Precipitation, November 1984, Kyoto (Japan).
2. R.F.Altman, et. al., "Pilot - Scale Evaluation of ESP Wide
Plate Spacing", Proc. of_ the Sixth Symp. ori the Transfer
and Uti 1 ization o_f Particulate Control Technology, November
1986, New Orleans (USA)
3. L.A.Hawkins, et . al., "Characterization of Discharge
Electrode Performance: Results of Laboratory and Pilot
Plant Experiments", Proc. of the Sixth Symp. on the
Transfer and Uti1ization o£ Particulate Control Technology,
November 1986, New Orleans (USA).
4. G.Dinelli, et al . , "Enhanced Precipitation Efficiency of
Electrostatic Precipitators by Means of Impulse
Energization", IEEE-IAS Annual Meeting, October 1988,
Pittsburg, (USA).
Table 1
DESIGN DATA OF THE PILOT ESP
Collection plates height ( mm )
Collection plates length ( mm )
Number of electric fieldes ( n )
Aspect ratio
Nominal gas velocity ( m/sec )
n
Nominal gas flow ( Nm/h )
Nominal gas temperature (°C )
Total collection surface ( m )
S.C.A. ( at nominal flow rate ) ( m^rn/sec )
Number of gas passages ( n ° )
Collection plates spacing ( mm )
250
300
400
500
3670
4980
2
2.7
1.12
12000
145
290
56
4
290
56
4
218
43
3
145
33
2
6-9
-------
Table 2
Pilot ESP - Elettrical power supplies design parameters
Conventional power supply
Nominal power
Output voltage ( mean value )
Output voltage ( peak value )
Output current ( peak value )
Narrow pulse power supply
D.C. voltage section
Nominal power
D.C. voltage
Current ( mean value at 30 kV
kVA )
kV
kV
mA
kVA
kV )
mA
1.2
87
150
100
2.5
5-60
30
Impulse section
Nominal power
Impulse amplitude
Impulse width ( 300 mm spacing
Impulse width ( 400 mm spacing
Impulse frequency
Current ( mean value )
kVA )
kV )
( Hz '
( mA
3.2
0-120
75
64
10-500
40
Table 3
Pilot ESP - Experimental results
Collection plates spacing ( mm )
Type of emitting electrodes
Efficiency ( % )
3)
Outlet dust concentration (mg/Nm^
C
I
C
I
S.C.A. ( m2/m3/sec )
(i
250
dia.
3mm
94.3
95.8
731
533
64
(2
300
dia.
5mm
96.1
98.9
510
140
56
(2
400
dia.
5mm
96.6
99.2
440
105
43
(2
400
dia.
8mm
94.8
99.17
676
108
43
(2
500
dia.
5mm
94.6
98.4
700
210
33
(i
300
dia.
5mm
98
99.3
263
89
56
0
300
stmltd
strips
97.9
98.9
248
146
56
(i
300
helical
lire
97.7
99
307
128
56
C = Conventional power supply 1 = Impulse power supply 1) = S.A. coal ( Shell )
2) = S.A. coal ( Amcoal ) 3) = At an inlet dust concentration of 13000 mg/Nm
6-10
-------
300mm 400mm 500mm 600mm
spacing spacing spacing spacing
1680
1- Emitting elettrodes rapper
2- Collection elettrodes
3— H.V. impulse power supply
4- H.V. D.C. power supply
5- Load cell
6- Weighed hopper
7- Distribution plates rapper
8- Collection electrodes rapper
9- Switching box
Fig. 1 - Pilot ESP mechanical design
-------
24
20
16
12
8
Pilot ESP
1)- plates spacing
2\- plates spacing
3)- plates spacing
4)- plates spacing
250 mm
300 mm
400 mm
500 mm
10 20 30 40 50 60 70 80 90 100 110 kV
Fig. 2 - Voltage-current density characteristics win conventional power supply
24
« 20.
e
o
•§16.
12
8.
4.
Pilot ESP
1) - plates spacing = 250 mm ( base voltage
2) - plates spacing = 300 mm ( base voltage
400 mm ( base voltage
500 mm ( base voltage
1 - ~ IT — O
- plates spacing
- plates spacing
27 kV )
33 kV
42 kV )
50 kV
10 20 30 40 50 60 70 80 90 100 110 kV
Fig. 3 - Peak voltage - current density characteristics with impulse power supply
6-12
-------
24
M 20J
6
o
"S 16
12.
8-
4.
Pilot ESP
( plates spacing = 300 mm )
1) - round wires electrodes ( 5 mm Dia. )
2) - helical electrodes ( 2.5 mm Dia.)
3) - serrated strip electrodes ( 1= 10 mm )
10 20 30 40 50 60 70 80 90 100 110 kV
Fig. 4 - Voltage - current density characteristics with conventional power supply
e
o
20
•^16
12]
Pilot ESP
( plates spacing = 300 mm )
1) - round wires electrodes ( 5 mm Dia.)
2) - helical electrodes ( 2.5 mm Dia.)
3) - serrateistrip electrodes ( 1 = 10 mm)
Vh= 35 kV
7^= 30 kV
Vb= 28 kV
10 20 30 40 50 60 70 80 90
Fig. 5 - Peak voltage - current density with impulse power supply
100 110 kV
6-13
-------
99
95
50
10
5
0.02
Pilot ESP
( plates spacing-300 mm )
D Conventional power supply
B Impulse power supply
ill
1.4 2.5 4.6 8.3 15 27
Fig. 6 - Flay ash particles penetration versus particles diameter
m
99
^ 95
-I 90
50
10
5
0.02
1
1
Pilot ESP
( plates spacing-400 mm )
D Conventional power supply
H Impulse power supply
14 2.5 4.6 8.3 15 27
Fig. 7 - Flay ash particles penetration versus particles diameter
m
6-14
-------
a eo
^50
40
30
20
10
Pilot ESP
( plates spaging-300 mm ]
Conventional power supply
Impulse power supply
Hi Cta n. IZL-_
1 2 3 4 5 6 7 8 9 10 Hoppers
Fig. 8 - Flay ash collected per unit of collection surface overlaying each hopper
50 -
40 -
30 •
20
10
Pilot ESP
( plates spacing-400 mm ]
D Conventional power supply
H Impulse power supply
123 45 8 7 8 9 10 Hoppers
Fig. 9 - Flay ash collected per unit of collection surface overlaying each hopper
6-15
-------
DELAYING SODIUM DEPLETION IN
ELECTROSTATIC PRECIPITATORS AT GHENT GENERATING STATION
CARLA CLAY ROBINSON
KENTUCKY UTILITIES COMPANY
P.O. BOX 338
GHENT, KY 41045
ABSTRACT
The Ghent Station Unit 2 burns Eastern Kentucky low-sulfur (0.6
to 0.7%) bituminous coal that routinely impairs electrostatic
precipitator performance. The degradation of ESP performance is
believed to be caused by a sodium depletion phenomenon. An
intermittent energization sequence was implemented, via the
automatic voltage controls, to determine if sodium depletion
symptoms could be delayed.
Ghent Unit 2 has a Combustion Engineering tangentially fired
boiler, a 511 MW General Electric turbine/generator and is
equipped with a hot-side, 228 SCA Buell precipitator. During
normal operation, precipitator inlet temperatures range from 420
degrees F at low loads, to 620 degrees F at maximum load. Miner-
al analyses of flyash from coal suppliers typically indicate
sodium levels of 0.34% to 0.51% of total weight. The flyash
composition combined with the low operating temperatures results
in flyash resistivities of 3 Ell to 8 E12.
For the past five years, within 30 days after a startup, Unit 2
has required removal from service due to high opacity. Lengthen-
ing the time period before the precipitator becomes sodium de-
pleted would help lower sodium costs and lower the costs due to
unnecessary startups and shutdowns.
The results of using intermittent energization to delay sodium
depletion are that the unit has been able to remain on line for
104 days before opacity reached an average level of 17%. Sodium
conditioning was then used successfully to decrease the average
baseline opacity from 17% to 8%.
7-1
-------
DELAYING SODIUM DEPLETION IN
ELECTROSTATIC PRECIPITATORS AT GHENT GENERATING STATION
INTRODUCTION
About 5 years ago, Unit 2 coal supply began originating from the
coal spot market. With the addition of different vendors of
bituminous coal, precipitator performance suffered greatly. This
has impacted and sometimes dictated the operation of Unit 2.
Within 30 days after a startup, Unit 2 has required removal from
service due to high opacity. Repairs made to the electrostatic
precipitator (ESP) to improve performance have had only a minor
impact.
Symptoms displayed by the ESP, including the 30 day degradation
of voltage and current readings, and reduced collection efficien-
cy during low load and low temperatures, led to the sodium deple-
tion phenomenon being suspected as the cause of most of the
performance problems. (At low load, and low temperatures, col-
lection efficiency was poor-this sometimes led to the unit being
operated at higher loads to keep opacity lower.) Research indi-
cated that Ghent 2 coal flyash could cause sodium depletion under
the unit's operating conditions.
Repairs were made to the precipitator in an attempt to disprove
this suspicion. It was felt that the precipitator should be in
the best mechanical condition possible in order to rule out any
physical defects that could be causing symptoms of the sodium
depletion phenomenon. The turning vanes in 2 of the 4 boxes were
modified to help straighten the gas flow. An air blowing system
was installed in the inlet plenums to keep the floors clean of
ash and prevent casing cracks. A new rapping system was in-
stalled to maximize plate rapping performance. Automatic voltage
controls were installed to replace the original controls which
had become outdated and required excessive maintenance. The
inside of the precipitator was scrutinized during outage inspec-
tions with the hopes of finding something that could be causing
performance degradation. After these repairs did not lead to
optimum performance, the decision was made to try sodium condi-
tioning.
7-2
-------
SODIUM DEPLETION PHENOMENON
Resistivity of flyash has long been a design factor of ESPs. The
two main determining elements of resistivities in hot-side ESPs
are temperature and chemical composition of the flyash. Sodium
depletion occurs when the negative charge on the ESP wires pulls
the positively charged sodium ions out of the plates' inherent
ash layer. The "depletion" results in an ash layer of higher
resistivity to an electrical charge. A low voltage arcing in the
high resistivity areas causes the voltage controls to lower their
useful current and voltages, thereby degrading the performance.
Sodium is one of the few elements in the flyash that hold the ash
to the collecting plates. If the ash will not stick to the
plates, particulate losses are high.
This ESP was designed in the early 1970's, when sodium depletion
was not a known factor in flyash resistivity. The chemical
composition of the flyash of the coal proposed to be burned in
the unit was used to calculate an approximate specific collection
area requirement (SCA is collection plate area versus gas flow)
for the ESP. Since the sodium depletion phenomenon had not been
discovered, the design engineers did not consider its affect.
This resulted in the slow degradation of performance. Unit 2 was
designed with 228 square feet/1000 acfm. The sodium depletion is
exacerbated by the small size of the ESP.
Factors influencing the sodium depletion of Ghent Unit Two are
the low operating temperature of Unit 2 combined with the low-
sulfur coal that is used. Unit 2 operating temperatures range
from 420 degrees F at low loads to 620 degrees F at full load.
The low temperature at low loads has previously led to higher
opacity. Many times the unit was operated at maximum load to
keep the ESP temperature up and the opacity down.
PREDICTED ASH RESISTIVITIES FOR GHENT 2 COAL SUPPLIES
There are some known factors surrounding coal flyash analysis
that affect resistivity in a hot-side ESP. In Ghent Two's case,
there is a known combination of chemicals (1J that produces a
highly resistive flyash:
• A high percentage of alumina-silicates (over 80%)
• A low iron content (under 10%)
• A low sodium content (under 0.5% of the total weight)
7-3
-------
Using a computer program written by Dr. Roy Bickelhaupt, of Clean
Air Laboratory and recommended by Dr. Ralph Altman of the Elec-
tric Power Research Institute (EPRI), flyash resistivities can be
predicted for any flyash sample using a mineral analysis. Resis-
tivity predictions for Unit 2 are as high as 8 E12 ohm cm, when 1
E10 ohm cm would be the most desirable. See Figure #1. This
figure also predicts what ash resistivities would be if sodium
contents of that same coal were 1.34, 1.84 and 2.34%.
Dr. Ralph Altman of EPRI was contacted to assist in determining
possible solutions to the problem of higher resistivity coal
flyash. Dr. Altman explained, in a meeting held at Ghent Station
May 9, 1989, that sodium conditioning was one of the few alterna-
tives to this performance degradation over time. A decision was
reached to fabricate a test conditioning system. Rather than
conditioning the supply coal from the time of startup, it was
decided to begin conditioning when unit opacity averaged 15%.
INTERMITTENT ENERGIZATION SEQUENCE
Research concerning the sodium depletion phenomenon suggested the
precipitator secondary current as a possible cause of the sodium
ions being pulled from the plates. It was thought that using
intermittent energization would reduce the total current density
and lower the secondary voltage. If the current density could be
reduced, the sodium ions would remain on the plates longer and
the symptoms of the phenomenon would be delayed.
The new precipitator automatic voltage controls (Model P37s
manufactured by STI/Environecs) installed in the spring of 1989
had the capability to intermittently energize (IE) the ESP wires.
The concept of IE is that the precipitator acts much like a
capacitor and can hold an electrical charge after power is re-
moved. IE actually reduces the total power by blocking some of
the AC cycles.
An intermittent energization sequence program was initiated when
Unit 2 returned to service on August 27, 1989. The objective was
to attempt to use the IE sequence to delay the depletion degrada-
tion. The sequence initiated on the inlet fields was three
cycles off, two cycles on. The sequence selected for the front
middle fields was 2 cycles off, two cycles on. The back middle
field had one cycle off, two cycles on. The outlet field had
full power on. The 3-2-1-0 sequence is shown in Figure #2. This
program is also saving a small amount of the total power consumed
by the ESP, but at this time, it is not known exactly how much.
7-4
-------
SODIUM CONDITIONING TEST
By the August 27, 1989 startup date, it had been determined that
when the opacity increased to an average of 15%, sodium sulfate
would be added to the coal supply. To accomplish this, a small
flexible screw conveyor was purchased and installed in the crush-
er house.
With the intermittent energization sequence having been initiated
on the voltage controls, the initial 30 day period ended with the
Unit 2 opacity at 6%. Thirty more days of operating time ended
and the average opacity was 8%. After 90 days of operating time
the opacity was 11%. However, the decline in performance came
rapidly after the 90 day period. By day 104, the average unit
opacity was 15 to 17%. The ESP started displaying the typical
sodium depleted indications such as: the degradation of the
useful secondary current and voltages due to low voltage arcing,
and the higher opacities during low load operation.
By this time, it was thought that if the IE sequence could delay
the sodium depletion and if the plates were replenished with
sodium, conditioning could cease and the IE would help hold the
sodium on the plates. It was also theorized that the unit would
remain in compliance for 3 to 12 weeks.
Sodium conditioning began on December 6, 1989 at 1045. By 19:00
hours on 12-6-89, the opacity began to decrease. Average opacity
remained at 8 to 9% from 12-6-89 to 12-13-89. On 12-13-89,
opacity raised to about 11%. At that time, a "test" coal (15%
ash content) was introduced to the boiler, the opacity rose about
3-5%. Conditioning ceased at 0900 on 12-16-89. At that time, the
decision was made to let the unit run without sodium conditioning
to see if the opacity would remain low.
During the first 5 days of sodium conditioning, an average of 3.2
pounds of sodium per ton of coal was added. During the next five
days of conditioning, some problems arose with the feed rate of
the coal. We conditioned nearly 33,000 tons of coal and used 48
tons of sodium, for an average addition rate of about 2.9 pounds
per ton of coal.
From a previous mineral analysis of the flyash in the ESP, the
calculated desired rate of addition was 3.2 pounds of sodium per
ton of coal. This would have resulted in a 1.2% sodium content
7-5
-------
in the flyash and resistivity of 1 E10 ohm cm. (very desirable).
Due to errors in the feed rate of both the coal feeder and the
screw conveyor, and coal handling problems brought on by cold
weather conditions, the desired 3.2 pound sodium average was not
achieved.
During the sodium conditioning test, we used some coal from seven
of our eight low-sulfur coal vendors.
RESULTS OF SODIUM CONDITIONING
Using sodium conditioning, Unit 2 opacity improved within 24
hours after conditioning began. Opacity was 17% and above before
conditioning and dropped to 8 to 11% with conditioning. Condi-
tioning allowed the unit to stay in compliance and to remain on
line. As the test results indicated, an equilibrium point was
reached at the low of 8% opacity. See Figure #3.
Other indications of ESP improvement occurred besides the lower-
ing of opacity- The current density increased. Also increasing
with the secondary current was the primary current and voltage.
These are all indications that the resistivity of the flyash was
lowered.
The mineral analyses of the flyash taken from ESP hoppers during
the conditioning test indicate that the sodium level had in-
creased. A normal sodium level in the flyash would be about
0.41%. The samples taken during conditioning indicate a sodium
level of .9% average. The desired percentage of sodium calculat-
ed was 1.2%.
The sodium cost for 10 days conditioning was $8,272 (purchased in
100 pound bags). This makes the sodium cost of conditioning
33,000 tons of coal $0.25 per ton. Had the calculated rate of
3.2 pounds of sodium per ton of coal been reached, the cost of
conditioning 33,000 tons of coal would have been $0.32 per ton.
Using intermittent energization delayed the effects of the sodium
depletion phenomenon for 74 days beyond our previous typical 30
day period. The use of IE resulted in a calculated savings of
$61,200 in sodium costs (using the sodium costs incurred during
the conditioning test and based on 74 days not conditioning at
$827 per day, assuming that the load had been needed and condi-
tioning was required).
7-6
-------
CONCLUSIONS
The use of intermittent energization greatly enhanced the Unit 2
precipitator performance. Although the IE could not completely
halt the degradation due to the sodium depletion phenomenon, it
did result in a delay of 74 days. This in itself should change
the future operation of Unit 2. With the testing of sodium
conditioning now complete, it is logical to plan for using the
conditioning on an as needed basis.
Lowering the resistivity of the flyash in the precipitator by
using IE and sodium conditioning has changed the characteristics
of the ESP. Past history indicated an ESP that was very sensi-
tive to the low gas temperature experienced during low load
operation of the unit. Before using IE, opacity was high at low
loads and lower during full load operation with the higher pre-
cipitator temperatures. During this experiment, the precipitator
exhibited more "normal" operation wherein the higher loads pro-
duce higher opacities, and the low load operation produces lower
opacities, (a direct relationship with the amount of ash in the
coal). This ash to coal relationship now affects this precipita-
tor the same way that it affects any other non-depleted ESP.
Observations can now be made with a known type of coal being
burned and the percentage of ash in that particular coal to
clearly see the relationship between the ash content and the
opacity. For instance, when the "test" coal was burned, the
average 15% ash content of that coal produced higher opacity than
our typical coals with an average 10% ash content.
With the sodium conditioning having replenished the collecting
plates with sodium, the depletion phenomenon was delayed again
for six weeks. After six weeks Unit 2 ESP started displaying
symptoms of sodium depletion. Further testing is needed to
determine the time periods between depleted conditions.
The objective of using the intermittent energization was to see
if the sodium depletion could be delayed such that Unit 2's
output would be more reliable, and to decrease the costs of
frequent startups and shutdowns. To that end, the experiment was
successful.
Even though the level of sodium in the flyash did not reach the
target level of 1.2% (0.9% was reached), the experiment proved
that opacity could be lowered by conditioning the coal supply.
7-7
-------
Two options now available for Unit 2 future operation are:
1. Use intermittent energization to delay the sodium
depletion and then remove the unit from service to
reverse the depletion.
2. Use intermittent energization to delay the sodium
depletion. Then simulate the conditioning experiment
by using sodium conditioning as needed after the
unit is depleted, to reverse the depletion.
It is recommended that sodium conditioning be used after Unit 2
is depleted to reverse the sodium phenomenon effects. The bene-
fits would include keeping our sodium cost to a minimum and
reducing the need to remove Unit 2 from service due to rising
opacity.
ACKNOWLEDGEMENTS
Special thanks to Dr. Roy E. Bickelhaupt of Clean Air Laborato-
ries for the predicted resistivity work, Dr. Ralph Altman of
EPRI, Steve Hanlon of STI/Environecs, and to Ghent Generating
Station Personnel: Cecil VanDiver, Jim Ellington and Mike Lewis.
REFERENCES
1. Hall, H. J. Fly ash chemistry indices for resistivity
and effects on electrostatic precipitator design and performance.
Fourth Symposium on Transfer and Utilization of Particulate
Control Technology, 11-15 Oct. 1982, Houston, Texas. EPA-600-9-
84-025b, Nov. 1984.
-------
TEMPERATURE, °F
1000
°K
V.
10
B
c
4
2
,oL
B
6
4
2
10=1
6
4
•2
B
b
4
iou
8
6
2
10s
4
2
8C
-] —
ia
a
200
1 —
1 '
2.7
3 1C
1
3OO «XJ 500
1
1
2-8 2-5 2.4 ' 2.3 2.2
K> 150 2
1
1
a
\.
tOO 70O
i i i _ ' ._ i j -i i — i 1 — •—
- Units 2 h '4
TRUE DfKSITY l^or-1! .
C-XS fHAje TEST tr^ECTEO
MjO. V.*
tO> Vti
lliiDL.
AiH LAt IR DATA
MII r Dr«0
*
\
r
'-
t^
X
^
-
\
1
t]
• i
N
i.
T
1
I 1.6
4
M4 ••«
•_Srt CO-*"OS HON
L.,0
•7°)
.0,
IO)
LOI
-
[=!
r^-
-
\
4
•T
c
-
3
j
r:
\
\
P°1nw,
3.
29?
) .
1).
JL.
1C
5
3
r
i
1_
5
1
1
i_
»
n
•
-~r
-•
\
•
>^*
-p
4
-'
~l
I
"T
r 1.4 1.3
00 M
TEMPERATURE, °C
Figure 1. Predicted Ash Resistivities for Unit 2
CAE-72
June 1989
-------
Energization
^•ssd on
Figure 2. Intermittent Energization Cycles used on Ghent 2—typical for
all four precipitator boxes.
7-10
-------
UNIT 2 OPRCITV BEFORb, DUPING, HMD RFTER SODIUM RDDITIOh
JU
h-
1 — <
u
u_
a 15
0
R
en
a en n
U a: -
... . in t-
a v 5
I- ,r u n
M a a u
a a a cc
a H 'z g 3
'ft
'i
;
,il
,
l'| I
-fll
T T M 1
I 1 i
1
. HI
W 3 ,^ £
•n h- M a
T- n (-H
111 1-J ^_.
n CD a rn H
,,-i-n z s m uj 3
j £ z a -- a
|i n 3 Lli -i
ri m T. (f
1 - 3 "
1 1 tl CD f
n
I i "1
i ! 1 ll 1
1 I cu
1 I 'TTTTri'1' T TTTT [TrTFT ^
II !
||| 1
1 [II
0 ! £34587 89 !01 1£13 4 15 f 17 8
DflvS
UJ HI
1-
| • LL
r ! [][ ||
1 i
li 1
j
9cU£1 £2H3£^£5
Figure 3. Unit Opacity before, during and after sodium conditioning. 0-25 days actually a
condensed calendar of Nov. 28. 19B9 to Jan. 1 1. 1990.
7-11
-------
DeSOx and DeNOx by PPCP and SPCP
S. Masuda
Fukui Institute of Technology
No.415, 3-2-1, Nishigahara
Kita-ku, Tokyo 114
Japan
J. Wang
Anshan Research and Design Institute
of
Electrostatic Technology
Anshan, Liaoning
China
ABSTRACT
Unlike ordinary plasma chemical process performed in a low pressure gas atmosphere
the Pulse Corona Induced Plasma Chemical Process (PPCP) and Surface Streamer Induced
Plasma Chemical Process (SPCP) can produce highly non-equilibrium plasma having a
high electron temperature and low ion/molecule temperature even under ordinary
pressure to produce copious oxidizing radicals, such as 0, 02", OH, 03, etc.
As a result gaseous polltants from combustion gases, including NOx, SOx, mercury
vapour, etc. are easily removed in a form of higher level oxides, in particular by
the addition of fixing agents like ammonia gas.
PPCP is suitable to DeSOx and DeNOx of combustion gases from medium-size sources
(such as diesel engines for co-generation and marine use) and large-scale sources
(utility and industrial boilers). Whereas, SPCP is suitable to a small-scale comb-
ustion furnaces.
The principle, construction and test results of PPCP and SPCP are reported in this
paper.
5-1
-------
DeSOx and DeNOx by PPCP and SPCP
INTRODUCTION
The control of SOx in combustion gases is being made by its absorption to calcium-
based solvents in either wet or dry form, while the control of NOx has been made
primarily by its decompsition with the aid of catalyzers and addition of ammonia
gas. The use of high-energy electron beam has also been tested in combination with
ammonia. Common to these processes is the investment cost of the control devices,
with their operation cost also being substantial.
An attempt has been tested to use an extremely fast-rising, narrow pulse high volt-
age in a corona electrode system in consideration of its simple construction and a
low investment cost of the hardware of its reaction chamber. Both laboratory tests
and pilot plant tests indicated quite a promising performance for DeSOx and DeNOx,
with a problem of lowering its operation cost (energy efficiency) to be solved
(1-9). The authors named this process as "Pulse Corona Induced Plasma Chemical
Process (PPCP)" in consideration of the unique role of this particular mode of the
streamer corona discharge in the whole realm of plasma chemistry, including not only
control of gaseous pollutants (SOx, NOx, Hg-vapour, Freon, etc.) but also many other
innovative means of plasma chemical processing (surface treatment; CVD; ozone gener-
ation , etc.).
In parallel to the developmental efforts of PPCP for DeNOx and DeSOx from combustion
gases, one of the authors (S. Masuda) had made with his co-workers an effort to use
PPCP for surface activation of PP (polypropylene) bumpers so as to provide enough
adehesivity to keep color paint film, and could finally succeed to construct a
practical-scale PPCP equipment for this particular application.
The authors also examined a possibility of using the surface streamer discharge for
the same purpose of DeSOx and DeNOx as well as control of other gaseous pollutants
with the use of the ceramic-based surface discharge units(10), in consideration of
our finding that this mode of discharge is extremely cost-effective for ozone gener-
ation. The results obtained were quite encouraging.
In this paper are reported the fundamental aspects of PPCP and SPCP, primarily from
engineering view point, and the technical hardwares of both reaction chambers and
power supplies, and some of the test results we obtained so far.
GENERATION OF NON-EQUILIBRIUM PLASMA UNDER ORDINARY PRESSURE FOR RADICAL PRODUCTION
Most of the plasma chemical process uses a Non-Equilibrium Plasma which is a trans-
ient state from its thermal equilibrium in the sense that only electrons are in a
very high temperature (high electron energy) and both ions and neutral molecules are
in a near-oridinary temperature (low ion/molecule energy). So far as the generat-
ion of chemically active species and radicals by collision is concerned, only elect-
rons having a very small mass make a useful job, while ions do nothing meaningful.
Whereas, when ion and molecular temperature rises a pre-breakdown high-temperature
channel is formed, and it finally turns into sparking and arcing which can be self-
sustained by thermal ionization and is characterized by its very low potential drop.
In other words, the heating of ions and molecules destroys through generation of arc
8-2
-------
a high electric field to be sustained for accerelation of job-making electrons. So,
it can be understood why one or other of technical means to inject energy from elec-
tric field into electrons only and not into ions, and thereby to produce a highly
non-equilibrium plasma, has a crucial imprtance.
One of the method for accelerating only electrons and not ions is to use a very high
frequency electric field (MHz-GHz), since electrons having a very small mass can
oscillate in it to absorbe energy from the field, while massive ions remain almost
motionless without absorbing the energy. However, this selective heating is not an
enough condition for establishing the non-equilibrium plasma, as the energy of the
heated electrons is transfered through ccollision to ions and neutral molecules with
time to raise the ion/molecule temperature. Hence, another requirement is to avoid
or minimize this energy transfer. In the conventional plasma chemical process this
has been realized by using a low pressure gas atmosphere, where the mean free path
of electrons is greatly raised and its collision frequency with ions and molecules
is negligibly low. This "Low-Pressure High-Frequency Plasma Chemical Process" is
being widely used in semi-conductor processing, but the requirement for the low-
pressure is a prohibitive condition for its use in many chemical processing which
must be made under ordinary pressure environment. Once the high-frequency field be
applied in the inter-electrode gas spacing under ordinary pressures, we get immed-
iately arcing, as is widely known.
Another method of accerelating only electrons and not ions is to use a very narrow
pulse high voltage in combination with a corona electrode system, where its corona
electrode serves as a stable triger element of streamer corona. In this case the
duration of high field is so short that the massive ions cannot be accerelated. The
mechanism of minimizing the ion/molecule temperature rise through electron bombard-
ment is a very low duty ratio of the pulse (ratio of the pulse duration time to the
total pulse repetition period), which allows for an enough cooling off time of ions
and molecules which had been partially heated by electron collision. It has been
discovered that the pulse rise time plays a very important role in this process.
Faster the pulse rise, the more active is the streamer corona with a concurrently
higher DeSOx/DeNOx effect (1). This is the PPCP process as already described.
The greatest advantage of the PPCP is that it allows the use of the plasma chemical
specific reactions under the ordinary gas pressure condition. This process has some
analogy with diesel engine which uses a very high compression ratio to use a more
efficient heat cycle without causing melting of its cylinders through intermittent
combustion to allow heat dissipation.
An alternative method to PPCP is to use the surface streamer mode discharge which
occurs in a special electrode configuration in which an insulator sheet is sandwich-
ed by corona electrodes on its one side and a film-like induction electrode on the
other. When a high-frequency ac voltage or pulse voltage is applied between these
two electrodes, surface streamers develop from the side edges of the corona electr-
odes to proceed along the insulator surface, but they cannot bridge across the two
electrodes to turn into arc, as they are interrupted by the insulating barrier. So,
they streamers stop at a certain point where the tangential component of the electr-
ic field acting on the streamer tip drops below its sustaining level. The silent
glow discharge used for ozone production, which occurs between two parallel electr-
odes comprizing an insulating layer therebetween, has also a similar element for the
interruption of arcing. So, this can also be used in place of PPCP. The surface
streamer is a very fast process so that it fulfil the requirement of selective heat-
ing of electrons. The silent glow discharge is an assembly of many tiny intermitt-
ent pulsive glow discharges, distributed densely in space and time to look like a
steady-state glow discharge, so that it also fulfil the selective heating condition.
The difference exists, however, between the former (Surface Discharge Induced Plasma
Chemical Process = SPCP) and the latter (Silent Glow Discharge Induced Plasma Chemi-
cal Process = GPCP), in that the former allows a much more effective heat dissipat-
ion to avoid thermal decomposition of the reaction products through the insulator
surface while the latter has a limited heat dissipation through a gas volume. This
8-3
-------
difference has a great meaning technically, since the SPCP allows the use of a much
higher frequency to cause a substantially larger heat loss generation. So far as
an excessive temperature rise be avoided by an effective cooling of the system, the
generation of the reaction products is proprtional to the frequency. Hence, PPCP
allows a substantial reduction in costs of the reaction chamber and its power supply
compared at the same total output of the products.
Common to PPCP, SPCP and GPCP processes, one of the most essential factors to be
carefully considered is the value of the indivisual non-elastic collision frequency
of an electron with a neutral molecule for generation of each useful radical as com-
pared to that of elastic collision giving rise to the molecular temperature rise.
These parameters are functions of Ex or E/p and gas species, where E = local
electric field, X = electron mean free path, and p = absolute gas pre-
ssure .
The performance and energy efficiency of PPCP depends on the design of
both the corona electrode system (reaction chamber) and its pulse pow-
er supply to a great extent. For generation of such a fast-rising
narrow pulse high voltage (e.g. rise time Tr=10-50 ns; half-tail Th=
100-200 ns; peak field Ep=15-30 kV/cm) a completely new approach must
be made, leading to a new area of nanosecond high-voltage pulse techn-
ology.
Those for SPCP is critically dependent on the cooling efficiency of
the reaction chamber, and secondly on the design of the high-frequency
power supply which enables the application of an adequately high volt-
age at an adequately high frequency (Vpp=10-15 kV; f=10-30 kHz). The
limiting factor in this case is the occurence of a jumping phenomenon,
which is a typical phenomenon occuring in non-linear oscillation. The
load current jumps suddenly to an anomalous level at a certain voltage
to damage transformer winding? and other circuit components. The app-
arent capacity cf the surface discharge unit increases each half-cycle
when the surface discharge occurs, and this acts in combination with
the saturation of the transformer core. This difficulty could be
solved by one of the authers (S. Masuda) by a novel circuit design (11).
It should be noted that PPCP, SPCP and GPCP are to be considered from
a new aspect of "Gaseous Chemitronics" which further belongs to the
whole realm of the "Chemitronics" - an aspect of science and technolo-
gy to place emphasis in chemical reactions to the electronic and quan-
tum electronic interactions and manipulations (12).
PPCP PROCESS
Figure 1 illustrates a typical construction of PPCP, consisting of a
reaction chamber out of a corona electrode system, its pulse power
supply, and an after processing unit for a final removing of the reac-
tion products from gas phase. In most of the cases negative corona
is used, as its streamer easily streaches to occupy an entire inter-
electrode gas space. Hence, the utility factor of the space is about
10 times^higher than that in negative streamer corona which occupies
only a limited space around the corona electrode. However, there is
practically little difference between the positive and negative coronas
so far as the energy yield of reaction (kWh/kg) is concerned.
The use of a catalyzer (e.g. Ti02-silica based) after the reaction cha-
mber can enhace the PPCP reaction to drastically reduce the necessary
residence time of gas in the reaction chamber, and thereby reduce the
costs of both the corona system and its pulse power supply with a con-
current reduce of its running cost, too. It is likely that both the
8-4
-------
the pollutant molecules (SOx, NOx, etc.) and the radicals (with a
longer life time) are absorpt on the inner surface of the catalyzer
to react there in a two-dimensional condensed phase, thereby overcom-
ing a very low gas/gas reaction speed at low concentrations. It is
observed that the active spots of the catalyzer are gradually covered
by the reaction products (e.g. ammonium sulphate; ammonium nitrate) to
indicate a sudden rise of the pollutant emission. The catalyzer must
be regenerated at this point, either by passing a heated clean air or
the PPCP treated clean air.
It has also been observed that almost the same performance could be
obtained by passing clean air only through the PPCP reaction chamber
located outside of the main gas stream and mixing the PPCP-treated air
with the gas, with and without the downstream catalyzer.
Ammonia addition is usually made in front of the PPCP reaction chamber
at a lower gas temperature below, say 90 °C so that the solid reaction
products (ammonium sulfate and ammonium nitrate) can be obtained. The
after processing unit is an electrostatic precipitator in this case.
Figure 2 illustrates 2 examples of the power supply for use in PPCP
processing to generate an extremely fast-rising narrow pulse high vol-
tage. Both are the "Condensor-Strorage Types" where electrical ener-
gy is stored to the tank-condensor and suddenly fed to the corona sys-
tem using a very high switching element, like a stationary or rotating
spark-switch. The use of the rotary spark-switch allows the charging
of the tank-condensor during its "OFF" period, and its switching dur-
ing the blocking half-cycle of the diode. This selective switching
makes it possible to remove either the resistance or inductance to be
used in the charging circuit for avoiding restriking at the spark-
switch after its normal switching action. Any other kind of switch-
ing element, including a trigger-gap switch, laser-trigger switch, or
high-speed thyratron or semi-conductor switch, can be used.
Actually, the most essential factor to finally restrict the rising
speed and duration of the pulse is the "Stray Inductance" of the cir-
cuit. Hence, the corona electrode system, its power supply, and its
feeding circuit must be considered as an integral system for optimiz-
ation.
Figure 3 indicates an example of DeNOx effect by PPCP where the conc-
entrations of NO and N02 are plotted with increasing pulse peal volt-
age, Vp (Ep=peak value of average field)(1). NO begins to drop to turn
into N02 at the corona start (Ep=5 kV/cm) with a concurrent increase
in N02, and becomes zero at Ep=11.5 kV/cm where N02 indicates the max-
imum concentration. Then, NO2 suddenly drops to zero at Ep=12 kV/cm.
This test was made under high-energy electron beam irradiation to make
clear the cross-correlation effect, which, however, could not be obs-
erved. The effect of the electron beam irradiation only is indicated
by the drop of the initial concentrations of NO and N02 •
Figure^4^shows an example of NOx and SOX reduction by PPCP with ammon-
ia addition, plotted against the energy consumption per unit volume of
the treated gas (Wh/Nm3), measured at ENEL Pilot Plant (9).
Figure 5 shows an example of the removal effeciency of mercury vapour
from the combution gas of an incineration plant, plotted against the
pulse peak voltage for 2 s and 7.5 s residence time (7). It can be
seen that the efficiency rises with increasing pulse peak voltage and
decreasing gas temperature.
This temperature-dependence is likely to be its effect on the format-
8-5
-------
ion of the radicals responsible to the PPCP effects.
SPCP PROCESS
Figure 6 shows a photograph of the Ceramic-Based Surface Discharge
Unit for use the SPCP processes (10), and Figure 7 its surface disch-
arged occuring when a high-frequncy ac voltage is applied between the
corona corona and induction electrodes (Figure 8) .
Two sheets out of alumina/binder mixture (92 % purity alumina; organic
binder) are printed with tungsten ink (tungsten powder dispersed in
organic dolvent) to form corona electrodes and an induction electrode,
respectively, using the silk-printing method. Then, these sheets are
laminated and rolled to form a cylindrical unit as shown in Figure 6
or planer unit.Sintering is made in a hydrogen furnace at 1500 °C, and
a thin protection coat (99 % alumina) is attached by sintering to
cover the corona electrodes.
The outer surface of the cylinder is water-cooled or air-cooled with
a water-jacket or cooling-fins attached. The performace rises in
general with lowering temperature and water content of gas and incr-
easing its pressure (10). Its operation at a very low temperature
(-100 to - 200 °C) may give rise to a drastic increase in its perform-
ance, in particular energy yield (13) .
A series of preliminary tests revealed that the SPCP provides also an
equivalent DeSOx/DeNOx effects to those of PPCP, and also destruction
of freon and other gases under issues. SPCP also provides a very
good means for CVD as it enables an operation at an ordinary temper-
ature and pressure of gas.
A very unique feature of the Ceramic-Based SPCP units is that they can
be used under a very wide temperature and pressure range (-200 to 600
°C; vacuum to 6 atmG).
A great future is anticipated for SPCP for a small capacity DeNOx/
DeSOX applications.
CONCLUSION
A role of the non-equilibrium plasma for oxidizing radicals for DeSOx/
DeNOx, and PPCP and SPCP for exploiting this plasma under an ordinary
gas pressure are discussed.
The PPCP process has been successfully tested for DeSOx/DeNOx of com-
bution gases from medium to large scale boilers and emission sorces
(utility boiler; incineration boiler; diesel engine). The target for
developing the practical PPCP system for DeSOx/DeNOx is now directed
to realization of the Pulser/Electrode-Hybrid which can provide a much
higher energy yield at a short gas residence time.
The PPCP for surface treatment of plastic materials has already been
completed.
The SPCP process has been primarily used successfully for ozone gene-
ration, and it proved to be a very compact, cost-effective device.
A preliminary test series revealed its great potential for use in a
small scale DeSOx/DeNOx applications.
-------
REFERENCES
1. S.Masuda and H.Nakao."Control of NOX by Positive and Negative
Pulsed Corona Discharges'.1 Rec. IEEE/IAS 1986 Ann. Conf., pp.1173-
1182 (Oct., 1986).
2. J.S. Clements, A. Mizuno, W.C. Finney and R.H. Davis. "Combined
Removal of SO2, NOX and Flyash from Flue Gas Using Pulsed Stream-
er Corona" ibid, pp.1183-1190 (Oct., 1986).
3. L. Civitano, G. Dinelli and M. Rea. "Removal of NOx and S02 from
Combustion Gases by Means of Corona Energization, Proc. of 3rd
Int. Conf. on Electrostatic Precipitation, pp.677-687 (Oct., 1987
Padova, Italy).
4. S.Masuda, Y.Wu, T.Urabe and Y.Ono. "Pulse Corona Induced Plasma
Chemical Process for DeNOx, DeSOx and Mercury Vapour Control",
ibid., pp.667-676 (Oct., 1987, Padova, Italy).
5. S.Masuda and Y.Wu. "Removal of NOx by Corona Discharge Induced
by Sharp Rising Nanosecond Pulse Voltage", Electrostatics 87,
Inst. Phys (London) - Conf. Ser. No.85, pp.249-254 (April, 1987,
Oxford, UK).
6. S.Masuda. "Pulse Corona Induced Plasma Chemical Process - A Hori-
zon of New Plasma Chemical Technologies", Pure and Applied Chem-
istry, Vol.60, No.5, p.723 (1988).
7. S.Masuda, Y.Wu, T.Urabe and Y.Ono. "DeNOx and Control of Mercury
Vapour from Combustion Gase by Pulse Corona Induced Plasma Chem-
ical Process (PPCP)", Proc. 8th Int. Symp. on Plasma Chemistry,
pp.2222-2226 (Aug./Sept., 1987, Tokyo, Japan).
8. L.Civitano, G.Dinelli, F.Busi, M.Dangelantonio, O.Tubertini, I.
Gallimberti and M. Rea. "Flue Gas Simultaneous DeSOx/DeNOx by
Impulse Corona Energization", Proc. Int. Atom. Energy Agency Con-
sultants Meeting (Oct., 1988, Karlsruhe, West-Germany).
9. ENEL Research on Electrostatic Technologies for Pollutant Emission
Control - Results and Perspective. "Reduction of Nitrogen and Sul-
phur Oxide", pp.36-48 (1989).
10. S.Masuda, K.Akutsu, M.Kuroda, Y.Awatsu and Y.Shibuya. "A Ceramic-
Based Ozonizer Using High-Frequency Discharge", IEEE/IA Trans.,
Vol.24, No.2, pp.223-230 (March/April, 1988).
11. S.Masuda. US Patent No.4.644.457, US Patent No. 4.706.182, EPC
Patent Appl. No. 85106599.5, EPC Patent Appl. No. 87300152.3
12. S.Masuda. "Proposal for Chemitronics", Review Talk. Proc. 22nd
Ann.Meeting of Inst. Chemical Eng. Japan (Oct., 1989, Tokyo, Japan
- in Japanese).
13. S.Masuda, S.Koizumi, J.Inoue and H.Araki. "Production of Ozone by
Surface and Glow Discharge at Cryogenic Temperatures", IEEE/IA
Trans., Vol.24, No.5, pp.928-933 (Sept./Oct., 1988).
5-7
-------
—1~ digital
i r«cord«r
JCT
constant
alyze
blower AIR NO exchanger tempera ture
Fig. 1 Principle of PPCP
Fig. 2 (a) Pulse Power Supply
for PPCP (parallel-
capacitot type)
Pulse Forming Condensers
1 JL 1
Fig. 2(b) Pulse Power Supply for PPCP (series capacitor Type)
c 50 I
~[BO>],ni.
•*-rNOl i ni t i
1 A
(-["Vmu
ul
»l
Total NO
f
•
NO
i_-i
NO,
la] z
Q: 3Nl/min
Tq: 75 s
tq: 10 °C
qap : 5cm
d : 3mm+
fp: 50HI
*>*
_^A
fe"
L-"%
M
•V-
0 a 4 68 10 12"Eo(kV/cm)
0 10 20 30 40 50 60 Vp(kV)
Peak Voltage Vp and Peak Field Ep
Fig. 3 DeNOx Performance of
PPCP (NH3: 180 ppm;
negative; dry air)
200
Fig. 4
DeSOx/DeSOx Perform-
ance of PPCP at Pilot
Plant of ENEL (NH3;
coal combustion)
-------
10
10 30 40
VP UV)
Fig. 5 Removal of Mercury Vapour from
Combustion Gas of Incenerator
Fig. 6 Ceramic-Based Surface
Discharge Unit for SPCP
Fig. 7 High-Frequency
Surface Discharge
mtnil to Induction Electr.
Fig. 8 Construction of Ceramic-Based
Surface Discharge Unit
8-9
-------
The Destruction of Volatile Organic Compounds
by an Innovative Corona Technology
Geddes H. Ramsey, Norman Plaks, Chester A. Vogel, Wade H. Ponder
Air and Energy Engineering Research Laboratory
Environmental Protection Agency
Research Triangle Park, North Carolina 27711
Larry E. Hamel
Acurex Corporation
P.O. Box 13109
Research Triangle Park, North Carolina 27709
9-1
-------
The Destruction of Volatile Oganic Compounds
by an Innovative Corona Technology
ABSTRACT
The emerging concern for ozone non-attainment, for which many
volatile organic compounds are a precursor, and the need to
improve technology to control low concentration streams provided
the impetus for work on high intensity corona reactor devices.
EPA is currently testing a corona destruction reactor, and the
results show mueh promise.
In setting up the original feasibility-of-concept experiments, an
attempt was made to relate the work to the greatest extent
possible to current problems facing the Agency. The destruction
of both non-halogenated and halogenated compounds has been
evaluated. The non-halogenated hydrocarbons were toluene and
benzene. The halogenated hydrocarbons were methylene chloride
and trichlorotrifluoroethane (CFC-113). Two types of corona
reactors are being evaluated. The first type makes use of a bed
of ferroelectric pellets across which an alternating current
electric field is impressed. The second develops corona between
two electrodes that have been energized by a fast rise time
(nanosecond range) electric field (PPCP). The data from the
experiments to date allow the following conclusions:
• The hydrocarbon molecule was completely destroyed in the
reaction. No unreacted fragments were identified in the
analyses.
« The destruction efficiency for toluene at optimum conditions
was greater than 99 percent.
From these data, several possible mechanism hypotheses have been
proposed. The purpose of this paper is to present these reaction
hypotheses which are based on the excitation of the oxygen
molecule which initiates the reaction. Once initiated, the
reaction goes to completion.
9-2
-------
INTRODUCTION
The EPA interest in corona destruction of molecular species
started with modeling of a point-plane reactor for destruction of
toxic and other noxious substances [1]. This work was followed
by the development of an improved reactor using a triangular
electrode geometry that simulated the shape of the active corona
region [2].
In Japan work was being done on the use of reactors energized
with fast rise time, high electric field pulses (nanosecond
duration) for collection of SOX and NOX as well as mercury
emissions from municipal waste incinerators. The technology is
called pulsed corona induced plasma chemical process (PPCP) [3].
Other work there showed that corona and high electric fields
could be developed in a ferroelectric pellet-layer for particle
collection [4]. The purpose of this work is to determine if
volatile organic compounds (VOCs) can be destroyed in these high
intensity corona reactors.
The emerging concern for ozone non-attainment, for which many
VOCs are a precursor, the need to improve technology to control
low concentration streams, and the potential for improving
collection of SOX and NOX provided impetus for the work on high
intensity corona reactor devices.
In setting up the original feasibility-of-concept experiments, an
attempt was made to relate the work to the greatest extent
possible to current problems threatening the environment. The
destruction of both non-halogenated and halogenated compounds has
been evaluated. The non-halogenated hydrocarbon was toluene.
Toluene was chosen because it is a common solvent of the type
implicated in ozone non-attainment. A second consideration was
that toluene is a wood smoke surrogate. The halogenated
compounds were methylene chloride, which is a common solvent, and
CFC-113, a solvent commonly used in the electronics industry,
both of which have been implicated in stratospheric ozone layer
destruction.
This paper will present a description of the technology, a brief
description of the results, several hypotheses of the reaction
mechanisms occurring in the process, and a listing of additional
experiments proposed for the future to support these hypotheses.
TECHNOLOGY DESCRIPTION
Two types of corona reactors are being evaluated. The first type
9-3
-------
develops corona between two electrodes that have been energized
by a fast rise time (nanosecond range) electric field.
The pelletized bed reactor is shown in Figure 1. The pellets
must be Bade of a material with a high dielectric constant. In
this ease the material is barium titanate with a dielectric
constant of about €000. Porous stainless steel plates, through
which the gas enters and exits the bed, are the electrodes to
which the source of AC voltage is applied. Corona appears
between the contact points of the pellets with the application of
an external AC electric field that is as low as about 1 kV/cm.
Sparking across the bed occurs for fields of 6 to 8 kV/cm,
depending upon the size of the pellets. The best operating point
for the reactor is just under sparking. The static field from
the application of an external direct current (DC) voltage,
rather than AC, causes sparking across the bed rather than
corona.
Gas Inlet
Gas Outlet
Electrode
Etadirod*
Bed
Figure 1: Corona destruction reactor.
In a particle layer across which an electric field has been
impressed, the local fields between the particles are greater
than the average applied field. Furthermore the local electric
fields increase with increasing particle dielectric constant [5].
Very strong fields are developed between the pellets in the
direction of the applied field. An analysis based upon
application of Paschen's Law, to determine the voltage required
for spark breakdown for the pellet geometry, suggests that the
electric field in the interstitial spaces between the pellets
ranges from 20 to more than 600 KV/cm.
The second reactor, which is of the PPCP type, is shown in Figure
2. The gap space between the electrodes, through which a very
short (nanosecond pulse) high field strength pulse is impressed,
does not contain pellets or any other packing material. The
pulse energization system, as shown in Figure 3, consists of a DC
power source, a pulse forming network, IL and C., a rotary spark
gap acting as a high voltage switch, and a pulse length forming
resistor, Rj. The very rapid discharge of capacitor C. provides
the very fast rise time pulse. The pulse repetition rate is set
9-4
-------
by the rotational speed of the rotary spark gap. Based upon the
peak applied pulse voltage and the geometry of the reactor
electrodes, electric fields on the order of 20 kV/cm or more can
easily be developed.
Triton ttiM«Mt»MlTub«
Rgun2. Schematic of pulsed corona p4nma reactor.
FlgurtX Schematic of nanosecond pulM power supply.
Table I shows the concentrations of the major components in the
entering and exiting gas streams for a typical run in the pellet
bed reactor.
Table I. MAJOR COMPONENT CONCENTRATIONS
Gas
Component
Toluene
Carbon Dioxide
Carbon Monoxide
Concentration
ppm
Inlet Outlet
229
300
Not detected
Hoi detected
313
584
T«*t condffiocw: 800ccfminand14KV
RESULTS
The first phase of the corona destruction program was to
prove that the process could be used to destroy VOCs efficiently.
From the statistically designed experiment, the following
conclusions were evident:
• Of the three pellet sizes (1, 3, and 5 mm) tested, the
1 mm diameter pellets destroyed toluene most efficiently.
9-5
-------
Toluene was apparently destroyed in the reaction. No
products of incomplete reaction were identified in the
analyses.
The destruction efficiency for toluene at optimum conditions
was greater than 99 percent in both the packed bed and the
pulsed reactor.
The reaction favors the formation of carbon monoxide over
carbon dioxide.
MECHANISMS
Electrons undergo both elastic and inelastic collisions during
their capricious travels. In an elastic collision the electron
retains the majority of its kinetic energy. Under the influence
of the strong electric field, free electrons are accelerated.
They undergo an elastic collision at the end of each free path
length. The electrons continue to increase in energy until the
energy becomes high enough to allow the electrons to undergo an
inelastic collision. During an inelastic collision the electron
transfers all or part of its kinetic energy to the particle with
which it has collided. In addition, inelastic collisions result
in a change to the target particle or molecule such as
ionization, disassociation, or excitation [5]. In an inelastic
collision significant amounts of energy are transferred from an
electron to the target species. Examples of these inelastic
collisions are:
• Electron attachment by electronegative gases to form
negative ions,
• Disassociation of molecular species into smaller
fragments including formation of free radicals,
• Excitation of molecular and elemental species,
• Ionization to form positive ions and additional free
electrons (under favorable circumstances a Townsend
avalanche develops generating many additional free
electrons), and
• Breaking down of molecular species into their elemental and
atomic components.
The amount of energy required for the above events varies by type
of event and molecular/elemental species. According to the
literature, some examples of energy requirements for different
types of events are less than 5 eV for electron attachment and 5-
25 eV to form positive ions by electron removal [6]. An electron
volt (eV) is defined as the energy that an electron acquires (or
loses) in passing through a 1 V change. One electron volt is
equivalent to 1.6E-19 J or 3.8E-23 kcal. The electron volt is a
9-6
-------
particularly useful unit of measure for this work because it
allows an easily made comparison of electrical energy input to
energy requirements for reaction of the target molecules. The
probability of one of the above events occurring is expressed by
the collision cross section which is mainly a function of
concentration. Whether the event will occur is dependent on the
electron's having achieved the energy level needed for the event
to occur.
The general velocity of electrons in the direction of the
electric field is known as the drift velocity [7], At
atmospheric pressure and ambient temperatures, an electron's
energy level can increase by about 1 eV in one mean free path
length, if the mean free path length is parallel to an electric
field of 20-30 kV/cm. At atmospheric pressure and ambient
temperature a mean free path length is about 1E-7 m.
In addition to the above effects of inelastic electron
collisions there are also photoelectric effects (ionization,
disassociation, excitation, etc.) which are either activated by
or result in the emission of a photon. The events that can occur
are very much interrelated.
In total the picture that is presented is one of extremely
high complexity when the possibility of electron collisions and
the effects of photons are considered. The energy distribution
in a swarm of energetic electrons ranges from very low to very
high, with the majority centered around some median value.
Therefore electrons having a wide range of energies will undergo
various types of inelastic collisions such as attachment,
excitation, and ionization. For a mixture of gases the picture
becomes even more complex.
The major work on toluene was done in the pelletized bed reactor.
The power input to the reactor was measured, which provided a
calculation of the energy introduced for a unit of time. For a
typical toluene concentration of about 200 ppm the energy
introduced per toluene molecule was about 10 eV (230 kcal/g-
mole). The carbon by-products were CO and CO2. The ratio of CO
to C02 was usually about 6:1.
The energy input to the pulsed reactor was not measured. For a
theoretical maximum energy based upon the DC supply voltage, the
capacity of C, in Figure 2, and the pulse repetition rate, the
available energy was about 250 times greater than it was for the
pelletized bed reactor. What is not known is the residual charge
remaining in C1 after each pulse, the amount coupled to the
reactor, and the amount drained off through Rg. The energy
available for the reactor is expected to be considerably less
than the theoretical maximum.
A number of reaction mechanisms are possible in the corona
destruction of aromatic hydrocarbons. Three of the reactions are
presented and discussed. The first reaction mechanism involves
initial attack of the hydrocarbon molecule with an energized
9-7
-------
from one of the carbons in the ring, have a lower probability
[8].
A mechanism likely to occur in the destruction of toluene is the
oxidation of the methyl group of the molecule. Toluene has a
resonance structure where a proton is lost or gained at the
methyl group which should result in a more reactive site. The
methyl group serves as an electron donor to the phenyl group. An
oxygen would attack the methyl group and the reaction would
proceed:
+ CH2O
The CH20 radical rapidly reacts to form a CHO radical which goes
to CO. The benzene radical reacts:
or O-OC* radical + •OC-OO radical
The O»C or 0»OC radical reacts with oxygen to form CO2.
The other radical oxidizes rapidly to CHO and then to CO [5].
The reaction to form CO rather than CO, is favored at low
temperatures. A CO2 molecule is formed when the ring breaks but
the CO reaction is favored in the remainder of bond destruction
reactions. This accounts for the approximately 6:1 ratio of CO
to C02observed experimentally.
The energies of bond formation/destruction are [6]:
C-C 3.6 eV
OC 6.3 eV
OC (in ring) 5.5 eV
C-H 4.3 eV
C-0 3.7 eV
00 7.7 eV
00 (in C02) 8.3 eV
In the reaction, if a C-H bond is broken (by an electron or
reaction with oxygen), energy is required but a radical will form
which will react with oxygen as O, O~,O2", or excited O2. The
energy released when the C-O bond forms is enough to break the
adjacent C-C bonds with an excess of energy in all cases. There
is not sufficient electrical energy inputted to the reactor to
sever all of the C-C bonds. Therefore the energy for the
reaction once past the initiation energy must come from the
oxidation of the toluene itself [9]. Because 20 percent of the
gas is oxygen and since the oxygen molecule is one of the easiest
to excite, the excitation of the oxygen molecule is consequently
the most likely mechanism occurring in the process. The
potential energies for the ground state and the first four
electronically excited states of O2 are shown in Table II.
9-8
-------
TABLE II. THE POTENTIAL ENERGIES FOR THE EXCITED STATES OF
OXYGEN
States of Oxygen
Molecule
around state
1st excitation
2nd Mutation
3rd excitation
4th excitation
Energy
Required
•V
0
0.08
1.83
425
&00
Gonwntnti
Forbidden transition
AJtowed; naff-fife of about 10 seconds
cotttioaiBy relaxing to 1st *"**&*• state.
Theoretically forbidden
Allowed; creates two oxygen atoms In the
ground state via a non-radiative transition.
The addition of energy greater than 7 eV causes dissociation of
the oxygen molecule to one atom in the ground state and one in
the 1st excitation state [10]. Many excited states of oxygen are
possible inside the corona destruction reactor.
The benzene molecule should react similarly to the toluene
molecule but slightly more energy would be required to initiate
the benzene reaction. Instead of the excited oxygen attacking
the methyl group of the toluene molecule, the point of attack in
the benzene molecule would have to be the ring structure or a C-H
bond. The bond energies would favor an attack of the C-H bond
(4.3 eV for the C-H bond compared to 5.5 eV for the OC bond).
This is slightly higher than the C-C bond energy which would be
the point of attack in the toluene molecule (3.6 «V for the C-C
bond). Therefore, since a higher energy is required to initiate
the reaction of the benzene molecule, destruction of benzene
should be lower than toluene under similar conditions.
The second possible reaction is the breaking of a C-C bond in
the chain by a sufficiently energetic electron. This reaction is
of the type:
AB + • > [AB] + e
in which [AB] is a radical. For the toluene molecule, the ring
structure has a greater affinity for an electron than does the
methyl group. The energy required to break a C-C bond in the
toluene ring is 5.3 eV (123 kcal/g-mole). For toluene, cleavage
of an arbitrary C-C bond would be:
9-9
-------
CH3 H3
^ H H H C H H
iQJ -t- 5.3 ®V — »• • C-C-OC-OC-
The fragment on the right side of the above reaction equation is
a radical. The energy required for ionization of toluene is 8.5
eV (195 keal/g-mole). Consequently the collision cross section
for C-C bond cleavage is considerably less than that for
ionization.
Once a C-C bond is broken and the free radical is formed, it is
able to react with oxygen. The heat ©f oxidation of toluene
is 39 eV (901 kcal/g-mole) when going to CO2 and 22 ®V (497
kcal/mole) when going t© CO [6], This energy is sufficient to
sustain the oxidation of all carbon molecules once the reaction
starts.
For this second reaction to occur, electrons would have to
achieve 5.3 eV t© sever a C-C bond. Many electrons do achieve
the higher energy levels, but not all. In addition for this
reaction to dominate, as when the energy input is 10 eV per
toluene molecule, a significant portion of the energy inputted
would be required for breaking the first C-C bond. This is
unlikely.
A final reaction possibility is removing a hydrogen from the ring
structure by electron collision. The energy required to break a
C-H bond is about 4.3 eV. Once the C-H bond is broken the
reaction will proceed by the same mechanism as the primary
reaction suggested above. This mechanism would also account for
the CO/COZ formation.
The intermediate steps in the oxidation require the attachment of
an oxygen to a severed carbon bond or to a site where a
hydrogen was removed. The energy released by attachment of the
oxygen is sufficient to break an adjacent C-C bond, which
provides the site for the next oxygen attachment. The most
likely intermediate by-product is the radical CHO which is
favored at low temperatures. The CHO radical leads to the
formation of CO; higher temperature reactions favor the formation
of C02. Note that, once a C-C bond in the ring is severed, the
radical that is formed has two active ends for attachment of an
oxygen.
In all the suggested mechanisms, once the reaction ©f the
individual molecule is initiated, the destruction ©f the
molecule proceeds to completion since no other lower molecular
weight species are found during analysis. If the molecule were
not completely oxidized, other hydrocarbon by-products would
appear in the exhaust stream. The absence of lower molecular
weight hydrocarbons has been confirmed by gas chromatography/mass
spectrometry (GC/MS). For operating conditions in which less
than 100 percent destruction is purposefully achieved, the
unreacted toluene molecules remain intact, which is evident on
both the Residual Gas Analyzer (RGA) and GC/MS.
9-10
-------
unreacted toluene molecules remain intact, which is evident on
both the Residual Gas Analyzer (RGA) and GC/MS.
For the destruction of halogenated VOCs, it is hypothesized that
one or more of the halogens (Cl or F) must be removed from the
target molecule, for the reaction to proceed. The energy
required to sever Cl-C and F-C bonds is 3.5 and 5 eV,
respectively. For the small pelletized bed reactor, the energy
required to break one of the halogen-carbon bonds is a
significant portion of the inputted electrical energy. The
pulsed energy reactor appears to have considerably more energy
available than does the pelletized bed reactor. A greater
destruction efficiency for halogenated VOCs was evident for the
pulsed reactor than for the pelletized bed reactor. Therefore,
we believe that the destruction of a halogenated VOC is more
input energy dependent than is the destruction of a non-
halogenated VOC. This is consistent with the types of reactions
that are involved.
For the destruction work in which electrical energy input
measurements were able to be made, no differences were noted for
variations in flow rate or concentration. This suggests that,
for the pelletized bed reactor, all of the electrical input (not
including dissipation factor loss) goes to one sort or another of
elastic or inelastic collisions involving electrons (and some
ions). The total amount of electrical energy consumed by the
reactor is dominated by its electrical characteristics rather
than the target gas species and the energy required for
initiation of reaction.
DISCUSSION
The reaction mechanism and intermediate radicals produced may be
comparable to both catalyzed and combustion reactions. The major
difference is the low temperature at which it occurs. Catalyzed
oxidation reactions require initiation to occur; afterwards they
are generally self sustaining. Oxidation by combustion raises
the temperature of the bulk gas as well as the molecules to be
combusted. Combustion requires a sufficiently high concentration
of the fuel to sustain the combustion. Corona destruction, in
which the concentration is below the flammability limit, may
raise the temperature of individual molecules without affecting
the temperature of the bulk gas [9].
When the toluene concentration was about 200 ppm, and the energy
input was about 10 eV per molecule, the destruction was
essentially 100 percent complete. The energy required to
initiate the destruction reaction ranged from about 1.4 to
slightly over 5 eV which was about 14 to 50 percent of the total
supplied as electrical input. The fate of the other 50 to 86
percent of the input energy is unknown. Obviously some portion
was used to make ions which, because of their low energies, do
not contribute significantly to the reactions, but due to their
drift velocities do expend energy. There is an expenditure of
9-11
-------
energy as photons as evidenced by the visible corona. Some
unknown side reactions very likely occur that do not enter the
toluene destruction reaction. Finally some of the energy is
being dissipated as heat both from electron and ion bombardment
and from the dissipation factor in the dielectric pellets.
For higher concentrations of toluene(400 ppm), the energy
available for each toluene molecule would be about 5 eV; the
total power input does not vary with changing concentration. If
a significant fraction of the energy inputted goes to support
other activities that do not contribute to the destruction of the
toluene than it is probable that there will not be sufficient
energy available to initiate the destruction of all of the
toluene.
The data suggest that, for a non-halogenated VOC such as toluene,
the destruction of each molecule must be initiated, after which
the molecule will continue to oxidize. For higher
concentrations, which are still below the flammable limit, the
destruction efficiency will be energy input limited.
For halogenated VOCs the data suggest that, because the reaction
is endothermic, the efficiency of the destruction is directly
related to the electrical energy that is inputted. It is
necessary to strip the halogens from the molecule as part of the
destruction. Little is currently known about the intermediate
and final by-products.
ADDITIONAL EXPERIMENTS
In order to better define the corona destruction process, in-
house EPA experiments are planned to support or refute these
suggested mechanisms and ensure the validity of the final
theoretical mechanism. The experiments which are planned
include:
• Determine destruction with different compounds - The
experiments and analysis suggest that the destruction
efficiency depends upon either the presence of easily
attacked sites on the target molecule such as the methyl
group on the toluene, or the energy needed to break chemical
bonds. For some of the reactions, excited, ionized, or
disassociated oxygen, which resulted from energetic
electrons, is needed. The destruction reactions were self-
sustaining after this initiation energy. For the compounds
with carbon-halogen bonds (methylene chloride and CFC-113,
energy must be added to break the bonds and the energy of
bond formation with oxygen does not allow the reaction to be
self-sustaining to complete destruction. The more C-C1 and
C-F bonds in the molecule, the more energy intensive the
reaction process will be. Testing the following compounds
would be important to better understand the mechanism:
• Methane, which does not have any c-C bonds.
9-12
-------
• An aliphatic compound such as hexane containing neither
an aromatic ring nor double bonds.
• Benzene, an aromatic which does have double bond
character but no methyl group.
Another experiment of interest might be to evaluate the
destruction efficiency of straight chain VOCs having a double
bond in various positions.
For all future tests, measures will taken to carefully measure
and relate the energy required for destruction. Planned
additional tests and analysis are:
• GC/MS analysis - The GC/MS will be used to analyze the
emissions from the corona reactor. Any intermediate
compounds from the destruction reaction can be identified
from this analysis. This determination may also aid in the
identification of radicals formed in the process. This
will also show whether the reaction for an individual
molecule will go to completion once it is initiated.
• Moisture balance - No analyses have been run to determine
the fate of the hydrogen produced during the destruction
reaction. It has been assumed that all the hydrogen formed
reacts with oxygen, but the amount of water in the exit
stream was never determined. Water was observed on the RGA
but was never quantified.
• Carbon balance closure - Better instrumentation will be used
to determine the components in the exhaust of the reactor.
The Byron Hydrocarbon Analyzer has proved useful but a more
sophisticated instrument is required to quantify the final
products. A more accurate analysis of the C02/C0 ratio is
also necessary to complete the reaction mechanism.
• The effects of additional moisture will be evaluated. For
the preliminary evaluations dry carrier gas was used. In
real situations there would be water in the carrier gas.
Water would provide a collision cross section for additional
reactions, which would consequently be expected to affect
the mechanism. Of special interest would be the formation
of peroxide and hydroxyl radicals.
SUMMATION
• The energy supplied to the corona reactor is not sufficient
to break all the bonds in an aromatic hydrocarbon molecule.
• The corona reactor provides energy to initiate a reaction of
the toluene molecule. Once initiated, the reaction proceeds
to completion.
9-13
-------
« Once initiated the reaction of the toluene molecule proceeds
as a low temperature reaction with little or no effect upon
other toluene molecules. The reaction is unlike combustion
in which th© bulk gas temperature rises to a sufficient
level to initiate oxidation of other molecules.
• The most probable initiation reaction for toluene is attack
of the methyl group by an excited oxygen. Other possible
reactions require higher initiation energy.
• The low temperature of the corona destruction process favors
the formation of CO over C02.
• If the primary mechanism is valid, other VOCs having
different molecular structure may react differently from
toluene and the CO2/CO ratio may vary.
• Halogenated compounds require energy to strip the halogens
from the molecule for destruction of the molecule. The fate
of the by-products of the destruction requires additional
study.
REFERENCES
1. Yamamoto, T., Lawless, P. A., and Sparks, L. E. Narrow-Gap
Point-to-Plane Corona with High Velocity Flows. In;
Proceedings of IEEE-IAS Annual Meeting. Denver, Colorado,
September 1986.
2. Yamamoto, T., Lawless, P. A., and Sparks, L. E. Triangle-
Shaped DC Corona Discharge Device for Molecular
Decomposition. In: Proceedings of IEEE-IAS Annual Meeting.
Atlanta, GA, October 1987.
3. Masuda, S., Wu, Y., Urabe, T., and Ono, Y., Pulse Corona
Induced Plasma Chemical Process for DeNox, DeSox and
Mercury Vapour Control of Combustion Gas. In; Proceedings
of the Third International Conference on Electrostatic
Precipitation. Abano, Italy, October 1987,
4. Mizuno, A., and Ito, H., An Electrostatic Precipitator Using
Packed Ferroelectric Pellet-Layer for Dust Collection, In;
Proceedings of the Third International Conference on
Electrostatic Precipitation. Abano, Italy, October 1987.
5. White, Harry J., Industrial. Electrical Precipitation.
Addison- Wesley Publishing Co., Reading, MA, 1963.
6. Pimentel, G. C., and Spratley, R. D., Chemical Bonding
Clarified Through Quantum Mechanics, Holden-Day, Inc., San
Francisco, CA, January 1970.
9-14
-------
7. McDonald, J. R., Mosley, R. B., and Sparks, L. E., An
Approach for Describing Electrical Characteristics of
Precipitated Dust Layers, JAPCA. Vol. 30, No. 4, April
1980.
8. Miller, J. A., and Fisk, 6. A., Combustion Chemistry,
Chemical and Engineering News, Vol. 65, August 31, 1987.
9. Heinsohn, R. J., and Becker, P. M., Effects of Electric
Fields on Flames. In: Combustion Technology; Some Modern
Developments. Academic Press, New York, NY, 1974.
10. Finlayson-Pitts, B. J., and Pitts, J. N., Atmospheric
Chemistry; Fundamentals and Experimental Techniques. John
Wiley and Sons, New York, NY, 1986.
9-15
-------
APPLICATION OF CORONA-INDUCED PLASMA REACTORS TO
DECOMPOSITION OF VOLATILE ORGANIC COMPOUNDS
T. Yamamoto
P. A. Lawless
K. Ramanathan
D. S. Ensor
Research Triangle Institute
P.O. Box 12194
Research Triangle Park, North Carolina 27709
G. H. Ramsey
N. Plaks
U.S. Environmental Protection Agency
A1r and Energy Engineering Research Laboratory
Research Triangle Park, North Carolina 27711
ABSTRACT
Two laboratory-scale plasma reactors, an alternating current (ac) energized
ferroelectric (h1gh-d1electr1c ceramic) packed bed reactor, and a nanosecond pulsed
corona reactor, were constructed. This study was conducted to develop baseline
engineering data to demonstrate the feasibility of application of plasma reactors
for decomposing three volatile organic compounds (VOCs).
Destruction of toluene at 100 percent was obtained using two plasma reactors.
Conversions of methylene chloride at 95 percent and trlchlorotrlfluoroethane (known
as CFC-113) at 67 percent were achieved. The conversions were dependent on the
electron energy of the reactor and were also related to how strongly halogen
species were bonded with carbon.
10-1
-------
APPLICATION OF CORONA-INDUCED PLASMA REACTORS TO
DECOMPOSITION OF VOLATILE ORGANIC COMPOUNDS
INTRODUCTION
Controlling volatile organic compounds (VOCs) and various chlorofluorocarbons
(CFCs) is of increasing interest because of global environmental problems.
However, the destruction of CFCs has been regarded as especially difficult because
of their inertness and chemical stability. Besides, the products of CFC
destruction—in particular fluorine species such as HF, F2, and chlorine compounds,
or any other intermediates—are difficult to treat because of their corrosiveness.
The conventional technologies for controlling VOCs and CFCs are carbon adsorption,
catalytic oxidation, and thermal incineration. However, these technologies have
associated problems in some applications such as cost, energy requirements, or the
necessity for subsequently destroying the collected pollutant. For these reasons,
a novel technology using pulse-induced plasma discharge devices was investigated as
an alternative technical approach to control these pollutants.
Recently, plasma chemical processes have been investigated for promoting oxidation,
enhancing molecular dissociation, or producing free radicals to enhance a chemical
reaction. Plasma processes have been carried out either in a high-temperature
environment or at reduced pressure to provide higher electron and lower gas
temperature glow discharge plasmas. A typical reactor is the silent corona
reactor, which 1s energized with an alternating current (ac) ranging up to 40 kHz
at reduced pressure [1,2,3].
For atmospheric pressure applications, a narrow-gap, direct current (dc) corona
discharge device with polnt-to-plane [4,5] and triangle-shaped geometry [6] was
developed that would decompose the simulant dimethyl methyl phosphonate (DMMP).
The power requirements for this device were considerably less than those for the
silent corona device. One limitation of a dc corona reactor is that the effective
electrode distance is less than approximately 1.0 cm.
10-2
-------
A pulsed corona plasma reactor was used to produce streamer corona at atmospheric
pressure and temperature. A dc power supply was used to charge a capacitor that
was discharged with a rotating spark gap. The pulse formed by the discharge had a
rise time on the order of nanoseconds. Streamer corona discharge energized by a
fast rising time pulse voltage can produce an intensive plasma, which effectively
promotes gas-phase chemical reactions. The applications were ozone generation, or
decomposition of nitrogen monoxide, sulfur dioxide, and mercury from combustion
gas. The technology 1s called the pulse-Induced plasma chemical process (PPCP),
initially developed by Masuda et al. [7].
Other work showed that corona and high electric fields could be developed in the
packed bed reactor for particle collection [8]. This reactor employed an ac power
supply, which produced high-energy free electrons between the pellets.
In the present study, two laboratory-scale plasma reactors, one a ferroelectric
packed bed reactor and the other a nanosecond pulsed corona reactor, were
constructed. This study was to develop baseline engineering data on the
application of these plasma reactors to the destruction of various VOCs. Gas
retention time, concentration, corona power, and moisture content were varied to
determine their effects on destruction efficiency.
EXPERIMENTAL SYSTEMS
The packed bed reactor employed an ac power supply in conjunction with a tubular
reactor packed with a ferroelectric (high-dielectric ceramic) pellet layer. The
BaTi03 pellets, 1, 3, or 5 mm in diameter, were held within the tube arrangement by
two metal mesh electrodes (2.5 cm separation) connected to a high-voltage acpower
supply as shown in Figure 1.
When external ac voltage (60 Hz) was applied across the high-dielectric layer, the
BaTi03 pellets (dielectric constant of 5,000) were polarized, and an intense
electric field was formed around each pellet contact point, resulting in partial
discharge. As the applied ac vultage Increased beyond the corona onset voltage,
increase of plasma activity was visible as a light activity. The reactor was
filled with high-energy free electrons every half cycle. Emission spectra of this
packed bed reactor showed the presence of Ti and Ba ions. These emissions may
result from local heating near the contact point where electron bombardment and
electron temperature are excessively high.
observation of the current wave form with an oscilloscope showed that the current
consisted of the primary ac current and pulses with duration on the order of
10-3
-------
nanoseconds, which is equivalent to the time required for an electron to travel
between the pellets. These pulsing wave forms started to occur at the corona onset
voltage and disappeared immediately after the ac peak voltage due to the space-
charge effect. A similar occurrence was observed by Mizuno, Ito, and Yoshida [8].
The pulsed-corona reactor consisted of a wire-in-center where the high pulsed
voltage was applied, and a grounded stainless steel tube fitted onto the glass
tube, as shown in Figure 2. The wire-to-cylinder distance was 8.6 mm, and the
effective wire length was 57 mm. The pulsed reactor employed a positive dc power
supply and a rotating spark gap that was altered to produce a short pulse length
with an extremely fast rise time; i.e., nanosecond-order pulse. The detailed
operation of the pulsed-corona reactor is described elsewhere [9].
A sharp-rise pulsed corona produces streamer coronas, which has the advantage of
generating free electrons while producing a limited number of ions. Electrons can
be intensely accelerated without raising ion and gas temperature in such a short
pulse field. The electric field can be raised to an extremely high level without
causing sparking. The streamer coronas are further intensified by the rising rate
of applied pulse voltage because of very small space-charge suppression.
A series of experiments was conducted to determine the destruction efficiency of
toluene (CsHsC^), methylene chloride (CH2C12), and trichlorotrifluoroethane
(C2C13F3), known as CFC-113, using a packed bed plasma reactor and a nanosecond
pulsed corona reactor. Destruction efficiencies were based on reactant mass
balance as determined by a gas chromatography-flame ionization detector (GC-FID).
A schematic of the experimental and analytical system is shown in Figure 3. Gas
flow rate, concentration, reactor operating conditions (voltage, pulse repetition
rate), and moisture content were varied to obtain the reactor characteristics. Gas
chromatography was used for evaluation of destruction efficiencies and analysis of
reactant conversion for each VOC.
RESULTS AND DISCUSSION
Destruction of Toluene Using a Packed Bed Reactor
Relatively reactive hydrocarbons, such as toluene, a major component of VOC
emissions, were used. Figure 4 shows the voltage-dependent destruction efficiency
for toluene when the flow rate was set at a higher rate (800 cm3/min). The effects
of pellet size and toluene concentration are plotted.
Of the three pellet sizes (1, 3, and 5 mm) tested, the 1 mm pellets attained the
highest voltage (18 kV) and lowest current, resulting 1n significantly higher
10-4
-------
destruction efficiency. This high conversion may be attributed to the attainment
of much higher sparking voltage and to the achievement of more uniform plasma
activities throughout the reactor volume. This was confirmed with a high-
sensitivity camera. Note that higher efficiencies were obtained with the lower
concentration (57-60 ppm) than with the higher concentration (229-235 ppm) for all
pellet sizes tested.
Figure 5 shows the destruction efficiency for toluene when the flow rate was lower
(200 cm3/m1n). Almost complete destruction was attained with smaller pellet size
and lower operating voltages. The trends showed that the smaller the pellet size
and the lower the concentration, the higher the destruction efficiency. It is
clear that flow rate (or the residence time) has a significant effect on
destruction efficiency.
It is also evident that toluene decomposed mostly Into C02, H^O, and CO and no
unreacted fragments were identified. A large percent of hydrocarbon was decomposed
to CO (approximately 700 ppm) for 229-235 ppm toluene concentration, and a very
small fraction of CC>2 was formed only near the sparking potential for all pellet
sizes. CO concentration was proportionally lower for lower toluene concentration.
When the flow rate was higher (800 cm^/mln) and concentration was high, CO
concentration became pellet-size dependent—higher CO concentration resulted
fromsmaller pellet sizes. The produce analysis 1s incomplete at this stage of the
study, and hence a good carbon balance cannot be obtained.
When the pellets were replaced with 3-mm diameter glass spheres (dielectric
constant of 4), the nanosecond pulsed current was no longer observed, resulting in
no destruction of toluene. This Implies that the molecular decomposition is
strongly associated with electron activity.
Destruction of Toluene Using a Pulsed Corona Reactor
Figure 6 shows the destruction efficiency of toluene (50 ppm) plotted against pulse
repetition rate 1n pulses per second (pps) when the residence time was 1.5 s. When
the applied voltage was 22.0 kV, 100-percent conversion was obtained with 110 pps
or more. The efficiency sharply dropped as pps and voltage were further decreased.
Ozone concentrations were measured with 50 ppm of toluene and were as high as 220
ppm. In general, ozone concentration increased with increased applied voltage and
pulse repetition rate.
Figure 7 shows the effect of residence time on conversion efficiency for toluene as
a function of pulse repetition rate. Beyond 2.5 s residence time, complete
10-5
-------
conversion was achieved regardless of pps and pulse voltage above 18 kV. Pulse
repetition rate and pulse voltage became crucial factors when the residence time
was shorter than approximately 2.5 s. This implies that the conversion 1s related
to pulse Input energy or pulse rise time.
Figure 8 shows the effect of moisture on the destruction of toluene. The
concentration and residence time were set at 50 ppm and 1.4 s, respectively.
Adding 1.0 weight percent of water reduced the conversion efficiency significantly
in the whole range of pulse repetition rates.
Destruction of Methylene Chloride Using a Packed Bed Reactor
The chlorine 1n methylene chloride (CH2C12) is known to bond strongly with carbon.
It is more stable chemically than toluene and, therefore, higher electron energies
are expected to decompose methylene chloride.
Figure 9 shows the destruction efficiency for an ac-energized ferroelectric packed
bed plasma reactor with a pellet size of 3 mm. The destruction efficiency was
plotted as a function of flow rate (100, 200, and 500 cm-Vmin) when the concen-
tration was maintained at 500 ppm. When the flow rate was set at 200 cm3/min, the
effect of concentration was Investigated. Maximum destruction attained was 61
percent, and the conversion decreased with increased flow rate (decreasing
residence time) and higher concentration. A higher destruction efficiency was
obtained when the smaller pellet size (1 mm diameter) was used.
Destruction of Methylene Chloride Using a Pulsed Corona Reactor
Figure 10 shows the destruction efficiency for the pulsed corona reactor.
Methylene chloride conversion was plotted against pulse repetition rate for two
voltage and concentration levels when the residence time was set at 7.9 s. A
maximum of 95 percent conversion was achieved with higher pulse repetition rate and
pulsed voltage. The effect of concentration was not significant for higher
operating voltages (22.0 kV). but much higher conversion was obtained with lower
concentration for lower operating voltages (18 kV). The electron energies were not
sufficiently high to yield significant conversion when operating with lower pulsed
voltages and pulse repetition rates.
Figure 11 shows the effect of the residence time on the conversion of methylene
chloride. As the residence time and pulse voltage Increased, the methylene
chloride conversion increased. The drop in conversion at 18 kV, 1,000 ppm, and
7.9 s residence time cannot readily be explained at this time.
The reaction rate per pulse of energy input 1s plotted 1n Figure 12 for the
residence time of 1.5 s. The reaction rate per pulse Increased with decreased
pulse rate and pulse voltage. If the reaction rate per pulse were divided by the
10-6
-------
concentration, two lines for a given voltage would approach each other, resulting
of one line. In other words, the reaction rate for destroying 1 ppm of methylene
cnlorlde would depend on only two parameters (pulse voltage and pulse repetition
rate). Note that this argument 1s no longer valid when destruction reaches 100
percent. The data for the reaction rate will be useful for analyzing and designing
the reactor.
Destruction of CFC-113 Using a Packed Bed Reactor
Figure 13 shows the destruction efficiency of CFC-113 as a function of flow rate,
concentration, and applied ac voltage for a pellet 3 mm in diameter. The maximum
efficiency achieved was 28 percent with flow rate of 200 cm3/min and concentration
of 500 ppm. This was predictable because chlorine and fluorine in CFC-113 are more
strongly bonded and more stable than in methylene chloride. Obviously, it is more
difficult to decompose CFC-113 than methylene chloride and toluene.
When the smaller pellet (1 mm in diameter) was used, a higher operating ac voltage
(24 kV) was obtained, which resulted in much higher destruction efficiency (52
percent). Also, the ozone concentration was as high as 180 ppm.
Destruction of CFC-113 Using a Pulsed Corona Reactor
Figure 14 shows the destruction efficiency for two different concentration levels
(500 and 1,000 ppm) of CFC-113 when the retention time was set at 7.9 s. The
destruction efficiency Increased with Increased pulse repetition rate for all
operating cases. Again, the lower conversion achieved for 18 kV, 1,000 ppm, and
167 pps was not explainable. Higher conversion was achieved with higher pulse
voltage. The maximum efficiency obtained was 67 percent with 500 ppm and 7.9 s of
residence time. As seen with other VOCs, higher destruction was obtained with
lower concentration.
Figure 15 shows the effect of residence time on CFC-113 conversion. Although a
maximum efficiency of only 67 percent was achieved, it is believed that much higher
efficiencies could be attained by optimizing the power supply with higher voltage
and pulse rate. The present pulser is limited by pulse repetition rate and power
supply voltage. Apparently, much higher electron energies are required to achieve
Improved conversion. A black powder-Uke deposit and a high-molecular-weight com-
pound like tar were deposited on the reactor electrode and surface. However, the
reaction products have not yet been identified.
Figure 16 shows the effect of moisture on the destruction of CFC-113. Here, the
CFC-113 concentration was maintained at 500 ppm, and the residence time was set at
7.9 s. The difference between the conversions was approximately 5.0 percent
throughout the pulse repetition when 1.0 weight percent of water was added to dry
air. The presence of vapor was detrimental to the conversion efficiency.
10-7
-------
CONCLUSIONS
Laboratory-scale plasma reactors such as the ac-energ1zed ferroelectric packed bed
reactor and the nanosecond pulsed corona reactor were developed to evaluate the
destruction efficiency of various VOCs. Destruction of 100 percent was obtained
for toluene. Conversions of 95 percent for methylene chloride and 67 percent for
CFC-113 were achieved with the present design of plasma reactors. Conversion was
dependent on the electron energies in the reactor and was also related to how
strongly halogen species were bonded with carbon.
REFERENCES
1. Masuda, S.r K. Akutsu, M. Kuroda, Y. Awatsu, and Y. Shibuya. "A Ceramic-Based
Ozonizer Using High-Frequency Discharge." In Proceedings of the IEE/IES 1985
Annual Conference, Toronto, Canada, October 1985, pp. 1353-1358.
2. Inomata, T., S. Okazaki, T. Moriwaki, and M. Suzuki. "Brief Communication:
The Application of Silent Electric Discharges to Propagating Flames,"
Combustion and Flame, 50, 1983, pp. 361-363.
3. Gilman, J. P., J. G. Birmingham, and R. R. Moore. "Acetonitrile as a Simulant
for Cyanide Compounds for Plasma Testing." In Proceedings of the 1985
Scientific Conference on Chemical Defense Research, CRDC-SP-86007, 1986,
p. 435.
4. Castle, P. M., I. E. Kanter, P. K. Lee, and L. E. Kline. "Corona Glow
Detoxification Study," Westinghouse Corporation: Final Report, Contract No.
DAAA 09-82-C-5396, 1984.
5. Yamamoto, T., P. A. Lawless, and L. E. Sparks. "Narrow-Gap Point-to-Plane
Corona with High-Velocity Flows," IEEE Transactions on Industry
Applications. Vol. 24, No. 5, September/October 1988, pp. 934-939.
6. Yamamoto, T., P. A. Lawless, and L. E. Sparks. "Triangle-Shaped DC Corona
Discharge Device for Molecular Decomposition," IEEE Transactions on Industry
Applications, Vol. 25, No. 4, July/August 1989, pp. 743-749.
7. Masuda, S., Y. Wu, T. Urabe, and Y. Ono. "Pulse Corona Induced Plasma
Chemical Process for DeNOx, DeSOx, and Mercury Vapor Control of Combustion
Gas," Third International Conference on Electrostatic Precipitation, Abano,
Italy, October 1987.
8. Mizuno, A., H. Ito, and H. Yoshida. "AC Partial Discharge Characteristics of
the Electrostatic Precipitator Using a Packed Ferroelectric Pellet Layer." In
Proceedings of the 1988 Institute of Electrostatics Japan. October 1988,
pp. 337-340.'
9. Yamamoto, T., K. Ramanathan, P. A. Lawless, D. S. Ensor, J. R. Newsome, N.
Plaks, G. H. Ramsey, C. A. Vogel, and L. Hamel. "Control of Volatile Organic
Compounds by an AC Energized Ferroelectric Pellet Reactor and a Pulsed Corona
Reactor." In Proceedings of the IEEE-IAS Annual Meeting. Library of Congress
No. 80-640527, San Diego, California, 1989, pp. 2175-2179.
10-8
-------
AC Power Supply
Teflon
High Dielectric Fine Mesh Screen
Pellets
Glass Tube
\
Pulsed High Voltage T~ 1
Corona Wire O-Ring
Teflon
Stainless Steel Tube
Figure 1. Schematic of ac packed bed plasma
reactor.
Figure 2. Schematic of pulsed corona plasma
reactor.
100
80
GC Cam«< Gis
Pfl: T«o-tug« Prattur* BwjjUKx
FC: CMIwOTial Flo* Conm>0*r
CT: Cortroi Thannocoupit
BFUV: Soap Bubbto Flow M««r and Von
GC: Gas Phfnfliumjripfl
FID: Flanw-lonization Oetaoor
QMS: Quadwpoto Mm SpedronMtr
Figure 3. Apparatus schematic: corona destruction studies.
IS
60
. « 40
O yj
£ 20
800 cm 3 /min
•^H 60 ppm, 1 mm
-O- 229 ppm, 1 mm
-*- 59 ppm, 3 mm
-^- 231 ppm. 3 mm
••• 57 ppm, 5 mm
^D— 235 ppm. 5 mm
100
6 8 10 12
ac Voltage (kV)
16
57 ppm, 1 mm
237 ppm, 1 mm
231 ppm, 3 mm
59 ppm, 5 mm
234 ppm. 5 mm
6 8 10
ac Voltage (kV)
Figure 4. Toluene destruction efficiency for a flow Figure 5. Toluene destruction efficiency for a flow
rate of 800 cnvVmin. rate of 200 cm3/min.
10-9
-------
100
c
o
0
E
(f>
Q
o
c
o
"o
100
80
5?
~ 60
2*
c
£
= 40
UJ
20
0
- "7
- /
/
/
/
/
/
- ./
u
—
50 ppm, 18 kV
•<^ 167 pps
— •*>• 110 pps
-O> 33.3 pps
I I I I I I I
10 100 1000
Pulse Repetition Rate (pps)
Figure 6. Toluene destruction efficiency for
pulsed corona reactor.
246
Residence Time (s)
Figure 7. Effect of residence time on toluene
destruction as a function of pulse energy.
18 kV. no humidity
18 kV. 1 wt.% watar
22 kV, no humidity
22 kV. 1 wt.% water
100 1000
Pulse Repetition Rate (pps)
100
80
— 60
o
O
40
20
B 500 ppm, 100 cm3/min
«O» 500 ppm, 200 cm3/mm
HZH* 500 ppm, 500 cm3/min
^^» 1000 ppm, 200 cm3Tnin
3 mm barium tttanate pellets
5 10 15
Voltage (kV)
20
Figure 8. Effect of humidity on toluene destruc •
tion for pulsed corona reactor.
Figure 9. Methylene chloride destruction efficiency
for ac energized ferroelectric packed bed
plasma reactor.
100
80
60
40
20
18 kV. 500 ppm
18 kV, 1000 ppm
22 kV, 500 ppm
22 kV, 1000 ppm
t = 7.9 s
10 100
Pulse Repetition Rate (pps)
1000
18 kV. 500 ppm
18 kV. 1000 ppm
kV. 500 ppm
22 kV. 1000 ppm
246
Residence Time (s)
Figure 10. MeC!2 destruction efficiency for pulsed
corona reactor.
Figure 11. Effect of residence time on MeCI2
destruction as a function of pulse energy.
10-10
-------
10"
- io •
2 10'
I 10
5
10
18 kV. 500 ppm
18 kV. 1000 ppm
22 kV. 500 ppm
22 kV. 1000 ppm
1.5 s
10 100 1000
Pulse Repetition Rate (pps)
Figure 12. Destruction per pulse for methylene
chloride.
100
90
80
70
60
50
40
30
20
10
0
500 ppm, 100 cm3/mln
500 ppm. 200 cm3 'min
500 ppn .
500 ppm. 500 cm3 mm
-1000 ppm. 200 cm3'mm
ellels
— 3 mm barium tltanale pell
5 10 15
Voltage (kV)
20
Figure 13. CFC-113 destruction efficiency for 3 mm
barium titanate pellets.
100
80
60
" .o
O
40
20
- I = 7.9 s
• 18 kV. 500 ppm
• 18 kV, 1000 ppm
• 22 kV. 500 ppm
• 22 kV, 1000 ppm
10 100 1000
Pulse Repetition Rate (pps)
18 kV, 500 ppm
18 kV, 1000 ppm
kV, 500 ppm
22 kV, 1000 ppm
246
Residence Time (s)
Figure 14. CFC-113 destruction efficiency for
pulsed corona reactor.
Figure 15. Effect of residence time on CFC-113
destruction as a function of pulse
energy.
100
so
§ 40
o
20
18 kV, no humidity
•O 18 kV, 1 wt* water
•HH- 22 kV. no humidity
22 kV. 1 wt.% water
500 ppm, t = 7.9 s
10 100 1000
Pulse Repetition Rate (pps)
Figure 16. Effect of water on CFC-113 destruction
efficiency.
10-11
-------
REQUIREMENTS FOR A
PRECIPITATOR PERFORMANCE EXPERT SYSTEM
John G. Musgrove
Bechtel Power Corporation
3000 Post Oak Boulevard
Houston, Texas 77252-2166
Robert L. Jeffcoat
Southern Research Institute
2000 Ninth Avenue South
Birmingham, Alabama 35253-5305
ABSTRACT
An electrostatic precipitator performance expert system (PPES) can be valuable
to plant personnel in the diagnosis and remediation of deficient precipitator
operation. The peculiarities of ESP troubleshooting impose certain special
requirements on the design of the system, over normal good practice in software
and knowledge engineering. In particular, PPES must blend nonprocedural (e.g.,
rule-based) and procedural (conventional) program elements. It must permit
intelligent decisions based on complex information such as V-I curves; preserve
the option for continual inclusion of improved models and other information; and
be capable of acquiring and processing information in real time. The EPRIGEMS
interface specification is applicable to this task, as are several currently
available expert system shells. Economic considerations will also influence the
selection of the final design approach.
11-1
-------
REQUIREMENTS FOR A
PRECIPITATOR PERFORMANCE EXPERT SYSTEM
BACKGROUND AND NEED
The Electric Power Research Institute (EPRI) has supported considerable research
into operating electrostatic precipitators (ESPs) for fly ash collection.
During the 1980s, EPRI also began development of expert systems to provide power
plant operators a better way to access the substantial body of research and
apply it to their operations. This paper explores how expert system methods can
be effectively brought to bear on ESP operation and maintenance.
The advent of inexpensive micro- and minicomputers has permitted vendor-designed
control systems and their interfaces to advance in ease of use considerably in
recent years. Unfortunately, each vendor has pursued his own approach to data
collection, storage, and presentation. The control features to assist
unattended operation have advanced beyond spark rate control to include spark
squelching and voltage recovery, energy management controls, and intermittent
energization.
The availability of powerful software tools now permits the development of
user-friendly programs to exploit the knowledge available from research and
industry experience. These programs can present the information in a manner
suited to the task, "context sensitivity". They can also assist and direct
precipitator operators in the diagnosis of both short- and long-term operating
problems. An electrostatic precipitator performance expert system is an
appropriate, achievable, and cost-effective goal. The design must be based,
though, on a thorough understanding of ESP operation. This paper discusses some
of the issues that must be addressed during this development.
PROBLEM DIAGNOSIS
An effective precipitator performance expert system (PPES) will both guide and
assist plant personnel confronted by a suspected ESP problem. The user will be
led through an orderly sequence of measurements, analyses, and tests to identify
and correct the problem. The art and science of ESP troubleshooting have been
formalized to some degree, and Reference 1 provides a good starting point.
However, the required computer system should aim to be more than an "on-line
manual." In particular, an expert system methodology must be chosen which
permits analytical models, historical data bases, on-line data acquisition, and
rule-based reasoning to be strongly integrated into a single diagnostic tool.
It is convenient to divide the troubleshooting process into four phases, each of
which has distinct characteristics which drive the selection of expert system
methods. These phases are: identification (monitoring), isolation (diagnosis),
assessment (analysis), and remediation (advisory).
11-2
-------
Identification
The first step is to ascertain whether a problem exists which is attributable
to, or correctable by, the electrostatic precipitator. Conventionally, such a
problem is detected by increased opacity, obvious operating problems (such as
excessive failures in the electrical system), or a regulatory citation. An
on-line, properly instrumented expert system can provide the monitoring
capability to pick up some of these problems. Perhaps more importantly, it can
continually compare present operation with historical and theoretical data to
detect patterns indicating subtle, incipient, or slowly progressing
malfunctions. In this phase, PPES is functioning as a monitoring expert system.
To accomplish this function, the PPES needs the capability to acquire, log, and
interpret operating data. It is technically feasible to link the expert system
directly to process control instrumentation, allowing it to function as an
intelligent data logger. Such capability may well be provided in future ESP
designs, either integral to the controller or via a standardized interface.
However, the same purpose can also be served by manually entered data; the PPES
would prompt the user to measure and enter required data according to a
specified schedule.
It is a straightforward matter to record operating data for later recall. To
interpret this information to identify an ESP problem is, on the other hand, a
complex assignment. Aberrations in simple variables like opacity, spark rate,
and peak secondary voltage can probably be automatically interpreted. An abrupt
change or inflection would indicate an acute problem, whereas a secular trend
would simply be noted for eventual maintenance action. The V-I curve, by
contrast, is rich in information content and difficult to interpret. An
appropriate parameterization of the curve would allow the individual parameters
to be treated as simple variables, and monitored conventionally.
The end of the identification phase is the determination that something is
wrong, together with a cursory list of symptoms. In the next step, the problem
will be traced to its source.
Isolation
Once a problem has been determined to exist, the process of isolation can
begin. The PPES would act as a diagnostic expert system. It would collect
information on the ESP and progressively identifying the most likely cause or
causes of the observed behavior.
The ESP troubleshooting procedure is a complex blend of rules, checklists,
observations, quantitative analyses, and judgment calls. PPES will capture this
method as completely as possible, still prompting the human user for
observations, measurements, and assessments which are beyond the computer
programs skill. As in any efficient diagnostic system, the PPES search strategy
is a compromise among three approaches. These are: acuity (collecting the data
or asking the question which, at each step, will most effectively discriminate
among alternate hypotheses); rapidity (considering the most probable diagnoses
first); and cost (performing the simplest and cheapest diagnostic test first).
Various alternatives for diagnostic heuristics will need to be explored; a
cost-weighted Bayesian decision metric appears promising. For optimal
performance, the rules and metrics can be tuned to the particular ESP
installation on the basis of design and historical information. For example,
11-3
-------
electrode breakage is a much more common problem in weighted-wire ESPs than in
rigid electrode systems. Similarly, a cold-side ESP downstream of a
Ljundstrom-type air preheater is prone to cross-duct temperature nonunifonnity.
The diagnostic sequence typically begins with a survey of electrical conditions,
starting with qualitative phenomena (excessive sparking, abnormal tripping,
extreme fluctuations in voltage or current, etc.). The voltage-current (V-I)
curves may then be examined as described below. In this process it is important
to distinguish between generalized and localized abnormalities: generalized
problems tend to rule out isolated equipment failures (such as defective
rappers, misalignment, over-full hoppers, or isolated T/R defects), and
represent a strong diagnostic means of descriminating among alternatives.
The most useful single diagnostic measurement is the V-I cxirve. PPES must
provide the ability to store, recall, manipulate, and analyze such curves. This
represents a significant design challenge, and it may be necessary to progress
from human to machine interpretation in phases. The V-I curves contain
information which can be interpreted at several levels to yield diagnostic
information. An actual operating curve can be compared to as-built (or
as-overhauled) curves, operating and air-loaded, to gain quantitative
information on performance. The V-I curves of successive sections (and, if
available, of individual cross-duct zones) can be compared to isolate localized
failures such as shorts, broken wires, and bad rappers. They can also be
compared for nonuniform flow, temperature, and mass loading. The details of
curve shape are indicative of particle resistivity, mass loading, temperature,
and general alignment, as well as other factors.
As an integral part of the isolation process, the user will be prompted for
qualitative information on the condition of the precipitator and the plant
operation. For example, he may be asked to conduct a visual check of certain
rappers, or to determine whether there has been a change in fuel supply. The
solution strategy must consider the relative difficulty and execution time of
various measurements. The inspection of external equipment is relatively quick
and inexpensive, laboratory tests are time-consuming, and internal inspection
must wait for a plant shutdown.
The isolation process, up to this point, consists of both categorical and
quantitative comparisons of actual against nominal conditions. Often this is
tantamount to diagnosis, as for a defective rapper or a shorted insulator. In
other situations, the isolated effect may have more than one possible root
cause. To identify the underlying problem, an additional step of assessment is
needed.
Assessment
Mathematical models of ESP performance can be invoked to help isolate and
quantify the phenomena responsible for degraded operation. These models are
procedural, computationally intense, and complex; they have no inherent "expert
system" character. It would be unwise to impose a new and probably inappropriate
formalism on them. However, their analytical power is essential to PPES. In
turn, the special characteristics of PPES can make existing models much more
accessible and valuable to the average user. At this stage, PPES behaves as an
integrated system of analytical routines.
11-4
-------
Four existing models which are fundamental to precipitator performance
estimation are:
• ESP Performance Model (Ref. 2) This model derives ESP collection
efficiency by using the Deutsch equation. Calculations are made on
narrow particle size bands over short increments of length to track
changing conditions within the precipitator. The model allows the
user to choose either fast estimations, or slower and more accurate
rigorous calculations, of the collecting fields and particle charges.
The model has been thoroughly validated.
• ESP Operating Point Correlation (Ref. 3) This model derives V-I curves
and electrical setpoints for a five-field ESP. The are based on a
correlation of seventeen pulverized-coal utility ESPs. The required
input data are ash resistivity and duct width.
• Ash Resistivity Model (Ref. 4) Version 2 of Roy Bickelhaupt's
resistivity model will be used. This model determines resistivity
from ash composition. The effects of acid in the flue gas are
incorporated.
• Particle Size Distribution Model (Ref. 3) This model correlates the
size distribution of the suspended ash particles to the coal and
boiler types.
These and similar models are typically implemented as large programs in FORTRAN
or BASIC. Although of proven value in system design and analysis, the
acceptance of these models for routine troubleshooting has been limited. Their
limited use is due to the complexity of the input data required and the awkward
formats in which they must be supplied.
PPES can facilitate the use of analytical models by serving as an intelligent
front-end processor. Among the functions provided by the expert system would
be:
• Prompting for required data.
• Conversion from user-supplied input to the parameters and units
required by the model.
• Providing appropriate default values for unknown variables.
• Checking input data for consistency and validity
• Updating, and extracting data from, archives.
• Abstracting and displaying results.
In troubleshooting, as in design, parametric sensitivity data are more valuable
than point analyses. PPES can be especially useful in helping the user
organize, carry out, and interpret this kind of multi-case study. Because each
point run of a complex model may be time-consuming, some ingenuity will be
required to develop an efficient parametric variation scheme.
11-5
-------
Remediation
The objective of PPES is to help the user correct ESP problems rapidly and
efficiently. Once a diagnosis has been made, a list of actions should be
presented to the operator. Sequential actions may be interspersed with
questions to verify that the expected intermediate result has been achieved.
Multiple options for correction should be prioritized according to difficulty
and probability of success.
One useful function of a PPES is as a tool for planning shutdown activities.
Known or probable internal failures which were isolated in previous phases can
be given high priority for correction. Suspected problems, which can be ruled
out or in by inspection, might be ranked next. Routine maintenance and
inspection, along with possible sources of slow deterioration, would come next.
Finally, calibration testa which can only be done during shutdown (e.g.,
air-loaded V-I curves) can be scheduled.
PPES may be capable of detecting limitations in the inherent design of the ESP,
such as inadequate secondary voltage or non-ideal flow patterns. This
information should be made available to engineering personnel, together with an
explanation of the underlying reasoning, for inclusion in plant upgrade plans.
SYSTEM ARCHITECTURE
Based on the foregoing discussion, it is possible to sketch minimal requirements
for a PPES architecture.
User Interface
The user interface must be easy to use and rapidly learned. The EPRIGEMS
specification (Ref. 5) provides a methodology for developing a common interface
suited to many microcomputer-based applications. Use of this interface should
expedite the operators' access to the necessary information by functioning
similarly to other EPRIGEMS applications in the power plant. Figure 1 provides
a potential structure for menus using the EPRIGEMS format.
The user interface should eliminate the tedious, error-prone, and unnecessary
requirement to re-enter the same data for different applications. In
particular, the entry of static data should be a one-time event. Data sharing
among all modules associated with the expert system assures that conclusions are
drawn on the basis of internally consistent information.
Specifically, data sharing means that a common set of site-specific data would
be accessed by all of the analytical models. If a specific data item was not
available, there should be the option to call and use data from a similar fuel
or boiler. Access to an industry-wide data base, or its distribution with the
program, could be the basis of a suitable set of default values. These defaults
might include coal, ash, and particle size analyses for six to twelve types of
coals and pertinent information for four to six boiler types.
11-6
-------
Procedural and Nonprocedural Aspects
PPES will include a significant element of procedural computation (i.e.,
operations carried out in a well-defined sequential fashion). Existing
performance models are exclusively in this class, and much of the signal
analysis underlying performance monitoring can be efficiently expressed in
procedural terms. On the other hand, the diagnostic and planning tasks may be
best implemented in some nonprocedural paradigm.
For these reasons, PPES will require a flexible and efficient means to mix
procedural and nonprocedural modules. A multi-tasking environment can provide a
way to communicate between disparate programs (between an expert system shell,
for instance, and a FORTRAN resistivity model). This approach is preferable to
embedding the procedural operations within an essentially nonprocedural shell.
To embed the procedural operation would entail a major investment in
translation, compromise execution efficiency. It would likely sacrifice system
flexibility for enhancements.
An implicit long-term goal is to bring PPES on-line, minimizing the need for
operator entry of routine data. This means that the system must accommodate
real-time data acquisition and monitoring. Real-time expert systems are
notoriously difficult to implement; nevertheless, starting with a multi-tasking
operating system will facilitate the eventual transition to on-line operation.
Knowledge Representation
For expert system analysis and inferencing, one or more knowledge bases will
need to be developed. While many expert system paradigms exist, two appear'
especially useful for implementing this application.
Rule Based. For knowledge bases dealing with singular' instances of equipment or
events, each unique and single-valued, a production rule system using backward
chaining could be appropriate. Backward chaining expert systems are goal-driven
and would be suitable for problem diagnosis; a hybrid forward/backward chaining
system can be especially efficient. Production rules follow the conditional
logic form "if/then/else". These kinds of systems are among the easiest to
develop and implement. Many applicable expert system shells exist, some having
a limited ability to build rules from existing data bases or spreadsheets.
Others have tool-kits to aid the knowledge acquisition and rule building
process. The limitations of these production rule shells lie in their user
interfaces, external interfaces (process-process and process-environment), and
their ability to operate on several low-cost hardware platforms.
Frame Based. For knowledge bases dealing with the multiplicity of precipitator
hardware, cells, fields, rappers, hoppers, etc., a frame-based representation
system appears appropriate. The frame is a natural way to represent multiple
elements, each of which has the same set of characteristic descriptors (but
differing values for these descriptors). The benefits of frame representation
are twofold: the ability to write a single rule covering all instances of an
element, and sub-elements of a frame can inherit the structure and
characteristics of the parent frame.
A frame for a precipitator might have data fields, or slots, for
transformer/rectifier (T/R) identity, rapper system identity, field and cell
identity, and hopper identity. The sub-frames for T./R set, cell, and hopper
11-7
-------
might have slots for primary and secondary voltage and current, discharge and
collecting electrode rapper information, and ash handling system information,
respectively. Data entered at the parent level would be inherited down to the
element level.
Shells and Executives
For rapid prototype development of the initial knowledge bases, the use of
expert system shells is recommended. Shells — the "master carpenter tools"
(Ref. 6) of expert system development — provide extensive facilities for
creating knowledge bases and designing user interfaces, along with a reliable
general-purpose inference engine. Although the user interface might not be
fully consistent with the EPRIGEMS specification at every level, a good shell
nonetheless allows rapid development of appropriate and powerful interfaces.
Moreover, the first priority is to develop and test the knowledge bases for
their functionality and utility, turning attention to the final interface once
the internal operation has been proven. Finally, some development environments
allow developed knowledge bases and inference engines to be called (used) by
other programs. Thus, an EPRIGEMS interface program module could potentially
call the knowledge bases without having to invoke and use the shell's own user-
interface.
An important consideration in the selection of a shell is whether the resulting
expert system can be economically delivered to its intended audience. The
shells themselves, which provide a wide range of development tools not needed by
the end user, tend to be costly and large. Some software products generate
stand-alone executable code which can be invoked (but not modified) by custom
programs, and for which no license fees apply. Other products make available a
relatively inexpensive run-time kernel which must accompany the delivered code.
Still others make no provision to separate the development and execution
environments. For a product which is intended for the utility industry at
large, the considerations among functionality, hardware cost, and license fees
must be taken seriously.
THE KNOWLEDGE BASE
Definition
To represent the knowledge and information about the precipitator and its
operation properly, it is likely that multiple knowledge bases will need to be
developed. This will permit the use of several different domain experts without
introducing the problems of producing conflicting knowledge. It will also
permit incremental development, one knowledge base at a time, to limit project
complexity and risk.
Individual knowledge bases could be developed for rapper system, hopper and ash
handling system, or T/R set operation, for example. These knowledge bases would
then be called as necessary from the supervisory program.
11-8
-------
Acquisition
The central idea of knowledge engineering is to capture the knowledge of experts
in a domain (electrostatic precipitators in this case), and express (or encode)
it in a form manageable by the computer. Knowledge acquisition from the domain
experts is a challenging task which presents the greatest element of project
risk. Normally, knowledge engineers interview the experts to ascertain their
diagnostic approaches, from which they structure and develop the knowledge
bases. The design/build/test development cycle, Figure 2 (Ref. 7), is
incremental and iterative. The testing cycle is useful for verification and
validation of the acquired knowledge. It is also useful as a means of prompting
the expert for more knowledge where the existing system is incomplete. While
new software tools exist to assist or automate knowledge acquisition, none can
substitute for an intensive collaboration between knowledge engineers and domain
experts.
Encoding
Knowledge encoding, as with knowledge acquisition, would be incremental. The
use of expert system shells to build and test the prototype knowledge bases
rapidly is crucial to building confidence in the development of the system and
to limit the risk of financial failure.
Maintenance
An expert system like PPES, once built and delivered, cannot be considered
static. At the least, it must be customized to each ESP installation with the
provision of design information and nominal operating data. At a slightly
higher level, the system will be self-modifying (if only parametrically) as it
collects and classifies historical information.
At the highest level, however, there will be a continuing requirement to update
PPES with new information: improved heuristics, rules applicable to new
designs, and updated mathematical performance models. All these considerations
dictate an open-ended implementation which will permit upgrades to be made
efficiently. Rule-based expert systems make incremental modifications
relatively easy, at least technically. A modular software architecture with well
defined interfaces will also contribute to maintainability
CONCLUSION
Expert systems have proven themselves in a growing number of applications,
including several in the electric power industry. It is feasible to apply this
experience to electrostatic precipitators, and a practical, cost-effective
precipitator performance expert system can be developed using available
technology. To be successful, however, it must include capabilities which are
not commonly found in today's expert systems, such as integration of large
procedural components, signal processing and signature analysis, and real-time
data acquisition. These considerations, along with the need to make the product
widely and economically accessible to the industry, will strongly influence the
design approach.
11-9
-------
REFERENCES
1. DuBard. J.L., and Nichols, G.B., Electrostatic Precipitator
Guidelines. CS-5198, vol. 3. Troubleshooting. Palo Alto, California.
Electric Power Research Institute, 1987.
2. Faulkner, M.G., and DuBard, J.L., A Mathematical Model of
Electrostatic Precipitation (Revisions). EPA-600/7-84-069a,b,c.
Research Triangle Park, N.C. U.S. Environmental Protection Agency,
1984.
3. DuBard, J.L., and Dahlin, R.S., Precipitator Performance Estimation
Procedure. CS-5040. Palo Alto, California. Electric Power Research
Institute, 1987
4. Bickelhaupt, R.E., A Study to Improve a Technique for Predicting Plv
Ash Resistivity with Emphasis on the Effect of Sulfur Trioxide. EPA
Report EPA-600/7-86-010, NTIS PB86-178126. Research Triangle Park,
N.C. U.S. Environmental Protection Agency, 1985.
5. EPRIGEMS Team. EPRIGEMS™ PRODUCT SPECIFICATIONS. Palo Alto,
California. Electric Power Research Institute, November, 1988.
6. Davis, Randall. Knowledge-Based Expert Systems: Planning and
Implementation. Reading: Addison-Wesley Publishing Company, Inc.,
1987, p. 165.
7 Barrett, Michael L., and Beerel, Annabel C. Expert Systems in
Business: A Practical Approach. Chichester: Ellis Horwood Limited,
1988, p. 137
11-10
-------
FILE
ADVISOR
VIEW
SPECIAL
TOOLS
J
Form — |
Restart
r New
- Open
- Clear
- Save
- Print
- Import
- Navigator
• Knowledge Bases
- About forms
Save — r Session
L Report
Delete -,
r Session
- Report
- Rules
- Facts
- Goals
- Glossary
- Session
- Performance
Model
- Operating Point
Correlation
- Ash Resistivity
- Update Model
- Add Program
- Delete Program
- Edit K.B.
- Particle Size
Distribution
- Report
- Other
Print — T- Session
1- Report
Help
Set Defaults —T- Computer
- Graphics
- Boiler
ter —r-
ina L
- Coal
Monitor
Printer
Storage Drive
Manufacturer
- Draft Type
- Pulverizer Type
- Steam Conditions
Coal Analysis
Ash Analysis
Particle Size Analysis
- Precipitator —r- Manufacturer
- Fields
- Gas Passages
- T/R Sets
- Rappers
- Control System
- Discharge Electrodes
Figure 1. Potential Menu Structure Under EPRIGEMS
11-11
-------
clarification
and revisions
revisions
changes and
extensions
Defining the
system
Acquiring and
structuring
the knowledge
Building the
components
Refining the
system
further
requirements
reformation
Figure 2. Expert System Development Cycle
11-12
-------
AN INTEGRATED ELECTROSTATIC PRECIPITATOR MODEL FOR
MICROCOMPUTERS
P. A. Lawless
Research Triangle Institute
P. 0. Box 12194
Research Triangle Park, NC 27709
R. F. Altman
Electric Power Research Institute
516 Franklin Building
Chattanooga, TN 37411
ABSTRACT
Computer modeling of electrostatic precipitator (ESP) performance has reached a
high level of accuracy, but existing computer models still leave much to be
desired in ease of use. The model presented here is designed to overcome the
ease-of-use problems, but still give the user the benefit of all the
precipitator-related research that has been done in the past decade. A
microcomputer program has been developed to allow the user access to models that
predict ESP performance, fly ash resistivity, particle size, and flue gas
composition and volume. The models have come out of many separate research
efforts, but have now been put together in a close-knit working relationship. The
program has been designed for ease of use with a structured menu environment.
Standard cases are supplied in the documentation to help learn the use of the
program. It runs on most configurations of IBM PC-compatible computers. To
facilitate its use in the utility industry, the model has been limited in the
types of electrodes and configurations it can support, but within these limits, it
performs its functions well.
12-1
-------
AN INTEGRATED ELECTROSTATIC PRECIPITATOR MODEL FOR
MICROCOMPUTERS
INTRODUCTION
Computer models for evaluating electrostatic precipitator (ESP) operation have
been developed over more than a decade. Many of these models (1 - 7) were written
for mainframe computers to satisfy research needs. Over the years, many of them
have been converted to run on microcomputers (8) and others (9 - 11) were
developed with easy-to-use interfaces. However, the growth of these models in
separate research organizations was not conducive to a consistent method of
presentation or to the ability to use one model's output for input to another. In
addition, documentation on the use of these models has been poor (mainly for the
mainframe versions) to non-existent. In light of the potential usefulness of an
integrated approach to these models, the Electric Power Research Institute (EPRI)
has contracted with Research Triangle Institute to produce an ESP model that
combines these parts into a single, user-friendly whole.
The composite model contains the core calculations of combustion models, ash
resistivity models, particle size distribution models and correlations, V-I models
and correlations, and up-to-date ESP collection models with non-ideal effects. It
controls the data entry with a menu-structured interface, and makes the results of
intermediate calculations available to appropriate parts as input data. The
outputs are presented in a variety of forms, both tabular and graphical.
The rest of this paper discusses the model's structure from the standpoint of
input and output, the model's algorithms in each of its parts, and the assumptions
built into the model; and it gives some results showing the accuracy with which
the model can be used.
MODEL STRUCTURE
The model structure is best described by showing the menus which form the user
interface. From them, the overall structure and interactions with the user can be
seen. The Main Menu is the principal point of interaction with the user.
12-2
-------
Main Menu
The Main Menu choices are displayed in Figure 1. Each of these choices is
selected by moving a cursor to the proper line and pressing the key, or by
simply typing the first letter of the choice. Depending on the menu item
selected, sub-menus or direct actions may be produced. For instance, the File and
Data options lead to sub-menus, while the Calculate option initiates the
performance calculations.
The Utilities sub-menu contains certain program configuration choices that can be
activated, a path describing where the data files may be found, and a temporary
exit to the operating system. These are not of major importance to the program
and will not be elaborated further.
File Menu
The data files are loaded according to a grouped structure. The purpose of the
structure is to allow, for example, a single coal mine to be used at several
plants with different ESPs. The data are divided into separate files for
plant/boiler, coal, flue gas, ash and resistivity, particles, ESP design, and
non-ideal factors. These separate files are combined by being named in a Master
File, which is used to represent the entire ESP installation. Then, whenever a
given Master File is loaded, all the associated data files are loaded with it. If
any of the data files are modified or replaced, then the Master File is also
modified.
The menu options for the file menu, Figure 2, are mainly concerned with the Master
File. A Master File may be loaded by name or updated and saved by name from this
menu. Whenever a Master File is saved, its name is also stored for later use.
For example, if the user finishes the program and exits, the next time the program
is started, the same Master File will be automatically loaded. Master Files may
be changed from this menu, by either loading a new one or saving the present one
under another name. The contents of the Master Filc may be edited to change the
file names in it and to add a few lines of comments about the file itself. In
this way, alternate configurations for an ESP might be kept in separate Master
Files, with the differences between them noted in each.
The Browse option allows the user to see the raw data in the various files. It
leads to a sub-menu giving choices of the type of data to view. When a type is
picked, a list of the files of that type will be displayed. The contents of any
file can then be displayed on the screen. This facility allows the user a quick
review of the data. This might be, for example, for the purpose of changing an
entry in the Master File.
12-3
-------
Data Input Menu
The Data Menu (Figure 3) controls the selection of data to be edited. Each of the
menu selections leads to a detailed input screen for the type of data given by the
selection name. In each of the input screens, the user has the option to load and
save files of that specific type. If this is done, the contents of the Master
File are updated concurrently to reflect the changes made. This gives the user
another method for updating the configuration in a Master File.
Figure 3 also lists the data that are asked for in each of the input menus. The
goal of the menu design is to group important or essential data at the top of
menu, with less important data at the bottom. Although all the data asked for
have some relevance in the model, some are only used to calculate intermediate
results, which can be overridden by direct inputs.
For instance, the coal composition and consumption rate can be used to calculate
the volume of flue gas produced. The user may already have measurements of flue
gas volume and not need to calculate it. Under those circumstances, composition
and consumption may be bypassed on the input screens. However, if they are
present, the program will display the calculated gas volume alongside the volume
presently being used. If the two are substantially different, the user should
investigate the reasons for the discrepancy.
Output Selections
The user is given three output selections, tabular data to the screen, tabular
data to the printer, and graphical data to the screen. If the printer is an Epson
(TM) or Hewlett-Packard (TM) compatible printer, then the graphics screens may be
printed, although any color information will be lost.
The screen tabular results are arranged in several pages, with the most important
ones first, the less important ones later. The user may interrupt the viewing at
any point in the presentation. The tabular data to the printer is arranged
similarly, but includes the names of the data files used to generate the results,
and optionally, the contents of the files, since it is assumed that printed
results will be referred to independently of the program. The pagination and
format of the printed results are organized for the different page size of a
printer. The graphs are also presented by pages, with the user allowed to break
the presentation at any time. If the printers are compatible, as mentioned
before, the Print-Screen function can be invoked to transfer the graphs to the
printer.
12-4
-------
MODEL ALGORITHMS
The models used within the program should be considered conventional, because they
have appeared in other places. The core algorithms have been extracted for use in
the program. Although not a model, the coal rank classification (12) has been
included to give the user a way to select a broad range of coal compositions. If
a rank is specified, default proximate and ultimate analyses are introduced, from
which many other properties can be calculated. This was added to make it possible
to compare the major effects of a coal change, within broad limits. Because of
the inherent variability of coals, these default values are not representative of
any particular coal source or mine.
Combustion
The combustion products are calculated on the basis of stoichiometric conversion
of the carbon, hydrogen, and sulfur in the coal into C02, H20, and S02 (12). The
flue gas contains a specified fraction of excess air and the coal is supplied at a
given burn rate. When the user supplies a temperature at the inlet of the ESP,
the volume of flue gas produced can be computed.
The SO, content of the flue gas is based upon the conversion factors of
Bickelhaupt (3). That is, the conversion factor between S02 and SO, is 0.004 if
the combined atomic concentrations of magnesium and calcium is less than 5.0
percent, and 0.001 if greater. Consequently, a predicted SO, concentration in the
flue gas may have to be modified once the ash constituents are determined. The
program takes this into account by checking the levels of S02 and SO, for
consistency with the ash. Any SO, introduced deliberately as a conditioning agent
would not be subject to this limitation, because the effect seems to be related to
some catalytic action of the ash components.
Resistivity
The Bickelhaupt resistivity algorithms (2 - 4) are used to predict resistivity
versus temperature over a range from about 200 °F to 700 °F. The specified
operating temperature is interpolated on this curve to obtain the resistivity for
that temperature; if the temperature is changed, the resistivity will change. On
the other hand, if the SO, content of the gas is changed, the whole curve will
need to be recomputed.
If the ESP is a hot-side unit subject to the sodium depletion effect, the
resistivity curve is based on the hot-side volume resistivity, but has been
extended to lower temperatures with the normal surface resistivity component.
12-5
-------
This was done to provide a smooth curve over the entire temperature range, but
some caution in its use is required. Since the sodium depleted resistivity only
develops at elevated temperatures, the lower part of the curve is not meaningful.
If a hot-side unit were converted to cold-side operation, the normal resistivity
curve would be appropriate. Only at temperatures near about 500 °F would the
combined surface and volume resistivities be more accurate than the volume
resistivity alone would be.
Particle Distribution
The particle size distribution is calculated from coal and boiler data from two
models (6,7). One model is a set of correlations for boilers burning bituminous
and sub-bituminous coals; some of the correlations depend on the firing method
used in the boiler. If the conditions for using the correlations are met, then
the correlations are used. If the correlations cannot be used, then the coal
breakup model is used to generate a particle size distribution. The breakup model
uses a mean coal particle size and a Rosin-Rammler distribution exponent for
predicting the fly ash distribution. Note that these methods do not provide for
the addition of non-coal powders, such as wet or dry sorbents. As alternatives to
these predictive methods, direct input of a cumulative size distribution from
impactor data or log-normal parameters is allowed.
V-I Curves
The performance models generally operate quite successfully when the voltage and
current operating points of the ESP are known. The electric fields needed to
calculate particle charging and collection can be derived fairly accurately from
the voltages and wire-plate spacings, even if they are not calculated numerically.
If the voltage and currents are measured, they may be input directly into the
model. If direct readings are done, it is important to exclude the effects of
back corona. The best approach is to measure a V-I curve for the section and pick
an operating point where the slope of the curve increases rapidly.
If direct measurements are not available, then a voltage-current correlation that
depends on ash resistivity may be used (5). This correlation will set the
voltages and currents slightly differently for each electrical section in the ESP,
effectively accounting for particulate space charge in the ESP- The correlation
was developed by attempting to pick currents below the onset of back corona, but,
because of the difficulty of being certain, it probably includes some back corona
current at the high resistivities. The uncertainties of the correlation for the
fifth and sixth electrical sections led us to use values only through the fourth.
12-6
-------
If unusual dust loads or electrode spacings are encountered, it is possible to
calculate clean-plate V-I curves using the methods in References 10 and 11 and to
calculate the particulate charging according to References 9 and 13. This method
makes assumptions about the onset of back corona and sparking and may give results
different from the correlation. The agreement between the V-I model and the
correlation data is excellent for resistivities above 1011 ohm-cm, but is poorer
at lower resistivities, where other factors, such as peak-to-average voltage
ratios and electrode misalignment, come into play.
ESP Collection
The collection model has three parts: particle charging, particle collection, and
loss mechanisms. The charging algorithm uses an addition of field and continuum
diffusion charging (14), but if the charge added in any time increment is more
than that delivered by the corona wires, then the charging is truncated. The
particle collection is by the mechanism of Deutsch collection, where the particle
migration velocity may use either an empirical correction (15) or a turbulent core
parameter to account for the less-than-totally turbulent flow in an ESP. The loss
mechanisms are: the velocity maldistribution, applied as a modifier to the Deutsch
collection term; the sneakage, which bypasses a fraction of the flow around the
collection zone; and the rapping reentrainment, which adds a fraction of the
collected material to the gas stream for subsequent recollection.
The total collection is calculated in two steps. First, the collection without
rapping is calculated and the collection efficiency of each section is stored.
Then, rapping 1s applied successively to each section, where a specified fraction
of the material is reentrained with a certain size distribution. The reentrained
dust is treated as if it were passing alone through the remainder of the sections,
where most is collected and some reaches the outlet. The total outlet emissions
then consist of a steady component and rapping contributions from each of the
sections.
The total emission is computed as a sum over all particle sizes. An opacity is
also calculated for a given stack diameter. Finally, a weighting of the emissions
according to the PM10 standard is applied for a PM10 emission level.
ASSUMPTIONS IN THE MODEL
Several assumptions are made in the model to simplify the data input requirements.
The coal density is assumed to be 1.3 g/cm . This only affects the calculated
particle size distribution using the breakup model; the variation of density has
12-7
-------
only a slight effect. The dust layer thickness affects the effective voltage drop
between the corona wire and dust layer surface, particularly for high resistivity
dusts. Because the thickness is usually difficult to determine and changes with
time, the factors that are affected by it use an constant value.
Ion mobility enters into the charging and V-I calculations. Since the flue gas is
assumed to contain significant amounts of SOS. an ion mobility appropriate to S02
is used (16). The effects of varying amounts of water vapor are ignored. This
does mean that V-I curves measured on an air-loaded ESP may differ substantially
from those of the model's. Although some of the V-I models allow for complex
corona electrodes, this model assumes that all the electrodes are round wires of
the same diameter and same spacing with respect to one another.
The rapping emissions treatment assumes that the calculated emissions are averages
over a long period of time, i.e., several rapping cycles. Instantaneous peaks,
such as appear on the opacity meters, are not predicted by this model, nor are
changes in the rapping cycle times. In order to predict an opacity, the optical
constants of the fly ash are assumed. The values chosen are typical of fly ash,
but an ash with appreciably different levels of carbon could have a significantly
different opacity. The constants used do give good agreement with the few mass
and opacity data that are available. That is, if the mass emissions are in good
agreement with the model, so are the opacity measurements.
RESULTS OF FITTING TEST ESPs
The model's predictions can be compared with measured results in two ways. One is
to assume values for the loss factors and see how well the predictions agree with
measurements; the other is to use the measured data to obtain the best parameter
values for future use.
The second type of test will be included in the documentation for the program.
The results of three fits will be discussed here. The data for these three fits
come from Reference 15, where the plants are identified by number.
The model allows different parameters to be evaluated, because they affect
different parts of the particle spectrum. For instance, the rapping puff mean
diameter and standard deviation can fit the distinctive hump in the penetration
curve from about 3 fim to 20 pm. This hump can be seen clearly in Figures 4 and 5,
but is less distinct in Figure 6. The sneakage fraction and the turbulent core
fraction can be adjusted to fit the fine particle hump from 0.1 pm to about 2 pm.
12-8
-------
The rapping reentrainment factor can then be adjusted to match the height of the
rapping hump. Overall, the total mass penetration and opacity must be kept in
line with the measurements.
The fits of the model to these data give the values in Table 1. It can be seen
that some of the parameters show a wide variation, while others are quite
restricted in their range. It is particularly interesting to note the narrowness
of the rapping puff (ag parameter). In (15), a value of 2.5 is given, based upon
an analysis of several sets of impactor data. It appears that the variations
between sets broadened the average, leading to a rapping distribution that is too
wide to fit the actual measurements.
The rapping fractions of the Colbert and Johnsonville model fits are also well
below the average value of 0.12 estimated for a broad range of ESPs (17). Among
other things, this would indicate that changes to the rapping system might not
have much impact on the overall emissions. On the other hand, the Naughton ESP
might show significant changes if its rapping were improved.
CONCLUSIONS
The present program represents a major effort to integrate several ESP-related
models into a single, user-friendly whole. The goals of the program are to make
the diagnosis of ESP problems easier and more accurate, and to provide a tool for
the rapid evaluation of proposed changes of ESP or boiler operation. By bringing
together the model parts and by have complete documentation, the program is
designed to serve the ESP community. The model can give rapid insight into the
operation of an ESP and point the way toward improvements in operation.
REFERENCES
1. S. Oglesby, Jr. and G. B. Nichols. Electrostatic precipitation.
New York: Marcel Dekker, 1978.
2. R. E. Bickelhaupt. 1979 A Technique for Predicting Fly Ash
Resistivity. EPA-800/7-79-204. Washington, D. C.: U. S.
Environmental Protection Agency.
3. R. E. Bickelhaupt. 1986 Fly Ash Resistivity Prediction Improvement
with Emphasis on_Sulfur Trioxide. EPA-600/7-86-010. Washington,
D. C.: U. S. Environmental Protection Agency.
4. R. E. Bickelhaupt and R.F. Altman. "A Method for Predicting the Effective
Volume Resistivity of a Sodium Depleted Fly Ash Layer." JAPCA
vol. 34, 1984, p 832.
12-9
-------
5. J. L. DuBard and R. F- Altman. "Prediction of Electrical Operating
Points for use in a Precipitator Sizing Procedure." In Proceedings:
Conference on Electrostatic Precipitator Technology for Coal-Fired
Power Plants, 1983, pp. 4-2 - 4-31.
6. R. S. Dahlin and R. F. Altman. "Prediction of Mass Loading and Particle
Size Distribution for Use in a Precipitator Sizing Procedure." In
Proceedings: Conference on Electrostatic Precipitator Technology
for Coal -Fired Power Plants, 1983, pp. 4-32 - 4-51.
7. R. S. Dahlin, J. P. Gooch, and L. Y. Sadler, III. "Predicting the
Particle Size Distribution of Fly Ash." In Proceedings; Sixth
Symposium on the Transfer and Utilization of Particulate Control
Technology vol. 1, 1986, pp. 7-1 - 7-20.
8. M. K. Owen and A. S. Viner 1985. Microcomputer Programs for
Particulate Control. EPA-600/8-85-025a (NTIS PB86-146529) .
Washington, D. C.: U. S. Environmental Protection Agency.
9. P. A. Lawless and L. E. Sparks. "Modeling Particulate Charging in ESPs.
IEEE Transactions on Industry Applications, vol. 24, Sept/Oct
1988, pp. 922-927.
10. P. Lawless and L. E. Sparks. 1986 An Interactive Model for
Calculating V-I Curves in ESPs: Version 1.0. EPA-600/8-86-030
(NTIS PB87-100046/ AS). September 1986. Washington, D. C.: U. S.
Environmental Protection Agency.
11. P. Lawless and L. E. Sparks. An Interactive Model for Calculating
V-I Curves in ESPs: Version 2.0. In review, March 1987.
12. R. H. Perry and C. H. Chilton. Chemical Engineers' Handbook,
New York: McGraw-Hill, 1973.
13. P. A. Lawless. "Modeling Particulate Charging in ESPs II. Analytic
Approximations and Refinements." In Proceedings of IEEE Industry
Applications Society Annual Meeting, No. 89CH2792-0, 1989,
f:
pp. 2154-2162 (1989
14. R. A. Fjeld and A. R. McFarland. "Evaluation of Select Approximations
for Calculating Particle Charging Rates in the Continuum Regime."
Aerosol Science and Technology, vol. 10, no. 3, 1988, pp. 535-549.
15. J. P. Gooch and G. H. Marchant. 1978. Electrostatic precipitator
rapping reentrainment and computer model studies. FP-792.
Palo Alto, California: Electric Power Research Institute, 1978.
16. P. A. Lawless and L. E. Sparks. "Measurement of Ion Mobilities in Air
and Sulfur Dioxide-air Mixtures as a Function of Temperature".
Atmospheric Environment, vol. 14, 1980, pp. 481-483.
17. P. A. Lawless and L. E. Sparks. "A Review of Mathematical Models for
ESPs and Comparison of Their Successes." In Proceedings of the
Second International Conference on Electrostatic Precipitation,
S. Masuda, ed. Kyoto, 1984, pp. 513-522.
12-10
-------
MAIN MENU
I
File Operations
Data Input and Editing
Calculations
View Results
Print Results
Graph Results
Utilities
Exit to DOS
Figure 1. Main Menu Choices
12-11
-------
FILE OPTIONS
Load Master File
Save Master File
Edit Master File
Browse Data Files
BROWSE OPTIONS
J
Master Files
Boiler Files
Coal Files
Gas Files
Ash Files
Particle Files
Design Files
SRT Files (non-ideal)
Figure 2. File Option and Browse Options
12-12
-------
DATA OPTIONS
J
Boiler Data
Coal Data
Ash Data
Particle Data
Design Data
SRT Data (non-ideal)
Boiler Data; Plant Name & Unit, Type of Firing, Mean Coal
Particle Size, Rosin Rammler Exponent
Coal Data: Burn Rate, Rank, Proximate & Ultimate Analyses
Gas Data: Temperature at ESP, Gas volume. Gas Composition
Ash Data: Ash Analysis, Cold-side Resistivity, Hot-side
Resistivity
Particle Data; Particle Density, Grain Loading, Log-normal or
Cumulative Parameters
Design Data; Plate Area, Total Length, Number of sections.
Stack Diameter, Section areas and lengths. Corona wire diameters,
Elctrical operating points
SRT (Sneakage. Rapping & Turbulence) Data: sneakage by
sections. Rapping Reentrainment Factors, Rapping Distribution HMD
and Std Dev, Turbulence Core Fraction
Figure 3. Data Menu and types of data required.
12-13
-------
c 0.01
o
c
0.001:
0.
x
Expt
a
w/o Rapping
w Rapping
X
X
d
n
an
Xx
nun ODD
0.01
0.1
| | | M I 1 1 1—I I I II I 1 1 1—I I I I I
1 10 100
Particle Diameter (urn)
Figure 4. Size-dependent penetrations with and without rapping (Colbert)
0.1
0.01
c
0
0.001 =
c
a_
0.0001 E
1E-05
x
Expt
a
w/o Rapping
X
w Rapping
x r
I X
an x
n anon o n
I 1—I I I I 111 1 1—I I I I III
0.01 0.1 1
T I I I II I 1 1 1—I I I I I
10 100
Particle Diameter (um)
Figure 5. Size-dependent penetrations with and without rapping (Naughton)
12-14
-------
U. 1 =
-
"
0.01 E
c
*<£5
| 0.001 =
^ -
Q}
CL *
0.0001:
^ c nt;
X
^ &%> x
n EI g x
w/o Rapping g, >JfC H
x >X x
w Rapping ^ H
°x xX^
S
X
an x
a a x
rm rnnpg
0.01
0.1 1 10
Particle Diameter (um)
100
Figure 6. Size-dependent penetrations with and without rapping (Johnsonville)
Table 1
FITTED PARAMETERS
Parameter
Colbert Naughton Johnsonville
Rap Fraction
Rap MMD
Rap ag
Sneakage
Velocity cr(measured)
Turbulent Fraction
Mass Penetration
Measured
Calculated
Opacity
Measured (%)
Calculated (%)
0.045
6.0
1.6
0.07
0.55
0.24
0.004
0.0035
—
5.5
0.17
7.0
1.5
0.05
0.35
0.55
0.0008
0.0012
1.1
0.8
0.055
5.0
1.5
0.07
0.34
0.45
0.0015
0.0014
2.0
1.9
12-15
-------
AN ADVANCED MICROCOMPUTER MODEL FOR
ELECTROSTATIC PRECIPITATORS
P. A. Lawless
Research Triangle Institute
P. 0. Box 12194
Research Triangle Park, NC 27709
N. Plaks
Air and Energy Engineering Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
ABSTRACT
Computer modeling of electrostatic precipitator (ESP) performance has reached a
high level of accuracy when the electrical operating conditions are known.
Predicting voltage-current (V-I) relations has been a difficult problem, but
several different models have addressed it in the past. Still, the problems of
modeling the operation of ESPs under high loading and back corona conditions have
not been totally solved, although many advances have been made. Recently,
existing V-I models have been improved to include cold-pipe precharger and
flat-plate collector geometries. Important advances in the space charge
interaction now make possible a more accurate calculation of particle charging and
collection under high dust load conditions. The model presented here is similar
to the EPRI-sponsored ESP model described in another paper, but lacks several of
the modules which that model possesses, specifically those related to the
collection of coal fly ash. On the other hand, the technical flexibility of this
model is considerably higher, with user control over electrode geometry, physical
arrangements, environment, and other parameters not specifically related to
utility applications. This program has been designed for ease of use, with a
structured menu environment, but the operations performed are complex. A tutorial
manual is being developed to make it easier to use and understand.
13-1
-------
AN ADVANCED MICROCOMPUTER MODEL FOR
ELECTROSTATIC PRECIPITATORS
INTRODUCTION
Computer models for evaluating electrostatic precipitator (ESP) performance have
been under development for more than two decades. The most well-known of these
models (1) was written for mainframe computers. Over the years, the performance
model and several associated models were converted to run on microcomputers (2),
and others (3) were developed with easy-to-use interfaces. In more recent times,
new advances in the prediction of voltage-current (V-I) characteristics of
electrodes has been made, but the models incorporating them have been difficult to
use without considerable practice. In addition, documentation on the use of these
models has been poor to non-existent. In light of the advanced electrode modeling
capability and building on the development of an integrated model by the Electric
Power Research Institute (EPRI) (4), the U. S. Environmental Protection Agency
(EPA) has contracted with Research Triangle Institute to produce a well-documented
ESP model that combines very high technical capabilities with a user-friendly
interface.
This model contains the core calculations of the V-I models and up-to-date ESP
collection models with non-ideal effects. It is designed to encompass more ESP
applications than just coal-fired boilers, and so 1t requires more detailed input
of data. It does not contain coal-specific models, such as combustion or
resistivity models. The model also allows evaluation of various electrode
configurations, rather than being restricted to wire electrodes. This allows it
to evaluate ESPs which use the emerging multistage technology, which makes use of
separate charging and collecting sections. This new ESP technology provides a
significant advancement in collection potential over conventional designs.
This model controls the data entry with a menu-structured interface and presents
its output in tabular or graphical form. The rest of this paper discusses the
model's structure from the standpoint of input and output, the model's algorithms
in each of its parts, and some results showing the accuracy with which it can be
used.
13-2
-------
MODEL STRUCTURE
The model structure is best described by showing the menus which form the user
interface. From them, the overall structure and interactions with the user can be
seen. The Main Menu is the principal point of interaction with the user.
Main Menu
The Main Menu choices are displayed in Figure 1. Each of these choices is
selected by moving a cursor to the proper line and pressing the key, or by
simply typing the first letter of the choice. Each item on this menu leads to
sub-menus, differing in this respect from the EPRI model.
The Utilities sub-menu contains certain program configuration choices that can be
activated, a path describing where the data files may be found, and a temporary
exit to the operating system. These are not of major importance to the program
and will not be elaborated further.
File Menu
The data files are loaded according to a grouped structure. The purpose of the
structure is to allow, for example, different electrode designs to be placed
within a single ESP shell. The data are divided into separate files for flue gas
and resistivity, particles, ESP design, electrode design, and non-ideal factors.
These separate files are combined by being named in a Master File, which is used
to represent the entire ESP installation. Then, whenever a given Master File is
loaded, all the associated data files are loaded with it. If any of the data
files are modified or replaced, then the Master File is also modified. The menu
options for the File Menu, Figure 2, are mainly concerned with the Master File. A
Master File may be loaded by name or updated and saved by name from this menu.
When a Master File is saved, another file is also saved containing the Master File
name. Then, when the program is restarted, that Master File is automatically
loaded. The only convenient way to change Master Files is from this menu. The
Master File may be edited to change the file names in it and to add a few lines of
comments about the file itself. In this way, alternate configurations for an ESP
might be kept in separate Master Files, with the differences noted in them.
The Browse option allows the user to see the raw data in the various files. It
leads to a sub-menu giving choices of the type of data to view, and when a type is
picked, then a list of the files of that type will be displayed. The contents of
any file can then be listed on the screen. This facility allows the user a quick
review of the data, perhaps to include it in a Master File.
13-3
-------
Data Input Menu
The Data Menu (Figure 3) controls the selection of data to be edited. Each of the
menu selections leads to a detailed input screen for the type of data given by the
selection name. In each of the input screens, the user has the option to load and
save files of that specific type. If this is done, the Master File is updated
concurrently to reflect the changes made. This gives the user another method for
regrouping data in a Master File. Figure 3 also lists the data that are asked for
in each of the input menus. The goal of the menu design is to group important or
essential data at the top of the menu, with less important data at the bottom.
Calculations
The Calculations Menu, not shown here, controls the selection of calculations for
the V-I characteristics of the chosen electrodes or the use of them in the
performance model. Since a major part of this model is the ability to design new
electrodes, their V-I curves need to be predicted and presented before a decision
to use them can be made.
Output Menus
The output menus control the presentation of either the V-I curve calculations or
the resulting performance calculations. Both types of results are presented
similarly, but with obvious differences. The user is given three output
selections, tabular data to the screen, tabular data to the printer, and graphical
data to the screen. If the printer is an Epson (TM) or Hewlett-Packard (TM)
compatible printer, then the graphics screens may be printed, although any color
information will be lost.
The screen tabular results are arranged in several pages, with the most important
ones first, the less important ones later. The user may interrupt the viewing at
any point in the presentation. The tabular data to the printer is arranged
similarly, but includes the names of the data files used to generate the results,
and optionally, the contents of the files, since it is assumed that printed
results will be referred to independently of the program. The pagination and
format of the printed results is organized for the different page size of a
printer.
The graphs are also presented by pages, with the user allowed to break the
presentation at any time. If the printers are compatible, as mentioned before,
the Print-Screen function can be invoked to transfer the graphs to the printer.
13-4
-------
MODEL ALGORITHMS
The principal V-I prediction algorithms used in the model have been discussed
previously in some detail.(5) The capabilities built into this model will be
described in the following sections. Since gas, particle, and dust layer
resistivity data are all direct inputs, there are no models for these factors to
use. The user could turn to the EPRI integrated model or other models to obtain
these data, if desired.
V-I Curves
Although operating points may be directly input to the performance model, or the
V-I correlation (6) may be called, this model is fully intended to be used with
the V-I predictors developed in previous versions, with updated space charge
interactions. The predictors respect the differences between tuft and glow
corona. They also include the differences between filtered and unfiltered dc
power supplies. Comparison of the predicted operating points of a typical ESP
section with dusts of various resistivities shows excellent agreement with the
correlation data at high resistivities and good agreement with spark-limited
operating points at low resistivities, provided the pulsating dc crest factor is
properly accounted for.
The V-I portion of the model predicts V-I characteristics and electric fields for
three types of electrodes: round wires with arbitrary diameters and spacings, flat
plates, and wires-between-pipes (cold-pipe precharger geometry). Although the
flat plate does not strictly have a V-I curve, its use as a collection electrode
is widespread and its field needs to be included in performance models. The
models are designed to predict the clean-gas V-I curves of all the elements over a
range of common voltages. Recent modeling advances in space charge interactions
(7,8) have shown that the clean curve can be used to obtain the
space-charge-limited curve with a fairly simple approach that is physically
consistent.
ESP Collection
Most ESP performance models generally operate quite successfully when the voltage
and current operating points of the ESP are known. If fixed voltages and currents
are used, then the performance portion of this model will use them as direct
inputs. The electric fields needed to calculate particle charging and collection
can be derived fairly accurately from the voltages, but are not calculated
numerically. However, the interaction of the whole V-I curve with the particulate
charging and collection is a complex process that can only be carried out with
13-5
-------
detailed calculations. By incorporating this interaction into the model, the ESP
performance can be calculated using operating parameters that are beyond present
experience.
The collection portion of the model has three parts: particle charging, particle
collection, and loss mechanisms. The charging algorithm uses an addition of field
and continuum diffusion charging (9), with the charge calculated iteratively from
the corona current provided by the wires. The particle collection is by the
mechanism of Deutsch collection, where the particle migration velocity may use
either an empirical correction (10) or a turbulent core parameter to account for
the less-than-totally turbulent flow in an ESP. The loss mechanisms are: velocity
maldistribution, applied as a modifier to the Deutsch collection term; sneakage,
which bypasses a fraction of the flow around the collection zone; and rapping
reentrainment, which adds a fraction of the collected material to the gas stream
for subsequent recollection.(1)
The total collection is calculated in two steps. First, the collection without
rapping is calculated and the collection efficiency of each section is stored.
Then, rapping is applied successively to each section, where a specified fraction
of the material is reentrained with a certain size distribution. The reentrained
dust is treated as if it were passing alone through the remainder of the sections,
where most is collected and some reaches the outlet. The total outlet emissions
then consist of a steady component and rapping contributions from each of the
sections. This process neglects the fact that the rapping puff is a heavy load on
the charging capability of the ESP sections.
The total emissions are computed as a sum over all particle sizes. An opacity is
also calculated for a given stack diameter. Finally, a weighting of the emissions
according to the PM10 standard is applied for a PM10 emission level.
RESULTS OF FITTING A TEST ESP WITH A PRECHARGER
The model is well-suited for computing the electrical conditions for the Valmont
pilot scale ESP. ESP performance measurements were made under sponsorship of EPRI
on the EPA owned pilot scale ESP that had been at the Valmont Power Plant at
Boulder, Colorado (11). The ESP configuration tested was various combinations of
cold-pipe prechargers and wire-plate collectors.
The Valmont pilot scale ESP was designed and built for EPA as a research unit by
Denver Research Institute of Denver University. It consists of four collector
13-6
-------
fields, each of which is preceded by a precharger. The ESP was designed for
maximum configuration flexibility. The collectors could be of either the
wire-plate or flat-plate type. The plate-to-plate spacing and the corona
discharge electrode size were both variable. Both the cold-pipe and tri-electrode
prechargers had been evaluated before the EPRI test program.
The tests that had been conducted by EPRI were with various prechargers turned on
and off. Sulfur trioxide (SO,) conditioning was used for some tests to decrease
the resistivity of the fly ash. A description of the pilot scale ESP, the gas
conditions, and the particulate characteristics are provided in Table 1. It
should be noted that the fly ash resistivity was quite high, on the order of 5 x
1011 ohm-cm. The size distribution of the fly ash entering the ESP has a mass
mean diameter of 29.8 /im and a geometric standard deviation of 5.2, which suggests
that there is a large number of fine particles in the range below 1.0 pm diameter.
The fine particles will also contribute to a significant space charge effect even
though the grain loadings are relatively low. For the performance evaluations, a
particulate inlet loading of 2.43 g/m1 was used throughout for simplicity.
The tests run at Valmont are summarized in Table 2. The four baseline tests were
made with only the collector sections energized (no prechargers). The differences
in measured efficiencies suggest that the relatively high resistivity of the fly
ash caused back corona, which contributed to the widely varying efficiencies for
the four test runs. The relatively high efficiency measurement of the first
baseline test, 97.85%, is unexplainable. The two tests with SO, conditioning
reduced the fly ash resistivity to about 1 x 1010 ohm-cm, the efficiency
improvement is consistent with past experience for ESPs having similar SCAs.
Various combinations of prechargers were energized for the ESP/precharger tests.
The two-precharger tests had the prechargers upstream of the first and third
collector sections energized. The three-precharger tests had the prechargers
upstream of the second, third, and fourth collector sections energized. In the
four-precharger test, the prechargers upstream of each of the collector sections
were energized.
In using the model, it is necessary to estimate the non-ideal conditions in the
performance prediction portion. The non-ideal conditions are:
1. the particulate fraction that is reentrained from rapping;
2. the size distribution of the reentrained particulate
from rapping;
13-7
-------
3. the fraction of the gas, or sneakage, that bypasses
the electrified regions of the ESP;
4. the uniformity of the velocity in the ESP, expressed
as the standard deviation; and
5. a correction to the SCA, which makes the effect of
turbulence in the gas stream less than the Deutsch-
Anderson equation assumes.
The accuracy requirements for these non-ideal conditions become of greater
importance as the efficiency of the ESP is increased. For very high efficiency
ESPs, the efficiency becomes dominated by the non-ideal conditions.
With extensive comparison of the model with measured ESP performances, average
values for the non-ideal factors could be determined, especially for similar ESPs.
There has not been sufficient work done to adequately provide data to a specific
ESP such as the Valmont pilot scale unit. Instead, the approach chosen was to use
the standard default rapping fraction of 0.124 and a rapping reentrainment size
distribution with 6 /
-------
With its ability to model different electrode configurations at above normal
particulate loadings, it will allow evaluation of emerging multistage ESP
technology that has the potential for significantly advancing the state of the
art. With its flexibility and user-friendly approach, it can be learned quickly,
but it will still be able to address many of the subtleties of ESP performance.
REFERENCES
1. S. Oglesby, Jr. and G. B. Nichols. Electrostatic precipitation.
New York: Marcel Dekker, 1978.
2. M. K. Owen and A. S. Viner. Microcomputer Programs for
Particulate Control. EPA-600/8-85-025a (NTIS PB86-146529).
September 1985. Research Triangle Park, NC: U. S. Environmental
Protection Agency.
3. P. Lawless and L. E. Sparks. 1986 An Interactive Computer Model for
Calculating V-I Curves in ESPs; Version 1.0. EPA-600/8-86-030a
(NTIS PB87-100046). September 1986.Research Triangle Park, NC: U. S.
Environmental Protection Agency.
4. Previous paper in these proceedings. (P. A. Lawless and R. F. Altman.
"An Integrated Electrostatic Precipitator Model for Microcomputers.")
5. P. A. Lawless, N. Plaks, and L. E. Sparks. "An Interactive Model for
Analysis of Electrical Conditions in Electrostatic Precipitators."
In Proceedings: Seventh Symposium on the Transfer and Utilization
of Particulate Control Technology vol. 1, p. 32-1 32-16.
EPA-600/9-89-046a (NTIS PB89-194039). May, 1989 Research Triangle
Park, NC: U. S. Environmental Protection Agency.
6. J. L. DuBard and R. F. Altman. "Prediction of Electrical Operating
Points for use in a Precipitator Sizing Procedure." In Proceedings:
Conference on Electrostatic Precipitator Technology for Coal-Fired
Power Plants, 1983, pp. 4-2 - 4-31.
7. P. A. Lawless and L. E. Sparks. "Modeling Particulate Charging in ESPs."
IEEE Transactions on Industry Applications, vol. 24, September/
October 1988, pp. 922-927.
8. P. A. Lawless. "Modeling Particulate Charging in ESPs II. Analytic
Approximations and Refinements." In Proceedings of IEEE Industry
Applications Society Annual Meeting, No. 89CH2792-0, 1989,
pp. 2154-2162 (1989).
9. R. A. Fjeld and A. R. McFarland. "Evaluation of Select Approximations
for Calculating Particle Charging Rates in the Continuum Regime."
Aerosol Science and Technology, vol. 10, no. 3, 1988, pp. 535-549.
10. J. P. Gooch and G. H. Marchant. 1978. Electrostatic precipitator
rapping reentrainment and computer model studies. FP-792.
Palo Alto, California: Electric Power Research Institute, 1978.
13-9
-------
11. G. Rinard, M. Anderson, and R. Altman. "Proof of Concept Testing of
ESP Retrofit Technologies for Low and High Resistivity Fly Ash." In
Proceedings: Seventh Symposium on the Transfer and Utilization of
Participate Control Technology vol. 1, p. 19-1 19-14.
EPA-600/9-89-046a (NTIS PB89-194039). May, 1989 Research Triangle
Park, NC: U. S. Environmental Protection Agency.
Table 1
Valmont ESP, Gas, and Participate Data for EPRI Tests
ESP Geometry
Number of collector sections
Length of collector sections
Height of collector sections
Area of collector sections
Plate-to-plate spacing
Wire-to-wire spacing
Corona wire diameter
Wires per lane
Lanes per section
Precharger Geometry
Number of precharger sections
Pipe diameter
Pipe spacing
Corona wire diameter
Gas Conditions
Average volumetric flow rate
Average gas velocity
Average gas temperature
SCA, prechargers off
SCA, 2 prechargers on
SCA, 3 prechargers on
SCA, 4 prechargers on
Particulate Characteristics
Geometric mass mean diameter
Geometric standard deviation
Bulk resistivity
Bulk resistivity with S03
Density
Dielectric constant
4
1.8
3.7
80.3
22.9
22.9
3.2
7
6
m
m
m2
cm
cm
mm
(6
(12
(864
(9
(9
(0.125
ft)
ft)
ft2)
in)
in)
in)
4
6.1
22.9
3.2
cm
cm
mm
(2.4 in)
(9 in)
(0.125 in)
8.5 mVs
1.66 m/s
149 °C
37.8 mVmVs
38.4 mVmVs
39.2 mVmVs
40.1 mVmVs
(18000 acfm)
(5.5 ft/s)
(300 °F)
(198 ft2/kacfm)
(201 ft2/kacfm)
(206 ft2/kacfm)
(210 ft2/kacfm)
29.8 fim
5.2
5 x 1011 ohm-cm
1 x 1010 ohm-cm
2.27 g/cm3
100
13-10
-------
Table 2
Comparison of Model Calculations with Measured Efficiencies
Test,#
Inlet
Loading
g/m» (gr/ff)
Measured
Efficiency
Modeled
Efficiency
Baseline,
Basel ine,
Baseline,
Baseline,
Model
SO,, 1
SO,, 2
Model
1
2
3
4
Two Prechargers, 1
Two Prechargers, 2
Model
Three Prechargers
Model
Four Prechargers, 1
Four Prechargers, 2
Model
2.49
2.43
2.65
4.20
2.43
2.27
2.43
(1.086)
(1.063)
(1.156)
(1.833)
(1.063
2.50 (1.090)
3.21 (1.404)
2.43 (1.063)
1.75 (0.763)
3.89 (1.698)
2.43 (1.063)
(0.993)
(1.063)
2.09 (0.912)
3.36 (1.469)
2.43 (1.063)
97.85*0.5
90.37*2.44
89.75*1.65
94.42*0.48
98.75*0.58
97.94*0.58
98.42*0.43
97.83*0.31
98.25*0.46
99.08*0.39
98.86*0.17
89.84
98.06
98.18
97.91
98.62
13-11
-------
MAIN MENU
I
File Operations
Data Input and Editing I
Calculations
View Results
Print Results
Graph Results
Utilities
Exit to DOS
Figure 1. Main Menu Choices.
13-12
-------
FILE OPTIONS
J
Load Master File
Save Master File
Edit Master File
Browse Data Files
r
BROWSE OPTIONS
Master Files
Gas and Resistivity Files
Particle Files
ESP Design Files
Electrode Design Files
SRT Files (non-ideal)
Sneakage, Rapping,
and Turbulence
Figure 2. File Option and Browse Options.
13-13
-------
DATA OPTIONS
Gas and Resistivity Data
Particle Data
Design Data
Electrode Data
SRT Data (non-ideal)
Gas and Resistivity Data; Temperature at ESP, Gas
Volume, Gas Composition, Dust Resistivity
Particle Data: Particle Density, Grain Loading, Log-
normal or Cumulative Parameters
Design Data; Plate Area, Total Length, Number of
Sections, Stack Diameter, Section Areas and Lengths, Corona
Wire Diameters, Electrical Operating Points
Electrode Data: Type of Electrode, Wire Diameter, Wire-
plate Spacing, Wire Locations in Section, Cold-pipe
Diameter, Flat Plate Length, Corona Onset Factors
SRT (Sneakage. Rapping, and Turbulence) Data: sneakage
by Sections, Rapping Reentrainment Factors, Rapping
Distribution Mass Median Diameter and Standard Deviation,
Turbulence Core Fraction
Figure 3. Data Menu and Types of Data Required
13-14
-------
MEASUREMENTS INSIDE A MODEL PRECIPITATOR
Ad Braam and Wiebren Hiemstra
N.V. KEMA
Utrechtseweg 310
6812 AR ARNHEM
The Netherlands
ABSTRACT
Experimental (and theoretical) research has been conducted to understand the physical
mechanisms inside an Electrostatic Precipitator (ESP). For this purpose an ESP model was
built consisting of 0.9 x 1.0 m2 collecting plates with a spacing of 0.3 m. The Precipitator,
constructed with four round corona wires, could be operated with positive or negative
voltages up to 65 kV.
Extensive velocity and turbulence measurements were done by means of a Laser Doppler
Anemometer (LDA). For seeding monodisperse polystyrene particles were used. Charge-to-
mass and aerodynamic size measurements of these particles were performed through
excitation with an alternating electric field within the measuring volume of the LDA.
Air velocities and turbulence in one gas passage of an ESP were recorded at ambient
temperature and flow conditions similar to actual installations. These recordings were made
both for positive and negative corona discharge. Charge characteristics depending on
particle size and position inside the ESP were determined.
Theoretical electric-field calculations and measurements led to the determination of the
appearing electric wind effects.
INTRODUCTION
Laser Doppler Anemometry is a useful tool to study the mechanisms that play a role in
Electrostatic Precipitation.
Several investigators used it for research of wire and plate-type ESPs. Jurewicz (1) measured
the velocity field behind a small Laboratory-scale Precipitator for a negative voltage of 9 kV.
Lawless (2) executed LDA measurements inside a Laboratory Precipitator, testing corona
wind hypotheses. Clark (3) performed measurements with a two component LDA inside an
ESP model. This model was in good agreement with commercial ESPs; however only little
attention was paid getting a low turbulence intensity and well defined uniform flow. For this
reason it is difficult to compare his experiments with theoretical calculations.
For a better understanding of the several mechanisms and because of its non-intrusive
nature we also used an LDA for the measurements inside an energized precipitator model.
The LDA has been used both for velocity measurements and the determination of aero-
14-1
-------
dynamic size and charge-to-mass ratio of the particles. This is a technique developed at the
Kema Laboratories (5).
Particle motion inside a Precipitator is determined by several physical mechanisms, such as:
e Particle motion due to the turbulent gas flow.
• Coulomb forces working on charged particles, drifting them toward the plates.
a Electric wind effects caused by the momentum added to the gas by the ion flux.
The infuence of electric wind on gas flow and the resulting impact on particle transport is
possibly the least understood of the phenomena that occur in Precipitators. The electric
wind-effect can be derived from the results of the measurements. Since the small polystyrene
particles of 1.65 ^m will follow the turbulent gasflow completely there is no difference in
motion between gas and particles in a non-energized precipitator. By subtracting the particle
velocities inside an energized and non-energized Precipitator the influence of the original
turbulent fluid flow will be eliminated. The remaining particle motion is caused by the drift
velocity and electric wind. The drift velocity is governed by the electrostatic force, the particle
inertia and viscous drag given by Stokes' equation. Under steady state conditions the drift
velocity is:
qEpc
w = (1)
6 TT r n
q is the charge on a single particle. Ep is the electric field at the position of the particle in the
field. The particle charge (q) is known by measurement and numerical calculations were
performed for the electric field (Ep). Drift velocities can be calculated now and electric wind
effects derived.
THE MODEL ELECTROSTATIC PRECIPITATOR
The Model Electrostatic Precipitator consists of two parallel collecting plates and four
discharge electrodes. The actual dimensions are given in Table 1.
TABLE 1
Model Precipitator Dimensions
Plate-to-plate distance (m) 0.30
Wire-to-wire spacing (m) 0.20
Diameter of smooth corona-wire (m) 0.003
Mean gas velocity (m/s) 1.25
Length of model (m) 1.0
Width of model (m) 0.9
The plate distance and wire spacing were chosen to match those of a commercial Precipita-
tor. The width (actual height) of the Model Precipitator was set to 0.9 m so that the assump-
tion of a two dimensional flow is allowed. Contrary to commercial ESPs the collection plates
and discharge electrodes were placed in a horizontal plane.
The discharge electrodes used in this investigation are smooth silver-steel wires. These wires
14-2
-------
were placed in the glass side-walls of the model.
The Model is part of a closed-circuit made of acrylate pipes with an inner diameter of 190
mm. The upper view of the measuring facility is shown in Fig. 1. A blower provides an
adjustable air flow in the facility with a velocity range from 0.5 to 4.5 m/s in the test section.
To get a low turbulence level and uniform flow condition, a lot of attention is paid to the inlet
and outlet arrangements. Ahead of the model a flow straightener, diffuser, a settling chamber
with three screens and an elliptically shaped constriction were installed. Fig. 2 shows the
constriction, the outlet section and the model.
Instead of fly-ash, monodisperse polystyrene particles with a mean diameter of 1.65 //m. and
a standard deviation of 0.16 ^m. were used in the experiments. These particles suspended
in ethanol are atomised by means of a Laskin nozzle and injected into the test section. The
injection takes place in front of the constriction at a height between the discharge electrode
and lowest collecting plate. The experiments were performed with ambient air instead of flue
gas. This will influence the corona discharge mainly by differing Voltage-Current Charac-
teristics in comparison with a Precipitator under industrial conditions.
LASER DOPPLER ANEMOMETRY
IDA measurements were conducted in forward scatter mode with a 35 mW HeNe laser as a
light source (Fig.3). Sensing of the velocity direction was effected by a rotating grating which
introduced a frequency shift of 819 kHz.
Lenses were chosen to form a measuring volume with a length of 14 mm, a width of 0.5 mm
and a fringe spacing of 9 /^m. A photomultiplier was used for the detection of scattered light.
The frequency of each Doppler burst was determined by zerocrossing detection after the
signal had been filtered, amplified and digitized by a transient recorder. Data acquisition was
performed by a computer.
DETERMINATION OF AERODYNAMIC SIZE AND CHARGE
Light scattering techniques with lasers determining the "optical sizes" are often applied for in-
situ measurements of particle dimensions in flue gas. These techniques, however, give
limited information about the aerodynamic sizes of particles, which determine their behaviour
in a gas flow. Especially this information is needed to characterize the motion of particles.
Although theoretical charging mechanisms have been described in literature, experimental
verification is hardly available (4,10).
For this reason a method has been developed to measure the aerodynamic size, charge-to-
mass ratio and velocity of a particle. For a theoretical background of the method we refer to
Hunik (5). A brief description of the principle is given below.
The method involves the application of a homogeneous alternating electric field over the
measuring volume of an LDA. In order to realise this two paperclip-shaped electrodes are
used, each on one side of the measuring volume (Fig.4). The induced oscillatory motion of a
particle passing through this measuring volume parallel to the interference fringes causes a
frequency-modulated LDA signal. The characteristics of the oscillations and hence the
aerodynamic size, charge-to-mass ratio, as well as the velocity of the particle, can be
determined after Fourier Transformation of this signal.
The phase shift between motion of the particle and alternating electric field is a direct
measure for the aerodynamic size (the phases in the frequency spectrum must be corrected
for phase shifts due to photomultiplier tube, amplifiers, filters, etc.).
Charge-to-mass ratio can be calculated from the aerodynamic size and the amplitude ratio
between zero- and first-order peaks in the frequency spectrum. The particle velocity is
14-3
-------
determined by the position of the zero-order peak.
SETTING UP MEASUREMENTS
Velocity, size and charge measurements were carried out symmetrically between both glass
sides of the Precipitator Model and between the second and third discharge electrode and
lower collecting plate.
To investigate the difference between positive and negative corona, several velocity measure-
ments were done parallel to the discharge electrode and perpendicular to the collecting
plate. Fig. 5 shows the measuring area in the model. The measurements were performed
under the following experimental conditions:
• Constant gas velocity (1.25m/s).
• Zero voltage, positive voltage of 65 kV with current density of 625 /^A/m2 and
negative voltage of 61.2 kV with current density of 654 /zA/m2
• Stability of air temperature (32 - 35 °C).
• Stability of the current density.
Velocities as well as particle sizes and charges were determined. For the determination of
the velocity in the main flow direction (x) and the collecting plate direction (y) the laser
beams have to be aligned in a horizontal and vertical plane respectively. Displacement of the
measuring volume in the y- and z-directions is provided by two stepping-motors (precision
0.01 mm). For the displacement in x-direction the LDA-arrangement has to be moved. Each
velocity value is an average of two thousand samples. The measuring time varies from 1.5 to
2 minutes
RESULTS
VELOCITY MEASUREMENTS
In zero voltage situation the main flow is uniform with a flat velocity profile (Fig. 6). In
designing the inlet section, a lot of attention was paid obtaining such a smooth profile for a
good evaluation of the electric wind. Near the collecting plate the start of the boundary layer
is visible. The velocity profile close to the plate cannot be measured because of optical
blockages of the laser beams.
Fig. 7 shows the velocity profile for a positive voltage of 65 kV. The transverse velocities (y-
direction) are largest at the discharge electrode and collecting plate. The maximum trans-
verse velocity amounts to 8% of the mean velocity.
The velocity profile for a negative voltage of 61.2 kV is shown in Fig. 8. The flow pattern for
negative voltage is much more irregular. In mainflow direction just after the discharge
electrodes there is an area with negative transverse velocities, more downstream positive
transverse velocities were measured.
Positive and negative voltages show substantial differences in influencing the primary gas
flow. These differences are caused by the characteristics of the two corona discharges.
The ionisation process for positive corona takes place regularly around the discharge wire.
Both the charging process of the particles and the electrostatic forces acting on them are
uniform; the process can be considered to be two-dimensional.
Negative corona discharges are distributed irregularly around the discharge wire. Visual
judgement of the corona tufts shows that these tufts have different appearances. They keep
turning around the wire more or less, moving along it and varying in intensity. This results in
14-4
-------
a dynamic and three dimensional mechanism. The irregular velocity distribution in collecting
plate (y-) direction is shown in Fig. 9.
TURBULENCE MEASUREMENTS
Turbulence is calculated by taking the RMS-value of two thousand velocity samples at each
measuring location:
s = /I xu? U2 (2)
u,= velocity of each particle
U = mean velocity
N = number of samples.
Fig. 10 shows the turbulence profiles both for the velocity in flow direction and collecting
plate in the case of positive voltage.
An isotropic uniform turbulence was measured. Only behind the wire and along the collecting
plate a raised turbulence and non-isotropy has been observed. Comparing the zero and
positive voltage situation it can be concluded that positive corona hardly contributes to an
increasing turbulence level. For negative corona turbulence increases radically (Fig. 11).
Fluctuations amount to 0.5 m/s at a mean velocity of 1.25 m/s for both main and transverse
flow directions.
SIZE AND CHARGE MEASUREMENTS
Measurements of aerodynamic size and charge were performed with the method developed
at the Kema laboratories. A distribution of aerodynamic size of the polystyrene particles is
shown in Fig. 12. Charge measurements at the locations 25 mm. and 60 mm. beneath the
discharge electrode are given in Figures 14 and 15.
Studying these figures it appears that the particle charges as well as their variation increases
considerably the nearer the particles are at the wire and values of 500 el. charges occur
frequently. This is in contrast with the conclusions of Lawless (2). The mean charge as a
function of particle size has been determined by curve fitting charge measurements of one
thousand particles. Charge measurements were performed on eight positions in the xy plane
for a positive corona of 65 kV. The results of the mean charges of 1.65 pm particles have
been plotted in Fig. 16 (triangles and dots).
CALCULATION OF PARTICLE CHARGE
The Deutsch Model uses a mean charge for the calculation of the migration velocities of all
particles of a certain size. In fact, charges on those particles also depend on their position in
the electric field. To determine this drift velocity at every measuring location, it is necessary
to know each local mean particle charge.
Particles in a corona field are charged simultaneously by both field and diffusion charging.
Field charging dominates for particles with a diameter greater than 1 /im and diffusion
charging for those less than 0.2 ^m. Because the measurements were carried out with
particles greater than 1 ^m, field charging is the principal mechanism.
14-5
-------
The charge obtained by field charging alone is:
(4)
£0 = permittivity of free space
er = permittivity of particle material
Ep = electric field
rp = particle radius
i = current density
The charge of a particle of a certain diameter depends on the residence time and the electric
field.
The residence time can be calculated from the particle velocity and the position within the
model Precipitator.
The electric field distribution has been calculated with a mathematical model. This model
solves the Poisson and continuity equation by numerical techniques (finite difference
method) for a two-dimensional configuration. For more information about this mathematical
model we refer to Koopmans (6).
Fig. 16 shows the results of the electric field calculations beneath a discharge electrode for a
voltage of 65 kV and a configuration which is in agreement with the experiments. Field
charges on particles with an aerodynamic diameter of 1.65 ^m were calculated for x=0, 100
and 200 mm. and are shown in Fig. 17
The agreement between experiment and theoretical calculations of the charge distribution is
surprisingly good.
CALCULATION OF THE ELECTRIC WIND
Velocity measurements have proven that two-dimensional flow exists when a positive corona
is used. Theoretical calculations of corona wind have been performed by Koopmans (6) and
McDonald (9). In this investigation a comparison has been made between theoretical corona
wind calculations (6) and electric wind determination from experiments for a two-dimensional
flow. This comparison will be explained in the next few paragraphs.
DIFFERENTIAL VELOCITY
It has already been stated that by subtracting velocities in the energized and non-energized
ESP Model the turbulent flow wil be eliminated. Remaining particle motion is caused by
electric drift and corona wind. The differential velocity for a positive voltage of 65 kV. is
shown in Fig. 18. The maximum velocity amounts 0.13 m/s opposite to the original main flow
direction.
The largest velocities in the direction of the plate are found near the discharge electrode and
the collecting plate and are in the order of 0.1 m/s.
14-6
-------
DRIFT VELOCITY
Calculation of the drift velocity of 1.65 ^m. particles in each measuring position is possible by
solving Equation (1). Fig. 19 shows the drift velocity for a positive voltage of 65 kV. They are
large near the discharge electrode and collecting plate and amount to 0.1 m/s.
ELECTRIC WIND
Subtraction of the differential velocity and drift velocity will result in an image of the electric
wind and is shown in Fig. 20.
Between the discharge electrodes a flow is generated opposite the main flow direction. At x
- 50 mm. the flow faces the collecting plate and at x = 150 the discharge electrode. This
will influence the main flow in such a way that upstream of the discharge electrode the flow
will be deflected in the direction of the wire and downstream of the electrode in the opposite
direction. These results are in fair agreement with results obtained by Yamamoto (7) and
Leonard (8).
The electric field and space charge were calculated seperately from the turbulent flow for a
theoretical determination of the electric wind at conditions comparable those of the experi-
ments. Voltage and space charge were calculated with a two-dimensional model. With these
results the electric forces acting on the flow could be derived and used together with the k-e
model to solve the flow equations. The computer code PHOENICS based on finite volume
discretisation was used for. At the collecting plate logarithmic wall boundary conditions were
applied. Further information is given in (6).
The results of this calculation are shown in Fig. 21. Close to the discharge electrode a small
eddy turning counter clockwise is found and between the discharge electrodes and the plate
a large eddy turning clockwise appears. For the experimental electric wind effects several
assumptions had to be made. For instance the measurements of particle charges have been
limited to 8 locations whereas for the calculation of the drift velocities it had to be known on
56 locations. Nevertheless there is qualitative agreement of the electric wind effect from
experiment and theoretical calculations.
CONCLUSIONS
With a positive voltage for round discharge electrodes the influence of the electric wind on
the flow and particle migration is small. Negative corona causes much extra turbulence in the
flow. This will have a negative influence on the transport of particles to the collecting plates.
On the other hand, particle charging will be faster in the case of negative corona. A sig-
nificant domination of one of those effects cannot be proven. The experiments were per-
formed under ideal circumstances (smooth collecting plates and round electrodes). The
collecting plates and discharge electrodes used in industrial Precipitators will be enormous
sources of turbulence and will affect fluid and particle dynamics much more than the electric
wind does.
ACKNOWLEDGMENTS
This work was supported by the Dutch Electricity Companies
14-7
-------
REFERENCES
1. J. T. Jurewicz, D. E. Stock and C. T. Crowe. "Particle velocity measurements in an ESP
with Laser velocimeter." AlChE Symposium Series No. 165. Vol 73. 1977
2. P. A. Lawless and E. J. Shaughnessy. "Laser Doppler Anemometer measurements of
particle velocity in a Laboratory Precipitator." J. Aerosol Sci Annual Meeting., 1981
3. W. T. Clark, R. L. Bond and M. K. Mazumder. "Measurement of the electrokinetic
transport properties of particles in an ESP". In Proceedings of Symposium on Particu-
late Control technology.,pp.30-1
4. J. K. Horrocks a.o. "Two colour Laser Doppler measurements of particle size and
charge in the wake region of a Precipitator." IChemE Symposium Series No. 99. pp.
101
5. R. Hunik and W. Hiemstra. "Combined aerodynamic size, charge to mass and velocity
measurements of aerosol particles.
"International conference ESP. 1987, Padua, Italy.
6. G. Koopmans and Q. Hoogenboom. "Coronawind in a plate-wire Electrostatic Precipita-
tor."lnternational Conference ESP 1987, Padua,ltaly
7. T. Yamamoto and H. R. Velkoff. "Electrohydrodynamics in an Electrostatic Precipitator.
J. Fluid Mech..1981 vol. 108. pp.1
8. G. L. Leonard, M. Mitchner and S. A. Self. "An Experimental study of the electrohydro-
dynamic flow in E.P 's."J. Fluid Mech.,1983. vol. 127 pp. 123.
9. J. R. McDonald a.o. "A mathematical model for calculating electrical conditions in wire-
duct Electrostatic Precipitation Devices."J. Appl. Phys.,1977 48 pp.2231-2243.
10. J. R. McDonald a.o. Second Symposium Particulate Technology Volume II, Electro-
static Precipitators, Denver 1980.
11. K. L. McLean. "Electrostatic Precipitators". IEE Proceedings. Vol. 135, No.6, 1988.
14-8
-------
1 = Blower
2 = Pressure diff. cell
3 - Valve
4 = Bypass
5 = Particle injection
6,7,8 = Inlet section
9 = Model
10,11 = Outlet section
12 = Flow straightener
13 = Screens
Figure 1. Upper view of the measurement facility.
Flow direction
1 = Discharge electrode
2 = Measuring area
3 = Isolation transformer
4 = High voltage supply
5 = Earth plate
Figure 2. Laboratory Model with high voltage supply.
14-9
-------
Flow direction
1
2
3
4
5
6
HeNe Laser
Rotating grating
Mask
Lens 200 mm.
Lens 600 mm.
Model
7 = Mask
8 = Lens 250 mm.
9 - Photomultiplier
10 = Optical bench
11 = Discharge electrode
12 = Glass side
Figure 3. Laser Doppler Anemometry arrangement
~ 5.5 kV
19.5 kHz
0
0
1 = Laser beams
2 = Alternating field probe
3 = Mask
4 = Lens
5 = PM
Figure 4. Measuring volume with paperclip shaped probes.
14-10
-------
Main flow
direction
Rel. Particle
injection pos.
Figure 5. Position of the measuring area inside the Model.
Collecting plate
—' Scale: 1.0 m.s'1 Discharge electrode
Figure 6. Velocity profiles in a non-energized situation.
14-11
-------
Collecting plate
-Scale I.Om.s1 Discharge electrode
Figure 7. Velocity profiles for a positive voltage of 65 kV.
Collecting plate
-Scale: 1.0 m.s 1
Discharge electrode
Figure 8. Velocity profiles for a neg. voltage of 61.2 kV.
14-12
-------
CNJ
d
0)
CD
O
CO
25
40
50
60
70
80
90
100
-isn
. . t t . y
HJT
H|i'
A
mr
A
ini*
|Hl*
HII*
MM*
MM*'
>
.A.
'MM *
. t, .
, """•
* *MIM*
4 A .
' MUq*'
i i
•»M|Mr
A i
•'UP!*'
A
• «u|;ui*
A
•*UH*H*'
400
425
450 475 500
Walldistance (mm)
Figure 9. Transverse velocity (y-direction) in yz-plane at x = 150
for a neg. voltage of 61.2RV.
14-13
-------
Collecting plate
• Scale 0-10 m.s
Discharge electrode
Transverse
Main
Figure 10. Turbulence for a positive voltage of 65 kV.
Collecting plate
• Scale 0.10 m.s
Discharge electrode
•
Transverse
Main
Figure 11. Turbulence for a negative voltage of 61.2 kV.
14-14
-------
to
2 160 -
y
k_
ro
CL
o 120 -
i_
0>
.a
E
^ 80 -
40 -
0 -
— *—
• —
M = 1.65 x 10"6m
O = 1,60 x 10"7m
\^__
1.2 1.45 1.7 195 2.2
Diameter (yum)
Figure 12. Distribution of the aerodynamic diameter of polystyrene particles.
Figure 13. Microscopic picture of polystyrene particles.
14-15
-------
CD 600-
01
CO
_c
0 500-
400-
300-
200-
100-
0
1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2.0 2.1 2.2 2.3
Diameter (/urn)
Figure 14. Charge of polystyrene particles as a function of aerodynamic
diameter for a pos. voltage of 59.8 kV. halfway the wire and plate (x=200, y=60).
^900-
CD
cn
JS 800-
O
700-
600-
500-
400-
300-
200-
100-
0-
+ +
+
+
+ -f
f + + +
0. + + J.
+ + T + ++* +++ +
+ + /+1++ + + T + « \ *
;+;-+;r^++ + v+ ^+ ; V + ***
^4. "*"+^" "T^*" ^**" "'W^'ti. "*" + ++*"*" +
^+ *h- "*" "tlfej) .4- +-T "JL ^_ ^ . ,^" +
+ + +*" -p\. + + 1*^1^4+ j_ ^i + + +
+ +t+ '+ +
+
1-4 1.5 1.6 1.7 1.8 1.9 2.0 2.1 2.2 2.3
Diameter (/urn)
Figure 15. Charge of polystyrene particles as a function of aerodynamic
diameter for a pos. voltage of 59.8 kV. in the vicinity of wire and plate (x = 200,
14-16
-------
T 12.5-1
E
10.0-
7.5-
5.0-
2.5-
Discharge electrode
, - , - , - , - , - r
0 25 50 75 100 125 150
Distance between discharge and collecting electrode (mm)
Figure 16. Electric Field distribution as a function of wire-plate distance.
1 = ion mobility of 0.00012 m2/Vs
2 = ion mobility of 0.00022 m2/Vs
oT 700 -
600 -
500 -
400 -
300 -
x = 100
x = 0
1 1 1 | '|
0 50 100 150
Distance between discharge and collecting electrode (mm)
Figure 17. Charge of 1.65 urn. particle as a function of wire and plate distance
for a pos. voltage of 65 kV.
14-17
-------
Collecting plate
-Scale: 0.10 m.s 1 Discharge electrode
Figure 18. Differential velocity at a pos. voltage of 65 kV.
Collecting plate
- Scale: 0.10 m.s"1 Discharge electrode
Figure 19. Drift velocity at a pos. voltage of 65 kV.
14-18
-------
-Scale: O.IOm.s"
Collecting plate
Discharge electrode
Figure 20. Electric wind at a pos. voltage of 65 kV.
Collecting plate
Discharge electrode
-Scale: 0.75 m.s
Discharge electrode
Figure 21. Electric wind calculations with theoretical Model (pos. voltages 65
kV.).
14-19
-------
THE EFFECTS OF FIRESIDE PROCESS CONDITIONS
ON ELECTROSTATIC PRECIPITATOR PERFORMANCE
IN THE ELECTRIC UTILITY INDUSTRY
Herbert J. Hall
H. J. Hall Associates, Inc.
1250 State Road
Princeton, New Jersey 08540
ABSTRACT
Fireside conditions include all those factors involving fuel-ash properties, boiler
type and load, combustion conditions, flue gas properties and particulates entering
the precipitator. Knowledge of, and experience in, these process factors and their
effects are essential in such endeavors and areas as: (1) new ESP design, (2) up-
grading existing performance, (3) hot to cold-side ESP conversion, (4) optimum
boiler operation, (5) changing fuel sources, (6) long-term coal contracts, (7) spot
market purchasing, (8) long-term ESP performance reliability, (9) dealing with
changes in environmental codes and regulations, (10) applicability and use of new
technologies. Increasing emphasis is being placed upon high ESP performance for
ultra fine particles and upon emissions' reduction of S02 and NOX. Other recent
developments, e.g., dry S02 scrubber ahead of ESP, boiler or system injection of
chemicals for S02 control or for ash resistivity modification, pressurized fluid-
ized bed combustion, etc., can have important effects on ESP performance. In this
paper we emphasize the fundamentals and the overall systems nature of the electro-
static precipitation process, while reviewing some of the major fireside process
factors and their important influences on the design and performance of ESPs.
15-1
-------
THE EFFECTS OF FIRESIDE PROCESS CONDITIONS
ON ELECTROSTATIC PRECIPITATOR PERFORMANCE
IN THE ELECTRIC UTILITY INDUSTRY
INTRODUCTION
The successful development, circa 1920, of pulverized coal as a basic fuel for
large steam-electric generating plants in the United States was soon followed in
1923 by the first use of an electrostatic precipitator (ESP) to solve the difficult
fly ash collection problem in this service. Today there are about 1500 utility fly
ash ESP installations in the U.S.A. treating a total of over 600 million ACFM
(actual cubic feet per minute) flue gas derived from the combustion of pulverized
coal. The early requirements of 90% efficiency and a unit size of about 270K ACFM
have increased mightily over the past 66 years, and fly ash collection long ago be-
came by far the largest single application for electrostatic precipitators world-
wide.
Modern ESPs are required to perform reliably at overall collection efficiencies
99.5-99.9+% by weight inlet ash, including micron and submicron diameter particles,
in unit plant sizes typically about 500 MW to a maximum of 1300 MW, corresponding
with gas flow rates from -2 to nearly 5 million ACFM at gas temperatures circa 300F.
The ESP is a fairly complex system by itself, but it is always part of an equally
complex industrial process system. Thus, process effluent gas and particulate
properties, along with variations in system factors and operating conditions, not
only strongly affect performance capabilities, but also determine the proper ESP de-
sign approach and specifications.
In this paper, the term "fireside conditions" refers to process factors and condi-
tions in the coal-fired boiler utility industry. It includes all those factors in-
volving coal-ash properties, boiler type and load, combustion conditions, flue gas
properties and particulates entering the ESP. Full knowledge of, and esoteric ex-
pertise in, said fireside conditions and their impacts upon precipitator perform-
ance are essential in such endeavors and areas as: (1) new ESP design, (2) retro-
fit or upgrading performance of an existing ESP, (3) conversion of a troublesome
hot-side ESP to a cold-side ESP, (4) optimum boiler operation, (5) changing coal
sources, (6) long-term coal contracts, (7) spot market coal purchase, (8) long-term
ESP performance reliability, (9) dealing with changes in environmental codes and
regulations, (10) applicability and use of new technologies. Increasing emphasis
is being placed upon high precipitator performance in the 0-5 ym range and espe-
cially in the 0-2 pm particle diameter range where public health, some heavy metals
collection, stack plume and atmospheric visibility effects become most critical.
Other recent developments in fireside conditions, for example, dry type SO2 scrub-
ber ahead of an ESP, boiler injection of additives for S02 removal, boiler or sys-
tem injection of chemicals for ash resistivity or acid plume control, pressurized
fluidized bed combustion, can have important effects on ESP performance.
In the following sections we try to emphasize the fundamentals and overall systems
nature of the electrostatic precipitation process1'2'3, while reviewing some of the
major fireside variables and factors and their important influences on the design
and performance capability of precipitators.
15-2
-------
BASIC SYSTEM FACTORS IN ELECTROSTATIC PRECIPITATION
The ESP is an integral part of an electric power generating system. The coal is
finely ground in the pulverizers and is blown with some hot air into the furnace/
boiler, wherein additional hot air is injected for proper combustion. The products
of combustion with some excess air comprise the flue gas and its entrained residue
fly ash/particulate which pass into the ESP for treatment. The most important gen-
eral case is the cold-side ESP following the air preheaters. The hot ESP system
ahead of the air heater was introduced late 1960's in the U.S.A. as a potential so-
lution to high ash resistivity problems with low sulfur coals (-1% S) at common ESP
cold-side gas temperatures 300-350F. Successful results were indeed achieved with
many Eastern bituminous ash coals at gas temperatures 670-700+F. However, the un-
fortunate transfer of these early results to highly lignitic type ash from certain
low sulfur, Western coals caused great difficulty with high resistivity ash prob-
lems and poor performance1*. Several such troublesome hot ESPs have been converted
to cold-side operation with excellent results . No new hot-side fly ash precipita-
tors on Western coals have been sold in the U.S.A. during the last decade.
Table 1 outlines twelve basic system factors critically involved in the design, per-
formance and operation of electrostatic precipitators. The first five may be con-
sidered basic fireside conditions and process factors including boiler type, load
and various operating conditions which are not under the control of the ESP designer.
Items 6-12 are system factors mainly controlled by the ESP designer, but neverthe-
less influenced and affected by the basic process factors. In Figure 1 are depicted
some interrelationships among principal process and precipitator factors. We have a
certain gas flow rate containing a concentration of fly ash particles of known size
distribution which are to be removed at a prescribed efficiency at a certain effec-
tive particle migration velocity or precipitation rate factor shown as oi/wj,. These
define a specific collecting area (SCA) which, with the gas flow rate, determines
the ESP collecting plate area. Said migration velocity is derived from particle
properties, gas properties and electric field strengths which are controlled by the
fundamental resistivity of the fly ash being treated. At the bottom of Figure 1 we
then have the ESP arrangement with its high voltage electrical energization, auto-
matic controls and rapping. A modern microprocessor TR (transformer-rectifier) set
control system is indicated with energy management to maintain a desired stack
opacity.
An ESP design cannot be any better than the extent of knowledge which its designer
has of the applicable process factors and their effects on performance. Failure to
appreciate the importance of process knowledge and experience can lead to unreliable,
poor ESP performance, often requiring a costly field fix. Just as a coal-fired
boiler type and design must be properly matched to the properties of the coals to be
burned, so also must the ESP design be properly matched to the coal-ash properties,
flue gas characteristics, and boiler/system operating conditions involved. Likewise,
a good working understanding of the primary ESP performance determinants in relation
to basic plant process factors and conditions can greatly ease the work of utility
company personnel in getting boilers, coals, precipitators and system operating con-
ditions that optimally fit together for good results not only for making power, but
also for meeting applicable environmental codes and regulations with high reliabil-
ity.
Major ESP Performance Determinants
Critical precipitator design and performance evaluation factors have been reviewed
in recent papers2'6'7'8'14. Among the most important factors affecting precipitator
performance capability and compliance strategy are these :
15-3
-------
1. Ash Resistivity9'1Q'l^'12 - as controlled by ash chemistry, gas tempera-
ture and density, gas composition - especially H20, SOj/H^SC^ or the pres-
ence of other specific conditioning agents. Resistivity basically controls
the allowable ESP operating current densities, hence operating voltages,
useful power input and size of ESP equipment required.
2 . Inlet Particle Size Distribution and Concentration - as influenced by
boiler type, fuel properties and process factors.
3. Electrical Energization - the heart of the electrostatic precipitation
process comprising the electric field strengths for particle charging and
collection by means of electric forces applied directly to the particles,
per se. Said electric fields and copious supply of negative gas ions for
particle charging are provided by a high voltage corona discharge main-
tained between suitable electrodes, e.g., parallel plate ducts with co-
planar discharge electrodes centered therein. Ion mobilities, voltage-
current characteristics and electric fields are influenced by gas and ash
properties. The electrical energization system and conditions most
strongly influence effective particle migration velocities achievable.
Uniform current distribution yields a premium in ESP performance when every
square foot of collecting surface can do its equal share.
4. SCA (Specific Collecting Area) - ESP total collecting area per unit gas
flow rate. Important influences on SCA include efficiency requirements,
electrode geometry, ESP duct width and electric energy density.
5. Dust Loss Factors - as influenced by particle adhesive and cohesive prop-
erties, gas distribution quality, gas sneakage, local air inleakage, gas
turbulence conditions, gas velocity erosion and rap reentrainment, exces-
sive ESP sparking, saltation, and long term reliability factors.
Thus, we see that fireside or process conditions are intimately involved in deter-
mining the performance of ESPs. These and other factors are strongly and synergis-
tically interrelated yielding performance capability as described by the well-known
Deutsch and Matts-Ohnfeldt13 equations. In high performance, dry-type precipita-
tors, the effective attainable collection efficiencies in practice are often con-
trolled by the variable dust loss factors previously mentioned. It is important
that these losses be minimized by design.
FIRESIDE, PROCESS FACTORS AND EFFECTS
With the preceding background discussion to fix our ideas, we now review in some-
what more detail some of the major fireside conditions and process factor effects
on the performance of electrostatic precipitators in the coal-fired utility
industry.
Gas Flow Rate_
Performance effects are basically determined via the SCA term in the ESP formulae
mentioned above. Treating more gas than necessary tends to reduce precipitator per-
formance capability. For example, precipitators optimally rated at 99 and 99.5%
efficiency could drop to minima of 98.5 and 99.2%, respectively, for a gas flow
rate increase of 10%. The gas flow rate is determined by the composition and amount
of coal burned, the combustion excess air, system air inleakage, and the gas temper-
ature, pressure. Principal air inleakage is usually via the Ljungstrom air
15-4
-------
preheater, and other leakage can be associated with the boiler, flues, expansion
joints, access doors, etc. Sufficient allowance for any expected variations in coal
quality or for wear and ravages of time must be made in original design. Plant
site elevation has significant effects; e.g., at 5000 feet atmospheric pressure is
only about 83% of that at sea level, hence gas flow is increased by -20%. The
seasonal effects of gas temperature can also affect gas flow rate - highest in sum-
mer months. In older, smaller boiler systems designed for relatively high average
ESP gas velocities -6+ ft/sec, the gas flow distirbution quality, per se, becomes
very important in the securing of optimum performance.
Coal-Ash Characteristics - General
Information required includes as-received ultimate coal analyses (% wgt) - C, H2,
N2, 02, S, ash, moisture - plus high heat value BTU/lb; ash mineral analyses (%
wgt) Si02, A1203, Ti02, Fe2C>3 , Ca0- M9°' Li2°> Na20, K20, P2C>5, 303. Average values
of these parameters are insufficient; statistical variations of the factors are re-
quired in order to design ESP performance with some known risk assessment on choice
of coal-ash constituents for what is termed a most probable worst case (MPWC) coal5.
Said MPWC might be one that defines SCA to satisfy required ESP performance for say
95, 98 or 99% of a coal mine field. Analyses of many coal bore samples can define
areas of a mine having poor coal-ash properties to be avoided. Examination of pos-
sible coal sources and properties can determine suitable choices for given perform-
ance of an existing ESP, or requirements for upgrading. Knowledge of critical
factors is clearly valuable for coal specification and purchase as well as for mon-
itoring quality. The slagging and fouling ash characteristics should also be
checked for compatability with the boiler.
Inlet mass concentrations along with emissions regulations determine ESP performance
required. The amount of particulate leaving the boiler varies with coal ash and
heat contents, boiler design and operations. Carryover from a dry bottom boiler
may be 70-85% of total ash generated (including normal few per cent of carbon or
coke type particles), while only 15-25% carryover might be typical of a cyclone
boiler. Increased inlet mass loading generally means increased ESP outlet loading,
possible opacity problems.
Major Ash Resistivity Factors and Effects
Si02, Al^Oj, CaO, MgO. Increase ash resistivity.
Li20, Na20. Reduce resistivity via charge carriers, alkali metal ions. Li20 is
often present in very small amounts, if at all, so that Na20 becomes a major con-
ductive constituent.
Fe2C>3. Enhances charge carrier transport and reduces resistivity.
Gas Temperature. Strong effect on ash resistivity; in the presence of moisture, a
classic bell-shaped curve having typical maxima circa 300-350F with resistivity
falling off quite rapidly at lower temperatures and somewhat less rapidly at higher
temperatures. At low temperatuees (cold-side <-300F+), we have corona current con-
duction mainly over the surface of particles collected on ESP plates; at higher
temperatures current conduction is mainly through the volume of the dust, although
both surface and volume conduction are active.
Moisture in Gas. Reduces resistivity; derived from moisture and hydrogen in the
coal. 5-15% H20 volume in flue gas can easily reduce resistivity by an order of
magnitude. Note that excess air inleakage reduces H20 concentration in the gas,
hence can increase resistivity and reduce ESP performance.
15-5
-------
Availability in Gas. A natural conditioning agent for fly ash resistivity con-
trol which is derived from the sulfur in the coal and 303 generation in the boiler.
SO-, readily combines with H20 and exists as H2S04 gas above the acid dewpoint tem-
perature. H2S04 is stable <400-500F. Ash surface conduction derives from the ad-
sorption of H20 and H2S04 on the surface of fly ash particles. Adsorption processes
become very effective and more efficient at reduced temperatures; S03/H2S04 condi-
tioning effects essentially disappear in region 350-375F depending upon ash chemis-
try. If insufficient S03/H2S04 is present due to low sulfur coals and other site
specific factors, it can be injected into the flue gas to reduce resistivity to op-
timum levels =:l-3xl010 ohm cm. Only small amounts are needed - typically 5-20 ppm
by volume flue gas. 803 conditioning can be effective on both bituminous and lig-
nitic type ashes.
Basic Types of Ash Bituminous Lignitic
Definition Fe2°3 > Ca° + M9° Ca° + M9° > Fe2°3
Sources Typically eastern and Many western coals espec-
midwest bituminous ially Power River Basin,
coals; some western Wyoming; sub-bituminous,
coals lignites
Most difficult ash for ESP Si02 + A1203 >80% ash High CaO + MgO (20-30+%)
Low S coal <1% Low S coals <0.5%
Low Fe203 <5% Low Fe203 <5%
Low Na20 <0.5% Low Na20 <0.5%
Higher temps 330-350F Free lime available
Higher temps 330-350F+
Some Critical Resistivity Factors.
1. Critical resistivity for insipient back corona in ESP.
Cold side <~350F l-3xl010 ohm cm
Hot side -650-750F 2-5xl09 ohm cm
2. Good ESP performance - modest size ESP - cold side, 300F, minimum -0.5%
S04 by wgt of ash in water soluble portion of fly ash. This reflects ad-
sorped acid, favorable resistivity, high power input, high ESP perform-
ance .
3. Sulfur-ash index a =? — - - — - , a useful coal index11.
% ash
For gas temperatures ==300-330F, a less than about 0.1 generally indicates
good possibility of high resistivity ash and limited ESP current densi-
ties in eastern coal, bituminous type ash, standard 9" duct ESPs . Other
things being equal, the effects of 1% sulfur coal and 10% ash would be
about the same as 2% S, 20% ash coal.
4. Na20 ash contents typically minimum 2-3+% wgt at the ESP are required to
yield optimum ash resistivities <1010 ohm cm for low S coals. Addition
of Na2C03 to the coal belt to produce conditioning effects has been and
is being done for some of the low sulfur, lignitic western coal ashes
using hot-side ESPs. Success and happiness can be an iffy thing. Loss
of valuable Na20 due to fouling in the boiler back passes, the sodium de-
pletion phenomena10, thermal electrode distortion problems, and require-
ments for periodic ESP/boiler cleaning have been troublesome and
15-6
-------
expensive. At low boiler loads, hot side, ash resistivity increases and
can reduce ESP performance. On the cold side, ESP performance generally
improves at reduced boiler load.
303 Generation
Sulfur in the coal burns to SC>2 and a small amount is converted to SO^ in the boiler
system. For dry bottom boilers, some reasonable average conversion might be
-1±0.5%. Highest conversion typically occurs in cyclone boilers which run higher
temperatures. Factors affecting the conversion and amount of S03/H2SO4 availability
for conditioning the ash include these:
1. Fire ball temperature.
2. Amount of excess air in boiler and uniformity of distribution. Measurable
increase SO^ from 3 to 4% 02 volume - more 0 atoms available.
3. Presence of catalytic agents - e.g., Fe2O3 in the ash is very effective
-HOOF, or exposed rusty iron surfaces in boiler. ^2^5 also a good cata-
lyst, but little available in fly ash - unlike certain oil-fired boilers
producing significant SO,.
4. Higher S content of coal - more 303 availability.
5. Gas residence times in boiler system.
S03/H2S04 Losses
1. Dropout in cold end of air heater.
2. Absorption in carbon which is a good sink for 303/^304 - high specific
surface.
3. Good sources of carbon are (1) boiler flame hitting back wall of furnace
generates well; (2) >1% wgt coal grind on 50 mesh screen (one mill out of
four can cause a lot of unburned coal with volatiles burned out - lacy
coke type); (3) poor combustion, fluctuating fuel/air ratios at burners.
4. Condensation due to local cold air inleakage; possibly poor or torn insu-
lation, cold winds.
5. Chemical reaction with CaO or MgO producing, for example, a thin layer of
CaS04 which is a high resistivity material, per se. In high CaO ashes,
low S coal, natural 303 may be all used up reacting with free lime. How-
ever, such ashes can be readily conditioned by injecting additional 803
sufficient to adsorb on the thin films of CaS04. At gas temperatures
suitably low, excellent results have been obtained in reducing resistivity
with <10 ppm volume 303 injection for high performance in hot to cold side
ESP conversion cases.
Graphs and Data Illustrating Some of the Effects Discussed Above
1. Figure 2 - shows typical effect of reducing sulfur in coal on performance
of an ESP. Although coal sulfur by itself is not a good predictive meas-
ure of ash resistivity or ESP performance, nevertheless the trends
15-7
-------
indicated based on field results are valid.
Figure 3 - shows fly ash resistivity vs coal sulfur content for several
flue gas temperatures. This illustrates the important effects of gas
temperature at a given coal S content and emphasizes the improved results
available at lower temperatures <300F. Ash resistivity increases rapidly
below =1% S which we classify as low sulfur coal. At SxlO1 •'--lO12 ohm cm,
well developed back corona can be established and ESP performance dras-
tically reduced.
Figure 4 - illustrates the critical effects of ash chemistry on the re-
sistivity of fly ash. Ash samples were taken from field ESPs. Case A is
a high silica + alumina ash with low Na20, but fairly high Fe2C>3 content
of 13.3%. Note that maximum resistivity is -2x10 ohm cm without 803
and the ash is very susceptible to 803 conditioning with 1011, 4 ppm S03
at 375F, dropping to 5x10 9 ohm cm at 300F. Case B has the same peak re-
sistivity with very high silica + alumina, low Na20 but low Fe2C>3 content.
This combination yields a hydrophobic ash not very sensitive to H20 and
S03 conditioning at the higher temperatures. We have seen many cases of
this type in the field. Even at 10 ppm SOj in the gas, the resistivity
remains very high 1012 in the 325-350F range and does not fall to the
optimum level =10 10 ohm cm until <300F. Again, we see the powerful ad-
vantages of being able to control flue gas temperature below 300F with
low S coals. At certain levels, 15-20F change in gas temperature can
mean the difference between poor and good ESP performance.
Figure 5 - illustrates resistivity vs gas temperature and sodium depletion
effects in a hot ESP. The effects of 503 conditioning at cold side con-
ditions are also shown. This is a low S coal bituminous ash with very
high SiC>2 + A1203 = 88%, low Na20 = 0.34%, low CaO + MgO and also low
Fe2C>3 - 5%. Again, we see hydrophobic ash properties. This ESP in the
hot mode as low as 630-650F where performance starting with clean plates
would be all right for a while, but after a few weeks would degrade with
sparking at reduced current densities, requiring lower voltages and re-
duced performance yielding stack opacity problems. As shown in Figure 5,
sodium depletion effects could raise net ESP ash resistivity about an
order of magnitude to 2 or 3x10 10 ohm cm exceeding critical levels for hot
ESP operation. In this case, good ESP performance was restored by chang-
ing the coal supply to get more favorable ash chemistry including a much
higher Fe203 content.
Gas Temperature AT Air Heater Outlet
Due to air heater rotation, gas temperature differentials across the inlet face of
an ESP may be 40 to 100F depending upon plant size and condition of seals. For ex-
ample, the cold side may be running at 260F and the hot side at 350F giving a nomi-
nal average 305F which might be read on a chart in the control room (depending upon
location and number of thermocouples feeding the instrument) . From our discussions
on resistivity vs gas temperature, it is clear that the cold side can have a low
resistivity while the hot side will have a high resistivity likely near maximum.
Cold side might operate at a current density 4-5 times that possible in the hot
side. In this case we can have some different scenarios depending upon ash chemis-
try, coal-ash properties, ESP design.
1. Resistivity on the cold side might be about optimum with increased resis-
tivity on the hot side. Current densities will be controlled by hottest
temperature at minimum levels; so performance capability will depend upon
15-8
-------
how many parallel cells with separate TR sets are used across the inlet
face. Worst case is one large TR set.
2. If one tries to improve matters by 803 conditioning injection in the
total gas stream, the cold side would likely be below acid dewpoint,
leading to (a) very low resistivity and exposure to dust erosion + rap
reentrainment - heavy stack discharge; (b) possible acid condensation,
crusty dust buildup, local corrosion. Reduced resistivity and improved
performance on the hot side would occur, but not enough to overcome
losses on cold side.
3. One has to use graded SO^ injection according to prevailing temperatures
and ash resistivities perhaps none on cold side, increasing to maximum
in hottest area.
4. Best solution might be to mix the gases to get a reasonably uniform gas
temperature across the ESP, and use 803 conditioning as required but much
more efficiently - possibly only in summer months when gas temperatures
are highest.
Particle Size Distribution
Particle size has an important effect on ESP performance via its effect on achiev-
able migration velocities. Finer particles are harder to collect. The minimum per-
formance of an ESP lies in the typical range 0.2-1.0 ym diameter particles. It is
also interesting to note that maximum stack opacity occurs for particles in the
same range encompassing the wavelengths of visible light (0.4-0.7 ym).
Particle size is influenced by coal grinding. For typical bituminous eastern and
midwest coals, desirable grind is typically 70-80% coal through 200 mesh screen and
«1% coal on a 50 mesh screen. This gives a typical statistical average ash as
shown in Figure 6 at geometric median size -12 ym diameter. In general, for these
coals a coarse grind will yield a coarser ash and a finer grind a much finer ash.
For high moisture western coals, coarser grind can yield finer sized ash due to
moisture of coal going quickly to steam and blowing the coal particles apart into
finer size fractions. This is called the popcorn effect. In addition to retaining
much ash in the boiler slag, the cyclone boiler can produce a fine ash for inlet to
the ESP. We have seen cases where an ESP was designed in error on the basis of a
DB boiler ash sizing but used for ash collection on a cyclone boiler. Not surpris-
ing, performance was less than expected.
Figure 6 illustrates some typical particle size distributions in the electric util-
ity industry. Oil-fired boiler, very fine ash -90% <1 ym and typical atmospheric
dust in a large city are shown for comparison. For order of magnitude particle size
effects, a standard coal fly ash ESP design SCA at -99% (no resistivity effects)
would be expected to deliver 85-90% efficiency on oil-fired boiler ash with proper
electrical energization.
Based upon Figure 6, Table 2 summarizes some pertinent fly ash ESP inlet particle
size distribution parameters of interest in ESP design and performance capability.
Figure 7 shows the calculated penetration of a fly ash ESP as a function of particle
size14. Conditions: distribution A (Table 2), 305 mm ducts, 4 fields, gas flow
rate 1900K ACFM, SCA = 297 ft2/1000 ACFM, average ESP current density J = 30 ya/ft2
(no high resistivity problems); overall efficiency 99.68% wgt. Note that minimum
performance -94% occurs with 0.3-0.4 ym diameter particles. The general shape of
curves remains the same with penetration rising or falling according to operating
15-9
-------
current density, SCA, and useful energy density. At the ESP current density J =
10 ya/ft2, it is estimated that an SCA -700 ft2/1000 ACFM might be required to
achieve 99.83% efficiency on a long term reliable basis - e.g., with an ash resis-
tivity approaching about 10 ohm cm.
Other Comment - Process Factors
1. High sulfur coals in combination with low gas temperatures can yield low
resistivity ash difficult to retain - especially with older ESPs where
high gas velocities cause erosion and rap reentrainment losses. Larger
carbon, partially burned coal particles, under these conditions can sal-
tate out of the ESP. Although easy to collect, such particles quickly
lose their charge at the grounded plates, acquire a positive charge, and
head out into the gas stream toward the negative discharge wires where
they get recharged negative, and are recollected. This process can con-
tinue until the particles bounce along out of the collector - a process
called saltation.
2. In cases as in (1) above, NH3 injection following an air heater ahead of
the ESP can be used to neutralize some H2SC>4 and provide some additional
adhesive-cohesive properties to retain ash and improve ESP performance.
3. We have also used NH3 successfully to control acid stack plumes commonly
associated with high sulfur coals (3-4% S). Adequate ESP performance must
be available to collect the ammonium sulfites and sulfates formed.
4. SO^ over-conditioning is as bad as under-conditioning. Automatic control
based upon holding suitable ESP current densities can be effective.
5. Calcium compound injection into the boiler for some SC>2 removal is being
tested for possible application. These measures can increase ESP inlet
dust concentrations by factors 2 to 3 and may increase bulk ash resistiv-
ity by factors 10 to 100. Effects on ESP performance must be carefully
checked and necessary improvements carried out as required. For example,
gas humidification and temperature reduction can be considered. Dry SC>2
scrubber ahead of an ESP can be very effective due to reduced gas temper-
ature and increased moisture content - no high resistivity ash problems in
spite of increased dust loading with calcium reaction products. The con-
comitant reduction in gas flow rate can also prove beneficial.
6. Increasing rotation speed of air preheater can reduce temperature AT
across ESP face in critical cases.
7. Severe boiler tube leaks can cause excess moisture and sticky ESP ash
buildup problems.
8. Cenosphere silica presence in ESP outlet gases can increase opacity due to
low density material (sp. gr. <1) .
9. Pressurized fluid bed combustion would be considered favorable for main-
taining high ESP electric field strengths at relatively high gas tempera-
ture conditions for high performance on fine particles.
10. Boiler light-off with #6 heavy oil and poor combustion may cause unburned
hydrocarbon carryover and deposits on ESP electrodes. After subsequent
heating, said deposits can polymerize into a tough varnish-type layer
having high resistivity properties - hence poor ESP performance on coal
15-10
-------
ash. Use of #2 light oil for light-off cured the problem.
CONCLUSIONS
An ESP in the coal-fired boiler electric power industry is a many splendored thing.
It is capable of performing reliably at high performance to meet present and future
requirements. Much of its splendor derives from the proper integration into its
design of the major fireside conditions and process factors reviewed in this paper.
Said conditions and factors are important parts of an overall basic systems approach
which must be adopted to insure the performance capabilities required to fit with
the variable coal-ash properties, flue gas characteristics, and boiler/system oper-
ations that must be dealt with.
REFERENCES
1. H. J. White. Industrial Electrostatic Precipitation. Addison Wesley, Reading,
Mass., 1963.
2. H. J. White. "Electrostatic Precipitation of Fly Ash." Journal Air Pollution
Control Assn., Vol. 27, Nos. 1-4, 1977.
3. H. J. Hall. "Summary Overall Concepts for the Design, Energization and Control
of Electrostatic Precipitators." Paper No. 57-6, presented at 76th Annual
Meeting of APCA, Atlanta, Ga., 20-24 June, 1983.
4. J. A. Hudson and P. B. Crommelin, Jr. "Hot-Side Precipitators? (How Utilities
Got Burned)." Presented at Workshop on Hot-Side ESP Technology, Birmingham,
Ala., 18-20 May 1987, sponsored by EPRI.
5. H. J. Hall and N. W. Frisch. "Hot Precipitators - How to Cool Them Off."
Paper presented at Workshop on Hot-Side ESP Technology, Birmingham, Ala.,
18-20 May 1987, sponsored by EPRI.
6. H. J. Hall. "A Selected Review of Critical Parameters and Formulae for Design
and Performance Evaluation of Electrostatic Precipitators." Presented at Sixth
Symposium on the Transfer and Utilization of Particulate Control Technology,
New Orleans, La., 25-28 Feb. 1986. Proceedings EPRI CS-4918, Vol. 2, Project
1835-12, Nov. 1986, p. 12-1.
7. H. J. Hall. "Some Electrode Geometry, Electric Field, and Performance Effects
in Electrostatic Precipitation." Proceedings of the Second International Con-
ference on ESP, Kyoto, Japan, Nov. 1984, p. 354. Published by Air Pollution
Control Assn., Pittsburgh, Pa.
8. H. J. Hall. "High Voltage Power Supplies and Microprocessor Controls for
Electrostatic Precipitators." Proceedings First International Conference on
ESP, Monterey, Calif., Oct. 1981, p. 668. Published by APCA, Pittsburgh, Pa.
9. H. J. White. "Resistivity Problems in Electrostatic Precipitation." J. Air
Pollution Control Assn., 24:314, 1974.
10. R. E. Bickelhaupt. "Electrical Volume Conduction in Fly Ash. J. Air Pollution
Control Assn., 24:251, 1974. Also, "An Interpretation of the Deteriorative
Performance of Hot Side Precipitators," ibid, 30:882, 1980.
15-11
-------
11. H. J. Hall. "Fly Ash Chemistry Indices for Resistivity and Effects on Electro-
static Precipitator Design and Performance." Fourth Symposium on Technology
Transfer, Houston, Texas, 11-15 Oct., 1983. EPA-600-9-84-Q25b, Nov. 1984.
12. R. E. Bickelhaupt. "An Improved Model for Predicting Fly Ash Resistivity."
Proceedings Sixth Symposium Technology Transfer, New Orleans, La., 25-28 Feb.
1986 - EPRI CS-4918, Vol. 2, Project 1835-12, Nov. 1986, p. 11-1.
13. S. Matts and P. Ohnfeldt. "Efficient Gas Cleaning with SF Electrostatic Pre-
cipitators." Bulletin of A. B. Svenska Flakt Fabricken, Stockholm, Sweden.
14. H. J. Hall. "Critical Electrostatic Precipitator Technology Factors for Very
Fine Particle Collection." Proceedings of the Third International Conference
on Electrostatic Precipitation, Abano, Italy, October 1987, p. 763-782,
Massimo Rea, Editor, University of Padua.
15-12
-------
Process
Factors
Keyboard
Entry
Central
Computer
Monitor
CRT
Vendor
Bid
Analyses
Selection
Other Digital
Inputs
Figure 1. Factors in Electrostatic Precipitation
15-13
-------
23
O
H
EH
U
100
95
90
85
W
o
K
W
I
>H
m
s 80
PH
PH
H
75
8 70
65
60
Constant ash content and gas flow,
gas temperature -300+F.
Initial design @ 99%, 3% S, typical
for eastern coals.
I
0.5 1.0 1.5 2.0 2.5
% SULFUR IN COAL
3.0
3.5
Figure 2. Typical Effects of Sulfur Content in Coal on
Performance of an Electrostatic Precipitator
COAL SULFUR, % WGT
Figure 3. Resistivity vs Coal Sulfur
Content for Several Flue Gas Tempera-
ture Ranges (after White2)
-------
Computer calc. based
on work of Bickelhaupt
100
150
200 250 300
TEMPERATURE, F
350
400
450
Figure 4. Resistivity vs Temperature and
Two Different Ash Chemistries, Low S Coals
Effects for
15-15
-------
10
13
10
12
O
10
11
10-
Computer data on Bickelhaupt's work
200
Ash Analyses
. No SO
300
CaO
Fe203
MgO
SiO2
Na2°
^12°3
S03
K2°
Li2O
P2°5
Ti02
LOI
0.917 % wgt
4.97
0.634
58.80
0.34
29.20
0.61
2.81
0.023
0.07
1.68
Sodium depletion
case - hot ESP
400 500
TEMPERATURE, F
600
700
800
Figure 5. Ash Resistivity vs Gas Temperature and Sodium Depletion
Effects in a Hot Precipitator
15-16
-------
99.9
10
0.1
1 10
PARTICLE DIAMETER, pm
100
o
dp
§1
H
W
ft
0.1
.1
PARTICLE DIAMETER, ym
Figure 6. Typical Fly Ash Particle Size Distributions ESP Inlet
Figure 7. Fly Ash Precipitator Pene-
tration vs Particle Size
Conditions: distriution A, Table 2;
305 mm ducts; SCA = 297; -300F; ESP J =
30 pa/ft2; overall efficiency 99.
-------
Table 1
PRECIPITATOR BASIC SYSTEM FACTORS
1. Efficiency requirements, dust concentration in plus out, applicable codes
and regulations.
2 . Gas flow rate.
3. Gas composition, temperature, pressure, density, viscosity.
4. Coal-ash properties, sources.
Particle size distribution, ash chemistry, resistivity.
5. Other process factors.
6. Gas distribution quality.
7. Physical design of ESP, type, gas velocity, electrode system and arrangement.
8. HV power supply design - automatic control.
9. Electrical energization quality.
10. Ash buildup on electrodes, rapping, reentrainment, dust loss factors.
11. Hoppers and ash removal system.
12. Reliability - maintenance.
Table 2
TYPICAL FLY ASH ESP INLET PARTICLE SIZE DISTRIBUTION FACTORS
_ _ Norn Nom
Xg Xp Xa S0 Inlet S % wgt
Application ym ag ym ym m2/g g/m3 m2/m3 <1 ym
A. DB boiler, avg
bituminous ash 12 3'33 2'82 5'82 °'45 6'31 2"84 2'°
B. DB boiler
lignitic ash 8 3'3 l'92 3'92 °'67 4'67 3•13 4'°
C. Cyclone boiler 3.5 3.0 1.05 1.91 1.25 1.62 2.03 13
Xg = geom. median diam., ag = geometric stnd. deviation, Xp = modal diam., Xa =
average diam., SQ = specific surface, S = specific particle surface density.
Nominal ESP gas temperature 150C (-300F).
15-18
-------
OBSERVATIONS OF MODELED AND LABORATORY MEASURED RESISTIVITY
Roy E. Bickelhaupt
Clean Air Engineering, Inc.
207 North Woodwork Lane
Palatine, Illinois 60067
ABSTRACT
Users of the resistivity model, EPA-600/7-86-010, have expressed some confusion
when evaluating potential supplies of low-sulfur coal. This paper explains certain
characteristics of the model and offers useful techniques. Laboratory measurements
leading to an unexpected result regarding a hot-side precipitator are also discussed
16-1
-------
OBSERVATIONS OF MODELED AND LABORATORY MEASURED RESISTIVITY
INTRODUCTION
The challenge to industry regarding present and future limits for emissions of
sulfur and nitrogen oxides has produced a variety of approaches with which the
limits can be met. Among the approaches with respect to sulfur oxides is the
simple switch to a coal of lower sulfur content. This often involves the use
of a resistivity model (1) to evaluate candidate fuels. Some users have experienced
difficulty employing this aid. Several facets of the resistivity model will be
discussed to illustrate certain characteristics and useful techniques. The second
subject of this paper relates to what might be described as a serendipity experience
regarding an ash resistivity measurement.
MODELED RESISTIVITY
Switching to certain low-sulfur coals can yield fly ash having low concentrations
of both calcium and iron. Under these circumstances, the resistivity model is
very sensitive to these elements. Figures 1 and 2 are the modeled resistivity
values for two generally similar ashes that differ principally in iron concentration.
Examination of these figures reveals that the resistivity of the ash containing
a moderate level of iron is attenuated to a greater degree by sulfuric acid vapor.
Whether this is due to the affinity of the ash for acid adsorption or the mobility
of the charge carrying species is not known. It suggests two areas of required
research. First, an investigation of the adsorbed acid conduction mechanism,
and second, experimentation to make given ashes more susceptible to acid
conditioning.
Users of the resistivity model often do not have available data in graphic form
and rely on tabulated data. Sometimes an individual resistivity value for a given
temperature and acid concentration is computed. This approach seems to be used
when coals are screened for a switch to lower sulfur. If enough coals are
considered, the user is often surprised to find that, for a given set of conditions,
16-2
-------
the resistivity of two ashes can differ greatly even though ash chemistry has
changed an apparently negligible amount.
Figures 3 and 4 illustrate this situation. Resistivity was calculated for an
ash composition containing 3% alkaline earth elements and an iron concentration
varying from 3% to 7%. The increase in iron was concomitant with a decrease in
aluminium. This variation is common for ashes from Eastern U.S. bituminous coal.
In Figure 3, resistivity as a function of iron concentration is presented for
a given temperature and various levels of acid. In Figure 4, the acid concentration
is constant and the effect of temperature is illustrated. Both presentations
demonstrate the deleterious effect of low levels of iron. It is obvious that
unusually large acid concentrations or operational temperatures near the acid
dew point would be required to derive a satisfactory resistivity level. The
situation is usually combated by injection of proprietary conditioning agents
or the dual injection of ammonia and sulfur trioxide.
Although Figures 3 and 4 show the effect of iron on acid resistivity to be a step
function, this is unlikely. The experimentation to develop the model was based
on ten ash compositions and established slightly more than 100 acid resistivity
data points. Two of the ashes contained low levels of iron and alkaline earth
elements and produced the data illustrating the poor response to acid conditioning.
A larger data base would perhaps alter the appearance of the data shown by Figures
3 and 4. Additional factors may also have influence on the end results. However,
data obtained subsequent to the establishment of the resistivity model have not
refuted the general trend shown above.
The important point regarding the above discussion is that low-alkaline ashes
containing nominally 10% iron have a high probability of being successfully
conditioned either naturally or by injection of sulfuric acid vapor. When selecting
a low-sulfur coal, it is simply not prudent to use a fuel that produces a low-iron
ash unless one is prepared to deal with conditioning and precipitation problems.
On the subject of the resistivity model, two other points will be made. First,
in Table 5 of reference 1 the criteria are listed for the selection of the slope
of the acid resistivity curve as a function of reciprocal absolute temperature.
Observations made since the model was published suggest that, in the case of all
three ash compositional characteristics, the magnesium plus calcium concentration
should be 5.0%. With this change, the model shows greater agreement with
16-3
-------
observation, and it becomes somewhat more conservative.
The second point concerns the estimation of sulfuric acid in the flue gas. The
resistivity model is designed to predict resistivity as a function of temperature
and environmental conditions. If sulfuric acid plays a part in the conduction
process, the estimation of the concentration of this agent is critical to an
accurate resistivity prediction. The model offers an estimation of sulfuric acid
vapor to be found at the precipitator inlet based on a combustion calculation
using the ultimate coal analysis to determine sulfur dioxide and field measurements
to establish a ratio between sulfur dioxide and sulfur trioxide for acidic and
basic ashes. Although this technique gives reasonable results a high percentage
of the time, it can also be inaccurate (2). As more observations were made, it
became clear that the ratio of sulfur trioxide to sulfur dioxide is site specific
and can vary more than an order of magnitude. The effect can be even greater
in terms of resistivity.
The author has encouraged utility users of the resistivity model to determine
the ratio of sulfur oxides as a function of load for each unit. Accurate knowledge
of this environmental factor along with information regarding temperature and
environmental stratification can be of great service. An example of how this
can be put to use by control room and precipitator personnel is shown in Figure
5. This type of aid can be constructed using the resistivity model and precise
information regarding the sulfur oxide ratios. With this type of figure and
continuous monitoring of temperature and sulfur dioxide, one can immediately
evaluate the effect of resistivity with respect to precipitator problems.
LABORATORY MEASURED RESISTIVITY
Hayden Station of Colorado-Lite burns a low-sulfur coal and collects the ash with
precipitators located on the hot side of the air heater. The ash contains a low
level of sodium and, in the operating temperature range of 340°C (644°F) to 370°C
(698°F), has an inherent resistivity of about 2 E10 ohm cm. After sodium.depletion
has occurred, the effective resistivity is >1 Ell ohm cm. The Hayden Station
was one of the first at which the symptoms of the sodium-depletion phenomenon
were documented.
Both units at Hayden are conditioned with anhydrous ammonia which is injected
into the flue between the economizer and the precipitator inlet. This method
has been in use for years, and, while possessing limitations and disadvantages,
it has met with some success. Several months ago the ash sodium concentration
16-4
-------
reached a very low level straining the ability of the ammonia conditioning system.
Sodium conditioning of the coal came under consideration, and the writer became
reacquainted with the problem.
Baseline laboratory resistivity data were obtained for comparison with modeled
sodium-depleted values in the process of determining a desirable level of sodium
to be added to the coal. The results are shown in Figure 6. Data shown with
square symbols represent the high-temperature part of an IEEE Standard 548-1984
resistivity test (3) using the descending temperature technique. This test calls
for the ash to be annealed at 450°C (842°F) in dry air for about 14 hours. Modeled
(4) inherent and sodium-depleted resistivity values are also shown. The
agreement between the measured and predicted inherent data is very good. When a
laboratory test was being set up to substantiate the modeled sodium-depleted
resistivity, an ash sample was equilibrated with the test environment at 362°C
(684°F). The sample temperature was raised from room temperature to the test
temperature in two hours. It was anticipated that the initial resistivity value
for this test would be about 2 E10 ohm cm. Unexpectedly, the measured value was
about 2 E9 ohm cm. This is shown as a triangle in Figure 6. Although it was
known that ammonia was injected, this gas had been found ineffective with respect
to resistivity attenuation (5).
The unusual result was further examined with the Standard IEEE 548-1984 test
using both the ascending and descending modes. Between the ascending and
descending measurements, the sample is annealed for about 14 hours as previously
stated. The elapsed time for an ascending and descending temperature test is about
5 hours for each. The data are given in Figure 7. It is apparent that a rather
uniform hysteresis developed between the ascending and descending temperature
resistivity curves. Except at temperatures above 400°C (752°F), the ascending
temperature values are about one order of magnitude lower than the descending
data. Above 400°C (752°F), the effect is somewhat diminished as the conditioning
effect is destroyed. The desirable effect is obvious. At 370°C (698°F), the
inherent resistivity would be about 2 E10 ohm cm, and the effective resistivity,
after sodium depletion, would be about one order of magnitude greater resulting
in performance problems. As shown, the conditioning effect produces an initial
resistivity level of 2 E9 ohm cm.
To determine whether the conditioning effect could be destroyed at temperatures
lower than the usual annealing temperature of 450°C (842°F) associated with the
IEEE resistivity test, an additional experiment was conducted. An ash sample was
16-5
-------
quickly brought from room temperature to 265°C (508°F), and resistivity was
determined. The sample remained at this temperature for 24 hours without applied
potential after which the resistivity was again determined. Temperature was then
increased to the next level and then to a final value as shown in Figure 8. At
each temperature, the sample was annealed for 24 hours. It appears that the
deterioration of the conditioning effect occurs to some extent at 352°C (666°F).
Apparently in the operational temperature range for the Hayden precipitators there
is a contest between the formation and destruction of the conditioning agent.
A limited investigation was conducted to identify the cause of the observed effect.
Specific ion electrode results indicated very low levels of ammonia in solutions
prepared from Hayden ash. The concentration increased as the fly ash particle
size decreased. It also was noted that resistivity decreased with decreasing
particle size. The Kjeldahl procedure for nitrogen determination yielded a value
of about 0.04% with the nitrogen expressed as (NH.)-SO.. Finally, a single mass
spectrum was determined. The major phases released on heating the ash were ammonium
sulfate (NH ) SO., ammonium valerate NH C H 0 , and their decomposition products.
The ammonium valerate was released spontaneously at <150°C (302°F), and the
ammonium sulfate was released over a long period of time at a temperature >350°C
(662°F).
The collection of observed conditioning mechanisms includes space charge enhance-
ment, increase in ash cohesiveness, and modification of resistivity. For the
subject case, one would not intuitively assume the ammonia injection would benefit
the process by resistivity modification. Ammonia per se does not influence
resistivity, secondary voltage/current curves, or suppress positive ion generation
at hot-side precipitator temperatures (5). On the other hand, it is known that
ammonium compounds co-precipitated with fly ash can appreciably attenuate
resistivity (5). It has also been shown that the ammonium ion readily serves as
an ionic charge carrier in fly ash (6).
Previously, resistivity attenuation due to ammonia injection was not seriously
considered because the handbook decomposition temperature is 280°C (536°F),
considerably lower than the Hayden precipitator temperature. However, this means
of conditioning was suggested in reference 5 for cold side precipitators operating
at temperatures too high for successful acid conditioning even though the flue gas
contained significant sulfuric acid vapor.
Studies of ammonium sulfate decomposition (7) indicate that, under conditions of
16-6
-------
simulated flue gas and short reaction times, the decomposition is not complete in
the temperature range of 315°C (600°F) and 371°C (7006F). The present study
suggests that the ammonia injection leads to the formation of ammonium sulfate
either on the ash surface or as particulate to be co-precipitated.
Assuming that the observations detailed herein withstand additional research,
one can suggest the combined use of ammonia and sulfuric acid vapor as a condition-
ing procedure for cold-side precipitators operating at higher than normal
temperatures and hot-side precipitators operating at lower than normal temperature.
Injection of both agents is presently used to condition ashes of low iron concen-
tration that are collected at normal cold-side temperatures. In the case at Hayden
with a flue gas containing little sulfuric acid vapor, one contemplates the effect
of an injection of sulfur trioxide to compliment the ammonia injection.
ACKNOWLEDGMENT
The writer greatly appreciates the encouragement and cooperation given by Charles
Nelson, Curtis Elwood, and Charles Hogue of the Hayden Station of Colorado-Ute
Electric Association, Inc.
REFERENCES
1. Roy E. Bickelhaupt. "A Study to Improve a Technique for Predicting Fly
Ash Resistivity With Emphasis on the Effect of Sulfur Trioxide." EPA-600/7-
86-010, NTIS PB 86-178126, 1986.
2. W. A. Harrison et al. "Medium-Sulfur Coal and Fly Ash Resistivity." JAPCA
38 (2) pp. 209-216.
3. IEEE Standard 548-1984. "Criteria and Guidelines for the Laboratory
Measurement and Reporting of Fly Ash Resistivity." The Institute of
Electrical and Electronic Engineers, Inc. 345 East 47th Street, N.Y., N.Y.
10017.
4. Roy E. Bickelhaupt, "A Method for Predicting the Effective Volume
Resistivity of Sodium-Depleted Fly Ash Layers in Hot-Side Electrostatic
Precipitators." CS 3421. Electric Power Research Institute, Palo Alto,
California, March 1984.
5. Roy E. Bickelhaupt, et al. "Flue Gas Conditioning for Enhanced Precipitation
of Difficult Ashes." FP-910. Electric Power Research Institute, Palo Alto,
California, October 1978.
6. Author's unpublished data file.
7. Ralph F. Altman et al. "Method for Flue Gas Conditioning With the
Decomposition Products of Ammonium Sulfate or Ammonium Bisulfate." United
States Patent #4,533,364 August 6, 1985.
16-7
-------
FIELD
INTENSITY
ASH
COMPOSITION
O W/0 S03
D WITH 2.0 PPM S03
A WITH 1.0 PPM S03
•f WITH 4.0 PPM S03
1120
NA20
K20
MGO
CAO
FE203
AL203
S102
T102
P205
S03
E - 12 kV/CM
FLUE GAS
X WITH 10.0 PPM S03
1 .4
441 *C
826 «F
2.2 ' 2.'0
182 227
359 441
TEMPERATURE
Figure 1. Modeled resistivity for an ash of moderate iron concentration
produced from a low-sulfur coal
16-8
-------
FIELD
INTENSITY
ASH
COMPOSITION
W/0 S03
WITH 2.0 PPM S03
WITH 1.0 PPM S03
0 PPM S03
E « 12 kV/CM
FLUE GAS
1120
NA20
K20
MGO
CAO
FE203
AL203
S102
T102
P205
S03
10.0 PPM S03
2-'2 ' 2.0 1.8
182 227 283
359 441 541
TEMPERATURE
1 -4
441 «C
826 «F
Figure 2. Modeled resistivity for an ash of low iron concentration
produced from a low-sulfur coal
16-9
-------
1012
108
% Fe203 in Fly Ash
Figure 3. Effect of iron on resistivity at 152°C (306°F) in a flue gas
containing 9% water vapor and the acid concentration shown
16-10
-------
% Fe203 in Fly Ash
Figure 4. Effect of iron on resistivity in a flue gas containing 9%
water vapor and 4 ppm of acid vapor for the temperatures shown
16-11
-------
0
0
500
2
1000
4
1500
6
2000
2500
10
SO
SO & SO in ppm
L. O
Figure 5. Resistivity modeled for troubleshooting based on a specific
coal and furnace operation
16-12
-------
10
11
u
5
o
—
tn
UJ
E
101°
109
108
1000/°K
°C
°F
1.8
283
541
1.7
315
599
• Modeled, Sodium Depleted
• Modeled, Inherent
1.6
352
666
1.5
394
741
1.4
441
826
• Lab-Measured, Annealed
A Lab-Measured, As-Received
Environment: Air + 9% Water E - 4 kV/cm
Figure 6. Modeled and lab-measured resistivity of typical Hayden fly ash
16-13
-------
TEMPERATURE,°F
KT1
8
6
4
2
10
8
6
4
2
10 '
8
6
4
8
6
8
6
io8
8C
4
£*
) »
2OO
* i __t i a-
f
p
y
2.7
3 1C
^-
/
/
/
'
\^
s*
'*
\
300
^— — -
•^^
'
-•-
— ^
,
' .
*s
| -^^,
2.6 2.5 2.4 ' 2.3 2J
10 150 2(
4OO
1
i-^.
^
^
^
s* ,
x
50
JK,
0
"V
X
s
1 (
\
600 700
Hayden Fl v Ash
Collected With
Ammonia Conditioning
MEAN PARTICLE SIZE (^Jn
TRUE DENSITY g/cm3|
il
GAS PHASE TEST EXPECTED
O2. Vo %
C02. Vol
H20. Vol
S02, Vol
S03. Vol.
:EEZ
% -
ASH LAYER DATA
BULK DENSITY (g/cm3!
POHOSITY Icm^/cm3]
ELECTRIC FIELD IkV/cml
»RE
\
MARKS
S,
^
2.1 2.0 1.8 1.8
» 30
\
N
4
8OO 9OO
WEIGHT PERCENT
L.jO
N.,0
KjO
MjO
CsO
Al2<>3
S,02
T,02
S03
LOI
N
k^
W
s
\
\
i
1.7 1
«
t
N
1
\
.
\
s
\
\
s,
6 1.6
41
Sd
s.
^
n
^
N
K-
p
1.4
w
k
^
1.3
&
DO
1000
TEMPERATURE. °C
Figure 7. Resistivity in accordance with IEEE Standard 548-1984, ascending and descending modes
-------
10"
5
o
5
I
O
v>
LU
oc
1000/°l<
°c
°F
1.8
283
541
1.7
315
599
1.6
352
666
1.5
394
74l
1.4
441
826
Figure 8.
• Initial data at a given temperature
• After 24 hours without applied potential
Effect of time on the dissipation of the conditioning
phenomenon
16-15
-------
COMPUTER MODEL DEVELOPED TO PREDICT ESP PERFORMANCE
BASED ON COAL QUALITY
Wm. Borowy
(No paper provided)
17-1
-------
ENGINEERING STUDY ON WIDE PLATE SPACING
ELECTROSTATIC PRECIPITATORS
K. S. Kumar
Dr. P. L. Feldman
Environmental Services & Technologies
Research-Cottrell, Inc.
P.O. Box 1500
Somerville, NJ 08876
ABSTRACT
Results presented in this paper are derived from a study spon-
sored by East Kentucky Power Cooperative and Electric Power
Research Institute in 1988. The study included a) a literature
search on wide plate spacing precipitator technology and b) a
technoeconomic study on the upgrade of an existing electrostatic
precipitator at EKPC's Dale Station for various plate spacings
from 9" to 24"-
This paper begins with a survey of the various process factors
affecting electrostatic precipitator performance at wider plate
spacing. Technical guidelines in the choice of high voltage
power supplies and electrical clearances are then recommended.
Results of the cost analysis at plate spacings up to 24" are
presented using 12" spacing as a reference point. The cost
analysis discusses the various elements affected by the departure
from 12" spacing to larger spacings. The associated changes in
material costs, installation costs and power supply costs at
different plate spacings are discussed.
BACKGROUND OF WIDE PLATE SPACING ESPS
Until recently it had been common practice in the U.S. utility
industry to use electrostatic precipitators for flyash collection
with parallel collecting plates spaced about 9" apart and wires
as discharge electrodes. Several thousand electrostatic precipi-
tators were built this way, collecting dust from a variety of
sources under varying conditions. This "conventional" ESP had
served the U.S. utility industry for several decades. However,
during the late 1960's,, changing environmental regulations
signalled the era of high efficiency, high reliability precipita-
tors. Utilities were required to adhere to strict emission
control requirements on a continuous basis. Any failure of ESP
could increase emissions and necessitate a load reduction in
18-1
-------
order to comply with regulations. Thus, the market was looking
for ESP manufacturers who could provide a new level of reliabili-
ty in ESP hardware.
European and American manufacturers filled this need by offering
rigid discharge electrodes between plates spaced about 12" apart.
The apparent success in performance of the 12" plate-spacing
precipitators in the U.S. kindled sufficient interest among ESP
manufacturers to probe the effect of wider plate spacing. There
is now little question that the economic optimum plate-spacing is
greater than 12". What the optimum is and how it varies with
operating conditions are questions which must be answered on a
case-by-case basis until more experience with wide-plate spacing
is gained.
LITERATURE SEARCH ON WIDE PLATE SPACING TECHNOLOGY
In this section, we discuss the theoretical, experimental and
full scale results reported by researchers from U.S.A., Europe
and Japan. The reports cover a period from 1974 to 1987.
Wide Plate Spacing Studies By Research-Cottrell
Research-Cottrell has been involved in the study of wide plate
spacing effects since the early seventies in order to understand
the possible benefits. Feldman1 studied the theoretical effects
of wide plate spacing in 1975 and made the following observa-
tions:
1. Wide plate spacing results in lower plate area per
volume. Qualitatively the amount of reentrained dust
per volume of gas could be expected to decrease as
plate spacing increases. This effect could be impor-
tant in high efficiency precipitators where reentrain-
ment accounts for a significant portion of precipitator
loss.
2. Use of higher voltages with wide plate spacing, coupled
with increased space charge due to larger inter-
electrode spacing results in greater field strength at
the collecting electrode. Thus the migration rate
parameter can be expected to increase with plate
spacing, allowing collection efficiency to be main-
tained with reduced collection area.
3. Wide plate spacing benefits for high resistivity are
probably less than those for lower resistivity because
of a theoretically lower rate of voltage rise and space
charge with plate spacing at high resistivity.
4. Turbulent diffusion of particles may play a large role
in determining precipitator performance. Feldman's
model pointed towards a wider optimum plate spacing
than indicated by the Deutsch model which assumes
complete mixing of particulate.
18-2
-------
In 1975, Kumar, K. S.2 conducted experimental investigations in a
single duct laboratory precipitator using 9", 12" and 14" plate
to plate spacings with flat plates using 0.109" diameter wires.
Tests were conducted at particulate resistivities of 109 and 1012
ohm cm. These tests showed that the modified migration velocity,
wk, varied roughly proportionally with plate spacing for both
resistivities.
Later, Research-Cottrell worked jointly with a Japanese company3
exploring the effect of plate spacing up to 24" in a clinker
cooler cement dust environment. Several discharge electrode
geometries were studied for moderate, intermediate and high
resistivity conditions. In all cases, the migration rate parame-
ter increased with plate spacing beyond 9", but depending on
resistivity and electrode geometry, the parameter sometimes
reached an optimum between 12" and 24". Both electrode geometry
and resistivity had a significant impact on the migration rate
parameter and its relationship to plate spacing.
More recent experimental work4 confirmed past results that op-
timum plate spacing varies depending on discharge electrode
geometry and resistivity.
Work done by G. Coopeirman5 at Research-Cottrell on a review of
non-classical trends in wide plate spacing pointed out that
changing electric field profiles in the ESP due to changing plate
spacing result in increased collecting field strengths. He also
described the importance of resistivity in determining the effect
of wide plate spacing: a) high dust concentration can cause
severe corona suppression, especially at wider plate spacings.
Thus, it is possible that particle charge limitations under
severe corona suppression could seriously limit efficient par-
ticulate collection, in spite of increased collecting field
strengths due to space charge effects. Also, the finer the
particle size distribution of the dust, the closer would be the
optimum plate spacing; and b) the effect of corona suppression
with wider plate spacing would be further aggravated if high
resistivity dust conditions prevail in the precipitator. The
precipitator would be limited by low current densities due to
back corona effects and the reduced ion densities may be further
limited in particle charging by the corona quenching problem at
larger plate spacings.
Cost comparison studies on the use of wide plate spacing were
made by Research-Cottrell's E. DeHaas and D. V- Bubenick7, and
established significant cost advantages by going to wider plate
spacing. At wider spacings, mechanical elements in the precipit-
ator are significantly reduced in number, leading to both labor
and material cost savings. However, increased costs of power
supplies and insulators and greater electrical clearances tend to
offset the savings. It was found that, for reasonable assump-
tions on the relationship between migration and velocity and
plate spacing, the economic optimum probably lies between 16" and
20" plate spacing. The optimum is probably in the low end of the
range for high resistivity and may be higher than 20" if optimis-
tic values for migration velocity are used.
18-3
-------
Work Performed By Researchers Other Than Research-Cottrell
The conclusions of Japanese and European work do not differ
significantly from Research-Cottrell's analysis of wide plate
spacing effects:
« Space-charge effects must be addressed during the
choice of discharge electrodes and plate spac-
ing.16
« All suppliers offer 300 mm precipitator plate
spacing. Earlier standard offerings were in the
200-250 mm range.20' 21
• 400 mm plate spacing ESP's are now common in
Europe for most applications.19
• Both theory and practice have shown that migration
velocity in the precipitator can increase propor-
tionally with plate spacing.9
» In almost all cases, collection efficiency at 300
mm was maintained as plate spacings were increased
to 500 mm.19
• In fact, collection efficiency could be maintained
with half the plates in a given field, if same
residence time could be maintained. In these
cases, the total power to the ESP was kept ap-
proximately the same as at lower spacings.20
• Theoretical analysis and operating data suggest
that wider spacing will result in higher average
field strength in the interelectrode space and at
the collecting plate.22
• Large dust concentrations and finer particle size
adversely affect wide spacing ESP's. This is
because space charge increases at wide plate
spacing and tends to suppress corona generation.
The resulting inability to maintain overall power
level leads to adverse performance. In such
cases, closer plate spacing is recommended in the
inlet fields. ' 20
• Sub-pilot scale research at Arapahoe Power Station
on coal-fired flyash sponsored by EPRI suggests
that better control of back corona and operating
power is achieved at wider plate spacings than
Q tl "
• Better electrical control at high resistivity was
found due to better corona current distribution on
collecting plates.
• Optimum wire to wire spacing (if wire discharge
electrodes are used) exists for difference plate
spacings.
18-4
-------
At high concentrations of fine dust, high space
charge induced field strength leads to sparking
beyond a critical plate spacing level, disallowing
a proportional increase in operating voltage with
plate spacing.
It may be necessary to change the geometry of
discharge electrodes at wide plate spacings to
maintain the overall power consumption and hence
efficiency.
In full scale ESP's practical benefits, such as
reduced severity of misalignment and better cur-
rent distribution, can lead to the application of
power levels higher than at lower plate spacings,
and hence to overall superior collection efficien-
cies. '
WIDE PLATE SPACING COST STUDY RESULTS
Research-Cottrell recently completed the rebuild of existing
electrostatic precipitators of the 9" spacing, weighted wire
design at Dale State Units 3 and 4. The Dale Station is located
on the Kentucky River near Boonesboro, Kentucky. Poor internal
condition of these precipitators, built more than 25 years ago,
had resulted in performance problems necessitating a rebuild.
The rebuild of Dale Station Units 3 and 4 was completed in 1987
using the 12" spacing precipitator technology that utilized the
more durable rigid discharge electrode design. Cost for this 12"
spacing ESP was used as a basis to project the estimates at wider
plate spacings. Table 1 describes in detail the design condi-
tions and specifications for the existing ESP's at Dale Station
units 3 and 4.
While performing the cost comparisons for wider spacing technol-
ogy one of the two units (Dale Station 3 or 4) was selected for
study. It was assumed that the entire ESP internals need to be
removed and replaced with new components. For increased plate
spacing, all the power supplies and associated high voltage pipe
and guard were replaced with higher rated TR's. All associated
high voltage support bushings and feed through insulators were
also replaced. No changes were made to precipitator shell,
associated ductwork, hoppers and support steel. The entire
precipitator was replaced with wider duct passages. The length
of ESP (treatment time) was maintained the same for all plate
spacings.
Cost analysis was performed at two levels of electrical sec-
tionalization: a) one that maintained the same number of TR's
(8) per ESP for all plate spacings, and (b) another that used
only four TR's per ESP for the wider spacing ESP's. The costs
associated for each ESP include the following elements: a)
engineering and project management, operation and start-up b)
material costs that included ESP internals, power supply and
associated insulators and c) labor costs. Labor costs included
demolition of ESP internals and removal of old TR's, installation
of ESP components, and raising of precipitator roof for the case
18-5
-------
of 24" spacing. The total installed costs for plate spacings of
9", 16", 18" 24" were then compared with 12" spacings. The
accuracy of estimates are within ± 15"%.
The cost of wide plate spacing ESP is affected by the requirement
to apply higher voltage on discharge electrodes in proportion to
inter electrode spacing. Attendant with wide spacing is the
requirement to maintain minimum clearances between all high
voltage to ground surfaces. Figure 1 plots the KV peak values
for rating the power supply and minimum electrical clearances at
various plate spacings. It should be pointed out that Figure 1
represents a conservative upper limit in estimated voltage and
electrical clearances for various plate spacings and, therefore,
offers a conservative estimate of cost savings. It is seen from
Figure 1 that 24" spacing corresponds to a 200 KVp supply and an
18" clearance between high voltage to ground surfaces. An 18"
spacing precipitator on the other hand will require 150 KVp
precipitator power supply and a clearance requirement of 13.5"
from high voltage to ground.
The clearance requirement guides the layout for precipitator
internals inside the existing precipitator shell. The operating
voltage guidelines dictate the power supply design and all high
voltage insulator design considerations. Larger insulators, both
in diameter and height result for the higher voltage operation at
wider plate spacings.
The packaging of precipitator internals inside a given shell
results in fewer gas passages for progressively larger spacings.
However, no increase in the length of ESP is needed or assumed
because of the findings reported in literature that larger
spacings maintain the collection efficiency achieved at 12" plate
spacing as long as the same residence time and the same total
corona power can be maintained. Wider plate spacing therefore
results in: a) reduction in the quantity of precipitator inter
nals such as collecting plates, discharge electrodes, plate
rappers and associated support frames and the attendant decrease
in the cost of internals; b) an increase in the ratings and as-
sociated costs of power supplies and the increase in costs as-
sociated with electrical components and clearances at higher
operating voltages; and c) progressive reduction in the installa-
tion costs of precipitator internals in line with the decrease in
quantity.
In order to estimate the total installed costs of various plate
spacings for Dale Station unit, the material and labor costs for
each spacing are developed, and are shown below:
Internals. There is a decrease in the cost of precipitator
internals due to reduced number of plates, discharge electrode
rappers and associated support frames and assemblies. Figure 2
shows this effect as a function of plate spacing. It is seen
that there is a proportional decrease in the cost of precipitator
internals associated with plate spacing. The cost of ESP inter-
nals at 24" spacing goes down by 47% when compared to the 12"
spacing.
18-6
-------
Electrical Components. As shown earlier, the increased plate
spacing results in the requirement for larger power supplies and
insulators and hence higher cost. This effect is shown in Figure
3. It is seen that the cost of power supplies and associated
equipment rises more rapidly than the decrease in costs of
precipitator internals. A sensitivity study of sectionalization
on the cost of power supplies was performed. The cost of TR sets
and associated high voltage gear needed for a 16" spacing was 75%
over the 12" spacing costs while it rose to 320% over the 12"
spacing costs for the 18" spacing.
Because of decreased sensitivity to misalignments, the perfor-
mance of a wide spacing ESP should not be adversely affected by
reduced sectionalization. Therefore another estimate was prepar-
ed with the configuration of 4 TR's per precipitator for wider
gas passages. From Figure 3 it is seen, as expected, that the
cost increases with spacing are less dramatic compared to the 8
TR configuration. This sectionalization aspect could prove very
important in the cost effectiveness of small wide spaced ESP's.
Installation Costs. Figure 4 presents the change in installation
costs as a function of plate spacing.
The installation cost of precipitator should generally decrease
with increased plate spacing. However, at a plate spacing of 24"
the electrical guidelines shown in Figure 1 could not be met at
locations above the precipitator high tension frames. Therefore,
the roof would have to be raised by 10 inches to meet the re-
quirements. The installation costs for 24" spacing are therefore
increased significantly due to this requirement. It should be
noted that all installation costs also include the cost of
demolition of old internals.
Total Installed Costs. The total installed costs of the precipi-
tator that include the engineering, material and labor costs for
the various plate spacings are plotted in Figure 5. Two cases
have been plotted in this figure; one with 8 TR's per ESP and
the other using only 4 TR's per ESP on the wider duct ESP's.
The increase in power supply costs for the 8 TR's per ESP case
was more than compensated for by the decrease in material and
installation costs for the 16" and 18" spacing ESP's, and there
was a net reduction in total installed costs of 9.7% and 9%
respectively when compared to the 12" spacing ESP. However, for
the 24" spacing ESP, the requirements of roof extension and the
increased power supply costs resulted in a net increase of 3%
over the 12" spacing ESP. Clearly there was an economic optimum
at 16" spacing for the 8 TR's/ESP configuration.
A subsequent case study was conducted using the 4 TR's/ESP
configuration. For this configuration the decrease in costs was
more than the previous case. The 16" spacing showed a 16.5%
reduction in costs while the 18" spacing showed a 17-7% reduction
18-7
-------
when compared with the 12" spacing ESP rebuild. The 24" ESP in
this configuration showed a 10.8% decrease in costs compared to
the 12" spacing ESP. The reduction in costs with the 24" spacing
was not as high as the 16" and 18" spacing. Though the economic
optimum was in the 16"-18" spacing range, there was a net cost
decrease at 24" spacing at the 4 TR/ESP configuration. The
impact of TR costs on*wide spacing precipitators is thus clearly
seen for the Dale Station ESP.
CONCLUSIONS
Application of wide plate spacing technology has shown sig-
nificant cost benefits for retrofit applications. Savings in
total installed costs in the range of 16-18% is possible at plate
spacings of 16-18 inches. For smaller ESPs such as the one
studied at East Kentucky's Dale State unit, cost of power sup-
plies and extent of electrical sectionalization can have a major
impact on overall cost savings.
Because of the merging of technical and economic optimum plate
spacings at the 16-18 inch levels, future ESP upgrades and new
installations on utility applications are likely to center at
these wider spacing levels, as compared to the previous standards
of 9" and 12" spacings.
ACKNOWLEDGEMENTS
The authors wish to thank Dr. Harry Enoch of East Kentucky Power
Cooperative and Dr- Ralph Altman of EPRI for the financial
support and overall guidance in the conduct of this project.
We also thank the following people who cooperated under this
project in the dissemination of wide plate spacing technology to
industry:
Dr. H. H. Petersen & Dr. P. Lausen: F. L, Smidth, Denmark
Dr. G. Mayer-Schwinning: Lurgi, Germany
Mr. H. J. Hall: Hall Associates, USA
Dr. L. Russell Jones: Lodge-Cottrell, Great Britain
Prof. E. Weber: University of Essen, Germany
Mr. S. Matts: Flakt Inc., Sweden
REFERENCES
1. Feldman, P. L., The Effect of Plate Spacing on Precipita-
tion, Research-Cottrell Technical Memorandum, 1974
2. Kumar, K. S., Experimental Results of the Wide Plate Spacing
Study. Research-Cottrell Technical Memorandum, 1975
3. Clinker Coolers, Internal Report Research-Cottrell
18-8
-------
4. Menegozzi, L. N., et al, Laboratory Investigation of 14, 12
and 9" Plate Spacing, Research-Cottrell Technical Report,
1981.
5. Cooperman, G., Non Classical Trends in Wide Plate Spacing,
Research-Cottrell TEchnical Memorandum, 1977.
6. Brown, R. F., Interim Report. Full Scale Wide Plate Spacing
Evaluation at ..., Research-Cottrell Project Report, 1975.
7. DeHaas, E., et al, Significant Potential Cost Reduction
Through Wide-Plate Spacing, Research-Cottrell Technical
Memorandum, 1976.
8. Wide-Plate Spacing Commercialization Program, Research-
Cottrell Internal Memo, 1976.
9. Masuda, S., Present Status of Wide Spacing Type Precipitator
in Japan, EPA Symposium of Particulate Control Technology,
Denver, 1979.
10. Heinrich, D. O., Staub, (38 Cll), 446-451, Nov. 1978.
11. Matts, S., Some Experiments with Increased Electrode Spac-
ing, Proc. CSIRO Cont. on Electrostatic Precipitator, Paper
#13, 1978.
12. Cooperman, P., Non-Deutschian Phenomenon in Electrostatic
Precipitation. APCA 69th Annual Meeting, 1976.
13. Feldman, P. L., Kumar, K. S., Cooperman, G. , Turbulent Dif-
fusion in Electrostatic Precipitators, A.I.Ch.E. Symposium
Series #165, Vol. 73, 1977.
14. Williams, J. C. and Jackson, R. (1962), The Notion of Solid
Particles in an Electrostatic Precipitator. European Federa-
tion of Chemical Engineering pp 282-288.
15. White, H.J., Editor, Proceedings International Conference on
Electrostatic Precipitation, October 1981, Monerey, CA.
16. Noso, S., Performance Characteristics of Electrostatic
Precipitator with Wide Spacing, pp 654-657, International
Conference on Electrostatic Precipitation, Monterey, CA,
1981.
17. Darby, K., The Use of Pilot Precipitation in the Field and
Laboratory, pp 592-626, International Conference on Electro-
static Precipitation, Monterey, CA, 1981.
18. Feldman, P. L., Effects of Particle Size Distribution on the
Performance of Electrostatic Precipitator, APCA Meeting,
1975.
19. Mayer-Schwinning, G., Electrostatic Precipitators with Wide
Passage Spacing. Lurgi Information from Research Laboratory,
August, 1982.
18-9
-------
20. Heinrich, D. et al, Panel Discussion on Wide Spacing Preci-
pitators, pp 134-144, Second International Conference on
Electrostatic Precipitation, Kyoto, Japan, November, 1984.
21. Hall, H. J., Some Electrode Geometry from an Electric Field
and Performance Effects in Electrostatic Precipitation, pp.
354-361, Second International Conference on Electrostatic
Precipitation, Kyoto, Japan, November, 1984.
22. Wheeler, H. L. et al, An Investigation of Precipitator Wide
Plate Spacing. EPRI Project 1835-5, Final Report, December,
1987.
18-10
-------
00
SPECIFICATION OF EXISTING DALE 3 AND 4 ESP'S
Design Gas Flow Per ESP at 300°F, acfm 291,000
Design Collection Efficiency, %** 98
Plate Spacing, Inches 12
Number of Chambers/ESP 2
number of Gas Passages/Chamber 22
Number of Fields 4
Number of TR's/ESP 8
Collector Plate Geometry 31'H, 9'L
Total Collecting Area, Ft2 98,208
Discharge Electrode Type Rigid
Number of Electrodes per ESP 880
T-R Set Voltage Rating, kVaVg/kvp 55/85
** Based on operation with SO3 conditioning
TABLE 1
18
16.
14 .
12.
J10.
200
150
100
50
8
20
24
12 16
Plate Spacing, Inches
Estimated Power Supply Voltage and Clearance Requirements
as a Function of Plate Spacing
FIGURE 1
-------
CD
I—'
ro
50
45
40
35
30
25
20
15
10
5
12 16 18
Plate Spacing, Inches
24
Decrease in Cost of ESP Internals as a
Function of Plate Spacing Based on 12"
Spacing = 100?
Actual Value = $271,000
4 TR's ESP
16 18
Plate Spacing, Inches
24
Change in Power Supply and Associated
Cost as n Function of Plate Spacing
Based on 12 Inch Spacing = 100%
Actual Value = $111,000
FIGURE 2
FIGURE 3
-------
00
I—I
CO
30 .
20
5 10
10
12
14
16
18
16
18
Plate Spacing, Inches
Recent Change in Installation Costs
as a Function of Plate Spacing Based
on 12 Inch Spacing = 100%
Actual Value = $650.000
10
7
6
£
4
3
2
1
0
-1
-2
-3
-4
_ c
-S
-7
-10
-11
-12
-13
-14
-IE
-16
-17
-18 1
4 TR's ESP
12 16 18 21
Plate Spacing, Inches
Plate Spacing Versus Percent Change
in Total Installed Costs Based on
12 Inch Spacing = 100%
Actual Value = $1,290,000
FIGURE 4
FIGURE 5
-------
INCREASED PLATE SPACING IN ELECTROSTATIC PRECIPITATORS
K. Darby
Lodge-Cottrell U.K. Operations
Dresser Industries Inc.
George Street, Parade
Birmingham B3 1QQ
England
David Novogoratz
Lodge-Cottrell North American Operations
Dresser Industries, Inc.
601 Jefferson Street
Houston, Texas 77002
ABSTRACT
It has been the custom for users to evaluate the performance of electrostatic
precipitators by comparison of the specific plate area and the expected collection
efficiency. This is based on the assumption that the effective migration velocity
used in the Deutsch equation is constant and largely dependent on gas and dust
properties. The conclusion drawn would be that the collector plate spacing should
be the minimum possible (as little as 8" has been used and more commonly 9" to 12")
to obtain the largest collector area in the minimum space.
This paper reports operating experience on coal-fired power plants and and other
processes with collector spacings of 16". The indications are that the time the gas
is exposed to the electrical field is a more valid comparison than the specific
collector area. The implications and limitations of 16" and wider plate spacing are
discussed.
19-1
-------
INCREASED PLATE SPACING IN ELECTROSTATIC PRECIPITATORS
INTRODUCTION
It is not possible to calculate the performance of a precipitator from first
principles. There are accepted formulas which enable the velocity of a charged
particle in an electric field to be calculated. The force due to electric field
activity on the particle is opposed by the viscous drag of the gas in which the
particle is suspended. These calculations enable the so-called migration velocity
of the particle to be calculated, but this only applies to static gas conditions and
is of little use in designing a precipitator. It must not be confused with the
effective migration velocity (EMV) which is used in precipitator design calculations
and is calculated using an empirical formula. Effective migration velocity is
calculated from the size of the precipitator, the gas volume and a measured efficiency
of dust removal. It is probably misleading to give the factor the dimensions of a
velocity which tends to increase the risk of confusion with the theoretical migration
velocity. It is more an index of precipitator performance and is influenced by a
large number of factors related to the design of the precipitator, the properties of
the dust and gas, the concentration of dust, the temperature of the gas, and the
efficiency of dust removal required. This has been discussed in detail elsewhere
(Ref. 1).
The earliest empirical formula predicting precipitator efficiency was derived by
Anderson in 1919. This was in the form given in Table l(a), which indicated that the
efficiency of dust removal was related to the length of the precipitator for an
instant gas velocity. Since increasing length would increase also the contact time
of the gases exposed to the electric field, it could be concluded that this in fact
related efficiency to contact time.
The most commonly known formula is that derived by Deutsch in 1922; the derivation
of this is well documented in many published papers and books and is shown in its most
commonly used form in Table l(b). This indicates that efficiency of dust removal is
related exponentially to the product of the effective migration velocity and the
specific collector area (A/V) which is defined as the area of collecting electrode
per unit of gas volume passing through the precipitator.
If it is assumed that effective migration velocity is a constant, dependent on gas
and dust properties, and this has been the most commonly accepted view for many years,
then the precipitator efficiency is related to the specific collector area, the ratio
of the area of collecting electrode to the gas volume treated. This is at variance
with the Anderson formula which related efficiency to precipitator length, hence
contact time. A most important implication of the specific plate area formula is that
the greater the amount of collecting plate which can be packed into the electrode
housing, the more efficient the precipitator will become. There are practical
constraints on the minimum spacing based on electrical consideration, namely the need
to maintain the strongest possible electrical field. The lower limit of plate spacing
19-2
-------
is a practical decision dependent on such factors as the problem of aligning the
electrodes and the accuracy of the manufacturing process.
With reducing plate spacing, any errors in component manufacture and alignment and
any effect of dust build-up on electrodes or plates have an increasing effect on
precipitator efficiency. The result of this is that 8" (200 mm) is generally regarded
as the low limit for successful construction of single stage precipitators for
industrial application.
Until about twenty years ago, the situation was such that commercial precipitators
were built with collecting centers ranging from 8" (200 mm) to 12" (300 mm) according
to the design philosophy of the manufacturer. This presented no problems to
manufacturers provided that the customer or consultant engineers responsible for
comparing competitive tenders based their assessment of a particular design of
precipitator on the performance track record of the manufacturer.
Then came a period some twenty years ago when the customer/consultant engineer became
more directly involved in specifying the precipitator design. It was assumed that
effective migration velocity was a constant related to fuel and process conditions
and it became almost standard practice for the specific collector to be used as the
basic criterion for comparing different precipitator effective sizing and hence
potential dust collection efficiency. Figure 1 illustrates the effect of the plate
area concept on calculated gas cleaning efficiencies. For the same contact time, if
the 16" plate spacing gave 90% dust removal then the efficiency at 8" spacing would
be 99%. This approach discounted manufacturers' previous practice and experience with
the result that manufacturers who had traditionally based their designs on the wider
spacings, for example 12" (300 mm), found themselves in a situation where for the same
collecting electrode area as a competitor using 8" spacing, the casing enclosing the
electrode system was more than 50% larger in volume. The problem this caused in
competitive pricing is illustrated in Table 2 which shows the relative percentage cost
of the three major parts of the plant casing, electrode system and T/R sets.
The electrode system for the same plate area would be closely similar in cost
regardless of spacing, but the cost of the electrode housing would increase
significantly for the wider spaced plate. There would also be a smaller penalty on
the T/R sets due to the higher electrode voltage needed with increased electrode
spacing.
In combating this problem the initial reaction of the manufacturer using the wider
spacing was to reduce their plate spacing to the minimum which their design would
accommodate, i.e., without seriously altering critical electrical clearances and
manufacturing tolerances.
The result was that an undesirable situation arose since precipitators at 8" to 9"
centers are best suited to the weighted wire discharge electrode design. This design
has a number of disadvantages and has been superseded by rigid designs of electrode
giving greater reliability and longer life. Narrow center precipitators, apart from
requiring more accurate alignment and tighter manufacturing tolerances, also limit
the height of collecting electrodes which can be used, as any increase in height
further increases the alignment problem. This has resulted in the height of collector
electrodes at low spacings being limited to 30' to 36' compared to 45' to 50' with
19-3
-------
12" and larger spacings used with rigid discharge electrode systems.
With the modern trend to high efficiency precipitators, particularly on boiler plant,
there are a number of limitations imposed on the ground space available for the
precipitator, such as boiler centers and distance from air heater outlet to stack.
Increasing collecting electrode height was one of the few practical ways of providing
the size of precipitator needed.
There had been opposition to the plate area concept for many years, as illustrated
by the wide divergence in the plate spacing used by different manufacturers. In 1960
D.O. Heinrich filed British Patent No. 845331 (Ref. 2), the implications of which were
that effective migration velocity increased proportionately with plate spacing and,
therefore, the correct comparison would be the contact time of the gas in the electric
field. This supported the views of the manufacturers who had worked with plate
spacings of 12" or greater, but were now being penalized by the specific plate area
concept. The cost comparison would now favor wider plate spacing (Table 2), since
the wide spacing would fit in the same sized housing, but the quantity of plates and
discharge electrodes would be less.
Following the publication of this patent, a plant was constructed in Germany on the
semi-dry cement process where there was a particular problem of corrosion
necessitating the use of expensive alloys. By doubling the collector spacing from
10" to 20", the amount of electrode material was halved and it was possible to
construct the precipitator in the alloy needed without the very substantial increase
in cost which would otherwise have been incurred.
For some years following this, no further wide spacing plant was constructed because
the specific plate area concept dominated as the comparison criteria. In the last
ten years, numerous papers on the use of wider spacing have been published both in
Japan, where there has probably been the greatest experience, and in other countries
throughout the world, particularly in Europe.
DATA SUPPORTING THE WIDE SPACING CONCEPT
Acceptability of Wide Spacing
The problem of convincing engineers of the viability of the wide spacing approach
is readily understandable. Any failure of a precipitator, with wider than normal
plate spacing, to meet its design efficiency would most likely be blamed on the
reduced plate area this involved. The natural reaction would be to demand the
installation of additional plates involving considerable cost to the supplier of the
precipitator, and a large loss of revenue to the power plant operator due to the
fairly lengthy shutdown of the boiler for this to-be done. If the wide spacing
concept is correct, this would not improve efficiency.
If the wide spaced plant met its guaranteed performance on the other hand, there
would still remain some doubt about the influence of the fuel (possibly more favorable
to precipitation than anticipated?). The only totally convincing test would be:
• Twin flows on the same boiler, tested simultaneously, each of the
flows having a different plate spacing. But this, while giving
the evidence needed, would involve extra costs and is a
difficult concept to sell to users.
19-4
-------
This is the dilemma which has faced manufacturer and user alike and has resulted in
long delays in getting the matter resolved.
Laboratory Pilot Precipitator Tests on the Influence of Plate Spacing
An alternative way to demonstrate the influence of spacing is the use of pilot
precipitator units. Lodge-Cottrell constructed such a unit in the late 1970's; this
was described in detail in a previous paper (Ref. 3). A schematic layout of the
installation is shown in Figure 2. It consists of a gas fired combustion unit to
provide the gases; and at the outlet of the chamber, dust in the required quantity
is injected. Following this is a water spray tower for cooling, a valve for bleed
air inlet, and an electrical heating system to control the final temperature at the
precipitator inlet. The result is a system in which gas and dust properties can be
maintained for long periods with a degree of constancy not possible on a full-scale
plant.
The gases are fed into twin flow pilot precipitators for the purpose of investigating
the effect of spacing. The control unit is fitted with the standard electrode system
(10" or 12" centers) while the other unit is fitted with wider spaced plates (16" to
24").
Since the conditions for both flows are closely similar, efficiency of dust collection
will represent the effect of change in plate spacing only. Results of the effect of
plate spacing were reported previously in 1984 (Ref. 4). A criticism of the data
obtained then was the relatively low range of dust collection efficiencies obtained.
Following this, the contact time of the pilot unit was increased to give efficiencies
of dust collection similar to that required on commercial plant, i.e., above 99%.
Wide Spacing Data
Included in this paper are the following reports on commercial plant.
• Latest tests on precipitator optimization unit (1989).
t A power plant precipitator which, during its half-life rebuild, was
converted from 10" to 16" plate centers.
• Power plant operating on Australian low-sulfur, low-alkali dust in
Japan.
• Data obtained on dry process cement plant and cement clinker coolers.
• Collection of fine fume on steelmaking plant in a totally saturated
environment.
All wide spacing tests, pilot or full-scale, are for 16" plate spacing. This was
selected on the basis that the increase from 12" would represent a substantial
increase, yet still remaining within the predictability of design considerations and
any technical constraints.
The basic Lodge-Cottrell precipitator design accomodates this wide spacing
configuration, as experience has shown that electrical clearances and insulator design
19-5
-------
were adequate for the increased voltage to be applied. The transformer-rectifier
(T/R) sets were designed for a 30% increase in HT voltage, but similar current output
to 12" designs. The wide spacing concept is regarded as valid if the ratio of the
EMVs for the two spacings is equal to or greater than the ratio of the plate spacing.
PILOT PRECIPITATOR RESULTS
The unit was used to assess the variation in collection efficiency obtained with the
plates spaced at 12" and 16". The choice of 16" was decided on the basis that this
would involve no modification to insulators or electrical clearances on full-scale
plant as these were known to have an adequate safety margin to allow for the increased
voltage which would need to be applied.
The first test series was carried out on fly ash from a U.K. power station. This
is a dust which is relatively easy to precipitate. The operating conditions for these
tests and the results obtained are given in Table 3. From this the indications are
that an increase from 12" to 16" gave an increase in migration velocity of just under
50%, with a corresponding increase in collection efficiency.
In the next series of tests, cement from a semi-wet process kiln was used. Again,
Table 3 gives the operating conditions which are favorable to precipitation, and also
the efficiencies and relative migration velocities obtained. This shows that the
33% wider spacing gave similar efficiencies with an increase in migration velocity
of roughly 40% compared to the standard spacing. The test series was repeated using
a lower dewpoint gas (18°C compared to 30°C). In this case the wide spacing gave a
significantly higher efficiency with an increase in migration velocity for the wider
spaced unit of 60%.
Considering these two series of tests, both support the view that effective migration
velocity increases at least proportionately with spacing. Furthermore, there are
indications that with higher resistivity dust, the advantage many be increased rather
than decreased. This supports extensive earlier tests on this installation with a
wide variety of dust and resistivities but using only a single field. In contrast
the dust emissions and collection efficiencies obtained during this series of tests
were comparable with commercial requirements of greater than 99%.
FULL-SCALE PLANT
Longannet Power Station (Scotland) 600 MW
The precipitators were originally installed by another manufacturer in the early
seventies and acceptance tests took place in 1972. The plate centers for this
construction were 10". Two of the four precipitators were rebuilt in 1989 by Lodge-
Cottrell. On a trial basis, the South of Scotland Electricity Board agreed to
substitute 16" plate spacing with the appropriate modification of the Lodge-Cottrell
design to fit the casing. New T/R sets were also fitted with 60% increased HT voltage
output which also required replacing the suspension insulators to accommodate the
higher voltage.
19-6
-------
A comparison of the test results for the 10" and the 16" electrode system is given
in Tables 4A and 4B. The decision to compare the tests using the 1972 results on the
10" spacing and the 1989 results on 16" spacing was based on the argument that for
both series, the precipitators were new and in good order. Although this raises the
question of fuel variation with the seventeen year gap, it was considered to be the
best compromise. It should be noted that the Longannet Power Station is located
adjacent to the Longannet Mine which is part of the Hirst Seam.
The tests on the wide spaced units in 1989 were carried out by the Board's staff with
only dust emissions being measured. Since Hirst coal is consistent and the ash
content is still closely 18%, an inlet burden of 8 gr/scf (20 g/Nm3) was assumed for
the purpose of calculating dust removal efficiency and hence effective migration
velocity.
Shown in Table 4B is the 1972 test data on 10" spacing and the 1989 test data on 16"
spacing. For comparison purposes the test data was converted to a constant efficiency
of 99.5% using the modified Deutsch formula.
To achieve the same collection efficiency when plate spacing is increased from 10"
to 16", the effective migration velocity must increase by the same ratio.
Examining the results corrected to constant efficiency of 99.5%, the contact time for
the 16" centers is lower than for 10" (10.5 versus 12.1 seconds) indicating a more
efficient use of the existing gas chamber by using wider spacing.
Alternatively the increase in effective migration velocity is in the ratio, on
average, of 10.1 to 5.5, an increase of 1.84 times. The required value to ensure the
same performances at 16" as at 10" would require a ratio of 1.6 times.
In contrast, if the specific plate area concept were true and effective migration
velocity constant, the test efficiencies would have fallen to 94%.
With the assumptions made on inlet dust burden and precipitation characteristics of
the fly ash from the Hirst seam, which is consistent in its supply, the fact that the
EMV increases much more than proportionately with increase in spacing of plates
supports the wide spacing concept.
Power Stations - Japan
In the period 1987 to 1989, three precipitator installations were constructed by
Lodge-Cottrell licensees in Japan on pulverized coal boiler plant, using the 400mm
(16") plate spacing. A summary of the results of these tests is presented in Table
5. These tests took place on two different Australian coals, a 0.9% combustible
sulfur for two of the plants and the third with a 0.45% sulfur content and a low
sodium ash content of 0.3%. The latter coal is known to give adverse precipitation
conditions. The design of these precipitators was based on experience of these
Australian coals at 12" spacing, with the design migration velocity being increased
proportionately to the spacing increase.
19-7
-------
The boilers concerned are relatively small industrial boilers (50 MW) and it is
particularly interesting in view of the size of the boiler that it was elected to use
48' collector heights, this being to suit the site space available. In the case of
the first two boilers with the higher sulfur fuel, emissions were generally
substantially below 10 mg/Nm3 compared with a guarantee of 40 mg/Nm3.
The third plant working on the worst of the two Australian low sulfur fuels gave
emissions averaging 30 mg/Nm3 against a design of 100 mg/Nm3. These results further
substantiate the use of wide spacing on pulverized fuel boilers.
Dry Process Cement Plant
Kiln Operation. This dry process cement kiln was designed for a feed rate to the kiln
of 80 tons/hour. In common with this type of plant, the kiln gases were designed to
pass either through an evaporative cooling tower to reduce temperature from 370°C to
150°C, or in normal operation through the raw mill which would give the benefit of
heat recovery and reduce the gas temperature to the same order, 140°C. The design
migration velocity was based on a strict proportional increase of spacing to 16" from
the normally used 12" centers for this process. Table 6 presents the comparison of
design and test efficiency and dust emission. All test results exceed design by a
substantial margin. The voltages applied to electrodes ranged from 86 to 71 kV across
the precipitator fields. Generally, power consumptions were of a similar order to
those which would be expected from a 12" center plant.
C1inker Cooler. This unit treats the gases resulting from air being used to cool the
cement clinker produced by the kiln. Results are summarized in Table 6; gas
temperature design was 150°C minimum. With clinker coolers the temperature varies
widely and rapidly according to the operating condition of the kiln, and during the
tests averaged 250°C. The dust emission obtained was 0.0008 gr/scf (1 mg/Nm3) with
this 16" center plant. Inlet burdens were assumed to be 8 gr/scf (20 g/Nm3); the
corresponding efficiency would be 99.95%. Due to the kiln not yet operating at full
capacity, the gas volume through the precipitator was low. The test efficiency
(99.995%) exceeded the design efficiency (99.92%) by a considerable margin as a
result. Figure 3 shows contact time plotted against efficiency; the solid line is
an average performance line for 12" center plant. This line, shown projected, passes
through the test efficiency for the 16" spacing plant, providing further support for
the use of wide spacing.
Basic Oxygen Recovery Plant - June 1989
This 16" center precipitator follows a high energy scrubber and treats part of the
gas from the steelmaking process to reduce the dust content to a level suitable for
burning in process plant on the steelworks, i.e., 0.004 gr/scf (10 mg/Nm3). All of
this dust, having passed through a high energy scrubber, is well into the sub-micron
range. During acceptance t^sts, dust concentrations of 0.002 gr/scf (5 mg/Nm3) were
obtained with an inlet burden below maximum specified. This plant operates in a wet
system and is not subject to rapping cycles and re-entrainment. The dust inlet
concentration is low but the particle size is extremely fine. It might be expected
that corona suppression would occur, but this was certainly not so. This provides
evidence that at low concentrations of very fine fume, wide spacing is effective.
There may be limitations on very heavy concentrations of fine fume.
19-8
-------
COMMENTS ON THE TEST RESULTS
The tests reviewed in this paper cover a fairly wide range of situations. The
precipitator optimization unit gives a strict comparison of the performance of
different electrode spacings under identical gas and dust conditions and for this
reason can be considered the best comparison of the effect of spacing alone.
Efficiencies obtained are in line with those for commercial plant, thus eliminating
one of the criticisms of earlier data of this type when the precipitator was
considerably smaller and the efficiencies proportionately lower.
The major criticisms of the optimization unit are:
• Re-entrained and de-agglomerated dust from commercial precipitators
is used and hence, although the dust came from industrial processes it
may not represent fully the condition on the original industrial
precipitators.
t The collector height is essentially small compared to that used on
commercial plant, which ranges from 30' to 50'.
The commercial plant results reviewed are to Lodge-Cottrell totally convincing since
each plant is based on the concept that increasing the plate spacing to 16" would
result in migration velocities increasing proportionately compared to data from our
records on similar processes, but with narrower spacings usually in the range of 10"
to 12".
Lodge-Cottrell chose 16" spacing on the basis that the increase was substantial but
still remaining within the range of predictability of design considerations and any
process considerations. Furthermore, the basic Lodge-Cottrell precipitator design
readily accommodates wide spacing configuration in terms of electrical clearances and
insulator design required by the higher voltage. Consequently, the economy resulting
from the reduction in quantity of electrodes used by the wider spacing was easily
evaluated and set against the increased cost of the higher voltage T/R sets.
In past considerations of the use of wider spacing, it has been felt generally that
attention would have to be paid to processes with high concentrations of dust and fume
and that wide spacing might not be applicable in this sort of situation. The
experience obtained, although limited, indicates that on ordinary dusts such as
pulverized fuel and cement dust, no limitations on the use of wide spacing at 16"
are expected.
Earlier experience referred to in the 1960's supported the view that on cement dust
20" spacing raised no particular difficulties and this is possibly the next step in
the development of the concept.
The construction of the precipitator for the wider spacings and hence higher voltage,
presents no special problems. Power consumption generally is of the same order with
wide plate spacing despite higher operation voltages.
19-9
-------
It is felt that the information listed endorses the work of other research concluding
that specific collector area is not a valid comparison of precipitator sizing. The
alternative is to use contact time as the comparison, although there are indications
that wide spacing actually enhances performance providing the design of the electrode
system is correct and the T/R sets are adequately sized.
CONCLUSIONS
• Contact time, the time the dust laden gases are exposed to the
electric field, is the correct comparison of different precipi-
tators, not collector plate area.
• There may be performance advantages with increasing spacing.
t Wider spacing gives better access for maintenance, and alignment
is less critical.
s The upper limit for spacing depends technically on process conditions that
the dust or fume not give corona suppression. It is otherwise determined
by engineering considerations relating to casing, insulator, and T/R design
to accommodate higher voltage and the resulting overall economics of the
precipitator, i.e., Casing + Electrode System + T/R Sets.
ACKNOWLEDGMENTS
The Authors wish to express their appreciation to the South of Scotland Electricity
Board for agreement to use their test data for Longannet Power Station, and to thank
their colleagues for assistance in the preparation of this paper and the Management
of Lodge-Cottrell, Dresser Industries, for their encouragement and support in
publishing this paper.
REFERENCES
1. K. Darby and C. Whitehead. "The Use of Electrostatic Precipitators in Current
Power Station Practice." Institute of Fuel Symposium Changing Technology of
Electrostatic Precipitation. Adelaide, Australia, 1974.
2. British Patent No. 845331, assigned to Lodge-Cottrell Ltd, 1960.
3. K. Darby. "The Use of Pilot Precipitators in the Field and Laboratory."
International Conference on Electrostatic Precipitation. Monterey, California,
1981.
4. K. Darby. "Plate Spacing Effect on Precipitator Performance." Second
International Conference on Electrostatic Precipitation. Kyoto, Japan, 1984.
19-10
-------
V
(a) PLATE SPACING 16" (b) PLATE SPACING 8'
Figure 1
According to specific plate concept and using Deutsch equation,
if (a) had efficiency of dust collection 90%,
then (b) would have efficiency of dust collection 99%
or 2(a) field = 1 (b) field for 99% efficiency.
In contrast, if effective migration velocity is proportional to plate spacing, then
efficiency of (a) = efficiency of (b)
Both would have same contact time for gas in electric field.
CONDITIONING
CHAMBER
DUST
INPUT
GAS FIRED AIR HEATER \
GAS FIRED AIR HEATER Z
DESULPHURISER
BAG
FILTER
PRECIPITATORS
ELECTRIC .
HEATER \
TO
ATMOSPHERE
FAN
INLET SAMPLING POS'N'S OUTLET SAMPLING POS'N'S
& OPACITY METERS
FLUIDISED
BED COMBUSTOR
Figure 2. Flow chart of precipitator optimization plant.
19-11
-------
Average test data 12" centers plant
-*- Test on 16" center plant
Note: High efficiency due to operation at low
gas volume. Design efficiency 99.92%.
- 99.999
- 99.99
- 99.9
L- 99
CONTACT TIME
Figure 3. Cement - clinker cooler precipitator
19-12
-------
Table 1
EQUATIONS
(a) Anderson Equation (1919)
Efficiency of Dust Collection = 1 e"
where e exponential
c constant
L length of precipitator
(b) Deutsch Equation (1922)
Efficiency of dust collection = 1 e
where w effective migration velocity
A collecting area
V gas volume treated
-wA/V
Table 2
CURRENT PRACTICE AND RECOMMENDED PRACTICE OF ESP COMPARISONS
(with Various Plate Spacings)
Performance Parameters
Plate Spacing
Collection Efficiency (%)
Relative SCA
Relative Contact Time
Relative Cost Comparison
Internals
Casing
TR Sets and Controllers
TOTAL
Current Practice
Comparison of SCA
8"
99.00
1.00
1.0
40
45
12"
99.00
1.00
1.5
40
50
16"
99.00
1.00
2.0
40
65
Recommended Practice
Comparison of Contact Time
8"
99.00
1.00
1.0
40
45
15
12"
99.00
0.67
1.0
27
45
16
16"
99.00
0.50
1.0
20
45
}7
100
106
122
100
88
82
19-13
-------
Table 3
PILOT PRECIPITATOR OPTIMIZATION
Fly Ash U.K. Power Station
Gas Velocity in full
Gas Temperature
Dewpoint
Inlet Burden
Precipitator Plate Spacing
Efficiency %
EMV cm/s
Ratio EMV's
Ratio Plate Spacing
1.1 1.2 m/s
120°C
30°C
10-12 gr/Arn3
12"
99.37
16.8
1
1
16"
99.70
26.8
1.49
1.33
Semi-Wet Process Cement Kiln
Gas Velocity in full
Gas Temperature
Inlet Dust Burden
Gas Dewpoint (a)
(b)
Precipitator Plate Spacing
Efficiency Condition (a)
(b)
EMV
Ratio EMV's
Ratio Plate Spacing
(a)
(b)
(a)
(b)
1.1 1/2 m/s
160°C
8-12 g/Arrf
30°C (Favorable Dust Resistivity)
18°C (Higher Dust Resistivity)
12"
99.78
98.78
21.
14.
1
1
1
16"
99.80
99.10
29.3
22.7
1.39
1.61
1.33
19-14
-------
Table 4A
LONGANNET POWER STATION (SCOTLAND) 600 HW
ORIGINAL PLANT TESTED 1972
Plate Spacing 10" (250 mm)
Plate Area/ESP 20,423 m2
CONVERTED IN 1989 TO LODGE-COTTPFLL DESIGN ELECTRODES
Plate Spacing 16" (400 mm)
Plate Area/ESP 13,161 m!
ALL TESTS ON LOCAL COAL HIRST SEAM
TYPICAL ANALYSIS
COAL ASH 18%
VOLATILES 25%
SULFUR 0.47%
ASH SiO; 46% MgO 1.6%
Al,0j 31% Na,0 0.25%
CaO 1.7% K,0 1.1%
ORIGINAL PLANT
100% Load
75* Load
LODGE-COTTRELL
3 Fields
4 Fields
* Assumed
Ratio Increase
Ratio Increase
SCA
mVmVs
1972
96.5
129.3
1989
51.1
55.3
68.7
67.55
Plate
EHV:
Contact
Time
s
10" SPACING
12.1
16.2
16" SPACING
10.2
11.1
13.7
13.5
Spacing:
Table 4B
TEST RESULTS
CALCULATED FOR
MEASURED CONSTANT EFFICIENCY 99.5%
Inlet Emission Eff'y EHV
a/Nm3 mq/Nm1 % cm/s
20.9 127 99.39 5.28
22.2 38 99.83 4.93
20 * 107 99.46 10.24
20 * 93 99.53 9.71
20 * 42 99.79 8.97
20 * 48 99.76 8.93
Contact
SCA Time EHV
mVmVs s cm/s
104.0 13.0 5.09
89.4 11.2 5.92
52.4 10.5 10.1
53.8 10.8 9.8
50.7 10.15 10.4
52.1 10.4 10.2
1 . 1.60
1 1.84
19-15
-------
Table 5
LODGE-COTTRELL JAPANESE LICENSEES PF BOILER PLANT
16" Plate Spacing
COAL PLANT
A Site 1
(Average 5 tests)
Site 2
(Average 5 tests)
15.88
DESIGN
TEST
Inlet Emission Eff'y Inlet Emission Eff'y
g/Nm3 mg/Nm3 % g/Nm3 mg/Nm3 %
20.6 40
40
99.8 22.3
99.75 16.5
8.0
3.6
99.96
99.98
B Site 3
(Average 3 tests)
19.4
100
99.485 18.1
29.0
99.84
All designed on basis of increasing design migration velocity used for 12" spacing by
ratio 16:12. All have three electric fileds in series.
COAL A Australian Bituminous
COAL B Australian Bituminous
0.9% Sulfur
0.25% Na2 0
0.4% Sulfur
0.3% Na,0
Table 6
DRY PROCESS CEMENT KILN AND CLINKER COOLER
Plate Spacing 16"
KILN AND MILL OPERATION
Efficiency of Dust Removal
Dust Emission
Design
99.96%
0.05 gr/scf
Test
99 . 99%
0.01 gr/scf
2. KILN WITH COOLING TOWER
Efficiency of Dust Removal
Dust Emission
99.97%
0.05 gr/scf
99.985%
0.012 gr/scf
3. CLINKER COOLER
Efficiency of Dust Removal
Dust Emission
99.92%
0.05 gr/scf
99.99%
0.0025 gr/scf
19-16
-------
MECHANISMS OF PERFORMANCE ENHANCEMENT IN
WIDE PLATE ELECTROSTATIC PRECIPITATORS
H. Elshimy and G.S.P. Castle
Department of Electrical Engineering
The University of Western Ontario
London, Ontario N6A 5B9, Canada
ABSTRACT
In recent years it has been widely demonstrated that it is possible to reduce the
collecting area of an electrostatic precipitator by increasing the plate to plate
spacing without experiencing any significant reduction in collection efficiency.
This apparent contradiction of the Deutsch relationship has been addressed
previously by a number of authors but there is no general agreement on the
mechanisms. This paper reviews the factors involved in the two most accepted
reasons for this improvement, namely electrical (field enhancement) and
mechanical (reduced re-entrainment). This review leads to the prediction that
under normal plant conditions, with the operating voltage set close to sparkover,
mechanical effects are the most probable reason for the improved performance.
Published data of collection efficiency and electrical conditions from a number
of different full scale and pilot plant precipitators are analyzed to determine
if electric field enhancement can explain the results. It is shown that
electrical effects can only explain the improvements in less than one half of the
cases leading to the conclusion that reduction in mechanical re-entrainment is
the more probable reason for improved performance of wide plate precipitators in
many situations.
20-1
-------
MECHANISMS OF PERFORMANCE ENHANCEMENT IN
WIDE PLATE ELECTROSTATIC PRECIPITATORS
INTRODUCTION
The practical use of wide plate precipitators is well established with many
published results showing that under a broad range of conditions the effective
migration velocity w , generally increases in proportion to the wire to plate
spacing [1] [2] [3J . However, there is still no generally accepted understanding
of why this relationship occurs. It is recognized that many non-linearities are
included in the value of the effective migration velocity since it is determined
experimentally from the measured collection efficiency and the specific
collection area of the precipitator. Clearly there are electrical, mechanical
and aerodynamic effects which can affect the collection efficiency and may depend
upon the plate spacing. It is generally agreed that the most significant
variables are probably the electric field strength, rapping losses, and re-
entrainment losses.
In a previous paper it was argued that the increase in effective migration
velocity could be predicted from classical theory provided that as a minimum
condition the corona current density at the collector plate surface was held
constant [4J. In practice it is known that many other electrical criteria are
used including constant total corona current, constant average electric field,
constant total corona power, etc. Some results have shown a correlation between
measured electric field near the collector surface and increased migration
velocity [5J .
However, a number of other cases have been cited where it is unclear whether or
not electrical effects can be used to entirely or even partially explain the
results. In particular, Matts [6] has argued that in high efficiency
precipitators which are operated close to breakdown, electrical effects cannot
explain all the improvements that are found and that it is probable that re-
entrainment and/or rapping are more important.
The purpose of this paper is to discuss the circumstances under which electrical
effects can be used to explain the improved collection and to briefly review the
reasons which could explain improvements due to mechanical and aerodynamic
effects.
Following this discussion a number of experimental results found in the
literature are analyzed to determine whether electrical effects can explain the
efficiencies obtained.
ELECTRIC FIELD CONDITIONS
It is generally accepted that oj depends upon electric field as
u) a (E E ) (1)
e p c v '
20-2
-------
where E = electric field at the plate surface
E s charging field.
Usually one of two assumptions is made regarding the value of charging field
i.e. :
the charging field is equal to the average field E
av
.-. 0) a (E E ) (2)
e p av v '
b) the charging field is equal to the field at the plate.
•"• ^e a. (Ep2) (3)
Clearly in either case the effective migration velocity will increase with an
increase in the electric field strength at the plate surface.
As shown previously [4_] for negligible particle space charge and constant corona
current density, the field at the plate surface increases in proportion to -fh due
to ionic space charge thus leading to the prediction
to a h (4)
e
assuming equation [3] describes the electric field dependency of the effective
migration velocity.
However, clearly this will only be true if the field is not limited due to
breakdown. Since most precipitators (narrow or wide spacing) are operated at
close to their breakdown limit, the question which needs to be addressed is
whether or not the breakdown field increases with spacing.
BREAKDOWN FIELD STRENGTH
Sekar et al. [7J tested a laboratory scale wire-plate precipitator under clean
air conditions for three wire to plate spacings in the range 150-250 mm. They
reported that at the breakdown voltage the electric field at the plate is
approximately constant, independent of spacing, at a value in the range
kV
7.2 - 7.9 —. By analysing their data it can also be shown that E is also
kV
approximately constant in the range 6.7 6.0 —.
Further support for these observations can be found in the much more extensive
literature dealing with point-plane, wire-plane and rod-plane configurations [E3]
[9] [_10]. These studies show that for the case of negative corona in clean air,
kV
at breakdown, E remains constant at a value of approximately 6 — as the point
av cm
to plane spacing is increased.
For the case of a dust loaded precipitator it is well known that the maximum
operating voltage is normally well below that of a clean precipitator [ 11]. This
can be ascribed to the presence of the dust layer on the collecting plate, the
electrode at which electrical breakdown is initiated. The most commonly reported
20-3
-------
kV
value for the average electric field at breakdown is approximately 5 — although
it can fall significantly below this in cases of high resistivity dust depending
upon the degree of back corona present. Noso [12] has shown that the breakdown
voltage as determined in an operating precipitator undergoes a linear increase
with spacing providing the dust loading is not too large. His data show that
kV
this is equivalent to a value of E =3.4 — independent of wire-plate spacing.
-1 a.v cm
This can decrease substantially if dust is present in quantities sufficient to
create significant particle space charge.
From this review it can be concluded that if the operating conditions of the
precipitator are below the sparkover condition, as spacing is increased it is
possible to achieve increased values of E and E up to the limits imposed by
the breakdown conditions. If however the precipitator is operating at close to
breakdown then as spacing is increased under the best of conditions the value of
E and E will remain constant at a value of approximately 5 kV/cm (or less,
p av
depending upon conditions). This then limits the effect that E and E can have
on uj as spacing is increased. It is possible for the breakdown fields to
e
increase slightly as a result of improvements in corona current distribution on
the plate and improved alignment of electrodes etc. but these are probably second
order effects when compared to predicted electric field increases due to ionic
space charge.
MECHANICAL AND AERODYNAMIC FACTORS
The two most important non-electrical factors in wide plate precipitators are
believed to be re-entrainment and rapping losses with sneakage losses being of
lesser importance. Although much less work has been reported in the literature
on quantifying these effects there is some direct and indirect evidence that they
are affected by wire to plate spacing.
Most obviously, particle re-entrainment due to aerodynamic forces is proportional
to the surface area of the deposited dust. Thus it would be expected that as the
specific collection area is decreased due to increased spacing, the amount of
surface re-entrainment also should decrease. On the other hand, for the same
quantity of collected dust, as the specific area is decreased, the thickness of
the dust layer will build up at a faster rate and care must be taken to ensure
the appropriate rapping cycle is used to minimize rapping losses. There is
evidence [13] that rapping efficiency increases with layer thickness. This of
course is also dependent upon the adhesive properties of the dust which are
affected by both electrical and chemical factors. However it is clear that in a
highly efficient precipitator where re-entrainment losses comprise a major
portion of the particles which escape, aerodynamic and mechanical factors must be
a significant factor governing the effective migration velocity.
EXPERIMENTAL DATA
In order to determine the relative importance of the electrical compared to
mechanical and aerodynamic effects it is proposed to check the correlation
between the electric field relationships defined in equations (2) and (3) and
migration velocities for selected published data. Unfortunately, although many
papers have reported improvements in effective migration velocity for wide plate
precipitators, not all provide sufficient data to allow the calculation of the
electric field. Furthermore, in order get meaningful comparisons among different
precipitators etc. it is necessary to have data that can be compared for at least
20-4
-------
two plate spacings in a given unit operating under similar "well behaved"
conditions i.e.: no back corona etc. A total of sixteen different sets of
suitable experimental data have been identified as reported by three different
workers. These include a number of tests on full scale plants as summarized by
Heinrich [14] pilot plant testing reported by Dubard et al. [15] and the results
of pilot plant tests provided by Matts [ 16] .
In each case they have provided values of effective migration velocity along with
sufficient electrical data for tests carried out for at least two different plate
spacings under similar test conditions.
The electric fields were calculated for each case as follows.
AVERAGE FIELD (E )
av
This is defined by convention as simply the applied voltage (V) divided by the
wire to plate spacing (h) i.e.:
E = T (5)
av h v '
FIELD AT PLATE (E )
The electric field at the plate surface is governed by three components, a) the
electrostatic field due to the applied voltage, b) the space charge component due
to the ions and c) the space charge component due to the charged particles.
From the voltage and current information coupled with the geometrical and
operating conditions it is possible to estimate the value of the particle space
charge density, p, , by using the expression for the voltage current relationship
as given previously and onset voltage with dust load given as
V ' = V + — — (6)
o o 2e
o
where V is the corona onset voltage without dust present [11] .
o
This then allows the field E to be calculated using the expression reported
P
in [17] modified as suggested by Lawless [18] to give
2J h p h2
^f- + ( — ) (7)
010
nr E
where E,
,
1 2c
J = ionic current density at the plate
P
b. = ionic mobility
r = corona wire radius
o
20-5
-------
E = corona onset field
o
c = one half wire-wire spacing
The data for the various tests are suinmarized in Table I which also shows ratios
?
of effective migration velocity, plate spacing, the product E E and E
P clV p
for different pairs of plate spacings. These normalized parameters can then be
used to compare the relative effects among the different precipitator tests.
Figure I shows a plot of the ratio of effective migration velocity as a function
of the ratio of the plate spacing. The solid line on this figure shows the
direct correlation that would be expected if migration velocity increased in
direct proportion to the plate spacing. The dashed lines define the limits of
agreement within +15% of the direct correlation. The degree of agreement is
summarized in Table II which shows that over 75% of the results fall within these
limits whereas only 12% and 13% of the values lie below or above these limits
respectively. Figure II shows a plot of the ratio of migration velocity as a
function of the ratio of the products (E E ) /(E E ) . Once again the solid
p av z! p av 1
line represents the direct correlation results with the dashed lines giving the
+_15% limits of agreement. From Table II it can be seen that in this case only
31% of the results show direct correlation. None of the results fall below the
correlation limit shown on the curve whereas 69% fall above the upper assumed
limit. This means that 69% of the results had an increase in migration velocity
greater than that predicted by the increase in field strength on the assumption
that the dependency given in equation (2) applies.
Figure III shows a similar plot except that the dependency predicted by equation
(3) is used as the abscissa. These again show that 38% of the data agree with
the predicted dependency with 44% falling above and 18% falling below the
prediction.
DISCUSSION
Although the data reported above are based upon a limited sample their legitimacy
is re-enforced by the fact that each comparison was made from tests covering a
broad range of conditions but in which special care was taken to ensure similar
conditions were maintained between each pair of tests.
The results of Figure I confirm the general observations reported by many workers
that show the effective migration velocity increases in direct proportion to the
wire-plate spacing since 75% of the observations agree with this prediction.
These results suggest that the data is representative of those typical for wide
plate precipitators.
The results of Figure II show that although 31% of the data agrees with a
predicted increase in effective migration velocity as defined by equation (2),
69% of the results show increases greater than that predicted. There are no
observations showing effective migration velocities less than that predicted.
This indicates that the electric field dependency as defined by equation (2)
clearly underestimates the increase in effective migration velocity. This
presumably indicates that either a) the dependency given by equation (2) is
incorrect and/or b) other factors, such as mechanical and aerodynamic, may
account for much of the improvement in effective migration velocity.
20-6
-------
The results of Figure III show that 38% of the data agrees with a predicted
increase in migration velocity as defined by equation (3) whereas 44% are
underpredicted and 18% are overpredicted. The fact that there is some
underprediction in this case as compared to Figure II, suggests that within
experimental error, equation (3) is perhaps a better predictor of the electrical
dependency of effective migration velocity. However, the fact that 44% of the
data are still underpredicted according to this model, suggests that mechanical
and aerodynamic factors have as much if not a greater effect on the effective
migration velocity than the electric field conditions. It can also be observed
that the majority of the data points which are underpredicted by the electric
field dependency have field value ratios in the range 1-1.5 i.e.: very little
increase in electric field strength.
CONCLUSIONS
1. Under normal conditions the electric field strength (E and E )
p av
remains approximately constant at breakdown as wire-plate spacing
increases. This implies that there is a limit to the potential
improvement in effective migration velocity as wire-plate spacing is
increased if the precipitators are operated close to breakdown
conditions.
2. The test data reported shows good agreement of the relationship
w ah (75% agreement)
3. Only 31% of the reported data shows agreement with the electric field
dependency as defined in equation (2) and 38% agree with equation
(3). Equation (2) underestimates the improvement in 69% of the cases
and equation (3) in 44% of the cases leading to the conclusion that
mechanical and aerodynamic re-entrainment effects are as important if
not more important than the electric field dependency in some
circumstances. This confirms the views of Matts [6] and reaffirms
his belief that further work on re-entrainment losses needs to be
undertaken.
ACKNOWLEDGEMENTS
The authors acknowledge with thanks the financial support of the Natural Sciences
and Engineering Research Council of Canada who have provided a grant in aid of
research which has been used to support Mr. Elshimy.
20-7
-------
MIGRATION
VELOCITY
RATIO 2
O HEINRICH I
HEINRICH
A HEINRICH
A HEINRICH IV
O HEINRICH V
V DUBARD
DUBARD
O MATTS I
MATTS
0 1
RATIO OF PLATE SPACING
Figure I. Relationship Between the Ratios of
Migration Velocity and the Wire to
Plate Spacing
2
20-8
-------
4
3 —
MIGRATION
VELOCITY
RATIO 2
w«
0
O HEINRICH
• HEINRICH II
A HEINRICH
A HEINRICH IV
D HEINRICH V
V DUBARD
V DUBARD
MATTS I
4MATTS
1
0
1
RATIO OF PRODUCT OF AVERAGE ELECTRIC FIELD
AND ELECTRIC FIELD AT PLATE
(E E )
p av 1
Figure II. Relationship Between the Ratios of Migration Velocity and
the Product of the Average Electric Field and the Electric Field at
the Plate.
20-9
-------
4
MIGRATION
VELOCITY
RATIO 2
O HEINRICH
HEINRICH II
A HEINRICH III
A HEINRICH IV
D HEINRICH V
V DUBARD
DUBARD
OMATTS I
*MATTS
0
0123
RATIO OF SQUARE OF ELECTRIC FIELD AT PLATE (E ^/E ^)
Figure III. Relationship Between the Ratios of Migration Velocity and
the Square of the Electric Field Strength at the Plate.
20-10
-------
REFERENCES
1. D.O. Heinrich. "Review and Some Practical Aspects of Wide Duct Spacing."
Proc. Int. Conf. on Electrostatic Precipitation, Monterey, CA, 1981, pp.
639-659.
2. K. Darby- "Plate Spacing Effect on Precipitator Performance." Proc. of
Second Int. Conf. on Electrostatic Precipitation, Kyoto, Japan, 1984, pp.
376-383.
3. S. Masuda. "Present Status of Wide-Spacing Type Precipitators in Japan."
EPA Symposium, Denver, CO, 1979, pp. 483-501.
4. G.S.P. Castle et al. "Electric Field Conditions in Conventional and Wide
Duct Precipitators." Proc. of Third Int. Conf. on Electrostatic
Precipitation, Abano-Padova, Italy, 1987, pp. 65-80.
5. T. Misaka et al. "Electric Field Strength and Collection Efficiency of
Electrostatic Precipitators Having Wide Collecting Plate Pitches." CSIRO,
Conference on Electrostatic Precipitation, Leura, 1978.
6. S. Matts. "Re-entrainment - The Major Problem in Precipitators Designed
for High Efficiency." Proc. Third Int. Conf. on Electrostatic
Precipitation, Abano-Padova, Italy, 1987, pp. 103-105.
7. S. Sekar et al. "On the Prediction of Current-Voltage Characteristics for
Wire-Plate Precipitators." Journal of Electrostatics, 1981, pp. 35-43.
8. L.C. Thanh. "Negative Corona in a Multiple Interacting Point-to-Plane Gap
in Air." IEEE Trans, on Industry Applications, Vol. IA-21, #3, 1985.
9. Q. Vuhuu et al. "Influence of Gap Length on Wire-Plane Corona." IEEE
Trans, on PAS, Vol. PAS-88, #10, 1969.
10. W.L. Lama et al. "The Sparking Characteristics of Needle-to-Plane
Coronas." IEEE Trans, on Industry Applications, Vol. IA-12, 1976, pp. 288-
293.
11. G.W. Penney et al. "Sparkover as Influenced by Surface Conditions in D.C.
Corona." AIEE Trans., Pt. I (Communications and Electronics), Vol. 79,
1960, pp. 112-118.
12. S. Noso. "Performance Characteristics of Electrostatic Precipitators with
Wide Spacing." Proc. Int. Conf. on Electrostatic Precipitation, Monterey,
CA, 1981, pp. 654-667.
13. S.A. Self et al. "Experimental Study of Plate Rapping and Ash
Reentrainment." Proc. of Third Inf. Conf. on Electrostatic Precipitation,
Abano-Padova, Italy, 1987, pp. 401-420.
14. D.O. Heinrich. "Excerpt from Internal Report of Test Results on Six
Plants." Personal Communication, Dec. 1987.
15. J.L. Dubard et al. "An Investigation of Precipitator Wide Plate Spacing."
Research Project 1835-5, Final Report, Dec. 1987
16. S. Matts. Re: Influence of Field Strength Test Data, Personal
Communication, Apr. 1988.
20-11
-------
17 W.S. Bracha and G.S.P. Castle. "Electrical Conditions in Dust Loaded Wire-
Duct Precipitators." Conf. Record IEEE-IAS Annual Meeting, Pittburgh, PA,
1988, pp. 1652-1656.
18. P.A. Lawless. Discussion of [18], IEEE-IAS Annual Meeting, Pittsburgh, PA,
1988.
20-12
-------
Table 1
EXPERIMENTAL DATA
ro
o
TEST
HEINRICH
I
II a)
b)
III a)
b)
IV a)
b)
V
h
(cm)
10.8
12.6
15.8
16.4
32.8
16.4
32.8
15.1
25.2
15.1
25.2
25.2
16
32
16
32
16
28.5
CONDITIONS
-voltage at
sparkover
-fixed gas
velocity
-normal gas
velocity
-high gas
velocity
-constant total
current
-constant total
current
-normal gas
velocity
-2 fields
-normal gas
velocity
-single field
-normal gas
velocity
-normal gas
velocity
V
(kV)
37
46
61
36
63.5
38
66
40
70
42
55
54
52
110
51
122
36
70
J
E
av
((iA/m)2 (kV/cm)
200
250
320
44.7
96.2
53.9
107.6
41.6
74.3
114
189
200
86
205
74
99
100
162
3
3
3
2
1
2
2
2
2
2
2
2
3
3
3
3
2
2
.43
.65
.86
.2
.94
.32
.01
.65
.86
.78
.18
.14
.25
.44
.19
.81
.25
.46
E
P
(kV/cm)
2
3
3
1
2
2
2
2
3
2
3
2
3
3
3
4
1
2
.87
.32
.74
.88
.12
.03
.33
.76
.17
.66
.34
.63
.52
.7
.48
.63
.87
.59
We h2 (EP2}
(cm/sec) h1 2
24
33.5 1.17 1.34
45 1.46 1.70
11.6
25.8 2 1.27
14.8
35.8 2 1.32
4.9
8.8 1.67 1.32
4.9
6.5 1.67 1.58
9.5 1.67 .98
6.06
13.2 2 1.11
9.3
12.8 2 1.77
4.73
15.47 1.78 1.92
(E E )„ co_
p av 2 2
(E E ) CO.
p av 1 1
_
1.23 1.4
1.47 1.88
-
.99 2.22
-
.99 2.42
-
1.24 1.8
-
.98 1.33
.76 1.94
-
1.11 2.18
-
1.59 1.38
- _
1.51 3.27
-------
Table I (continued)
ro
o
TEST
DUBARD
I
II
MATTS
h
(cm)
11.43
15.24
22.86
11.43
15.24
22.86
15
20
15
CONDITIONS
-low gas
velocity
-constant
current
density
-high gas
velocity
-constant
current
density
-constant
current
density
V
(kV)
37
50
67
33
49
62
= 34
=47
= 47
J
(uA/m)2
415
410
390
256
410
433
100
100
300
E
av
(kV/cm)
3.24
3.28
2.93
2.89
3.22
2.7
2.26
2.35
3.13
E
P
(kV/cm)
2.76
3.12
3.43
2.27
3.09
3.6
2.37
2.86
3.87
a) hn
e 2
(cm/sec) h
17.8
24.4 1.33
30.3 2
21.6
29.2 1.33
38.1 2
62*
105* 1.33
58*
(E _) (E E )_ en
p2 p av 2 2
,T> X2 (E E )l W1
(Epl) p av'l 1
1.28 1.14 1.36
1.54 1.12 1.70
1.85 1.52 1.35
2.52 1.48 1.76
_
1.46 1.25 1.7
_
20
= 68
300
3.4
4.64
115*
1.33
1.44
1.30
1.98
-------
Table 2
COMPARISON OF EXPERIMENTAL RESULTS WITH PREDICTIONS
Comparison
w n
2 2
— VS. r—
Ul hl
Percent within
+15% of Direct
Correlation
75%
Percent below
Correlation
12%
Percent above
Correlation
13%
vs.
(E E )„
p av 2
(E E ).
p av 1
31%
69%
38%
18%
44%
20-15
-------
CHARACTERISTICS OF RAPPING ACCELERATION OF
PRECIPITATOR COLLECTING PLATES BEFORE AND AFTER THE INSTALLATION OF
PLATE STRAIGHTENING DEVICES
Jeffrey Cummins
Neundorfer, Incorporated
4590 Hamann Parkway
Willoughby, Ohio 44094
ABSTRACT
The installation of plate straightening devices on bowed or damaged collecting
plates in an electrostatic precipitator can improve plate to electrode
clearances and extend useful plate life. This paper discusses the effect of
the installation of plate straightening devices on the acceleration of
collecting plates resulting from rapping forces. The acceleration of
collecting plates due to rapping was measured in a precipitator before and
after the permanent installation of plate straightening devices. The test
results indicated a decrease in the rapping accelerations produced by an
electric vibrator and a pneumatic rapper. However, the acceleration of the
collecting plates produced by a gravity impact rapper increased by an average
17 percent at three different levels of rapping intensity following the
installation. The results of this evaluation indicate that collecting plates
can be successfully straightened without a significant degradation of rapping
acceleration.
DISCLAIMER
The work described in this paper was not funded by the U. S. Environmental
Protection Agency and therefore the contents do not necessarily reflect the
views of the Agency and no official endorsement should be inferred.
21-1
-------
CHARACTERISTICS OF RAPPING ACCELERATION OF
PRECIPITATOR COLLECTING PLATES BEFORE AND AFTER THE INSTALLATION OF
STRAIGHTENING DEVICES
INTRODUCTION
This paper addresses the results of an evaluation of the effect of the
installation of plate straightening devices on the rapping acceleration of
collecting plates. The data and conclusions which were derived from this
analysis represent the first phase of a continuing evaluation. The objective
of the initial phase of the project was to establish baseline data of the
acceleration of collecting plates which was produced by various rapper types
before and after the installation of plate straightening devices.
BACKGROUND
The optimization of plate-to-electrode clearances in electrostatic
precipitators is critical to precipitator field strength. Sparkover occurs at
low voltage in the case of reduced plate to electrode clearances. Because
precipitator collecting efficiency is a function of corona power, a reduction
in voltage can produce a significant effect on precipitator performance(l).
Close plate-to-electrode clearances in a precipitator can occur as the result
of bowing due to thermal forces or mechanical damage to collecting plates. In
the event that damage to collecting plates results in a significant reduction
in clearances, corrective measures may be necessary such as the removal of
adjacent emitting electrodes or the repair or replacement of the damaged
collecting plates. An alternative to the removal of electrodes or replacement
of the plates is the straightening of damaged plates either by heat treating
and crimping or by the permanent installation of plate straightening
devices(Z).
21-2
-------
Plate straightening systems consist of spacers which are affixed to the
internal baffles of the collecting plates, spreading the damaged plate
sections and maintaining the plate panels at their former spacing. The plate
straightening devices are typically held in place in the precipitator by the
interlocking and suspension of multiple straighteners, by clamping or
frictional forces, or by mechanical fasteners or welds. A representation of a
plate straightening device is shown in Figure 1. The installation of a plate
straightening system results in the physical connection of the collecting
plates at a number of points, creating a rigid framework. The movement of the
individual plates may be restricted by the installation of a straightening
system.
The effect of the installation of a plate straightening system on the
performance of the collecting rapping system is a factor in the evaluation of
the potential benefits of such a system. However, a limited amount of
information is available concerning the relationship between rapping system
performance and plate straightening system installations. This paper
addresses the effect of the installation of a system of plate straightening
devices on rapping acceleration in a single plate group. The conclusions
which are drawn from this analysis are specific to the installation and
conditions under which the field data was collected. The results of further
analyses will be affected by the condition of the collecting plates, plate
supports and rapping assemblies, the collecting system rapping density, and
the type of rappers and straightening devices which are evaluated.
TEST INSTALLATION
Rapping acceleration data was collected from an early 1970s vintage
electrostatic precipitator located downstream of a cyclone-fired boiler on a
500MW generating unit. The ESP has a design SCA of 300 ft2/1000acfm, and
operates at a temperature of 300°F. The collecting plates are composed of 20
gauge carbon steel, measuring 12 feet by 36 feet. The plates are supported in
groups of eleven plates and are rapped by electric vibrators at the leading
2
and trailing edges. The installed plate rapping density is 4752 ft /rapper.
21-3
-------
Because the configuration of the collecting plate groupings and rapping
transmission systems is consistent throughout the precipitator, the
measurement of rapping acceleration was limited to one representative group of
collecting plates in the second field. A visual examination of the test plate
group revealed bowing of the collecting plates to 1 inch from vertical plumb.
No mechanical damage to the collecting plates, plate support assemblies,
alignment members, or rapping transmission assemblies was discovered. The
collecting plates were water washed prior to the collection of acceleration
data and installation of the plate straightening devices. Acceleration data
were collected from eighteen points on three plates. The locations of the
test points are included in Figure 2.
EQUIPMENT
The collecting plate accelerations were measured by Bruel & Kjaer Model 4375
piezoelectric accelerometers. The accelerometers each weigh 2.4 grams bare,
with an effective weight of 2.7 grams. The frequency response extends to 16
_2
KHZ with a sensitivity of approximately 0.30 pC/ms . The transducers were
affixed to the collecting plates and rappers using magnetic bases. The signal
produced by the accelerometers was amplified and conditioned by a Bruel &
Kjaer Model 2635 charge amplifier. The conditioned signals were analyzed on a
Scientific Atlanta model SD 380 spectrum analyzer and recorded on a TEAC Model
R-81 recorder.
Stationary accelerometers were mounted on the rapper and at three points on
the rapping anvil beam. An additional accelerometer was repositioned at each
of the test points on the collecting plates.
SCOPE
The scope of the evaluation included the following:
1. Baseline measurements were collected of the acceleration produced by the
existing electric vibrator before the installation of plate straightening
devices.
21-4
-------
2. Measurements were collected of the acceleration produced by a 3 inch
diameter pneumatic rapper operated at 30 psi. Measurements of
acceleration were recorded from each of the test points and the rapper.
3. Measurements of acceleration produced by an electromagnetic lift/gravity
impact rapper were collected at the rapper and each of the plate test
points at rapping forces of 6.7 ft-lb, 13.3 ft-lb, and 20.0 ft-lb.
4. A system of plate straightening devices was installed in the test plate
group. The straightening devices consist of two interlocking carbon steel
members which were clamped to the internal plate baffles and held in place
by frictional forces. Straightening devices were installed at three
elevations on each internal plate baffle.
5. Plate acceleration measurements were collected at the rapper and at each
plate test point following the installation of the plate straightening
system for each of the three rapper configurations outlined above.
FINDINGS
Rapping impact creates a tension-compression stress which is conducted through
the collecting plate at the speed of sound. This wave creates an in-plane
acceleration in the plate. In response to the compression-tension wave, a
flexural wave is created which travels at a lower rate of speed through the
plate, creating an acceleration perpendicular to the surface of the collecting
plate. Accelerations which were measured during the course of this evaluation
were perpendicular to the surface of the plate. Zero-to-peak acceleration was
2 2
measured in gravities (g) with 1 g measured at 9.81 m/s (32.2 ft/sec ).
The mass of the accelerometer has been shown to influence the measurement of
plate acceleration in controlled tests(3). Because the objective of this
evaluation was to obtain comparative values of acceleration, the effect of the
mass of the accelerometers and the magnetic mounting bases was not considered
in the results.
21-5
-------
The acceleration of the collecting plates under each test condition is shown
in the attached Table 1.
Following the installation of the plate straightening system, the acceleration
produced by the electric vibrator was reduced by an average 4.7 g,
representing a 41 percent decrease in acceleration. The accelerometer mounted
on the vibrator measured an acceleration of 100 g.
The acceleration produced by the 3 inch pneumatic rapper decreased an average
of 8 g following the installation of the plate straightening system,
representing a 31 percent decrease in acceleration. The accelerometer mounted
on the rapper measured an acceleration of 255 g.
The average acceleration of the collecting plates which was produced by the
gravity impact rapper increased by the following degrees: 3 g or 15 percent at
6.7 ft-lb; 4.5 g or 17.7 percent at 13.3 ft-lbs; 5 g or 17.5 percent at 20
ft-lbs. The accelerometer mounted on the rapper measured accelerations of
2050 g at 6.7 ft-lbs, 2500 g at 13.3 ft-lbs, and 2700 g at 20 ft-lbs.
CONCLUSIONS
The following conclusions are based on the evaluation of the effect of plate
straightening on rapping transmission:
The acceleration of the collecting plates which was produced by the electric
vibrator and pneumatic rapper were significantly reduced following the
installation of the plate straightening system.
The acceleration of the collecting plates which was produced by the gravity
impact rapper increased an average 17 percent following the installation of
plate straightening devices. The acceleration characteristics of the
collecting plates were consistent at the three levels of input of rapping
force.
21-6
-------
Under the conditions of the evaluation, the results indicate that collecting
plates can be successfully straightened without a significant degradation in
rapping acceleration. Successive phases of the evaluation will address the
acceleration characteristics of alternate plate and straightening device
designs under laboratory and field conditions.
REFERENCES
1. C. A. Gallaer. Electrostatic Precipitator Reference Manual. Palo Alto,
California. Electric Power Research Institute, 1983, pp. 5.09-5.13.
2. J. Katz. The Art of Electrostatic Precipitation. Pittsburg: S&S
Publishing Company, Inc., 1979, pp. 86-90.
3. Effect of Accelerometer Mass on the Flexural Vibration of Plates.
FP-1006, vol. 2. Electrostatic Precipitator Plate Rapping and
Reliability. Palo Alto, California. Electric Power Research Institute,
September, 1980, pp. 7.1-7.31.
21-7
-------
ro
i—'
CXI
Collecting Plate
Section
Figure 1. Typical Plate Straightening Devices
Figure 2. Rapping Acceleration Evaluation
Test Point Locations
-------
A
C
C
O
• BEFORE
D AFTER
1A 3A 6A IB 3B 6B 1C 3C 6C ID 3D 6D IE 3E 6E IF 3F 6F
Figure 3. Collecting Plate Acceleration Produced by
Electric Vibrator Before & After Plate Straightening
-------
A
C
O
H BEFORE
D AFTER
1A 3A 6A IB 3B 6B 1C 3C 6C ID 3D 6D IE 3E 6E IF 3F 6F
FIGURE 4
COLLECTING PLATE ACCELERATION PRODUCED BY
PNEUMATIC VIBRATOR BEFORE & AFTER PLATE
STRAIGHTENING
-------
'10
o
• BEFORE
11J AFTER
11,11111.1,1,1
1A 3A 6A IB 3B 6B 1C 3C 6C ID 3D 6D IE 3E 6E IF 3F 6F
FIGURE 5
COLLECTING PLATE ACCELERATION PRODUCED BY
GRAVITY IMPACT RAPPER AT 6.7 ft./lb. BEFORE
& AFTER PLATE STRAIGHTENING
-------
A
C
C
E
L
E
R
A
T
I
O
N
BEFORE
AFTER
1A 3A 6A IB 3B 6B 1C 3C 6C ID 3D 6D IE 3E 6E IF 3F 6F
FIGURE 6
COLLECTING PLATE ACCELERATION PRODUCED BY
GRAVITY IMPACT RAPPER AT 13.3 ft./lb. BEFORE
& AFTER PLATE STRAIGHTENING
-------
A
C
C
E
L
E
R
A
T
I
O
N
• BEFORE
D AFTER
ll.llll
1A 3A 6A IB 3B 6B 1C 3C 6C ID 3D 6D IE 3E 6E IF 3F 6F
FIGURE 7
COLLECTING PLATE ACCELERATION PRODUCED BY
GRAVITY IMPACT RAPPER AT 20.0 ft./lb. BEFORE
& AFTER PLATE STRAIGHTENING
-------
TABLE 1
COLLECTING PLATE RAPPING ACCELERATION DATA
TEST ELECTRIC
POINT VIBRATOR
BEFORE AFTER
1A
3A
6A
IB
3B
6B
1C
3C
6C
ID
3D
6D
IE
3E
6E
IF
3F
6F
11
11
20
5
7
10
9
4
15
16
10
27
5
8
20
12
7
13
8
8
15
4
R
5
5
5
6
9
11
14
5
5
4
6
4
6
PNEUMATIC
VIBRATOR
BEFORE AFTER
25
28
32
13
20
20
12
15
28
29
33
34
27
23
25
23
23
31
14
22
41
10
12
14
9
9
13
25
21
31
6
10
19
15
15
17
GRAVITY IMPACT
RAPPER
6.7 ft./lb. 13.3 ft./lb.
BEFORE AFTER BEFORE AFTER
2
33
60
9
19
21
6
10
16
20
16
16
5
8
8
13
6
13
41
60
70
12
16
18
15
16
13
16
7
14
5
12
5
9
8
21
31
37
76
11
23
24
8
12
19
26
21
19
6
11
10
18
8
16
47
77
99
15
25
22
18
20
16
19
9
15
6
16
6
11
10
26
20.0 ft./lb.
BEFORE AFTER
33
38
83
14
24
27
8
15
22
28
23
21
7
15
12
19
8
17
53
79
104
16
29
23
20
21
19
22
11
21
7
17
7
12
11
30
21-14
-------
TEMPERATURE DEPENDENCY OF MAGNETIC IMPACT RAPPERS
Michael W. Neundorfer
Karl M. Artz
Michael A. McNabb
Neundorfer, Inc.
4590 Hamann Parkway
Willoughby, Ohio 44094
ABSTRACT
The output force of popular magnetic impact rappers varies from 25% to 50% due
to daily and seasonal operating temperature fluctuations. This variation can
significantly affect precipitator performance. Designers and users of these
rappers have apparently been unaware or unconcerned about this problem until
now.
Magnetic impact rappers are widely used for cleaning electrodes of coal fired
steam generator precipitators. Rappers are installed in weather enclosures or
outdoors. In both cases, rapper coil operating temperatures are uncontrolled.
Solar heating and day to night ambient temperature change and seasonal ambient
temperature change can cause a 70°F to 90°F change in coil operating
temperature. The resultant change in coil resistance is 16% to 20%. Tests
conducted on popular rappers reveal a typical 25% to 50% variation in output
force over this temperature range. This variation can cause deteriorated
precipitator performance and/or internal structural damage.
This paper describes rapper output/temperature test data on three popular
magnetic impact rappers. The paper discusses the precipitator performance
effects of uncontrolled rapping force variations. It also reveals electronic
and software methods (patents applied for) designed to provide consistent
rapping output corrected for temperature induced f i'1: Luau'ons.
DISCLAIMER
The work described in this paper was not funded by the U.S. Environmental
Protection Agency and therefore the contents do not necessarily reflect the
views of the Agency and no official endorsement should be inferred.
22-1
-------
TEMPERATURE DEPENDENCY OF MAGNETIC IMPACT RAPPERS
INTRODUCTION
Good performance of dry type electrostatic precipitators depends on efficient
electrode cleaning. Efficiency of electrode cleaning can be defined as the
percentage of electrode collected dust reaching the hoppers (or other lower
containment) divided by the total electrode collected dust. Many factors
affect efficiency of electrode cleaning: rapping energy input, rapping energy
transmission, electrode excitation, dust resistivity, dust agglomeration
characteristics, and gas flow. This paper focuses on rapping energy input of
magnetic type rappers. The purpose of this work was to measure, explain, and
compensate for magnetic rapper output dependence on operating temperature. We
will discuss the theory of operation of magnetic lift, gravity impact
(Magnetic Impact) rappers, report results of lift height versus temperature
tests on three popular magnetic impact rappers, and discuss the implications
of these findings on precipitator performance.
BACKGROUND; OPERATION OF MAGNETIC LIFT, GRAVITY IMPACT RAPPERS
Magnetic impact rappers are used on the majority of weighted wire
precipitators serving coal fired steam generators. We estimate that there are
over 1,000 precipitators in the United States with magnetic impact type
electrode rappers. Acceptable performance of most of these precipitators
depends on adequate, consistent rapper energy input.
Magnetic impact rapper performance varies up to thirty percent of a nominal 5
to 8 inch lift height when exposed to a 52°F operating temperature difference.
Daily and seasonal ambient temperature changes can produce rapper output
levels which vary significantly from operator set levels. These unanticipated
changes in rapper performance can cause precipitator performance degradation
and possible structural damage.
22-2
-------
The two main components of magnetic lift, gravity impact rappers are the coil
and the slug. They are analogous to the coil and piston of a common solenoid.
The coil is typically a hollow, cylindrical winding of insulated copper wire.
It is firmly mounted in a rigid housing, with its longitudinal axis vertical.
The slug is a cylindrical bar of magnetically soft material (i.e. a magnetic
material with a high magnetic permeability and a low magnetic retention) such
as mild carbon steel. The slug is placed with its longitudinal axis vertical
and its top partially within the coil. It rests directly on top of an impulse
transmission rod or coupler. (See Figure 1.) Magnetic impact rappers are
rated by the impact energy they can deliver. A twenty foot-pound rapper can
lift a 20 pound slug one foot (or a 10 pound slug 2 feet, etc.) and let it
drop onto the transmission rod or coupler. The actual energy delivered by a
rapper is controlled by lifting the slug to a selected height. Control of
power supply phase angle and/or on-time is used to control slug lift height.
Gravity accelerates the slug when the coil is deenergized.
The coil is wired in a direct current circuit with a controller as shown in
Figure 2. The controller energizes the circuit for a selected number of "half-
cycles". (A "half-cycle" is one half of the period of the alternating current
supplied to the full wave rectifier within the controller; for 60 Hz power a
half-cycle is 1/120 of a second.) Typically the energization period is
between 4 and 18 half-cycles, or 0.0333 to 0.150 seconds.
A coil lifts a slug as a result of the interaction of the magnetomotive force
(mmf) of the coil with the mmf force of the slug. The mmf of the coil is
equal to the total number of coil turns times the current through the coil.
The mmf of the slug is a function of the number of the coil's turns which
surround the slug, the current through the coil and the relative permeability
(Mu ) of the slug material.
The relative permeability of the slug material is not a constant even at
constant temperature. It is dependent upon the magnetizing force to which the
material is exposed. A plot of magnetic flux density vs. magnetizing force
for a material is called its magnetization curve, and it is used to determine
the relative permeability (Mu ) for the material.
Mu for most materials is equal to or very close to 1. However, for the five
ferromagnetic elements, which include iron, values of Mu can be hundreds or
even thousands. Mu is an indication of the degree of alignment between the
circumferential electron flow in the coil and the electron spins of the
material's atoms. The sum of the individual electron spins aligned with the
coil current can be considered to act as a single, simple circumferential
current entirely within the slug. This "induced current" is the magnetomotive
force of the slug. Its direction is the same as the direction of the current
in the coil. Just as two parallel wires carrying current in the same
direction are attracted to each other, as a result of the interaction of the
current in each wire with the magnetic field generated by the current in the
other, the slug with its mmf current is attracted to the coil with its mmf
current. This attractive force lifts the slug into the coil. When the coil
is deenergized, the magnetic fields rapidly decay, the slug falls and imparts
an impact blow to the impulse transmission rod or coupler.
22-3
-------
The sequence of events is:
a) Current is passed through the coil.
b) Coil current induces a magnetic field in the coil.
c) Coil magnetic field creates circumferential current within the slug.
d) Mutual attraction between the coil and the slug (concentric
conductors with one current direction) lifts the slug into the coil.
e) Current flow to the coil is stopped.
f) Gravity accelerates the slug onto the rapping transmission rod.
Current flow through the coil is impeded by the electrical resistance of the
coil circuit conductors and the induction of the magnetic fields in both the
air space within the coil and the slug itself. Figure 3 is an oscilloscope
tracing which shows the typical current rise during operation of a 20 foot-
pound rapper. The sinosoidal profile is a carry-over of the 60 Hz AC current
provided to the controller, and provides a means of determining current levels
at various half-cycle on-times. Due to the unfiltered DC excitation of the
inductive rapper coil, the peak current does not reach the voltage divided by
circuit resistance level that would be present in a non-inductive circuit.
Since the current level in the coil is the source of the magnetomotive force
in both the coil and the slug, the performance of the rapper is directly
affected by changes in current.
The position of the slug within the coil affects both the coil circuit current
and the permeability of the slug material. As the slug moves into the coil,
an increasing number of the coil's turns surround the slug and increase the
magnetizing force to which the slug is exposed. This results in a larger mmf
induced in the slug because of both a larger magnetizing force and the higher
permeability. The mmf of both the coil and the slug is considered to act at
their longitudinal centers. The vertical force lifting the slug into the
coil increases as the slug moves into the coil and is at a maximum when the
longitudinal centers of both the coil and slug are nearly aligned. However,
this force is equal to zero when the centers are aligned because the
attractive force between them is horizontal. Should the momentum of the slug
moving through the coil carry the slug's center beyond the center of the coil,
the attractive force will slow the slug and pull the slug and coil centers
into alignment.
TEMPERATURE EFFECTS
The resistivity of the coil conductor and circuit lead wires and the relative
permeability of the slug material are the two temperature dependent material
properties which may cause temperature dependent rapper performance.
22-4
-------
The resistance of the coil circuit is directly proportional to the resistivity
of the conductor material. For the operating temperature range of a rapper
may be exposed, say -20°F to 180°F, the change in resistivity of copper is
linear. If a rapper is installed and its lift height is set with the ambient
temperature at 68 F this linear relationship indicates there will be a 19.2%
decrease in circuit resistance at -20°F, and a 24.4% increase at 180°F.
Resistivity change is considered to be the significant contributor to the
change in rapper performance as a result of temperature change.
Coil temperature rises due to repeated operation. However, this temperature
rise is not significant. Typically, magnetic impact rappers are operated no
more frequently than two sequential raps every five minutes. Temperature rise
per rap is about 0.33°F.
Data are not available for the temperature caused changes in the magnetization
curves for mild steel, the most common slug material. However the curve for
mild steel is very similar to that for iron, and data for iron indicate only a
very small temperature caused change at the high magnitization levels typical
in electric rappers. For this reason, the change in relative permeability
(Mu ) caused by ambient temperature change is not considered to contribute
greatly to the change in rapper performance.
TESTS/METHODS
1. The first series of tests was intended to measure the effect of
temperature on the performance of three commercially available
magnetic lift rappers. For the low temperature test, all three
rappers were left outdoors on a cold night. The following day, they
were connected to a rapper controller, circuit resistance was
measured, and lift-height versus on-time measurements were made.
The temperature at the time of all the tests was 9°F _+ 2°F. Each
rapper was operated at its design voltage. (2 rappers at 110 volts,
one at 220 volts). On-times were controlled in increments of 2 line
half cycles. For each on-time, five lift-height measurements were
taken, averaged, and recorded.
2. For the high temperature tests, all three rappers were wrapped with
a thermocouple controlled electric resistance heater. An additional
thermocouple was placed in the air space within each rapper above
the slug. The rappers were then heavily wrapped in multiple layers
of fiberglass insulation, and heated until the inner temperature
reading stabilized. Circuit resistance was measured and lift-height
vs. on time-measurements were taken under the same procedure as in
method #1.
3. Similarly, circuit resistance and lift-height vs. on-time
measurements were made with the rappers at ambient temperature,
approximately 61°F.
22-5
-------
The second series of tests was intended to determine any
correlation between rapper performance at both elevated and reduced
temperatures, and the circuit resistance present at these
temperatures. Only one of the rappers was tested. (Rapper #1).
Two lift height vs on time tests were performed at ambient
temperature, (61°F). For one test, resistance was added to the
coil circuit to simulate the increased circuit resistance present
when the coil was heated. For the other test, lead wire and
circuit resistors were removed from the coil circuit to simulate
the lower circuit resistance present when the coil was cooled.
RESULTS
See Tables 1, 2 & 3 and Figures 4, 5 and 6 for lift height vs on time test
results for each of the three rappers tested in the first series of tests.
See Table 4 and Figure 7 for the results of the second series of tests,
combined with the results for the particular rapper tested from the first
series.
In all cases, rapper energy levels decreased at the higher temperatures, and
increased at the lower temperatures.
The addition/deletion of circuit resistance closely simulates the effect of
increasing/decreasing coil temperature.
CONCLUSION
Rapper output differences caused by normal operating
temperature changes are large enough to affect performance of
precipitators in many installations.
Coil and lead resistance changes due to temperature changes are the
most significant cause of rapper output change with temperature.
Magnetic rapper output vs temperature characters are unique to each
rapper design.
Since lead conductor and coil resistance changes with temperature are the
primary cause of rapper output variation due to operating temperature
differences, software can be developed (and is being developed) to
automatically compensate for operating temperature differences. (Patents
being applied for by Neundorfer, Inc.)
22-6
-------
FUTURE STUDY
This study was limited to the evaluation of three popular magnetic lift,
gravity impact rappers. The tests were limited to rapper output force.
However, precipitator performance degradation can be caused by temperature
induced rapper output changes. This degradation can result in high spark
rates at low power levels or back corona due to inadequate cleaning of high
resistivity dust or high reentrainment of low resistivity, poorly
agglomerating dusts.
Future study should focus on;
1. Measurement of temperature effects on other magnetic rapper
outputs. Spring assisted impact, magnetic lift rappers are also
affected by operating temperature changes.
2. Measurement of precipitator performance changes due to temperature
induced rapper output changes.
3. Determine the effect of higher rapper energy at reduced
temperatures on the structural integrity of precipitator
components.
REFERENCES
1. David Halliday and Robert Resnick. Fundamentals of Physics. Copyright
1970 by John Wiley and Sons, Inc.
2. International Critical Tables of Numerical Data. Physics, Chemistry and
Technology^! Volume VI McGraw Hill, 1929
3. Metals Handbook. Vol. 1 Properties and Selection of Metals. Copyright
1961 by American Society for Metals
4. Herbert W. Spencer III. EPA Paper No. EPA-600/2-76-140. "Rapping
Reentrainment in a Nearly Full-Scale Pilot Electrostatic Precipitator".
May, 1976
22-7
-------
/-COIL
SLUG
HOUSING
TRANSMISSION
ROD
Figure 1: Typical Magnetic Impulse Gravity
Impact Rapper and transmission rod
Figure 2: Rapper Control Schematic
Figure 3: Oscilloscope trace of 20 ft-lb rapper
22-8
-------
10
12
13
12 (fill
11 -
O 9 F Actual Rapper Temperature
C] 61 F Actual Rapper Temperature
A 129 F Actual Rapper Temperature
y
8 -
7 -
I ! 1 1 I !
x0
/' /^
s /
/ /
$/ /
/ / /
// /
1 12
-11
-
-10
-9
-
-8
-7
h
H5
4
587!
On-Time, Half Cycles of Line
10
12
13
Figure 4: Lift Height vs. On-Time at Various
Temperatures for Rapper No. 1
ENERGIZATION
# HALF CYCLES
2
4
o
8
10
RESISTANCE
RAPPER
(DATA FC
9 DEC. F.
LIFT HEIGHT
INCHES
0.1875
2.5
5.5
9.125
\
7.45 OHMS
NUMBER ONE
)R FIGURE 4)
61 DEG. F.
LIFT HEIGHT
INCHES
0.125
1.75
4.875
8.5
11.375
8.25 OHMS
129 DEG. F.
LIFT HEIGHT
INCHES
0.0625
1.625
4.125
7.375
10.625
9.1 OHMS
Table 1:
22-9
-------
t; R 7 g 9 10 11 12 13 14 15 15 17 18 19 £0 21
-iTIT1II>''J'''£
/ P /
O 9 F Actual Rapper Temperature
D 61 F Actual Rapper Temperature
A 131 F Actual Rapper Temperature
/
/
y
/
1234567
/
-5
— 4
3
10 11 12 13 14 15 16 17 18 19 20 21
On-Time, Half Cycles of Line
Figure 5: Lift Height vs. On-Time at Various
Temperatures for Rapper No. 2
ENERGIZATION
# HALF CYCLES
2
4
6
3
10
12
14
16
18
20
RESISTANCE
RAPPER
(DATA FC
9 DEC. F.
LIFT HEIGHT
INCHES
0
0.125
1.0625
2.375
4.25
6.5
8.75
11.5
\
\
4.9 OHMS
NUMBER TWO
R FIGURE 5)
61 DEC. F.
LIFT HEIGHT
INCHES
0
0.0625
0.5
1.25
2.75
4.625
6.75
9
11.0625
\
5.3 OHMS
131 DEC. F.
LIFT HEIGHT
INCHES
0
0.0625
0.125
1.1875
2.5625
4.125
6
7.25
9.375
11.625
6.1 OHMS
Table 2:
22-10
-------
10 11 12 13 14 15
5
4
3
2 -
O 9 F Actual Rapper Temperature
CU 61 F Actual Rapper Temperature
A 143 FActual Rapper Temperature
12
11
10
9
B
7
6
6 7 8 9
On-Time, Half Cycles of Une
10
12 13
14 15
Figure 6: Lift Height vs. On-Time at Various
Temperatures for Rapper No. 3
RAPPER NUMBER THREE
(DATA FOR FIGURE 6)
9 DEC. F.
ENERGIZATION
# HALF CYCLES
2
4
6
Q
10
12
14
RESISTANCE
LIFT HEIGHT
INCHES
0.0625
0.75
2.25
4.75
8.25
11.75
V
3.72 OHMS
61 DEC. F.
LIFT HEIGHT
INCHES
0.0625
0.5
1.625
3.5
6.125
9.125
\
3.95 OHMS
143 DEG. F.
LIFT HEIGHT
INCHES
0.0625
0.375
1.375
3
5.5
8.25
11.5
4.8 OHMS
Table 3:
22-11
-------
O 9 F Actual Rapper Temperature
• 9 F Simulated Temperature
A 129 F Actual Flapper Temperature
129 F Simulated Temperature
587
On-Time, Half Cycles of Une
Figure 7: Lift Height vs. On-Time at Various
Temperatures for Rapper No. 1
ENERGIZATION
# HALF CYCLES
2
4
6
8
10
12
RESISTANCE
RAPPER
(DATA FC
9 DEC. F. (COLD)
LIFT HEIGHT
INCHES
0.1875
2.5
5.5
9.125
\
\
7.45 OHMS
NUMBER ONE
R FIGURE 7)
129 DEC. F. (HOT;
LIFT HEIGHT
INCHES
0.0625
1.625
4.125
7.375
10.625
\
9.10 OHMS
SIMULATED COLD
LIFT HEIGHT
INCHES
0.25
2.25
6
9.875
\
[ \
7.45 OHMS
SIMULATED HOT
LIFT HEIGHT
INCHES
0.0625
1.625
4.125
7.1875
10.125
11.75
9.10 OHMS
Table 4:
22-12
-------
Figure 8: Outdoor Testing of Rapper Performance
Rapper 1 Rapper 2 Rapper 3
Figure 9: Rappers Tested for Temperature Effects
22-13
-------
EXPERIMENTAL STUDY OF ASH RAPPING OFF THE COLLECTOR
PLATES IN A LAB-SCALE ELECTROSTATIC PRECIPITATOR
D. H. Choi, S. A. Self, M. Mitchner, and R. Leach
Mechanical Engineering, Stanford University
Stanford, California 94305, U.S.A.
ABSTRACT
A comprehensive study of the dynamical response of the ash layer to collector plate
vibrations resulting from the rapping blows is reported. An extensive series of experiments
was performed using a simple plate-rapping system, for which essential parameters related
to plate rapping could be accurately controlled and reproduced. The experiments showed
that the maximum acceleration of the plate and the corona current through the ash layer
critically determine the fraction of ash dislodged and the fraction of ash reentrained into the
flow stream. Shear raps were more effective than normal raps, producing less reentrainment
as well. The low frequency bulk motion of the plate detached the ash more effectively than
high frequency vibrations. The mean flow velocity and the thickness and density of the
collected ash layer were also found to influence the ash dynamics. The implications of these
findings for more effective rapping strategies in full-scale systems are discussed.
23-1
-------
EXPERIMENTAL STUDY OF ASH RAPPING OFF THE COLLECTOR
PLATES IN A LAB-SCALE ELECTROSTATIC PRECIPITATOR
1. INTRODUCTION
Non-ideal effects, such as rapping reentrainment and flow sneakage, are known to account
for a large part of the total penetration in industrial electrostatic precipitators [1]. As the
collection efficiency was increased (> 99.8%) to meet current regulations for coal-fired power
plants, limiting the losses due to non-ideal effects becomes more critical. Currently, it has
become crucial to be able to design more effective plate-rapping systems, that optimize
ash removal with a minimum of associated ash reentrainment. There has been a lack of
understanding of the basic physical processes involved in ash detachment and reentrainment.
The present study was planned to provide such understanding and quantitative data on the
rapping efficiency and reentrainment loss as a function of the relevant parameters.
In full-scale precipitators, the design of collector plates and rapping mechanisms vary
widely with the manufacturer. However, most commonly, the plates are fabricated from
sheets of thin-gauge steel, stiffened by vertical ribs, suspended from rigid supports, and
rapped by hammer blows to the top edge. Measurements of plate acceleration on full-scale
systems indicate that the mechanical impact generates traveling stress waves, responsible for
a complex time-dependent distribution of accelerations across the plate surface. The out-
of-plane acceleration components (here referred to as the "normal acceleration") are usually
comparable to the in-plane components ("shear acceleration") even when the rapping blows
are delivered in the plane of the plate [2]. Furthermore, these accelerations are the result of
a superposition of a large number of high-frequency (high-order) modes of deformation, with
the fundamental (lowest-frequency) mode not being preferentially excited [3].The maximum
acceleration in the plates reached 100<7 (Ig — Q.Sm/sec2) and frequencies as high as 7kHz
could be seen in the wave forms.
With such detailed information on the response of collector plates to rapping blows al-
ready available, the present work was directed to studying the response of precipitated ash
layers to some simple modes of plate vibration. To this end a rapping rig was built, that
allowed a small section of the collector plate in the Stanford laboratory precipitator [4] to be
rapped in either primarily "normal" or "shear" modes. Five parameters were systematically
varied: (i) the layer thickness, (ii) the mean velocity of the gas stream flow, (iii) the corona
current, (iv) the maximum plate acceleration and (v) the rapping mode, following a rational
plan to keep the total number of experiments within the realm of the possible. The plate
motion is significantly changed by placing a soft rubber pad in front of the hammer. The
consequent change in ash removal characteristics was briefly investigated at the end of the
above systematic variations.
Our findings are generally in accord with the fundamental concept that, for detachment, a
given rap must provide sufficient acceleration to overcome the cohesive and adhesive stresses
23-2
-------
keeping the layer attached to the plate. A criterion for ash layer detachment was derived by
Choi et. al. [4] by analysing the equilibrium of a homogeneous layer, of uniform thickness,
subjected to a slowly increasing acceleration. The minimum acceleration needed for layer
detachment (i.e., the "critical" acceleration acr!#), is given by:
Max{(Tcoheslve,cradhesive}
a > acrit = (1)
Here, adhesive is the cohesive strength of the layer, cr adhesive is the strength of adhesion
between the layer and the plate, p is the density and I the thickness of the laj^er. This "quasi-
static" criterion applies equally well for shear and normal raps (provided normal raps are
tensional). In the latter case a connection can be made to Moslehi's work [5] (since the
relevant CT'S are normal strengths), which asserts both crcohesive and o-adfieaive increase with
current. However, Moslehi's theory is not applicable to shear raps (inspite of the fact that
shear strengths also increase with current) since the strength in shear involves not just a
clean separation but a disruption of the layer. Generally the critical acceleration is lower
for shear raps than for normal raps. Moreover, contrary to earlier observations by Juricic
and Herrmann [6], we find that the critical acceleration in shear mode increases markedly
with the plate oscillation frequency. As a fraction of the mass of ash initially on the plate,
the mass reentrained is always small, and is significantly less for shear raps than for normal
raps. Also, the fraction reentrained decreases with increasing current and increases with
increasing gas velocity.
A brief description of the experimental apparatus and techniques used in our work will
be followed by a full account of the rap efficiency and ash reentrainment data obtained by
varying the parameters listed above. The main conclusions from this work are discussed
and recommendations for future work are suggested in the last chapter.
2. APPARATUS AND TECHNIQUES
The laboratory wire-plate precipitator (Fig. 1) consists essentially of a low turbulence
wind tunnel, cross-section 25cm wide by 75cm high, preceded by an inlet plenum where fly
ash is redispersed. The test section is a 1.8m long, wire-plate precipitator, having seven
wire electrodes (2.77mm diam.) disposed along the vertical mid-plane of the channel, spaced
every 25cm..
The mean flow velocity can be varied up to Sm/sec. At a more typical velocity of 2m/sec,
the flow rate is 0.3Sm3/.sec and the specific collector area (SCA) is 7.2m2/(m3/sec}. The
Reynolds number based on channel width ranges from 3.4 x 104 to 1.3 x 105. There is a
contraction section and a honeycomb flow straightener at the entrance of the test section.
A turbulent boundary layer grows from that station, but the test section is not long enough
for the turbulence to become fully developed. The aluminum side walls of the test section
are grounded and the wire electrodes can be energized up to QOkV (either polarity) by a
DC power supply. Typical precipitation conditions are —55&V, corresponding to a linear
current density of 0.30mA/m and average collector plate current density of 0.60mA/m2
When the ambient relative humidity drops below 50%, a humidifier adds moisture to the
inlet air stream to inhibit back corona discharges.
23-3
-------
Humidifier
Clear Lexan Top
Corona Inhibitor
(Inlet Plenum)
\
T
\\ nappeo riaie / *»»€» i-iBv,«uue
\ \ / L
y
J
\
/
*-
High Voltage Busbar
Honeycomb Flow
Jet of Fly-ash and Air I Straightener
Figure 1 Schematic of the laboratory precipitator.
The fly ash used in this work is from the Arapahoe power plant precipitator and has a
relatively high electric resistivity. It has a reasonably log-normal distribution with a mass-
mean diameter of ~ 4// and a geometric standard deviation of ~ 1.5/x. The ash is stored in
a heated hopper, and is fed through a screw feeder into a sonic nozzle driven by compressed
air. Ash is redispersed in the sonic nozzle and injected counter to the main flow near
the entrance of the plenum inlet. This arrangement ensures good mixing of the ash with
the main flow. The rotation rate of the screw feeder controls the ash feed rate (typically
~ 5gm/sec) into the precipitation section.
Leaf Springs
Figure 2 Schematic of the Rapped Plate, showing the bridging member and
the counterweight in shear rap configuration.
23-4
-------
Corona Inhibitor
Lexan Top
Hammer Head with variable mass
Rapped Plate
Tungsten-Halogen
Lamp
Intermediate
Piece
Carbon Steel
Anvil
Soft Foam
Rubber pad
LED Trigger for
the A/D board
Collector Tray • ^Lexan Bottom
(insulator material) ^L
^^ High Voltage Busbar
Figure 3 Schematic of the Shear Rapping System.
Figure 2 shows an exploded schematic of the rapped plate system as configured for shear
raps. A 30cm wide by 25cm high rapped plate was made of 1mm thick aluminum sheets and
reinforced with aluminum beams and aluminum honeycomb, to minimize bending modes
while keeping its weight low. This plate is inserted into a window in the side wall of the
precipitator section, the front surface flush with the tunnel wall. The window is large enough
to accommodate the plate dimensions allowing a gap of 5mm on all sides. Corrugated thin
plastic sheets seal this gap. The rapped plate is attached to the two spring beams through
an aluminum bridging member. An anvil of carbon steel (not shown) was attached on the
lower edge of the bridging member to distribute the highly localized initial impact stress
over the cross section, as shown in Figure 3. A counterweight was used to keep the center
of gravity of the plate-bridge-counterweight system at the center of the bridging member to
minimize torsional vibrations of the springs. Figure 3 also shows the pendulum hammer used
to deliver the rapping blow. The hammer length and the weight of its head can be changed
in order to attain the desired plate acceleration. Two piezoelectric accelerometer heads are
placed on the rapped plate, as indicated, to monitor the shear and normal accelerations.
The soft rubber pad shown in Figure 3 was used to suppress the higher frequencies and
reduce the out-of-plane motion of the plate. The rubber pad reduces the maximum accel-
eration of the plate but increase the impact duration resulting in much cleaner acceleration
traces. The out-of-plane acceleration is also reduced by an order of magnitude when using
the rubber pad. Maximum accelerations around 100 g in the shear direction (and ~ 150 in
the normal direction) with bulk motion frequencies around lOQHz were typically obtained
using a 2cm thick foam rubber pad, as shown in Figure 4a. On the other hand, the hammer
impact without the rubber pad excited a large number of higher harmonic modes. A typ-
ical time trace of plate acceleration in this situation is shown in Figure 4b. Note how the
peak acceleration is high, but does not last for long (full width at half maximum around
23-5
-------
20
30 40
Time (msec)
50
60
Figure 4 Time trace of shear acceleration of rapped plate during shear ham-
mer impact: (a) Hammer stricking anvil with rubber sheet; (b) Hammer strick-
ing anvil without rubber sheet.
O.lmsec). The normal plate rapping system is very similar to the shear one, differing only
in the layout of the components.
Figure 3 also shows an opening in the opposite wall, facing the rapped plate. This opening
is closed by a metal plate during precipitation, and by a Lexan sheet during rapping to allow
high speed video recording of the rapping process. Still photographs of the rapped plate,
before and after the rap, are taken through this opening with the gas flow turned off. A tray
is also shown, positioned below the rapped plate. It is placed there only during rapping, to
collect the ash dislodged from the rapped plate and to thereby measure the rap efficiency.
To characterize ash reentrainment during rapping, we constructed an array of transmis-
23-6
-------
someters. A collimated beam of light from a halogen lamp is directed across the tunnel and
is detected by a silicon photodiode. The extinction (=fractional reduction in transmitted
power) due to the presence of ash was calibrated in terms of the mass loading of the input
ash using an array of sampling filters, as shown in Figure 5. Details of the transmissometer
calibration can be found in Reference [7].
Silicon photodetector
Iris
Lens
Glass window
ISOKINETIC ASH SAMPLER
Wire e/octrodes
L
Rapped plate
C/imet optical particle counter
Glass window
TRANSMISSOMETER
-Tungsten-halogen lamp
Figure 5 Schematic of the Transmissometer.
A rack of six transmissometers, spaced every 5cm vertically, was positioned 25cm down-
stream of the rapped plate, as shown in Figure 6. This allowed us to monitor for nonuni-
formities in the ash reentrainment across the height of the plate.
The outputs from the six transmissometers as well as the two accelerometers were digitized
at a rate of up to 18.75kHz per channel using an A/D conversion board and stored in a
personal computer. The transmissometer readings are used to monitor the penetration
during precipitation and to measure reentrainment during rapping. The same A/D board
also controls an electromagnet that is used to control the release of the hammer off from a
manually preset height, as shown in Figure 3.
A typical rapping experiment consists of two stages: the precipitation of an ash layer onto
the rapped plate, followed by the rapping of the same layer off the plate. The first two wire
electrodes upstream of the rapped plate were removed in order to reduce the precipitation
time, and all precipitation runs were made nominally with -5QkV, 0.5mA (across four wire
electrodes), gas velocity of 2m/sec with an ash input rate of 5gm/sec. The relative humidity
inside the laboratory recorded lows ~ 50% and highs ~ 80% and temperatures were between
20°C and 30°CI over the course of the investigation. Below we summarize the main steps in
the experiments:
23-7
-------
Corona inhibitors
Lexan top wall
Flow
Wire
electrodes
Rapped
plate
Hopper
Illumination
windows
Rack of seven
Transmissometers
Lexan bottom wall
High voltage busbar
Figure 6 Layout of the Rack of Transmissometers Relative to the Rapped
Plate.
1. Begin transmissometer data acquisition at a sampling rate of WHz per
channel. Start gas flow and turn corona power on.
2. Start ash feed and begin measuring precipitation time. Precipitation times
vary between 2mm and 20mm, depending on the ash layer thickness de-
sired.
3. After precipitation time has elapsed, turn gas flow and corona power off in
that order. Computer-driven data acquisition is interrupted.
4. Remove the metal plate from the opposite opening and take a still photo-
graph of the ash layer on the rapped plate. Install the collection hopper
tray beneath the rapped plate, as shown in Figure 3.
5. Install the Lexan window in the opposite opening and turn on the electric
power and the gas flow in that order. The current density and gas flow
speed are usually set to values different than those used in precipitation.
Ash feed is kept off during rapping.
6. Arm the hammer by manually turning the electromagnet on and raising
the hammer head until the magnet clamps it. The height of the hammer
head and its mass are set beforehand to produce the desired maximum
plate acceleration. Begin the data acquisition and the high-speed video
recording.
7. The computer turns off the electromagnet releasing the hammer. A prox-
imity sensor triggers the A/D board 2msec before hammer hits the anvil. It
takes Isec to acquire 4500 samples from each of the 8 channels sequentially
at a sampling rate of 4.5kHz per channel.
23-8
-------
8. Immediately after the rap, the gas flow and the electric power are turned
off in that order. Record the flow speed, the ambient temperature and
relative humidity, and the applied voltage and current density.
9. Open the Lexan window and take another still photograph of the ash layer
remaining on the rapped plate. Take the hopper tray out and weight its
contents. This is the mass of ash dislodged by a single rap, Mr.
10. Return the hopper tray to original position inside the tunnel and manually
scrape remaining ash off the rapped plate into the tray. Remove the tray
out and weigh its content. This is the total mass of ash initially precipitated
on the rapped plate, Mt.
11. Measure the thickness of the ash layer precipitated on the metal plate (that
was inserted into the opposite opening during precipitation). The thickness
ta is an average from 9 measurements.
12. Insert the metal plate into the opposite opening again. Clean up tunnel
and prepare for next experiment.
13. The compiiter calculates the rap efficiency, ?yr, and the fractional reentrain-
ment, 3?, from the transmissometer data and Mt, Mr. It also estimates the
average layci thickness on the rapped plate from Mf, M0 and 10.
The rap efficiency, rjr, and the layer thickness on the rapped plate, £r, are calculated from
the expressions:
l\'f A
n n Jl'-'r -"-O /0\
tr=C0x-r: -x— (3)
M0 Ar
We assume that the layer density is the same on both sides of the precipitator, and A0
and Ar are the areas of the opposite plate and rapped plate, respectively.
The concentration measured by the transmissometers is multiplied by the cross-section
area and by the mean flow speed, and then integrated over the total sampling interval to
get the total reentrained mass. This is in turn non-dimensionalized by the total mass of ash
collected on the rapped plate to obtain the fractional reentrainment, 3J:
,4,
00 i-lJ 1
Here, C(y,t) is the instantaneous ash concentration at the vertical position y and time i,
U(t) is the instantaneous flow velocity (averaged over space) at time t, U is the mean flow
velocity during rapping, Cji is the ash concentration measured by the j-th transmissometer
23-9
-------
at time i; = i/va, vs is the A/D board sampling frequency per channel, Ns is the total
number of A/D samples taken per channel, and b and h are the width and height respectively
of the tunnel. The concentrations are calculated from the transmissometer readings using
a calibration curve.
The record of a typical shear rapping experiment is shown in Figure 7. At the top are
shown conditions prevailing during the rap: voltage, current, mean gas speed, ambient
temperature and relative humidity. In addition, the average thickness and total mass of the
ash layer are listed, ending with the calculated rap efficiency. The upper six plots give the
transmission measured by each of the six transmissometers versus time. The seventh window
shows the shear acceleration of the rapped plate and the fractional ash flow rate measured by
the top four transmissometers, while the last window shows the normal acceleration and the
fractional ash flow rate measured by all six transmissometers. The scales for the fractional
ash flow rates are shown on the right side of the plots, and the acceleration scales on the
left. The mass flow rates are fractional values (fraction of the total collected mass on the
rapped plate per unit time, given in %/sec).
The distinction between the measurements from only the top four transmissometers and
from all six was based on repeated observations of an interesting feature. It was constantly
observed that the lowest two transmissometers, which are located below the bottom edge
of the rapped plate (as seen in Figure 6) registered a large reentrainment of ash beginning
approximately 300msec after the rap. Sometimes, the maximum reentrainment rate of this
delayed ash cloud was as much as three to four times the maximum of the prompt cloud
registered by the top four transmissometers. Video recordings showed this to be due to the
fact that the detached layer falling into the collecting tray created a large cloud of ash. The
origin of this cloud is not the rapping process per se, although it is an indirect consequence
of rapping.
Table 1. Values Assumed by Various Parameters in the Rapping Experi-
ments.
Layer
Thickness
±20%
(mm)
1.0
1.5
2.0
2.5
3.0
3.5
4.0
Layer
Density
±20%
(gm/cc)
0.5
0.6
0.7
0.8
Total
Current
±10%
(mA)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
Flow
Speed
±10%
(m/sec)
1.0
1.5
2.0
2.5
Rapping
Mode
shear
normal
Maximum
Acceleration
±5%
(g)
10
20
30
40
50
60
70
80
90
100
A base set of conditions was adopted, corresponding to a "base run". Subsequent vari-
ations consisted of changing the value of one parameter at a time, keeping all the others
23-10
-------
-47.5KV 0.80mA 1.50nVs 71.6F 55.5% Hum. 2.30mm 99.2gm 0.593gm/cc 98.2% RAP RAP266PLT
10
100
CO
E
10
100
10
100
Ti-
to
c
a
in
E
10
100
<£>
E
8
<
8
<
Total(4) = 0.009gm (0.0096%)
0 TO
006 •
006 !
004
002 ,
0
010
008
006
0>
-------
o
equivalent sets of variations were performed for both rapping modes. All experiments were
conducted with negative polarity during both precipitation and rapping. All raps used a
1.5cm thick pad of soft rubber, except where explicitly indicated otherwise.
For negative polarity, current emission from smooth wires occurs from isolated cathode
spots which are the source of visible discharge tufts projecting from the surface. These
cathode spots are rather randomly positioned along and around the wire surface, though
there is a tendency for them to be uniformly spaced due to the fact that the conducting
discharge channel lowers the electric field locally, inhibiting the formation of nearby spots.
As the current is increased, additional cathode spots appear and the tufts become closer
spaced.
Each discharge tuft is the source of a current tube of negative ions extending to the
grounded collector plates. The area of each tube increases (and the current density de-
creases) towards the collector, but the current tubes originating from separate cathode
spots do not merge. As a result, as shown by Tassiker [8], the current density distribution
at the collector is highly inhomogeneous, there being patches (or islands) having relatively
uniform current density separated by regions of essentially zero current. The fraction of
the plate area occupied by the islands increases as the total current is increased and the
intervening areas of zero current are "squeezed" into a smaller fraction.
In other work at Stanford [9,10] it was shown that the inhomogeneous pattern of current
density is manifested in a nonuniform pattern of the precipitated ash layer. In the island
areas (with current) the layer is more tightly packed (hence denser and thinner) than in
the intervening areas where it is loosely packed (less dense and thicker), though the mass of
ash per unit area is the same. The precipitated ash layers in our experiments exhibited this
inhomogeneous pattern as can be seen in Figs 9 and 10. This effect has a strong influence
on rapping because the tightly packed ash in the current carrying areas has a higher cohesive
strength than that of the loosely packed ash in the intervening areas of zero current. This
inhomogeneous pattern of ash layer strength, which is not reproducible from run to run,
causes a rather large scatter in the fraction of ash rapped off under nominally identical
conditions.
3. EXPERIMENTAL RESULTS
Each of the parameter variations is plotted on a single page having three windows stacked
vertically, sharing the same abscissa. The top window gives the rap efficiency, the middle
one gives the reentrainment measured by the top four transmissometers, while the bottom
window gives the reentrainment measured by all six transmissometers.
3.1 Variation of the Corona Current
Figure 8 shows the effect of the corona current on both the rap efficiency and the reen-
trainment. All parameters, except the current, were kept at their base values.The presence
of corona current during rapping affects the rapping process in two ways. It lowers the
rapping efficiency (undesirable), but also produces lower reentrainment (desirable). Fig. 8a
shows both rapping modes to have rapping efficiencies close to 100%, but the efficiency of
23-12
-------
Q~
cN
*~""^
5;
c
p
.c
5
03
03
cc
(^
CO
cc
-^
£
c
i
,c
TO
-fc
C
03
03
CC
cc
-^
^s.
"" —
X
g
03
'o
LU
Q^
ctj
cc
£
£
01
o
-a
1
(-»
1
a
K
IB
1
o
M
1
c
*
CO
==
1 3 • • • 1
90% o.l
80
70
60
50
40
30
20
10
0
0.06 1
§O
o
"
=- °
— C^
~—
\-
'- oo
=" 0
-- o
'- 1 1 r , 1 , , , , 1 , , , , | , , , , | , ,
t
I
0.05 1-
t
I
0.03
0.02
0.01
0
0.25;
0.20,
0.1 5(
4
0.1 0(
0.05
n
: o
- «
o
L • • o
- * m 5i • 8 °
7 • o • r
-
- i , , , i i i , , i , , , , i , , , i
i
r
-
r
, c\
r 8 *o
~ -.
- • Q
- 2 (Q • 9 „
: I »* f
1 1 1 1 1 1 I 1 1 1 1 1 1 1 1 1 1 1 1 1 _1_
1
o
o
o
(a)
1,1.111
•
°
•
(b)
,,!,!,,
•
£ (c)
,.!,,,,
0.05 0.10 0.15 0.20
Current per Wire (mA)
0.25
0.30
Figure 8 Variation of the Current during the Rap: Rap Efficiency (a), Rap
Reentrainment Measured by Top Four Transmissometers (b), by All Six Trans-
missometers (c); o normal raps, • shear raps.
the normal raps decreases more rapidly with increasing current than the shear raps. By the
time the current reaches the maximum possible (close to spark- over), normal raps are only
60% efficient, while shear raps are still 90% efficient. With increasing current, the fractional
reentrainment is reduced by a factor of five, which is a clear indication of the importance of
23-13
-------
Figure 9 Photographs of the ash layer: (a) before the rap, (b) after the rap;
Shear rap; Current = O.OmA/m
the corona current in increasing the cohesive strength of the particulate layer. The fractional
reentrainment measured by all six transmissometers is twice that measured by the top four
alone. While the top four transmissometers do not show a strong dependence on the rapping
mode, shear raps are seen to produce slightly less total reentrainment than corresponding
normal raps.
Figure 10 Photographs of the ash layer: (a) before the rap, (b) after the rap;
Shear rap; Current = Q.2QmA/m
Other evidence that sheds new light on the effect of corona current during the rapping
process is given by Figures 9 and 10. Each figure consists of two photographs of the ash
layer: (a) taken before the rap (right after the end of the precipitation) and (b) taken just
after the rap. The maximum acceleration of the plate was 60g in shear mode in both cases.
The average thickness and density of the two layers are the same. While Figure 9 (zero
23-14
-------
current) shows a clean plate after the rap, Figure 10 (0.28771^4/771) clearly shows a large
portion of the original layer that was dislodged, began to fall down but reattached to the
plate. It is important to notice that the layer was not broken into many small fragments
but into a few pieces of relatively large areas, and that those large areas exhibited very little
structural change except at the edges. The falling trajectories of a number of these small
unbroken fragments were measured from high-speed video recordings by Self et. al. [11].
These trajectories correspond to falling accelerations less than 0.5g even when reattachment
does not occur, suggesting the presence of a friction force much stronger than aerodynamic
drag force. This in turn implies that a normal force, related to the corona current, is
"compressing" (or attracting) the falling fragment onto the plate. Reattachment occurs
when this normal force is strong enough to produce a friction force larger than the weight
of the falling fragment. We believe that this reattachment phenomenon, repeatedly seen in
shear raps in the presence of current, plays an important role in decreasing the overall rap
efficiency, and is a direct effect of the corona current. It is a different kind of interaction
from the static layer strength enhancement factor analysed and observed by Moslehi [5],
and is clearly shown in a photograph obtained by Mitchner et. al. [10]. Normal raps also
show this reattachment phenomenon when the maximum acceleration is small.
3.2 Variation of the Layer Thickness
The nominal values of layer density and thickness could not repeatably be obtained, and
the data points are spread across the range. Although this makes it difficult to estimate
the scatter at each parameter value, the data are still useful in delineating trends, as can
be seen in Figure 11. Clearly the rap efficiency increases with the layer thickness. In the
limit of very large thickness, "natural" fall-off occurs which is known to have an adverse
effect on the overall efficiency of the precipitator. Again, normal raps are more sensitive to
the layer thickness than shear raps, the rap efficiency of normal raps falling almost to zero
for layer thicknesses below 1mm. The fractional reentrainment measured by the top four
transmissometers decreases with layer thickness, but the reentrained mass increases (but less
than linearly) with the total mass of the collected layer. The fractional reentrainment may
not be a good measure in this case, since the absolute reentrainment (total mass reentrained)
increases with layer thickness. The fractional reentrainment from all six transmissometers
appears to be independent of layer thickness; at least the trend is smaller than the scatter.
3.3 Variation of the Maximum Acceleration
For shear raps the maximum acceleration is taken as the absolute maximum of the ac-
celeration trace of the rapped plate. For normal raps, however, we define maximum ac-
celeration as the maximum acceleration of the rapped plate when it moves away from the
ash layer, i.e, when the layer experiences tension (rather than compression). In the normal
rap experiments, the rapping blow initially pushes the rapped plate into the tunnel, thereby
compressing the layer onto the plate. The maximum tension acting to separate the ash layer
from the plate occurs when the plate velocity first reaches zero and the motion is reversed.
Interestingly enough, the maximum acceleration with the plate pulling away from the layer
was higher than the maximum acceleration with plate compressing the layer.
The data in this subsection are divided into three groups: Figure 12 shows the results
23-15
-------
100
90
^ 80
o\
x 70
B 60
m
| 50
§ 40
§• 30
* 20
70
0
Or\ A
.04
w
i 0.03
E
o
«
.(o
i 0.02
to
K-
l^.
^
O
§• 0.07
K.
n
I • s» - 7 • •
L • ••
: • 0
L 0 • ° °
L o
E o o
r °
1 O
: O
r o °°
r o
h ra;
- , , . , 1 . , , , 1 , Q , . 1 . . . , 1 , , , , 1 , , , , 1 , , , , 1 , , , ,
: 0
-
:
L ° ° oo
: o •
0 0 0 •
- 0 Q* ° •
L • 9 • «o « 0
m • . £> •
: • • . %
^ * W
- . , i i 1 , , , , 1 , , , , 1 , , , , 1 , , , , 1 , , , , 1 , , , , 1 , , , ,
N0~
cN
CrT1
£*J
*~l
o>
p
c
c
1
c
-------
cJs
^~
SL
c
n\
vu
.c
1
Q)
cc
8-
cc
\o
s~
X
8
CD
|G
m
8-
cc
e
li
Q)
O
w
g
W)
c
CO
K
1
8-
£
IUU
90
80
70
60
50
40
30
20
10
0
0.06
0.05
0.04
0.03
0.02
0.01
n
- 9 9 O J0W ••'-' u fc M3 o O o
: * O 0 ° O
: O
r o
r o
_
"- ° a
: O ^
'- •• °
'- • o°
h • B
L • ° /„!
- o ( /
: , , fl ,nl I 1 1 1 I . . , , I , , , , I . , . , I , . , , I , , , , I , , , , I , , , ,
rri r^i
; ^
*~ O
r 00°
: • o
'- O O
: o CD
: o o
- ^ O « ° O
• • Q
- o5° ^ °*
^ • t * w
- , , , , 1 , 1 , , 1 , . . , 1 , , , , 1 , I , , 1 , , , , 1 , , I , 1 , , , . 1 1 , , 1
£
s
<
CO
^
u.^o
0.20
0.15
0.10
0.05
0
-. ° o °
~ 0 m °
I o
:
A
h o°
- o •
- ° 0
\J
-_ • o° 0 • o 9^Q
O w^^iKci o& A
^^ o^ (C)
I i T iTi i i i 1 i i i i 1 1 1 1 1 1 i 1 1 1 1 1 i i i 1 1 i i 1 1 1 1 1 1 1 1 I 1
0 10 20 30 40 50 60 70 80 9
Maximum Acceleration (g)
Figure V2 Variation of the Max. Acceleration of the Rapped Plate (Current=0.OmA):
Rap Efficiency (a), Rap Reentrainment Measured by Top Four Transmissome-
ters (b), by All Six Transmissometers (c); o normal raps, • shear raps.
of the current nonuniformity. For zero-current raps, it is seen that the detachment threshold
acceleration is about 7g for shear raps and I2g for normal raps. Shear raps stronger than 20g
completely remove the layer, while normal raps seem not to be able to produce complete
removal, no matter how large the maximum acceleration. However, normal raps above
23-17
-------
-^
o^
X
o
CD
'o
£
Uj
8-
CC
^
*-— -1 "5
*- E
C o
£ %
fc g
•E 8
03 5
1 1
CC 9-
&
CC
^
s g
c 1
CD o
SCO
w
C E
5 1
•*— j ^
c h-
CD
- o
= 9
r o
L *
I Q
L .0%
}- • *° ° (a)
= , , 0| f 9 , , , i , , , , i , , , , i , , , , i , , , , i , , , ,
~
_
_
—
_
-
—
i- •^ato?co°0.0 I
- S **• ° ° cPlP
r • .
: (b)
r , , , , I , , , , I , , , , I , , , , I , , , , I , , , , I , , , ,
:
-
-
-
_
-
—
_
- oo a ^ o
~ &y\ Q") i3 • ^
- % s^fiS1^^ o o ^ /Q\
— 1 — 1 — 1 — 1 — 1 — 1 — 1 — l_J — 1 — 1 — 1 — 1 — 1 — 1 — 1 — 1 — 1 — 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 — 1 —
0 10 20 30 40 50
Maximum Acceleration (g)
60
70
Figure 13 Variation of the Max. Acceleration of the Rapped Plate (Current=0.13m A/
Rap Efficiency (a), Rap Reentraiiiment Measured by Top Four Transmissome-
ters (b), by All Six Transmissometers (c); o normal raps, • shear raps
35g result in rap efficiencies around 90%. From Figure 14a, we see that current-on raps
with similar acceleration levels produce far lower rap efficiencies. Although the threshold
accelerations are similar, the rap efficiency rises slowly with maximum acceleration. Shear
raps reach 95% efficiency only above 60# and normal raps are even less efficient, removing
23-18
-------
only 70% with a 90
6
w
c
2
w
|
g.
e
®
|
01
<
IUU
90
80
70
60
50
40
30
20
10
0
0.06
0.05
0 04
0.03
0.02
0.01
0
0.25
0.20
0.15
0.10
0.05
n
Iff* o ° °
: • 0 * 0 0
L • * 0 o <° » °00*o&0<0 ° o °
: o • c? °° c9 (D
r^of**^ ^ i ^ °* ^
• i i i i i , i i [ i , , , , i i , i , i ,
-
-
; oo
o
- 8 °
o o o
'- ° XX&o ? ooo
I°WT i i i i i i 1 i i 1 i i i i i . .1. i i .. i
150
Figure 14 Variation of the Max. Acceleration of the Rapped Plate (Current
= 0.26mA/m): Rap Efficiency (a), Rap Reentrainment Measured by Top Four
Transmissometers (b), by All Six Transmissometers (c); o normal raps, • shear
raps
Besides the general trend of increasing reentrainment with increasing maximum
accel-
23-19
-------
eration, we observe that the rapping mode does not appreciably affect the reentrainment
behavior. Moreover, for current-on raps, reentrainment increases much more slowly with
maximum acceleration, compared to current-off raps. At QOg, current -off raps produce a.
total fractional reentrainment of 0.12%, whereas current-on raps produce only 0.03% - lower
by a factor of four. At 90;? this ratio drops to two, diminishing the initial advantage of
current-on raps over current-off ones.
3.4 Variation of the Flow Speed
Variation of the flow speed during rapping produced the results shown in Figure 15. The
rapping efficiency plot shows no obvious dependence on flow speed in the range investigated.
However, it is clearly seen that the rap efficiency for normal raps (around 70%) is lower than
that of shear raps (~ 94%). Fractional reentrainment, on the other hand, exhibits a strong
linear dependence on flow speed. Normal raps produce higher reentrainment than shear
raps (roughly a factor of two in Fig. 15c), and, again, the reentrainment measured by all six
transmissometers is about twice that measured by the top four.
3.5 Variation of the Rubber Pad Thickness
The primary goal of this variation was to study the effect of the impact duration on the
motion of the collector plate and its consequence on the ash detachment and reentrainment.
The impact duration can be changed by inserting elastic pads between the hammer head
and the steel anvil, as shown in Figure 3. The frequency spectrum of the plate acceleration
trace could be widely varied by using pads of different thicknesses and rubber materials.
The plate acceleration trace of the rap using a 1.5cm thick soft foam rubber pad (Figure 4a)
shows almost no high frequency components present: the plate executes a damped simple
harmonic oscillation with the fundamental frequency of I22Hz. On the other hand, the
acceleration trace of the rap without any rubber sheet (Figure 4b) shows a higher frequency
(S10.ffz) mode to dominate over the fundamental and some contribution from even higher
frequencies as well. Since increasing the rubber pad thickness also has the effect of decreasing
the maximum acceleration, the hammer head had to be raised to higher initial positions to
attain the same maximum accelerations. All results shown in this section were obtained
using a nominal maximum acceleration of 60g to rap ash layers 2mm thick and 0.7gm/cm3
density.
For shear raps, the rap efficiency decreases from 90% when the rubber pad was used to
20% when no rubber pad was used. The shear raps with the rubber pads also cause higher
fractional reentrainment than the raps without the pads. Normal raps using rubber pads
also produce higher fractional reentrainments than the raps not using the pads. However,
unlike the shear raps, the rap efficiency of normal raps using rubber pads does not show
any significant improvement over the raps not using the pads.
4. CONCLUSIONS
Although a quantitative model is not yet completely developed, the basic qualitative
features that were observed (some for the first time) can guide the field engineer as well
23-20
-------
sS~
(JN_
^
•+-*
c
CD
E
.c
en
s
c
CD
§
1
xp~
o^
X
0
C;
CD
[o
6
LU
c^
DC
%
CD
"qj
£
o
nsmiss
ca
^
^
o
i£
&
K-
IUU
90
80
70
60
50
40
30
20
10
' • * •
- • o
r o
'-
L °n° °
o
L oo
: O
'- °
'-
'- CO
-_
- O
L o
: 1 1 , 1 1 1 1 , , 1 I 1 , , 1 , , ,
n n 7 ,
u. u /
0.06
0.05
0.04
0.03
0.02
0.01
n
r
: o
—
~
•- •. •
»» ••
0
°0
o
0 0
o
o
o
(a)
1 1 1 1 1 , 1 , I I 1
0 00
0 0
0 f>
T.)
A
r
1 o °Q° °S
• **
i i i i 1 . i i i 1 .... 1 1 1 1 —
••
(b)
i i i i i i i i i i i
eg
+~l
c
CD
e
c
CD
I
e
-Cb
0)
o
<0
w
s
«
C:
&
.X
5;
U.^O
0.20
0.15
0.10
0.05
n
-
-
-
O
-_ 0 °
f~\
o o
t o o • o o
/-, \J ^.
^ 0 8° if
: .0-08° ° • (c)
1 1 1 1 1 1 ^ 1 ~ 1 1 1 1 1 1 I 1 ._!_ i 1 1 1 1 1 1 1 1 1 1
0
0.5 1.0 1.5 2.0
Flow Speed (m/sec)
2.5
3.0
Figure 15 Variation of the Mean Flow Speed during the Rap: Rap Efficiency
(a), Rap Reentrainment Measured by Top Four Transmissometers (b), by All
Six Transmissometers (c); o normal raps, • shear raps.
as the designer in better understanding the various aspects of ash layer detachment and
associated reentrainment. A list of the important practical implications follows:
1. In accordance with measurements by previous workers [1,6,12], shear raps
require significantly lower maximum acceleration than normal raps, for the
23-21
-------
Np~
0s-
*^
Efficiency
Q__
cts
cc
^
1 — i a
c I
CD w
11
5 2
CD l£
I*
§L is
frainmentl
•ansmissomet,
C i-
CD 35
DC ^
I?
1UU
90
80
70
60
50
40
30
20
10
0
0.06
0.05
0.04
0.03
0.02
0.01
0
0.15
0.10
0.05
0
r fc
~- A
E A o
\ A 00°
r A A
r A
L O o
~~ 0 A
=_ , , O
= i i i i 1 i i i i 1 i i i i 1 i i i A 1 Ai i i 1 i i i i 1 i i i A 1 lit i .1 i i i L
: o
-
- 00
: (b) A ^ A
: i i i 1 r i i i 1 i i r i 1 i i i i i i i i 1 i I I I 1 I I I I 1 I i t I 1 I I I I
:
- •
- O
- o o
; (c) A £ * A A A A
1 1 i r i 1 i i i i 1 i i i i 1 i i i i i i i i 1 i i I I 1 I 1 I I
0 10 20 30 40 50 60 70 80 9
Maximum Acceleration (g)
Figure 16 Variation of the Rubber Sheet Thickness (Current =
0.26m.4/m): Rap Efficiency (a), Rap Reentrainment Measured by Top Four
Transmissometers (b), by All Six Transmissometers (c); o normal raps with
pad, A normal raps without pad, • shear raps with pad, A shear raps without
pad
same layer thickness and rap efficiency. Shear raps produce rather less
reentrainment than normal raps, which is in agreement with some high-
speed photography results by Juricic and Herrmann [13].
23-22
-------
2. Total reentrainment is generally low (less than 0.1% of the ash initially pre-
cipitated) for either mode. However, a more useful way of looking at this is
to compare it with the continuous penetration occurring during precipita-
tion. Let us call this new parameter the fractional rapping penetration, p.
Our precipitation records show a typical continuous penetration of around
2gm/s. The peak reentrainment rate shown in Figure 7, for a typical shear
rapping experiment, is on the order of Q.Q8gm/s, which, however, comes
just from the small plate. The total collector surface area is some twenty
times larger than the area of the small plate, which brings the in total
rapping penetration of our setup (were the whole collector surface to be
rapped) to about 80% of the continuous penetration, i.e., p ~ 80%, while
3ft ~ 0.1%!
Rapping reentrainment rate
Continuous penetration rate (1~7?)
At
(I-*?)
(5)
where AT is the time lapse during which the bulk of the rapping reentrain-
ment takes place, and At is the precipitation time during which the total
mass, Min is fed at the precipitator inlet, r/ is the overall precipitation effi-
ciency (not to be confused with r/r, the rap efficiency. In the present case,
the precipitation efficiency was around 70%. At ~ 10mm, and Ar ~ Isec.
Thus 3? ~ 0.1% results in p ~ 150%, which is close to the value calcu-
lated above. In typical industrial precipitators r/ ~ 99%, which brings the
ratio p/5R to be on the order of 104. In addition, full-size precipitators
have several sections, and ash reentrained from upstream plates is largely
re-precipitated, so it is only the reentrainment from the last plate which
becomes of importance.
mo
m rr + n*i c
— •-
AT
t— —
AT
h- —
AT
h_
time
Figure 17 Simple model for ash penetration rate versus time.
3. The fractional rapping reentrainment, 5ft, can also be roughly related to
the average penetration, P, using the simple model shown in Figure 17.
On top of the continuous penetration rate, mc, rapping sets off an addi-
tional reentrainment rate, rnrr, which occurs for AT, at every At intervals.
One can define the average collection efficiency, fj (and the corresponding
average penetration, P), in terms of the "real" precipitation efficiency, r/
(and the corresponding penetration P = 1 — 77) and the fractional rapping
reentrainment, 3?:
rhc = rhin (1 - 77)
23-23
-------
min 77 K (A< + Ar)
mrr = — - AT -
mrr AT + rhc (At + AT)
m°ut = At + Ar
p = !^£ fj=l-P
TJ = 77 (1 - 3J) P = P + 3? +
As ft -> 0, P ->• P and r/ ->• 77. On the other hand, as 3? -> 1, P -> 1
and 77 — > 0. It is particularly interesting to see that when both P, 3ft 99.8%, i.e.,
P < 0.2%) makes the task of reducing the rapping reentrainment, 3ft, as
important as that of improving the precipitation efficiency (reduction of
P}-
4. A cleaner layer detachment occurred when the plate was subjected to lower
vibration frequencies (but having the same maximum acceleration), result-
ing in a smaller number of broken fragments (of correspondingly larger
sizes). Shear raps using a 1.5cm thick rubber pad (~ 13QHz fundamental
frequency) had rap efficiencies around 100%; the rap efficiencies dropped
to 20% when the rubber pad was taken out (~ 900^^). Normal raps with
and without the rubber pad did not show any appreciable difference in rap
efficiency. Both rapping modes produced more fractional reentrainment
when the rubber pad was used. Previous experimental studies of plate re-
sponse [2,3] show that dominant frequencies lie in the range of IkHz, which
we consider much too high to produce an efficient layer detachment. It was
shown that for shear raps the large impact duration (associated with us-
ing the rubber pads) has a significant beneficial effect on ash detachment.
This dependence of rap efficiency on the impact duration will be investi-
gated more thoroughly in the near future, when the length of the spring
beams will be varied (keeping the rubber pad on) as the means for varying
the fundamental frequency of the rapped plate.
5. For shear raps at high current densities, there is a reattachment phe-
nomenon: the layer breaks up into a small number of relatively large pieces
(sizes on the order of one half of the plate size). These fragments slide
down intact, but, for high currents, they slow down and reattach to the
plate. This process tends to produce lower rap efficiency as well as lower
reentrainment. This effect may be very important for the full-scale plates
which are much taller.
6. Two sources of reentrainment during rapping were distinguished. The
"true" reentrainment, that occurs when the layer detaches from the plate,
accounted for less than 50% (sometimes less than 25%, depending on the
current and flow conditions). The remaining reentrainment resulted from
the falling layer impacting the bottom of the collecting tray and redispers-
ing into the flow. We even observed some pieces being projected away from
the hopper tray, their momentum being provided by the momentum the
layer gained as it fell from the rapped plate. In practical precipitators,
23-24
-------
a similar phenomenon could occur when the falling layer reaches the up-
per parts of the collecting hoppers. One can expect the effect to be more
pronounced because of the larger plate heights involved.
While some of the conclusions of this work support earlier ideas and concepts, enough
evidence is provided here, that challenges the oftentimes held belief crediting normal raps
to be better than shear raps. Moreover, the new measurements regarding the effect of the
plate frequency on the rap efficiency is of importance due to its implications for the theory
of the electromechanics of ash layers.
Our results appear to indicate that the use of softly suspended plates, rapped by soft heavy
hammers producing dominantly low frequency bulk motion of the plates, is preferable to
rigidly supported plates, which are rapped by intense short impacts, producing a complex
superposition of high frequency vibrations.
The rather large degree of scatter seen in the experimental data can be explained in
terms of the nonuniform layer density and thickness due to the nonuniform current current
density. The current nonuniformity can be avoided by placing barbs on the wire electrodes
and forcing the corona tufts to remain at fixed points throughout a precipitation run. This
was investigated in the experiments by Self et. al. [9], when more or less elliptical patches
of ash layer with uniform thickness and density were precipitated. Better control over the
layer density and thickness can be achieved using thick (1.5cm diam.) wire electrodes with
barbs. Additional experiments using this new electrode configuration have been conducted
and the corresponding data show considerably less scatter. A full account of this work is in
preparation.
ACKNOWLEDGEMENT
This work was supported by the Electric Power Research Institute under Contract No.
RP533-1. The support and encouragement of Dr. R. Altman and Dr. R. Chang is gratefuly
acknowledged.
The work described in this paper was not funded by the U.S. Environmental Protection
Agency and therefore the contents do not necessarily reflect the views of the Agency and
no official endorsement should be inferred.
REFERENCES
1. Sproull, W.T. "Fundamentals of electrode rapping in industrial electrical precipita-
tors." J. Air Poll. Control Assoc., 15, 1965, pp. 50-55.
2. Billington, D.P., Hieber, G., and Rummel, S. "Full- scale, experimental study of rapped
electrostatic precipitator plates." Research Report Number 77-SM-2, Dept. of Civil
Engineering, Princeton University, 1977.
3. Eiiochson, L., and McKeever, B. "Preliminary experimental determination of pre-
23-25
-------
cipitator plate vibration modes by real time digital fourier analysis." EPRI Report
FP-1006, Vol. 3, Part 2, Research Project 1180-10, 1980.
4. Choi, D. H., Self, S. A., Mitchner, M., and Leach, R. "Experimental study of ash layer
detachment and reentrainment under normal and shear rapping of electrostatic precip-
itator collector plates." In Proceedings of the Eighth Symposium on the Transfer and
Utilization of Particulate Control Technology, Nashville, Tennessee, 1988.
5. Moslehi, G.B. "Electromechanics and electrical breakdown of particulate layers."
Ph.D. Thesis and HTGL Report T-236, Stanford University, 1983.
6. Juricic, D., and Herrmann, G. "Modeling and simulation of dust dislodgement on
collecting plates in electrostatic precipitators." In Proceedings of the Ninth Annual
Pittsburgh Conference on Modeling and Simulation, Pittsburgh, PA, 1978.
7. Choi, D. H. "Experimental study of the ash rapping of collector plates in electrostatic
precipitators." Ph.D. Thesis, Stanford University, 1990.
8. Tassicker, O. J. "Aspects offerees on charged particles in electrostatic precipitators."
Ph.D. Thesis for Wollogong University College, University of New Sourth Wales, July
1972.
9. Self, S.A., Mitchner, M., Fisher, M.J., Gere, D.S., and Leach, R.N. "Corona discharge
structure and its influence on precipitation and reentrainment." In Record of the
IEEE-IAS Conference, Philadelphia, PA, 1981, pp. 1128-1135.
10. Mitchner, M., Fisher, M.J., Gere, D.S., Leach, R.N., and Self, S.A. "Surface reentrain-
ment of collected fly ash in electrostatic precipitators." In Proceedings of the Third
Symposium on Transfer and Utilization of Particulate Control Technology, Orlando,
Florida, 1981.
11. Self, S.A., Choi, D.H., Mitchner, M., and Leach, R.N. "Experimental study of plate
rapping and ash reentrainment." In Proceedings of the Third International Confer-
ence on Electrostatic Precipitation, Abano-Padova, Italy, October 1987.
12. Ruckelshausen, K. "Uber die Beseitigung von Staubansatzer auf Technisch Glatten
Oberflachen durch Klopfen oder Vibrieren." Doctor's Thesis, Technischen Hochschule,
Stuttgart, No. DK 68511, Giese-Druck Kg. Offenbach/M., 1957.
13. Juricic, D., and Herrmann, G. "High speed photography of dust layer dislodgement."
EPRI Selected Workshop for Electrical Precipitator and Fabric Filter Manufacturers,
Denver, Colorado, 1978.
23-26
-------
MEETING EMISSION LEVELS THROUGH PRECIPITATOR UPGRADES
Sanford F. Weinmann
Lodge-Cottrell North American Operations
Dresser Industries, Inc.
601 Jefferson Street
Houston, Texas 77002
Kenneth R. Parker
Lodge-Cottrell U.K. Operations
Dresser Industries, Inc.
George Street Parade
Birmingham B31QQ
England
ABSTRACT
In the past 20 years, considerable experience has been gained, as a result of
installing precipitators worldwide, on plants firing a complete range of fuels from
lignites to anthracites. This experience has proved invaluable in determining the
design parameters of the precipitator required to meet a specific duty, particularly
where upgrading is necessary. This paper reviews various aspects of precipitator
design and how they impact on upgrade programs. Upgrades are typically required to
improve performance due to changes in emission specifications, fuel changes, plant
operating changes, and for equipment that has deteriorated from age or wear.
Precipitators are often designed to cope with fly ash from a wide variety of fuels
and boiler conditions. If the precipitator is designed to provide a specific emission
level for the worst of fuels, in terms of ash collection characteristics, then when
less demanding ashes are collected the process will behave as though it were over-
specified. With other fuels the ash characteristics may be such that the precipitator
behaves as if it were undersized.
It is often in these extremes that emission levels may be achieved with proper steps
to tailor the existing equipment to the problem. The scope of these reworks
generally involves both mechanical and electrical (controls) components of the ESP.
Details of typical changes are presented.
Optimization of the ESP process does not end with simple replacement of worn process
elements. Additional controls may be added to the existing process to fully modernize
a plant. Advanced control systems containing a supervisory control and data
acquisition (SCADA) element may be applied to units to better optimize performance.
Specifically, the SCADA system serves to match T/R (ESP) total power consumption to
real time ESP fly ash loading. Contemporary algorithms account for changing boiler
loads and best possible stack opacities on a dynamic basis. This system, known as
the Critical Point Control™ technique is also presented.
24-1
-------
MEETING EMISSION LEVELS THROUGH PRECIPITATOR UPGRADES
INTRODUCTION
In the past twenty years considerable experience has been gained, as a result of
installing precipitators world-wide, on plants firing a complete range of fuels from
lignites through to anthracites. This experience has proved invaluable in determining
the design parameters of the precipitator required to meet a specific duty,
particularly where upgrading is necessary.
Upgrades are typically required to improve performance due to:
• Changes in emission specifications.
• Fuel changes, for example lower sulfur coal.
• Plant operating changes, for example increased gas flow rate or
temperature.
• Equipment that has deteriorated from age or wear.
This paper reviews various aspects of precipitator design and how they impact on
upgrade programs formulated to address the above. Examples of both electrical and
mechanical upgrades are presented.
FACTORS AFFECTING PRECIPITATOR PERFORMANCE
Prior to reviewing aspects of an upgrade program, it is important to understand the
principle by which dust is collected, and the various factors affecting precipitator
operation and their impact on performance. These include:
Gas Distribution
Examination of the basic theory, proposed by Deutsch and later modified by Matts
Ohnfeldt, indicates it is important that the gas distribution is uniform across
collection areas because of the inverse logarithmic relationship between efficiency
and gas flow. Deviation from the ideal velocity profile results in a performance
fall, as shown in Figure 1 which relates efficiency to the coefficient of deviation,
and in Figure 2 which indicates the change in efficiency with variation in gas flow
rate.
In some older plants, gas distribution was achieved by means of back pressure from
the inclusion of a single perforated plate in the precipitator inlet. Distribution
testing on some of these units has shown this standard is poor. Hence it is now
common practice to use large models (for example, 1/8 full-scale) to determine the
position of gas distributors, splitters and deflectors to obtain an acceptable
standard in terms of coefficient of deviation. (The IGCI standard references a
coefficient of deviation of 15%.)
24-2
-------
Close attention is paid, during both model and site gas distribution testing, to
potential bypass of the field, either in the roof or hopper region, since any bypass
can limit the actual efficiency of collection.
Discharge and Collector Electrodes
There are many discharge electrode profiles in common usage. The main criterion is
that the radius of curvature of the emitter is small compared to the electrode
separation. This small radius is necessary to create the high electrical stress
required to produce ionization of the gas molecules. Generally, the ratio of
ions/electrons used in the charging of the dust particles is less than 10% of the
total corona produced. Where "high emission" electrodes are used, power may be
wasted, and if a difficult dust is encountered the additional corona promotes reverse
ionization more readily than a lower emission electrode.
To obtain a uniform corona field throughout the precipitator, it is essential that
the electrode alignment is of a high standard and remains such during all conditions
of operation. Otherwise the area of highest field strength will produce the highest
ionization, leaving other areas deficient, which may affect particle charging in spite
of the low overall corona utilization. Tests have shown that in general the optimum
spacing between alternate electrodes is equal to half the collector separation.
As with discharge electrodes, there are many forms of collectors. The main criterion
is that the profile is reasonably flat and hangs plumb. Some plates in present usage
are up to 16.5 ft wide x 48 ft deep, formed either from a fabricated sheet or a number
of cold rolled channels. It is important that the plate is free to expand in all
planes to avoid distortion and misalignment.
Collectors should have some form of inbuilt channels which are not only essential
stiffening members, but also act as anti re-entrainment baffles. These baffles create
a quiescent area in front of the collecting surface which prevents scouring and dust
loss at normal operation velocities.
Electrical Operation
The performance of the precipitator is dependent on maintaining the highest electrical
field strength and hence the highest voltage between the discharge and collecting
electrodes. From physics it can be shown that the force acting on a charged particle,
and hence particle migration, is proportional to the field strength or voltage
squared.
P Constant dependent on the dust properties
a Particle Radius
JJL Gas Viscosity
E Field Strength = V/d
S Particle Mobility
This demonstrates the importance of applying the highest possible voltage to the
electrodes. The principle of operating at the highest average discharge electrode
voltage to achieve the maximum efficiency has been proven from a large number of
plants. A typical curve presented in Figure 3 indicates that the minimum emission
(maximum efficiency) corresponds to the maximum mean discharge electrode voltage.
The figure also gives the corresponding relationships between discharge electrode
24-3
-------
voltage, secondary current and power input. It may be noted that both secondary
current and power increase rapidly once the peak electrode voltage has been reached.
This relationship forms an ideal control point on which an automatic voltage
controller can function.
Rapping
It is important that the precipitator remains in a reasonably clean condition in order
to maintain optimum electrical field strengths. To do this it is necessary to rap
periodically both the discharge electrode and collector systems to remove dust
deposits.
To maintain optimum power input in terms of current, it is particularly important
to keep the discharge electrodes free from build-up. Re-entrainment losses due to
rapping of the discharge electrodes are much less significant than from the
collectors, hence a higher frequency and intensity of rapping can be used with
advantage on the discharge system.
The collectors, which cope with the majority of the dust, need to be rapped once a
certain optimum thickness is achieved. A rapping blow is required to shear the
agglomerated dust particles off the collector for rapid transference to the hopper.
If the collector rapping is too frequent or of too high an intensity, the agglomerates
tend to explode from the collector plates to become re-entrained with the gas stream.
In essence all rapping is a compromise between having the collectors and electrodes
sufficiently clean, so as not to impact on the electrical operation and, on the other
hand, having the rapping not create a re-entrainment problem. Too frequent or intense
a blow can also lead to mechanical failures resulting in low plant availability.
Particle Size
Fraction efficiency curves have been determined for a large number of operating
precipitators. The results of these investigations indicate that the performance
of a precipitator remains fairly uniform, with a slight fall off for particles of
below 1 micron in diameter (Figure 4). For power plants, the percentage of particles
below 1 micron is usually too low to have a significant impact on the overall
performance, unless very low emissions are required. Due to natural selection, the
outlet fields are presented with dust having a predominance of fine particles.
Emission Requirements
It has been established that as the actual emission requirement is lowered, the size
of precipitator or contact time needs to be increased above that predicted from
accepted theory. This increase is brought about partly by the higher efficiency
requirement, by the slight but inevitable result of rapping re-entrainment, and by
the general reduction in particle size as the efficiency increases.
24-4
-------
Fuel and Ash Characteristics
The quantity of the ash in the fuel dictates the inlet loading to the precipitator
and hence the overall efficiency. In practice over a very wide range, the efficiency
of a precipitator is unaffected by dust loading. Although efficiency is maintained
as dust loading increases, it follows that the emission will also increase with dust
loading and hence could be outside the required limits. Therefore the maximum dust
load must always be considered when sizing any precipitator.
The moisture content of most fuels is typically around 8% to 12%. However, strip
mined fuels such as sub-bituminous or lignites can have moisture up to 35%, or even
60% in the case of brown coals. These high moisture levels raise the moisture in the
gas, resulting in higher operating voltage and reduced corona level (due to changes
in gas conductivity). This higher voltage may give rise to premature breakdown of
insulators, bus rings, or other close clearance points outside the actual field area.
It is well known that as the sulfur content in the fuel decreases then the amount
of self conditioning available falls. In practice this means that the resistivity
of the particles increases to a point where high resistivity impacts performance.
The effect of sulfur content in the fuel on plant size (contact time) is presented
in Figure 5 for a constant 99.5% efficiency.
As the sulfur content of a fuel falls, sodium in the ash becomes a more predominant
conditioning agent. It has been shown that for low sulfur fuels containing high
sodium in dust then high performance figures can be readily obtained. The impact of
sodium in ash on plant size (contact time) is indicated in Figure 5 for a constant
99.5% efficiency. The curve assumes the sulfur in coal is consistently low.
It will be appreciated that the above factors, many of which are empirical in nature,
have a significant impact on the sizing of any precipitator and must be understood
if the plant is to operate successfully at its design performance.
CONDITIONS DEMANDING UPGRADING/REPLACEMENT OF EQUIPMENT
In summary, cases which require upgrades to existing plants include;
• Precipitator too small, a consequence of
-- Changes in required emission levels.
-- Fuel changes for environmental or economic reasons.
-- Station operating changes (process changes).
t Precipitator unserviceable, a consequence of corrosion, worn
components, actual mechanical damage, or other shortcomings.
For changes in fuel supply, it is necessary to fully evaluate the coal and fly ash
composition, together with their impact on plant performance. Factors included in
this analysis are: ash content which determines the inlet loading to the plant,
moisture content, and sulfur levels which can determine the precipitability of the
dust. In the case of the ash composition, calcium and sodium levels modify the ash
resistivity and must be considered. For a precise assessment the ash analysis ideally
should be that presented to the precipitator and not the basic coal ash, especially
24-5
-------
for fuels which have high slagging characteristics. These slags modify the alkali
components of the fly ash, since slags are formed from alkali-rich, low melting point
eutectics.
For many plants, changes in fuel supply will not necessarily result in a change in
air preheater outlet conditions. However, a change to a fuel having a high moisture
content would give rise to additional gas flow and a higher back end temperature,
which would affect precipitation. Also, leaking access doors, failed expansion joints
and weld faults will result in excess volumes which could upset performance (see
Figure 2).
From knowledge of the efficiency, gas flow, gas temperature, and ash analysis, it is
possible to determine the design parameters required to meet a specific duty. If the
existing precipitator is to be considered for inclusion in the final upgrade
configuration, either in refurbished form, or the casing alone with revised
internals, then design factors must include such additional considerations as the
impact of gas velocity, dust re-entrainment, and retained internals. This assessment
will enable a recommendation to be made whether to utilize the existing equipment
or to go with a replacement unit.
If the existing precipitator is to be included with additional collecting fields, the
velocity through the existing casing will suggest if series or parallel fields are
best suited (if the upgrade velocity is in excess of 8 fps, a parallel unit is
recommended), but in most instances series fields invariably prove more economical.
The flue layout and space available will determine if a tail-end or front-end field
is possible. The tail-end unit typically has the advantage that it can be erected
without a boiler outage, except for the duct tie in.
Because of the required boiler outage for adding a front-end field, it is sometimes
more economical to replace the existing plant by a new unit, particularly if the
available space enables the plant to be erected along side the older one. An
illustration of this approach will be presented later.
MECHANICAL DESIGN
Reference has been made to the importance of obtaining a satisfactory standard of gas
distribution and its impact on performance. Hence, for any upgrade it is considered
important either to carry out a model test or at least establish that the distribution
is satisfactory within the existing plant from a site test.
While many precipitator suppliers claim advantages for specific component designs,
the main need is for the discharge electrodes to produce sufficient ions to charge
the particles and for the collectors to retain them. There must be a full complement
of discharge electrodes, and the collectors must be sound and correctly fixed. For
optimum efficiency, the profile of the collectors needs to be flat to ensure a uniform
electric field is obtained. For this it is essential for the discharge electrodes
to be in good alignment and have the ability to remain in alignment too all operating
conditions.
24-6
-------
The major drawback of existing plants having narrow spacings is that electrode
alignment is more critical in terms of maintaining a uniform field strength. If one
assumes a fabrication or erection tolerance of 1/2 inch, this represents a 12.5%
deviation on 8 inch collector centers, but only 6.25% on the wider 16 inch spacings.
This deviation is much more significant for the narrower spacings, since the
performance of the precipitator is equated to the field strength squared.
If the decision on upgrade is to use a directly coupled extension field, then
collector spacings have to be considered in light of possible maldistribution of
gases. Most collectors have some form of stiffener of measurable width which will
not line up with the adjacent field plates unless the spacings are the same.
To maintain the design emission characteristics it is important the discharge
electrodes operate in a "relatively" dust-free state. This is achieved by means of
adequate rapping. Hence this aspect of any design needs careful assessment to be
fully operative and to give long-term performance availability.
In the case of the collector, dust deposits can modify the electrical operating
conditions and it is important that the deposits are removed by a fully operative
rapping system without giving rise to re-entrainment. The secret of minimizing re-
entrainment is to allow the dust to build up to create a certain thickness which, when
rapped, shears off the collector as an agglomerate. As some re-entrainment is
inevitable, stage rapping where individual collectors are rapped in sequence, rather
than block rapping, is preferred where operating opacity peaks are to be minimized.
There is an argument in favor of block rapping, as it enables the optimum electrical
conditions to be re-established periodically. Extensive tests, however, have shown
that stage rapping, when correctly set up, maintains the performance while minimizing
opacity peaks.
ELECTRICAL DESIGN
First Generation Power Systems
Drastic changes have taken place over the past twenty years in the design and
operation of control systems applied to ESPs. Typically, older systems used a
combination of analog controls, saturable core reactors, and double half wave T/R
sets to control ESP power according to process conditions. These elements although
appearing effective have serious limitations with regard to precipitator performance,
especially so when the precipitator performance becomes marginal.
The saturable reactor relies upon inductive reactance in order to control T/R input
power. A primary winding is supplied with a variable d.c. current which in turn
creates a flux field, through which the high voltage T/R primary is coupled. The
voltage controller typically supplied the saturation current to the saturable reactor
in order to control the input to the T/R. This means of control is lacking in two
very critical areas of precipitator control.
First the saturable core reactor tends to respond slowly to changes in d.c. current
settings. The creation and collapse of the magnetic field in the saturable core takes
a relatively long period of time when compared with the 8.333 mS in a half line cycle.
Several sparks may occur before impedance levels could be increased sufficiently to
24-7
-------
quell the undesirable element of sparking on a bus section. Once established, an
arc could persist for a long period of time, which could cause damage to precipitator
internal elements, worsening the collection characteristics of the precipitator.
The second problem with the saturable core reactor is that it is virtually impossible
to shut the T/R input current down entirely. Even without a saturation current, a
primary T/R voltage exceeding 200 volts A.C. may be observed. Although the current
associated with this voltage is minimal, the T/R input may be significant enough to
cause sparking or arcing in the precipitator. Worse, the applied voltage which has
led to the electrical breakdown may not be controllable due to the limitations of the
present control system. Inevitably the uncontrolled sparking would lead to damage
within the bus section further worsening precipitator performance.
Analog Control Equipment
The use of analog control equipment was prevalent prior to 1980. Typically these
controllers used analog circuitry designed to condition T/R feedback signals as well
as provide output. Operational adjustments required such as spark rate, sensitivity,
and ramp rates were made via potentiometers. Over time these adjustments drifted.
The problem of drifting is accentuated when interactive adjustments drift in opposite
directions, magnifying error. As the equipment ages the drifting becomes worse in
terms of both magnitude and time interval (time to reach a point of maladjustment is
shortened) resulting in equipment which can no longer be effectively adjusted.
Another problem often associated with the analog controller is the way in which it
is coupled to the transformer rectifier bridge. Typically this vintage of control
equipment relied either upon a capacitively coupled signal from the bridge return or
high voltage divider to base controller actions. Although the adaptation of the
electronics to this application served its purpose, the capacitively coupled technique
does not provide a true vantage of T/R operation. The main concern of most
controllers of this era was spark rate.
Upgrading Power System and Controls
A linear reactor (fixed choke used in series with the T/R primary) and silicon
controlled rectifier (SCR) negate the problems described above. The SCR is a solid
state switch which permits control of T/R input over a wide range. If a bus section
is damaged, and sparking occurs at a low threshold, the SCR may be utilized to control
input power and minimize the ill effects thereof. Further, the SCR is a fast response
device which may be controlled on a half line cycle basis, hence arcing or sparking
may be extinguished at the fastest possible speed. These attributes help to maximize
the performance of the precipitator.
Computer based controllers have the facility to be directly coupled to the T/R. The
use of RMS converters along with high resolution analog to digital converters permit
the newer AVCs to sample accurately the average voltage of the bus section. Hence
modern controllers can accurately maximize voltage input, thereby maximizing
performance for a given process condition (Figure 3). Parameters used for control
can be stored in a non-volatile memory and not subject to change as with the analog
predecessor. Further, parameters can be entered using a key pad so that the accuracy
of these adjustments is not left to less precise circuitry.
24-8
-------
By coupling the elements described above it is possible to quicken controller response
such that arcs within the precipitator are extinguished on a timely basis, hence
upgrading precipitator performance. Further, modern control techniques, such as
intermittent energization, may be used in the presence of difficult dusts to enhance
precipitator performance.
Optimization of Power, Considering Best Available Technology Rules
In addition to the accuracy and speed of available replacement equipment, the T/R
control system may be coupled together by the addition of a supervisory control and
data acquisition (SCADA) computer or Minivisor™. Second generation personal based
computers are currently being utilized, replacing expensive "mini" class computers.
These devices often incorporate energy management systems. Environmental regulatory
bodies have recently raised objections to systems which target set-point (as opposed
to optimum) emission levels. As a result Lodge-Cottrell has applied the Critical
Point Control™ (CPC) algorithm.
CPC curtails only the excess power used by the precipitator (Figure 6). By
continuously monitoring opacity levels this routine assures that T/R power levels are
reduced only to the point of a defined opacity rise (as low as 0.1%). A set-point
is not used, and the system accounts for boiler load swings and possible changes to
fuel blends by searching for the optimum dust filtration levels based upon these
factors (Figure 7).
FACTORS NEEDED TO FULLY ASSESS AN UPGRADE SITUATION
Once established that an upgrade is required, the economics of accomplishing the
program must be studied. To fully appreciate this statement, one must consider the
economic value of plant production time and available space. Below is a list of
factors studied by technical personnel to provide the optimum solution both in terms
of economic and functional benefit;
Existing Precipitators
Number of Fields
Number of Ducts
Electrode Spacing
Rapping Systems
Insulator Design
Process details
Type of Fuel
Required Efficiencies
Gas Temperature
Collector Spacing
Total Plate Areas
No. and Size of T/Rs
Gas Distribution
Hopper Size & Design
Analysis of Fuel & Ash
Required Opacity
Plant Altitude
Collector Size and Form
Size of Discharge System
Number of Bus Sections
Duct Work
Existing Controls
Inlet Dust Loading
Gas Flow Rate
Available Space
PRACTICAL RESULTS OBTAINED FROM UPGRADE PROGRAMS
Plant A - Application of the Critical Point Control™
This plant is an 850 MW coal fired unit designed to handle a wide range of coals,
including low sulfur. Presently a coal with sulfur content in excess of 1.1% is used
24-9
-------
and as a result dust is easily precipitated. The net result is that precipitator
power consumption is much greater then need be. An existing SCADA computer was fitted
with the Critical Point Control™.
When the CPC is inactive, power levels in the precipitator hover between 2.5 and 3.1
MW. Once the routine is activated, power levels are reduced, first using Pulse
Modulation (PM), and then by reducing the T/R secondary voltage by the use of AVC
parametric settings.
After each power reduction step to a group of AVCs, the routine continues to monitor
the opacity signal. If the opacity is seen to rise by more than a predefined amount,
the logic determines that a critical point has been located. If a critical point is
found, the program reverses direction and increases power to the AVCs (in reverse
order) for one step, and holds that power level.
Once the opacity level stabilizes, the supervisor will hold power level constant
until :
• The routine is shut off. When this occurs the CPC automatically
restores 100% power levels to all AVCs.
• The opacity is seen to increase by the critical amount. When this
event occurs, the CPC will increase power levels to the AVCs in
reverse group order until the opacity is seen to fall to its
original base level.
• The opacity is seen to decrease. When this event occurs the CPC
will again begin reducing power and searching for a new critical
point.
Test Conditions. Due to the increased demand for electricity, the unit was operated
at output levels of 850 MW, approximately full load. Under these conditions the
precipitator is apt to consume the lowest amount of power due to the heavier dust
loadings. It has been well documented on other sites that a significant gain in power
savings is experienced when the boiler output is at a partial load. Hence the power
savings achieved during the test period were probably not optimal. A larger power
savings could be experienced as the boiler is cycled or boiler output falls to a lower
megawatt loading.
Results. On average the Critical Point Control™ system yielded an approximate power
savings of nearly 2,000 kW, with negligible impact on opacity. The bar chart
presented in Figure 8 shows the system being cycled on and off. Each bar element
represents a six minute average of power consumption as the system is being cycled
on and off over a short period of operation. The data presented is typical of
operation under base load conditions.
Assuming that:
Cost of fuel is $ 0.021 / kW-hr.
Plant will operate 328 days per year.
Power savings demonstrated during the tests can be maintained.
24-10
-------
... the system will provide a savings of $330,000 per year. As a result of the
addition of this control equipment, precipitator power consumption is optimized and
economic benefit is realized.
Plant B - Use of Wide Plate Spacing on Rebuild (1 boiler 660 MW)
In this case the existing precipitator met all emission requirements and fuel supply
had not varied. Due to corrosion, age, and other reasons it was decided to completely
rebuild the C and D flows of this four-flow precipitator. While it would have been
feasible to rebuild the two flows using the original design, it was possible, due to
the industry acceptance of wide plate spacing, to provide more economical equipment.
As such two flows were constructed using 16" centers, in place of the original 10"
centers. This new design was far more cost effective to construct both from the
standpoint of component count and installation time. The work was completed in the
space of a limited outage period.
During the outage flows A and B were overhauled and brought back to a satisfactory
condition with 10" centers. Following the corrective work tests from all flows
enabled a direct comparison of the equipment with different centers but like contact
times. Results are detailed below and confirm the wide spacing approach as not only
being more economical, but also producing improved performance level.
Design Flow A Flow B Flow C Flow D
Collector Spacing in. 10 10 16 16
Gas Flow Rate Am3/S 220.4 228.4 228.1 178.0 202.7
Gas Temperature °C 130.0 116.0 115.0 111.0 109.0
Inlet Loading g/NM3 19.61 15.00 15.00 18.37 12.21
Emission g/NM3 0.117 0.137* 0.026 0.013 0.009
Efficiency % 99.40 99.09 99.83 99.93 99.93
Sulfur in coal % 00.47 00.61 00.61 00.61 00.61
Ash in coal % 18.20 13.10 13.10 13.10 13.10
* Internal field fault with flow.
Plant C : Fuel and Emission Specification Changes - (3 boilers of 120 MW each)
This plant features a mechanical collector with a three field ESP unit originally
designed for a 98.5% efficiency with a 1.8% sulfur coal having a 15% ash content.
The fuel supplied was changed for both economic and sulfur content reasons to a coal
with 0.6% sulfur and an 18% ash content. Further, modification to legislation
required the plant to meet a 99.3% efficiency. In order to meet required emission
levels with the new fuel an upgrade program was enacted.
The lower sulfur content and the increased efficiency requirement demanded an
increased contact time of 60%. The existing precipitator was determined to be too
small. Since boiler output could be sold, and the cost of adding on to the existing
unit would be too costly in terms of time, a parallel unit was determined to be the
most economic solution. Plant layout permitted the construction of one new
precipitator (for one of the three boilers); space was not available for other
construction. Closer examination of the existing ESPs revealed that their condition
was too poor (due to age) to be reused.
24-11
-------
To comply with the requirement of maximum generation availability, the construction
project was arranged such that ductwork connections were made during short boiler
outages (typically during boiler inspections). The first of the three units was built
along side the original No. 3 precipitator. The original unit was demolished
following commissioning of the new unit. This made room for the second (new) unit
for boiler No. 2. After the second unit was connected the original No. 2 unit was
demolished leaving enough room for the last of the three new precipitators. Although
the placement of the new units did not coincide with boiler center lines, careful duct
design and the removal of the mechanical collectors enabled the original ID fans to
be reused, lending further economy to the project.
The results of the new precipitators are shown on the table below. The new units
utilized a configuration including 4 series fields with 15 x 30 foot collectors.
Maximum economy was achieved in terms of meeting emission levels thru a well planned
and executed upgrade program.
Design Unit 1 Unit 2 Unit 3
Gas Flow Rate Am3/S 203.0 227.0 196.7 202.0
Gas Temperature °C 138.0 163.0 138.0 151.0
Inlet Loading g/NM3 14.30 15.82 14.71 15.09
Emission g/NM3 0.10 0.034 0.027 0.048
Efficiency % 99.30 99.78 99.82 99.86
Sulfur in coal % 0.60 0.63 0.55 0.60
Ash in coal % 18.0 19.7 16.4 17.9
CONCLUSION
This paper describes an approach which can be followed to develop and execute a
successful upgrade/retrofit program. Goals of these programs are simple and straight
forward; meet the specified emission levels at the optimum economic cost. The
presentation indicates that in many instances an upgrade or retrofit can enable
existing equipment to meet stated emission levels as fuel or other factors change.
ACKNOWLEDGMENTS
The authors wish to thank our colleagues for assistance in preparation of this paper,
and the management of Lodge-Cottrell, Dresser Industries for their encouragement.
24-12
-------
^ 99.5 •
O
s
g 990.
LL
UJ
f. 98.5-
a.
uj
il
"~~~-~.^^
PRACTICAL DESIGN "^^
IDEAL ^^
"\
IOCI STANDARD
POOR
98.01 1 1 1 1 1 1 ( 1 (_
0 5 10 15 20 25 30 35 40 45
COEFFICIENT OF DEVIATION, %
FIGURE 1 . EFFECT OF GAS DISTRIBUTION ON PRECIPITATOR EFFICIENCY
100'
f
£ 99.5
UJ
a 99
p
o
UJ
_l
_l
O
~ —— OEUTSCH RELATIONSHIP
— _^~^^^^ EFFICIENCY = 1-e -54
WAITS OHNFELDT RELATIONSHIP ^v
EFFICIENCY = 1-e -(54)', ^^--^
N. ^^^^
\.
\
° 50 60 70 80 90 100 110 120
GAS FLOW,%
FIGURE 2. EFFECT OF CHANGES IN GAS FLOW
ON PRECIPITATOR EFFICIENCY
20
30 40 50
INPUT POWER, kVA
FIGURE 3. TYPICAL PRECIPITATOR OPERATING CHARACTERISTICS
20-
s,a
O 18 •
§ 16.
UJ
i 14'
i 12.
O
X
UJ 10-
i ,
I8'
^*'~
//
.
f
/
••"" \ /
I- \ /
\ /
\ ^S
.01 .1 10 10 100 1000
PARTICLE SIZE. MICRONS
FIGURE 4. TYPICAL PARTICLE SIZE PERFORMANCE RELATIONSHIP,
DUST NOT DIFFICULT TO PRECIPITATE
24-13
-------
05 10 15
SODIUM OXIDE IN ASH OR SULFUR IN FUEL, %
RGURE 5. EFFECT OF SODIUM OXIDE IN ASH AND SULFUR
IN FUEL ON PRECIPITATOR CONTACT TIME.
loo-
s'. 80-
Q.
Z
c
UJ
Q.
£ 40'
o
tr
Q.
§
i
\
\
\
\
\
\ CRITICAL POINT
VC
5 10 15 20 25
STACK OPACITY , %
FIGURE 6. PRECIPITATOH POWER INPUT VERSUS OPACITY
5 10 15 20 25
TIME HOURS
FIGURE 7 CRITICAL POINT CONTROL OPERATION.
2B-,
2 6- i
. 24-
1 22-
p 1 8-
| 16.
W 1 4-j
i
||
• 1
11
11
EMS CONTF1OL STATUS (* ON - OFF)
FIGURE 8. ENERGY MANAGEMENT SYSTEM TEST RESULTS
24-14
-------
OPERATING EXPERIENCE OF THE RIGID FRAME ELECTROSTATIC PRECIPITATORS
INSTALLED AT METROPOLITAN EDISON COMPANY'S PORTLAND STATION
Paul G. Abbott
Theodore C. Schafebook
GE Environmental Systems
Lebanon, Pennsylvania
John A. Brummer
Metropolitan Edison Company
Reading, Pennsylvania
ABSTRACT
The retrofit installation of rigid frame electrostatic precipitators (ESP) with
computer based control systems began in 1986 at Metropolitan Edison Company's
Portland Generating Station. The ESP for Unit No. 2 began operation in May
1987 while the ESP for Unit No. 1 began operation in January 1989. These units
have operated successfully since startup with particulate emissions and stack
opacity well below required limits.
This paper discusses the operating experience of these precipitator systems
with specific performance test results for various sulfur content coals being
fired. Also included is an update on the operation of the computer based
control system with the upgrade installation of Digital Power Saving
Energization(DPSE), commonly referred to in the Industry as intermittent
energization, and implementation of a specifically programmed automated control
strategy for energy management utilizing the DPSE control.
25-1
-------
INTRODUCTION
Metropolitan Edison Company's Portland Generating Station is located along the
Delaware River in Northampton County, south of the town of Portland,
Pennsylvania.
The station has two (2) pulverized coal fired boilers. Unit No. 1 has a
nominal gross generating capacity of 172 MW and began initial operation in
1958. Unit No. 2 has a nominal gross generating capacity of 252 MW and began
initial operation in 1962.
With construction commencing in 1986, GE Environmental Systems installed a new
rigid frame electrostatic precipitator system for each generating unit,
replacing the original smaller collectors consisting of combination cyclones
and electrostatic precipitators. New precipitator replacements were
necessitated due to the original equipment not being able to maintain
particulate emissions within present Regulatory requirements. The boilers were
frequently required to operate at reduced load in order to maintain a stack
opacity below 20%. The new ESPs for Unit No. 2 and Unit No. 1 were placed into
service in May 1987 and January 1989 respectively.
Regulatory requirements as well as contractual performance guarantees for each
unit are a maximum particulate emission of 0.10 Ib. per million Btu of boiler
heat input and a maximum stack opacity of 20%.
EQUIPMENT/SITE DESCRIPTION
Electrostatic Precipitator
The GE rigid frame electrostatic precipitator for each boiler is a two-chamber
unit with each chamber containing six bus sections in depth. The ESP on Unit
No. 1 has 32 12 inch wide gas passages per chamber while Unit No. 2 has 47
12 inch wide gas passages per chamber. Each ESP has six electrical fields
powered by six double outlet bushing transformer-rectifiers. The units are
designed to have five of the fields in operation to meet performance guarantee
requirements, allowing one redundant field.
The discharge electrodes are 3/16 inch twisted square bars, rigidly mounted in
tubular mast assemblies, supported by four point suspension through cylindrical
alumina insulators located on the precipitator roof. The discharge electrode
design has demonstrated superior voltage-current characteristics with optimized
current distribution and high reliability.
25-2
-------
The collecting electrode assemblies are integrated 18 inch wide roll formed
panels with integral edge stiffeners designed to provide uniform rapping energy
distribution for controlled, optimized plate cleaning. Test results have
demonstrated excellent distribution of acceleration response with minimum
values, measured at the bottom of the most distant plate, exceeding 60% of the
maximum acceleration measurements at the top of the collecting plate directly
under the rapper impact point. This low degree of attenuation eliminates the
requirement for excessive input forces at the impact point to obtain overall
effective plate cleaning.
The electrode cleaning system utilizes roof mounted electromagnetic impact
rappers and on-line control adjustment flexibility of individual rapper
intensity and operating frequency.
Control System
Each electrostatic precipitator has a GE Intelligent Precipitator (IP) computer
based control system. The control hardware local to the ESP is microprocessor
based transformer-rectifier controls (Digital AVC), a single board
computer-based rapping control system (Micro-Tapper) and several multiplexers
for data collection and system control. These devices are continuously
monitored by a microcomputer Host Workstation located in the main boiler
control room. This Host Workstation allows the operator pushbutton control
with a choice of manual or automatic mode of operation with various
pre-programmed control selections. The Workstation also has multiple tools for
trend analysis and troubleshooting such as automatic report generation and
graphing of data. The IP control system has been described in detail in a
previous paper. (1)
Scope of Supply
The scope of supply for each generating unit, in addition to the flange-flange
electrostatic precipitator, consisted of all ductwork from the air heater
outlet to the I.D. fan inlet and from the I.D. fan outlet to the stack
breeching, isolation dampers, all structural support steel, ESP roof weather
enclosure, hopper area enclosure, access facilities, insulation and lagging,
electrical controls and components, control room building, computer based
remote control system, demolition of existing cyclones-precipitator, associated
ductwork and structural steel, and complete erection of all new components.
All of the new equipment, with the exception of portions of the ductwork for
boiler and I.D. fan tie-in, were constructed while the boilers were on-line. A
twelve week boiler outage was separately scheduled for each boiler for
demolition and removal of existing equipment and to complete the new ductwork
tie-in to the boiler and I.D. fans and from the I.D. fans to the stack. The
boiler outage also allocated time for work on the I.D. fan housings and the
boiler.
25-3
-------
Site Layout
The precipitator layout is illustrated in Figure 1. Site constraints required
that the new precipitators be located on the North side of the plant with
extensive ductwork runs from the boiler to the precipitator and from the
precipitator to the ID fans. The precipitators are located side by side with
common center access, common ESP roof enclosure and common hopper enclosure.
FIELD TESTING
Test Series
The electrostatic precipitators were designed to operate at a gas flow of
584,000 ACFM at 266°F for Unit No. 1 and 868,000 ACFM at 27TF for Unit No. 2
and to provide a minimum collection efficiency of 99.2% in order to meet the
particulate emission limit of 0.10 Ibs/MMBtu and the opacity limit of 20%.
Source sampling test programs were conducted in order to evaluate precipitator
performance and compliance with the contractual guarantees. The tests were
conducted by an independent testing firm.
The Portland No. 2 ESP performance testing was conducted in four phases during
a time period between July 1987 to September 1987. Phase I, III and IV
evaluated ESP performance while firing three different fuels designated as
medium sulfur coal, low sulfur coal and medium sulfur-high ash coal. Phase II
was designated as State compliance tests.
The Portland No. 1 ESP performance testing was conducted in two phases during
April 1989. Phase I evaluated ESP performance while firing medium sulfur coal.
Phase II was designated as State compliance tests.
Fuel
The Portland Station is designed to fire Eastern bituminous coal with a sulfur
content range of 0.95% to 2.00%, a maximum ash content of 12.11%, a maximum
moisture content of 9.00% and a minimum heating value of 12,500 Btu/lb.
The average ultimate coal analyses for Phase I, II and IV performance tests on
Unit No. 2 are shown in Table 1. The fuels fired for the various test phases
demonstrated ESP performance for the complete design range of coal sulfur, with
average sulfur contents of 2.19%, 1.01% and 1.88%.
The average ultimate coal analysis for Phase I performance tests on Unit No. 1
is shown in Table 2. The fuel fired during this test series was a medium
sulfur coal with an average sulfur content of 2.00%.
25-4
-------
Test Results
Table 3 summarizes the performance test results of the three test phases for
Unit No. 2 with the redundant electrical field de-energized. Average
particulate mass collection efficiency and emissions were measured to be 99.62
percent (0.027 Ibs/MMBTU), 99.40 percent (0.052 Ibs/ MMBTU) and 99.71 percent
(0.031 Ibs/MMBTU) for Test Phase I, III and IV respectively.
The average measured stack opacities for Test Phase I, III and IV were 9%, 6%
and 9%. The Stack plume was observed to be slightly visible under Phase I and
IV test conditions and somewhat more opaque under Phase III test conditions.
The measured stack opacities of 9% and 6% for Phases III and IV tests are in
good agreement with opacities computed from the outlet dust loading and
particle size distribution. The measured stack opacity of 9% for Phase I tests
was greater than the computed opacity by a factor of almost 2.0. The large
error between measured and computed opacity was attributed to instrument error
and it was estimated that the actual opacity was approximately 5%.
Table 4 summarizes the performance test results for Unit No. 1 with the
redundant electrical field de-energized. Average particulate mass collection
efficiency and emissions were measured to be 99.56 percent (0.020 Ibs/MMBTU)
for test Phase I. The coal being fired for this test Phase I was similar to
that being fired during test Phase I for Unit No. 2 and resulted in very
similar precipitator performance based upon the calculated migration velocities
for each unit. For this reason it was determined that it was not necessary to
evaluate the Unit No. 1 ESP performance while firing the other two coals (low
sulfur and high ash medium sulfur) previously fired in the No. 2 unit, as
similar performance results were expected for Unit No. 1 as that measured for
Unit No. 2.
The average measured stack opacity for Test Phase I was 4%. The stack plume
was observed to be slightly visible under Phase I test conditions and the
measured stack opacity of 4% was in good agreement with opacity computed from
the outlet dust loading and particle size distribution.
The performance test results demonstrate that both precipitators are achieving
design performance with a conservative margin. This high level of performance
at a moderate SCA of less than 250 Ft /1000 ACFM is attributable to features
associated with equipment design as well as optimum operating conditions.
Typical chemical composition of the flyash samples obtained during the
performance tests for Unit No. 2 Phase I (medium sulfur) and Phase III (low
sulfur) are tabulated in Table 5. The flyash is characteristic of Eastern coal
with low calcium, moderately low sodium, and iron consistent with coal sulfur
content. The typical resistivity of the flyash resulting from the medium
sulfur coal was determined to be approximately 2x10 ohm'cm while that from
the low sulfur coal to be approximately 1x10 ohm'cm at operating
25-5
-------
temperatures of 268°F to 276°F. This range of resistivity is normally very
suitable for precipitator operation. The flyash composition and resistivity
for Unit No. 1 Phase I test samples were similar to that of Unit No. 2 Phase I
test samples.
The T-R operating electrical characteristics were excellent providing optimum
current densities with high applied voltage as illustrated by the typical
voltage current curves shown in Figure 2 while firing medium sulfur coal.
Average current density and applied voltage was 21 nA/sq. cm and 55 KV for the
medium sulfur coal tests and 12 nA/sq. cm and 55KV for the low sulfur coal
tests. The excellent voltage-current characteristics are not only a result of
the ideal range of flyash resistivity but are also indicative of the twisted
bar electrode rigid mast assembly which was designed to provide high voltage at
low current for maximum corona power utilization.
The contract required a second series of performance tests to be conducted
within 12 months of successful completion of the initial tests. The second
series of tests for each unit were waived on the basis of the excellent
performance results achieved in the first test series and continued good
performance thereafter as continuously documented by stack opacity monitors.
ESP OPERATION
After completion of minor adjustments during initial operation, both
precipitators have operated extremely well without any major equipment failure
or unscheduled downtime. The stack opacity of both units has consistently been
recorded in the range of 5% to 7% while operating at reduced T-R power input.
During a scheduled boiler outage in April 1989, a thorough inspection of Unit
No. 2 precipitator was conducted. The only item that required repair and
replacement were four rapper isolation insulators. This maintenance
requirement was attributed to improper tightening of a bolted connection during
construction. The only other component malfunction previously recorded was two
hopper throat heaters, which also were replaced during this outage.
During the initial startup of Unit No. 1, the boiler experienced feedwater
pump problems and could only achieve half load operation over a period of
approximately four weeks. During this period, the flue gas temperature to the
precipitator was in the range of 180°F which was below the acid dew point
temperature. The precipitator operated well during this period with stack
opacity at or below 2%. After the boiler achieved full load operation, the
precipitator continued to operate well below design levels without experiencing
any detrimental effects from the sustained sub-dewpoint operation. There have
been no reported equipment failures for Unit No. 1 precipitator.
25-6
-------
The precipitators have operated and continue to operate without equipment
failures, with the exception of those previously noted, demonstrating high
component reliability and providing 100% availability.
ESP CONTROL UPGRADE
GE Digital Power Savings Energization (DPSE) controls, also known in the
industry as intermittent energization, were installed in the T-R control
cabinets to compliment the Intelligent Precipitator Energy Management
capabilities.
The DPSE is a self-contained control module (7|" x 4 3/4") designed to be
mounted inside each T-R control cabinet or externally on the control cabinet
door. DPSE can also be incorporated with a door-mounted Terminal Display as an
option to the Digital AVC.
Each control module is furnished with a preconfigured plug-in cable to link the
DPSE with the Digital AVC T-R control. A 16 character alpha numeric liquid
crystal display indicates T-R electrical meter readings, spark rate, slow ramp,
secondary current setpoint and alarms.
The DPSE control has a wide range of adjustability by means of the switches
mounted on the module. It provides selection of from 1 to 16 ON half cycles
and from 2 to 32 OFF half cycles. A control mode switch allows selection of
local DPSE, remote DPSE and non-DPSE operation. In the local DPSE mode, the
Digital AVC gates the silicon-controlled rectifiers (SCRs) as adjusted by the
ON/OFF half cycle switches. The remote mode allows DPSE adjustment from the
remote Intelligent Precipitator (IP) Host Workstation located in the main
boiler control room.
The IP control system was originally furnished with six higher level algorithm
control strategies consisting of Energy Management, Automatic Startup, Power
Down Rapping, Rapper Optimization, Boiler Trip and Hopper Overfill. The
oroginal Energy Management control mode automatically controled T-R operating
levels utilizing SCR phase control (Set Point Control or SPC) to maintain the
measured stack opacity within an operator selected window of high and low
op.acity limits.
A new Energy Management control algorithm was developed to utilize DPSE as well
as SPC for T-R power level control consistent with stack opacity limits and
also incorporates back corona detection and adjustment to provide optimum
operation dependent upon specific operating conditions.
25-7
-------
The Energy Management control has four modes of operation consisting of Set
Point Control (SPC), Set Point Control with Back Corona Maintenance (SPCBC),
Digital power Savings Energization Control (DPSEC) and Digital Power Savings
Energization Control with Back Corona Maintenance (DPSECBC).
When any one of the control modes is first selected, V-I curves are performed
for all T-Rs in service. The curve information is used to calculate the slope
of the V-I curve, and, along with the respective collecting plate area and the
current operating point, is used to calculate an energy-saving parameter for
each T-R. The energy-saving parameters are used to select the optimum T-R for
increased or decreased power level adjustment to maintain stack opacity within
the selected high/low opacity window. The V-I curve information is also used
to flag any T-R's, if any, that are operating in a back corona condition. When
a flag is set for back corona, the level of secondary current associated with
initiation of back corona for that particular T-R is recorded for use with the
Back Corona Maintenance operating modes SPCBC and DPSECBC. Curves are retaken
every 24 hours or whenever a significant shift of the actual T-R operating
point from the expected operating point is detected.
After the V-I curves are completed for all of the T-Rs, a Critical Point
Adjustment routine is initiated. The initial measured opacity or reference
opacity is recorded. The T-R power input is then reduced until the opacity
increases to 2% above the reference opacity value (Critical Point) or the high
opacity limit for the opacity window is exceeded. When either one of these two
conditions occurs, the Critical Point search is complete and the normal
high/low opacity window control will resume using the selected operating mode.
The SPC operating mode will control the T-R power levels, as determined by the
measured opacity value, by varying the SCR conduction angle or the operating
setpoint of the T-Rs.
The DPSEC operating mode will control the T-R power levels, as determined by
the measured opacity value, by selectively varying the number of OFF half
cycles.
In either of these two operating modes, no back corona maintenance will take
place with all T-Rs eligible for power adjustment.
The SPCBC and DPSECBC control modes both incorporate back corona maintenance in
addition to the previously described control functions. The initial operation
of either of these control modes is to adjust those T-Rs flagged for back
corona to just below the secondary current levels previously recorded as the
initiation of back corona which is accomplished by utilizing the DPSE control
to vary the number of OFF half cycles.
25-8
-------
After back corona maintenance is complete, the routine will resume with normal
opacity feedback control. However, T-R's in back corona maintenance are not
eligible for power adjustment, but will be maintained at their optimal power
level or that level just below initiation of back corona. Operating
characteristics of the T-R's are routinely evaluated and re-adjusted if
necessary to maintain optimum power level.
All of the control modes incorporate defaults which abort all the T-R's, with
the exception of those in back corona maintenance, to maximum power settings to
avoid opacity excursions. These include a maximum opacity setpoint, which is
set at a level above the high opacity limit, and a feed-forward soot blowing
signal. After an abort condition, opacity feedback control is resumed only
after the stack opacity has returned to a value below the maximum opacity set
point.
Trial operation of the SPC and DPSEC control modes, with low and high opacity
set points of 9% and 10%, resulted in approximate power savings of 60% for SPC
control and 67% for DPSEC control. The initial base line opacity was
approximately 7% to 8%. During this particular evaluation period, DPSEC
control resulted in greater power savings than SPC control to maintain the same
stack opacity.
The new Energy Management control algorithm incorporates control strategies
that are not only intended to reduce energy consumption while maintaining a
specific stack opacity but also to manage the T-R energy or power levels for
optimum precipitator performance dependent upon specific T-R electrical
characteristics and operating conditions.
SUMMARY
Since startup, the Portland No. 1 and No. 2 electrostatic precipitators have
operated successfully with few problems and have operated well below
contractual requirements of 0.10 Ibs/MMBtu and 20% opacity. Based upon this
excellent performance to date, representing a combined total of 46 months of
operation, it is anticipated that these units will continue to operate well.
The IP computer based control system has been proven to be a reliable and
valuable asset, contributing to the overall electrostatic precipitator system
performance, the addition of the DPSE control, complete with automatic control
logic, has enhanced the overall effectiveness of the IP control system and
provides true energy management capability for optimum precipitator performance
as well as demonstrated energy cost savings.
REFERENCES
1. Abbott, P.G. and Schafebook T.C. Automatic ESP Control: Startup to
Shutdown. Seventh Symposium on the Transfer and Utilization of
Particulate Control Technology. Nashville, Tennessee. March 22-25, 1988
25-9
-------
ro
en
UhfYDBO BINJ I DUST BJN J
UNIT NO. 1
Figure 1. ESP General Arrangement
-------
TABLE 1
COAL ANALYSIS
UNIT NO. 2 PERFORMANCE TESTS
CARBON
HYDROGEN
OXYGEN
NITROGEN
SULFUR
ASH
MOISTURE
HEATING
VALUE
DESIGN
-
-
-
-
0.95% - 2.00%
12.11% (MAX)
9.00% (MAX.)
1 2,500 (MIN.)
PHASE
M
73.42
4.48
4.06
1.21
2.19
9.00
5.64
12,975
I
&&
0.94
0.23
0.56
0.04
0.16
0.87
0.53
260
PHASE III
M S^
72.32
4.22
4.18
1.23
1.01
11.28
5.76
12,706
0.99
0.16
0.75
0.07
0.04
0.52
0.61
126
PHASE
M
66.77
4.02
4.39
1.08
1.88
12.52
9.34
11,915
IV
5.E,
1.11
0.15
1.09
0.07
0.10
0.75
0.71
118
HEATING VALUE - BTU/LB
M- MEAN PERCENT
S.D. - STANDARD DEVIATION
TABLE 2
COAL ANALYSIS
UNIT NO. 1 PERFORMANCE TESTS
CARBON
HYDROGEN
OXYGEN
NITROGEN
SULFUR
ASH
MOISTURE
HEATING VALUE
DESIGN
-
-
-
-
0.95% - 2.00%
12.11% (MAX)
9.00% (MAX)
12,500 (MIN.)
PHASE I
M
74.36
4.77
3.73
1.34
2.00
9.07
4.73
13,123
SJX
0.85
0.10
1.65
0.04
0.22
1.85
0.32
303
HEATING VALUE - BTU/LB
M - MEAN PERCENT
S.D. - STANDARD DEVIATION
25-11
-------
TABLE 3
UNIT NO. 2 PERFORMANCE TEST RESULTS
TEST SERIES
DATE
LOAD, MW
GAS VOLUME, ACFM
GAS TEMPERATURE, °F
INLET DUST CONCENTRATION,
GRAINS/ACF
OUTLET DUST CONCENTRATION,
GRAINS/ACF
OUTLET EMISSIONS
LBS/MMBTU
PRECIPITATOR EFFICIENCY, %
POWER DENSITY, WATTS/SQ.FT.
OPACITY, %
DESIGN
252
868,000
271
3.740
0.030
0.10
99.20
20
PHASE I
8/25/87
256
809,639
273
2.290
0.0088
0.027
99.62
1.08
9<*)
PHASE III
7/29/87
257
881 ,983
276
2.582
0.0160
0.052
99.40
0.61
9
PHASE IV
9/1 0/87
250
810,211
268
3.439
0.0100
0.031
99.71
1.11
6
(*) ACCURACY OF OPACITY READING SUSPECT
ACTUAL VALUE ESTIMATED TO BE - 5%
TABLE 4
UNIT NO. 1 PERFORMANCE TEST RESULTS
TEST SERIES
DATE
LOAD, MW
GAS VOLUME, ACFM
GAS TEMPERATURE, °F
INLET DUST CONCENTRATION,
GRAINS/ACF
OUTLET DUST CONCENTRATION,
GRAINS/ACF
OUTLET EMISSIONS
LBS/MMBTU
PRECIPITATOR EFFICIENCY, %
POWER DENSITY, WATTS/SQ.FT.
OPACITY, %
DESIGN
172
584,000
266
3.71
0.03
0.10
99.20
20
PHASE I
4/26/89
159
599,796
276
1.521
0.0068
0.020
99.56
0.93
4
25-12
-------
TABLE 5
TYPICAL FLYASH ANALYSIS
UNIT NO. 2 PERFORMANCE TESTS
SILICA, Si02
FERRIC OXIDE, Fe2O3
ALUMINA, AI2O3
TITANIA, TiC-2
LIME, CaO
MAGNESIA, MgO
SULFUR TRIOXIDE, SO3
POTASSIUM OXIDE, K2O
SODIUM OXIDE, Na2O
PHASE 1
40.60
16.75
20.48
1.83
2.38
0.60
0.86
1.39
0.32
PHASE
50.65
9.98
23.40
2.50
1.31
0.90
0.37
2.95
0.25
Inlet
A Field
B Field
C Field
40
30
20
10
CURRENT DENSITY nA/sq. cm.
D Field
Outlet
E Field
A--S
30
40 50
APPLIED VOLTAGE KV
Figure 2. Typical Voltage - Current Curves
60
25-13
-------
MODERN ELECTRODE GEOMETRIES AND VOLTAGE
WAVEFORMS MINIMIZE THE REQUIRED SCAs
by: Kjell Porle, Sten Maartmann, Mats-Olof Bergstrb'm
Flakt Industriella Processer AB, Sweden
Keith Bradburn, Flakt Inc. USA
ABSTRACT
The size of an ESP required to reach a specified efficiency can be deduced in
different ways. Traditionally a SCA value is selected using a databank containing
a wealth of test results from ESPs designed more than 5 10 years ago. The result
may therefore be unduly conservative. Rather, we should take full advantage of new
developments in ESP engineering.
It is the intention of this paper to show that a modern plant design results in
reduced collecting area.
The energization mode and electrode geometry are key parameters to consider when
trying to increase ESP efficiency. Several examples from actual plants field tests
illustrate the importance of these variables.
The end user's main interest is to ensure that the future ESP plant will meet
emission requirements. He will carefully consider investment and operational
costs, reliability, space requirements, etc. He should then consider bids based on
each vendor's own experience. Which of course would have to be backed up by the
vendor's justification of the proposed plant. Such a procedure may lead to
considerable cost savings as well as justify incentives for further research and
development work in ESP technology.
INTRODUCTION
The estimation of the specific collecting area (SCA) required for an ESP is a
difficult task. All vendors and users know that it must be based on experience
rather than on theoretical considerations. This means that the SCA is selected
from a databank of test results from old and newer plants. Assuming that old
plants do not incorporate new technologies the chosen SCA value may be unduly
conservative. It is the intention of this paper to show that electrode geometry
and type of energization are two parameters that play a dominating role for the
selection of the ESP size. This is more pronounced the higher expected the dust
resistivity and may therefore be a major point to consider for example when
switching to low sulphur coals.
Evaluation of a bid from a vendor should therefore to a greater extent be based on
his own experience so that new technologies should be given credit. Further
research and development work in ESP technology is then justified. The paper will
give some historical background to experience with high resistivity fly ashes and
the choice of discharge electrodes. The difficulty in comparing results from one
26-1
-------
plant to another and from one electrode geometry to another resulted in extensive
in the 1980s. The importance of electrode geometry as well as
possible cost savings when
pilot testing
energization mode will be exemplified. An example of
using optimized technologies is also given.
FLAKT'S EXPERIENCE
Fig. 1 is an example of performance lines for various fly ash conditions.
Particulate collection efficiency is given as a function of relative specific
collecting area (SCA). Each performance line represents field measurements with
Flakt's standard pilot ESP system at a power plant firing a specific coal. For
comparison purposes performance line A, that was taken from a full-scale plant
firing Polish coal is also included. The size difference between the Polish fly
ash and a high resistivity ash from Australia, coal F, is about four times at the
tested prevailing conditions. In addition to coal ash composition, gas temperature
represents a major difference in operating data between the performance lines. The
results were obtained at actual rather low operating gas temperatures, and are not
valid for other higher temperatures. Table 1 shows some of the important
constituents of the respective fly ashes.
Fig. 1 Performance lines for various fly ashes.
TABLE 1
Coal
Ash content, %
(dry basis)
A
10
B
16
C
31
D
19
E
18
F
28
Sulfur, total, %
0.70 0.36 0.34 0.62 0.27
Na20 in coal ash, % 0.80 0.65 0.50 0.10 0.30
0.26
0.10
26-2
-------
Of course, the fly ashes can also be characterized on the basis of their
respective resistivities. Fig. 1 can be transferred in principle to curves like
the ones in Fig. 2. In reality, a curve represents a fairly broad band, because
resistivity is only one of the parameters affecting wk, the migration velocity.
For the sake of discussion Fig. 2 is appropriate, however.
W,,
Spiral
Plate
Resistivity
Fig.2. Migration velocity versus resistivity for spiral and plate discharge
electrodes.
Fig. 3. Spiral and plate discharge electrodes.
The performance lines in Fig. 1. were arrived at using the Flakt helical-wire, the
spiral, as discharge electrode. Fig. 3 shows two discharge electrodes - the spiral
and a plate discharge electrode with peaks pointing towards the collecting
surfaces. The latter type was introduced in the early 1960s in ESPs at the
Swanbank and Vales Point power stations in Australia. The sizing of the ESPs was
26-3
-------
preceeded by pilot testing with spirals. High resistivity conditions prevailed.
The full-scale performance with the plate discharge electrode did not full fill
expectations at all At Swanbank the migration velocity was more than halved. The
guaranteed efficiency was achieved only after rebuilding to spirals in two of
totally three fields. The dotted line in Fig. 2 reflects the migration velocities
obtained with the plate discharge electrode. The difference in wk between the
plate and the spiral is less for low and medium resistivity fly ashes than for
high resistivity ashes. Investigations and research work over the years including
current distribution testing in the laboratory have clearly identified the major
reasons for reduced efficiency:
* Back corona is increased by an uneven current distribution from the
plate electrode on the collecting surface. There was a high current
density opposite to the peaks of the discharge electrode and back corona
is also formed in these areas at very low average current densities.
There are also large areas with a current densities down to zero. The
result is inefficient charging and collection. The spiral electrode
shows a much better current distribution. Areas with extremely high and
low current densities are avoided. Examples of good and bad current
distributions have been published earlier by Porle, Ref. (1) and Sten
Maartmann, Ref. (7).
* Dust build-up, especially on the peaks results in reduced performance.
Small crusty balls grow slowly and change the electrical properties. The
effect of this may take a few days to develop. Consequently, a
representative test, either in a pilot ESP or in a full size unit, must
be run continuously for some time to create stable conditions. Matts,
Ref. (2) has shown how migration velocity can deteriorate because peaks
were clogged. The peaks were manually cleaned at certain intervals in a
pilot ESP, while other conditions were unchanged, and the migration
velocity was temporarily restored. A wire does not seem to be prone to
this effect, which may be explained by the "continuous corona surface".
The design of the discharge electrode is however not the only important factor.
Adequate current and voltage must be maintained in the bus section before
sparking occurs. Local high electrical field strength along the collecting surface
may trigger a spark. Akerlund, Ref.(3), has demonstrated that the configuration of
the collecting plate must be carefully designed and tuned to the specific
discharge electrode so that maximum electrical energy can be used in the field.
Ridges to stabilize the collecting plates can easily amplify the electrical field
several times.
DOUBLE CHAMBER PILOT ESP
As a consequence of the results with the plate type electrode, pilot testing with
various configurations was started. It is obvious however looking at fig 1. that
coal ash composition plays the major role in the migration velocity for a
configuration. Small deviations in coal ash composition may overrule the influence
of the design. A new approch for safe and reliable comparison between various
configurations was chosen - a double chamber pilot ESP.
A slip-stream from the flue gas after a 630 MW coal-fired boiler is taken to a
pilot ESP consisting of two parallel chambers of the same size. The slip stream is
isokinetically drawn out through nozzles spread across the main gas duct. A common
duct_ leads the gases to the inlet of the pilot, and the two casings. Thus,
realistic and equal gas and dust conditions are obtained in the two casings.
26-4
-------
Comparative testing has been performed with one electrode configuration in the
reference chamber, and another in the other. Fig. 4 shows results from five
series. The first series, designated A, was run with the same configuration in
both casings. Spirals were used as discharge electrodes. The results indicated
that the same performance could be obtained, and it was thus concluded that the
test procedure was reliable. At the end of the program series E was performed with
the same configuration in both casings to verify that the casings were still equal
in performance. All series were run with varying current densities and gas
velocities. .^As a result, the particulate emission varied in a wide range, from 25
to 300 mg/m NTP at an average inlet dust load of 10 g/m NTP. It was considered
important to have realistic efficiencies in order to be able to evaluate the
outcome of the future test.
Wk-test
W^-spiral
\.£.
1.1-
1-
0.9-
0.8-
0.7-
0.6-
0.5-
04-
•
• •
•
•
/".*-. .
.• * .-\
•*• •
* .•© ' .
u t
.'. ®"«.
'
0-
• *© **
« •"
Test series:
B
D
Fig. 4. Comparative testing of migration velocity for various electrode
configurations in a double chamber pilot ESP.
The included series, denominated B - D, were run with spirals as the reference
electrodes. Series B a mast-type electrode with sharp peaks. C and D both had
another mast with rounded peaks, but difference being the configuration of the
collecting surfaces was different. The reason for running field tests was that in
the laboratory, the discharge electrodes used in B D showed a substantially
better current distribution than the previously mentioned plate type, see Fig. 3.
Still, the current distribution from the spiral was the best.
Fig. 5 shows the basic difference in the collecting surface design in series C and
D. The wider ridge, series C, had a very pronounced negative effect on
performance.
26-5
-------
The spread in
closer study
increase the
electrode. It
flow pattern
data from this series is also larger than from the other series. A
of the results from series C shows that higher gas velocities
difference in migration velocity between the spiral and the mast
could be worthwhile to remember that it is a difference in the gas
between a collecting plate and a wire or a mast- or plate-type
discharge electrode, as the latter ones comprise a bigger obstacle.
300
o
D
300
O
H ' ~_n
50 C ' 100
Fig. 5. Electrode configuration for series C and D.
the pilot tests tally very well with those obtained from
Test serie^A - E were run with a fly ash of moderately high
resistivity, mostly around 10 ohm cm. The following conclusions can be drawn in
order to obtain maximum collection efficiencies:
The results from
full-scale plants.
The configuration of the discharge electrode
distribution that is as even as possible.
should give a current
Sharp and wide
even electrical
protrusions and ridges should be avoided
field strength along the collecting plate.
to obtain an
* The gas flow pattern in the electrode space influences the result.
The test results from the pilot should not be used as absolute values, but they
are considered to be the best guidance for determining the optimum electrode
geometry. Alignment, rigidity, tolerances, oscillation tendencies, the possibility
to avoid sneakage etc. have also to be considered when comparing two
configurations.
ENERGIZATION
The development of pulsed energization has been intense and successful during the
past ten years. Power savings and/or reduced emissions have been achieved in many
cases. This paper will only discuss the effect on emissions. Pulsed energization
implies that the corona current is not generated continuously. Various concepts
for generating pulsed corona are thoroughly described in literature.
26-6
-------
Reduced emissions are achieved when the dust resistivity is high, or when back
corona conditions prevail. Porle and Ekstrom showed, Ref. (4) that the
wk-improvement was greater, the higher the resistivity, and the shorter the pulse
width. Today's commercially available systems are either based on microsecond
pulses, e.g Flakt s Multipulse concept, or on millisecond pulses, - often
referred to as intermittent energization (IE) e.g, Flakt's Semipulse* concept
Millisecond pulses have gained wide acceptance mainly because ordinary
transformer/rectifier sets can be utilized with only minor modifications Only the
controller has to be modified. Microsecond pulses need a more sophisticated
T/R-set and are justified in special applications where high emission reductions
are necessary, e.g when limited space is available and lowest possible emission is
requi red.
mA ,,
AAAAAAAAAA
A
A
A
A
A A
A A
A Ar
mA , i
A A
r\ r\
AA
mA t L
A A
AA
Fig. 6. Conventional charging (upper diagram) and various intermittent
energization modes (lower four diagrams).
IE can be run in different modes, see Fig. 6. Landham et al, Ref. (5), showed in
1986 that the effect of various wave forms is a change in efficiency. This is in
line with Flakt's own experience. A mode, in which one single half wave of the
mains is used for charging, and a number of half waves are blocked, has given the
best performance. Porle, Ref. (1), has also shown that the time between the pulses
is important in itself. For high resistivity conditions it is likely that this
time should increase from the front towards the rear of the ESP. Landham also
demonstrated, Ref. (5), that the configuration of the discharge electrode was
important. Fig. 7 shows results from similar tests with the previously mentioned
dubble chamber pilot ESP. The two tested discharge electrodes gave not only
different migration velocities, but when applying Semipulse , the improvement was
higher for the spiral than for the mast-type electrode. The optimum waveform for
one electrode geometry may very well differ from that waveform for another
configuration.
The voltage waveform for the microsecond pulse is also an important factor.
26-7
-------
The Multipulse waveform is seen in Fig. 8. Fig. 9 sh^s the overall results
obtained from extensive investigations with Multipulse at the Swanbank and
Liddell power stations in Australia. Porle and Funnel 1, Ref. (6) showed that
optimum performance wai achieved for a given burst frequency, very often down to
about 10 Hz for severe conditions, and for a specific number of pulses per burst.
A higher wk-enhancemen:. was obtained with the prevailing conditions at Swanbank,
which had mixed electrode geometries, than at Liddell, which had spiral
electrodes.
Relative Wk
12
——- Spiral geometry
•^^ Other commercial
geometry
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40
Average current density (mA/m2)
R
Fig. 7. Effect of Semipulse on two various geometries as measured in the double
chamber ESP.
The overall conclusions from experience gained with pulsed energization is that
* the optimized waveform
* the electrode configuration and
* the resistivity or the dust properties
are the important factors for successful implementation of this technology.
Unless the vendor has a wide range of experience with a specific geometry and
energization method, it is almost impossible to predict the necessary SCA for
a given application. Pilot tests or a high degree of built-in safety margin
when sizing an ESP should be recommended if necessary back-ground material is
not available.
-------
EP voltage
Flashover limit
Time
TM
Fig. 8. Multipulse high voltage waveforms.
wk-enhancement
10 15
Relative wk, conventional energization
TM
Fig. 9. Enhancement of wk with Multipulse energization compared to conventional
energization.
26-9
-------
DISCUSSIONS
One way to illustrate the papers message is to assume that an enduser some years
ago put into operation a new large coal fired boiler equipped with an ESP sized in
all respects according to the requirements put forward by a consultant. But,
instead of ordering the ESP from one supplier, he oredered four casings one each
from a different supplier. The electrical equipment is however identical and has
IE capabi1ity.
Efficiency tests were run on each casing separately after a reasonable operating
time and when firing a coal giving difficult, high resistivity, to collect fly
ash. Each supplier was given the opportunity to ensure that the operation of his
casing was optimized to the conditions prevailing, before the test.
The design SCA and the required efficiency together give a design wk-value.
Similarly each test result can be expressed in such a value and as a consequence
as relative wk-value in per cent of the design wk-value. In the table 2 below
probable such values are given for each casing both for operation with and without
IE.
TABLE 2
ESP design
A
B
C
D
Rel . wk
base test
100
110
120
140
Rel . wk
IE test
110
125
140
170
wk improve
ment with IE
10
14
17
21
NOTE: The design denominations have no reference to the designs mentioned earlier
in the paper.
An enduser should draw the following conclusions:
* ESPs B, C and D could have been built with smaller SCAs.
* With IE all ESPs could decrease their size and the size difference
between the designs was accentuated. The influence on installation and
operating cost would be significant.
REFLECTIONS ON THE DIFFERENCE IN PERFORMANCE
Large sums have been spent to find trends in test results from ESP plants built by
different vendors. No doubt investigators have found that some designs appear to
be better or worse than others, but as the system works "everybody" shall be
accomodated. As a consequence the minimum SCA shall enable the worst design to
meet performance. Hence a difference in base performance is to be expected.
26-10
-------
REFLECTIONS ON SCA VALUES
A consultant relies, when suggesting a minimum SCA value, on a data base of test
values and trends in these. It will take at least five years, perhaps closer to
ten, before he knows that his recommendation was correct. In the reverse: The
results in his base are probably more than five, on an average perhaps more than
10 years old.
Here the vendors are at an advantage and particularly in a period when new
technology, in this case IE, is introduced. They will within say five years time
have installed IE equipment at a number of full-scale plants, tested them and
established new competition in markets where endusers tend to buy from the lowest
bidder that has "adequate" SCA. In such markets the SCA value is generally not
specified, but is chosen by the enduser from the bids he has received and
judgement of the justifications of the sizings each vendor has given.
It will take much longer for consultants to change their recommendations. They
rely primarily on tests from new full-scale ESPs and tend to wait until a new
technology has been proven. In the meantime plants will be built that are surely
oversized. Thus less efforts will be divoted to optimization and more results will
be fed to the experience bank that strenghten an attitude of "sizing as we always
have".
CONCLUSION
The efforts and the considerable sums spent by vendors on R&D to improve
performance of ESPs, warrants a more extensive evaluation procedure of bids than
is present practice. Endusers must allow bids with a SCA smaller than specified
and investigate throughly the evidence presented to justify each vendors choise.
Then vendors efforts to achieve even more with future investigations and
allocations for R&D will be spurred and be easier to warrant.
26-11
-------
REFERENCES
1. K. Porle, On Back Corona in Precipitators and Suppressing it Using Different
Energization Methods. Paper presented at the 3rd International Conference on
Electrostatic Precipitators, Albano, Italy, October 1987.
2. S. Matts, Pilot Scale Units for Precipitator Sizing. Paper presented at the
1st International Conference on Electrostatic Precipitators, Monterey,
California, 1981.
3. C.E. Akerlund, Calculation of Electric Field and Current Distribution along
Profiled Collecting Electrodes in a Wire-Plate Electrostatic Precipitator,
Paper presented at the 7th Symposium on the Transfer and Utilization of
Particulate Control Technology, Nashville, Tennessee, 1988.
4. K. Porle and B. Ekstrom, Pulsed Energization of Electrostatic Precipitators.
Paper presented on NCB International Seminar on Pragmatic Strategies for
Productivity and Modernization, New Delhi, India, January 1987.
5. E.G. Landham, Jr., J.L. DuBard and W. Piulle, Pilot-Scale Evaluation of ESP
Intermittent Energization. Paper presented at Sixth Symposium on the Transfer
and Utilization of Particulate Control Technology, February 1986.
6. K. Porle and P. Funnel!, Substantial ESP Performance Improvement achieved
with Microsecond Pulsing. Paper presented at the Third CSIRO-conference on
gas cleaning, Medow Bath, NSW, Australia, August 1988.
7 S. Maartmann, Current Distribution and Resistivity Effects in Electrostatic
Precipitators. Paper presented at the Centennial of Electrostatic
Precipitation, Adelaide, Australia, 1974.
26-12
-------
ESP DESIGN CONCEPTS FOR
IMPROVING PERFORMANCE AND RELIABILITY
ABSTRACT
The topic of this joint presentation will be the methods for improving the performance and
reliability of weighted wire electrostatic precipitators with specific emphasis on the concept
of dynamic alignment. The paper will feature the Nearman Creek ESP on the Kansas City
Board of Public Utilities System. The design concepts applied to the Nearman Creek
precipitator during its hot-side to cold-side conversion will be discussed in detail. These
design concepts as applied to Salt River Project's Navajo Station precipitators will also be
discussed. The operation and performance of these ESP's prior and subsequent to the
changes will be presented.
John Meinders
Kansas City, Kansas
Board of Public Utilities
Robert E. Jonellis
PrecipTech, Inc.
27-1
-------
ESP DESIGN CONCEPTS FOR
IMPROVING PERFORMANCE AND RELIABILITY
INTRODUCTION
The sun rises in the morning. The sun sets in the evening. It gets hot in the summer. It gets
cold in the winter. These are laws of nature which we rely upon each day. There are also
laws of nature which apply to electrostatic precipitators.
One of these, Coulomb's Inverse Square Law, can be written as:
This states that the force attracting a high voltage wire to a grounded collector plate is
inversely proportional to the square of the distance between the wire and the plate1.
Another formula used in connection with electrostatic precipitators is the Deutsch-
Anderson equation which can be stated as:
E=l-EXP(-wA/V)
Or translated: efficiency in a precipitator is a function of collection area, gas volume, and
migration velocity.2
There are other laws of nature, or characteristics, which impact precipitators. Many of these
make use of nice intriguing formulas to keep engineers like us interested. But we won't
take up your time discussing the formulas. It is not necessary to calculate the course of the
earth around the sun and prove your position mathematically to determine that if s time to
buy short sleeve shirts.
'F= attractive force between the wire and plate; K= a constant; Ql= the charge on the plate; Q2= the charge on the wire;
d= the distance between the wire and the plate
2 E=effidency; EXP= the base for natural logarithms; A= collection area; w= migration velocity; V= gas flow rate
27-2
-------
Some of the fundamental variables impacting electrostatic precipitator operation have not
yet found their way into the traditional science of predicting precipitator performance.
Among these are:
a. The stability of the dynamic alignment.
b. The consistency of static alignment.
c. The condition of collector plates.
d. The mechanical differences between precipitator designs.
e. The thickness of the ash layer on plates and wires.
f. The sneakage of gas around the treatment area.
g. The chemistry of the particulate.
h. The re-entrainment of the particulate.
i. The gas distribution through the precipitator.
j. The current density
k. The response of the controls.
This paper will concentrate on the "A-B-C's", leaving the rest for another time. We will
look at two specific projects and review the experiences of each.
THE NEARMAN CREEK PROJECT
Background
The first precipitator we will examine is Nearman Creek Unit #1, on the Kansas City Board
of Public Utilities system. Like many hot-side precipitators, this one had not performed up
to expectations. It was suffering from the typical hot-side syndrome: Emissions would rise
beyond acceptable levels; the unit would then be taken out of service every 3-4 months
and grain blasted. After cleaning, Unit #l's opacity would once again begin to rise from
8% to 20%.
Unit # 1's rappers were operated at peak intensity, and as often as possible. The intensity
was high enough to sometimes knock the top plate off of the rapper. Even with such
27-3
-------
diligent rapping, a buildup of from 1 to 1 1/2 inches could be found on the plates and a
uniform coating of ash from 1 3/4 to 2 inches in diameter would accumulate on the wires.
Removal of the buildup would reveal a tenacious film on the plates and a crust on the
wires.
The unit also suffered from the chronic problem of collector plate warpage and wire
breakage. Initial attempts to straighten the plates by crimping proved unsatisfactory.
The out-of-service and labor costs were becoming intolerable and there was considerable
concern for the environment. Something had to be done.
ESP Design Data
The precipitator is a negative pressure, weighted wire unit built in the late 1970's. The box
is respectably sized, with an original SCA of 370 and an aspect ratio of 1.2. Current density
is 81.7 ma./lOOO sq. ft. at the inlet fields and 98.0 ma./lOOO sq. ft. for all other fields. The
precipitator contains 32 cells, each holding 35 collector plates, 9 ft. wide by 30 ft. high,
spaced 9 inches apart. Original gas velocity was 4.32 FPS. The unit is 235 megawatts, with a
pulverized fuel boiler. The fuel is Powder River Basin coal with a sulphur content of 0.3%.
Upgrade of Precipitator
During the 1988 Spring outage, the precipitator was converted from hot-side to cold-side
operation. This was a matter of rearranging the ductwork to put the air heater upstream
rather than downstream from the precipitator. As a result, the operating temperature of the
precipitator dropped from 770 °F. to 335°F. Details concerning this conversion can be found
in the December 1989 issue of POWER ENGINEERING in an article entitled "Conversion
of ESP Boosts Reliability of Coal-Fired Power Plant"
Resistivity charts indicated that particle resistivity would rise and performance decline as
the flue gas temperature decreased (Figure 1). To compensate for this expected problem, a
flue gas conditioning system was installed. Calculations pointed to the strong need for such
a system but the traditional methods of analysis failed to predict what actually happened.
27-4
-------
200 300 400 500 600 700 800 900
TEMPERATURE, DEGREES FAHRENHEIT
Figure 1. Calculated fly ash resistivity using the Bickelhaupt Formula
for the coal normally burned at Nearman Creek
In addition to flue gas conditioning, the scope of the project at Nearman Creek included an
extensive modification and repair of the precipitator. This involved the following:
• Installation of a new anti-sway system using the "V" configuration
to prevent motion and fouling of the discharge electrode system.
• Widespread straightening of bowed collector plates using Ladder Bars.
• Rigorous wire-to-plate alignment and fine tuning to better-than-new
standards. This included moving upper frames by repositioning the
support insulators.
• Modification of the ductwork internals to provide a uniform gas distribution.
• Installation of a puff blower system to avoid ash buildup in the ductwork.
• Installation of an energy management/opacity feedback system.
27-5
-------
Surprising Results
As a result of the conversion and other repairs, BPU has found that the precipitator not
only operates extremely well, but that it maintains this performance without the gas
conditioning system. Opacity levels remain consistently between 2% and 7% (Figure 2).
o
<
Q_
O
o
ce:
Q_ .
TIME
MONTHS
Figure 2. Performance of Nearman Creek ESP before and after
applying the design concepts that were installed
during the hot-to-cold side conversion.
Once work was completed, normal rapping was sufficient to remove all but a 1/8 to 1/4
inch buildup on the plates. Wire buildup was found to range from clean wires to a 1/8 inch
diameter accumulation. The tenacious film on the plates did not reoccur. An underlying
crust still forms on the wires, but is easily removed.
27-6
-------
Although BPU had initially expressed concern that the installation of Ladder Bars might
hinder rapping efficiency, the opposite proved to be true, possibly due to a more evenly
distributed rapping impulse.
Another concern had been the necessity of removing wires at each anti-sway location.
However, experience is proving this not to be a factor. In fact, the performance is better
than when the unit was new. Reliability has also been improved. Wire breakage has been
effectively reduced to zero, and it is no longer necessary to bring the unit down for
cleaning. The gas conditioning system is not used for routine operation, but rather is kept
as a standby system.
How is it that the flue gas conditioning system should prove to be unneeded? What
characteristics of the precipitator had been changed without being accounted for in the
calculations? It is evident that the mechanical upgrade effort at Nearman Creek is largely
responsible for the surprising results. It appears that the mechanical condition of a
precipitator is much more important than previously realized.
Controls Confirm Alignment
New controls with an energy management/opacity feedback function were installed
during the conversion. This power optimization control mode maintains the stack opacity
within preset high and low opacity levels by controlling the power input to the transformer
rectifiers (T/R's).
The Nearman Creek energy management/opacity feedback high and low opacity set points
are set at 7% and 5% respectively. Often the stack opacity is below the 5% set point while
still in the energy management/opacity feedback mode.
27-7
-------
Table 1
NEARMAN CREEK PRECIPITATOR
PERCENTAGE OF POWER ON T/R'S
THROUGH TWO CHAMBERS
FROM INLET TO OUTLET
Energy management/opacity feedback mode controlling all T/R's at
20% of full power or less at 1:04 pm, February 5,1990:
T/R# ABCDEFGH
Actual % of Power 16 16 19 19 16 16 19 19
Energy management/opacity feedback mode taken out of service at
1:17 pm February 5,1990:
T/R# ABCD E F GH
Actual % of Power 15 28 44 36 29 49 59 59
Referring to Table 1, the capacity of the steam generator was at 75% of its maximum
continuous rating. The opacity was recorded at 3%. At this low opacity reading, the energy
management/opacity feedback has a 20% required set point for all the T/R's. The inlet
fields are low in the percentage of actual power while the outlet fields are being held at the
19% or 20% level. When the energy management/opacity feedback was turned off, the
opacity was virtually unchanged; however, the actual power levels increased dramatically.
Good dynamic alignment must be achieved to obtain power levels and opacity levels as
experienced in this unit.
Future Improvements
A number of additional repairs are scheduled for subsequent outages. Among these are:
• Conversion of 3/8 inch hanger rods which support the lower frame to
wire supports.
• Addition of link bars at original side spacer locations to better segregate
rapper groups.
• Installation of large diameter rapper sleeves and flexible boot seals to
prevent rapper binding at the hot roof.
27-8
-------
THE NAVAJO PROJECT
Background
Navajo Units #1, #2 and #3, Salt River Project, are identical units located near Page,
Arizona. Like the Nearman Creek precipitator, these had not performed to expectations.
Opacity would typically reach 20%. The plant is located on the Navajo Indian reservation
amidst national, state and tribal parks, and is committed to the cleanest possible stack. This
commitment goes beyond the EPA's standards.
A number of research projects have been conducted on these precipitators, including
airflow distribution modeling, slip stream treatment utilizing sodium and ammonium
products, large wire testing, rapper optimization programs, and currently, the application
of load cells to optimize rapping.
The staff is known for doing things right throughout the power plant. The precipitators are
routinely inspected at each outage using standardized procedures. Inspection and repair
data is kept meticulously, using a computer data base. Trends are spotted with graphs. The
plant typically performs any repairs as soon as damage is found. To squeeze more
performance from this extremely well-maintained precipitator, it was necessary to go
beyond the conventional approach.
ESP Design Data
The three precipitators are weighted wire units built in 1974,1975, and 1976 respectively.
The precipitators are hot-side units with an SCA of 250, and an aspect ratio of 1.2. Current
density is 40 ma./1000 sq. ft. for the inlet fields, 48 ma./1000 sq. ft. for the intermediate
fields and 63 ma./1000 sq. ft. for the outlet fields. The boxes are negative draft. Each
precipitator contains 96 cells, each holding 36 collector plates 6 ft. wide by 30 ft. high,
spaced 9 inches apart. Gas velocity is 5 FPS. The three units produce a total of 2400 MW
with pulverized fuel boilers. The fuel is 0.5% sulphur coal.
27-9
-------
Uggrade of Precipitator
A number of mechanical problems were found in the precipitators. Some of these
problems were designed into the unit; others had developed over time, but no practical
repair was available. These included slack wires caused by double-wire weights, discharge
electrode frames which swayed like porch swings, and collector plates which were warped
throughout the units.
The job of correcting these problems in 288 cells was spread over several outages. The
following repairs were performed:
• A new anti-sway system using the "V" configuration was
installed in all cells to prevent motion and fouling of the discharge
electrode system.
« Special oversized single-wire weights and lower discharge electrode
frames were installed in the inlet half of each precipitator to stop wire
whip and slack wires.
• Ladder Bars were installed to straighten the collector plates where required
throughout the precipitator.
• Alignment of internals was performed to better-than-new standards.
• Many other repairs and modifications were performed ranging from the
repair of structural members to the modification of anti-sneak baffles.
Surprising Results
After repairs, testing demonstrated that the "on-time" of power supplies was increased by
over 30%. In some ways this has the same effect as making the precipitator 30% larger.
Reliability has significantly improved. The three units are able to operate with no
precipitator related downtime, coming off line only for scheduled outages about every 12
months. Wire breakage in the precipitators is effectively reduced to zero.
27-10
-------
These results are surprising. It is believed, however, that similar results can be achieved in
other precipitators. The mechanical condition of a precipitator is fundamental to good
performance.
THE ALIGNMENT PUZZLE
The Concept of Dynamic Alignment
How is it that these mechanical repairs had such a significant impact on precipitator
performance and reliability? To repeat, there are laws of nature which apply to an
operating precipitator. One mentioned earlier, Coulomb's Inverse Square Law, says that
"the force attracting a high voltage wire to a grounded collector plate is inversely
proportional to the square of the distance between the wire and the plate". As we all
know, each time there is a spark in a given field, the controls temporarily quench the
power to that field giving the spark time to extinguish. What happens to the wires? For a
time they are drawn to the plates, then they are temporarily allowed to relax. So the wires
begin to wiggle and oscillate.
As the wires oscillate, the center of the wire comes into closer proximity to the plates,
much like the center of a jump rope, held by two children, comes into close proximity to
the sidewalk. As the wires move closer to the plates, the force of attraction is increased
dramatically. Eventually, the wires induce a slight pull on the lower frame. The sway of
the frame moves more wires in the same direction. With more movement comes more
pull on the lower frame. The result is much like what happens when you gently push a
little boy on a playground swing. He keeps moving further and further. This is exactly
what can happen to the lower frame and wires in a given field. Our traditional prediction
methods do not address this type of inefficiency in a precipitator. This is why, by
stabilizing the internals of a precipitator, it is possible to get better performance than
predicted.
It is time to come to grips with the fact that all collection area is not equal. There is such a
thing as "quality" of collection area as well as "quantity". Proper maintenance of a pre-
cipitator normally includes many manhours of painstaking alignment between wires and
plates. This is called "static alignment", or the location of the internals at rest. A good
static alignment will be destroyed during operation if the discharge electrode system is
27-11
-------
allowed to move back and forth like a porch swing, if .the wires are allowed to whip, or if
the weights are allowed to bottom-out. Unfortunately, even though equipment is usually
provided to disallow it, the discharge electrode system of many precipitators permits just
this type of movement.
"Dynamic alignment" is the wire-to-plate clearance which exists when the precipitator is in
operation. To achieve gains on performance and reliability, dynamic alignment must be the
same as static alignment. And static alignment must be near perfect. The efficiency
equation should be re-written. To get the ball rolling, the following idea is presented:
E=l-EXP(-wA/V)-(P-D)/C2
where "D" is minimum dynamic alignment, "P" is perfect alignment, and "C" is a factor
relating wire-to-plate spacing to precipitator efficiency. This equation states that the closer
dynamic alignment equates to perfect alignment, the better the performance will be. In
other words, the lower frame must not sway, the wires must not whip, and the weights
must not bottom out. The mechanical modifications used at Nearman Creek and Navajo
stations were designed to prevent these problems.
Solving the Alignment Puzzle
The Nearman Creek precipitator was originally built using the typical anti-sway design
which links the lower frame to the hopper. This system uses a vertical post insulator which
has a long lever arm that can permit excessive movement of the lower frame. The design
can foul with ash or bottom-out. Bottom-out is a serious condition which leads to large
numbers of slack wires and grounding of the field.
The Navajo precipitators were built using a porch swing type of anti-sway, which linked
the upper frame to the lower frame. This system basically did not work at all. In addition,
plates adjacent to the heavy porch swing hardware tended to warp in all three
precipitators.
The new anti-sway (Figure 3.) used at Nearman Creek and Navajo stations was designed
using four horizontal blade insulators per frame in two "V" shapes. This was found to be
exceptionally stable and prevented sway of the lower discharge electrode frame in excess of
+1/4 inch. Because of the design of the horizontal blade insulators and hardware, it was
almost impossible for the anti-sway device to foul or bottom out.
27-12
-------
The brackets linked the lower frame to the collector plates rather than to the hopper to
ensure that dynamic alignment was the same as static alignment. The blade insulators and
hardware were designed to prevent stress breakage. Special high temperature fiber
washers were used to prevent scoring of the insulators. The result was an anti-sway
system which performed well with a high degree of reliability. Insulator breakage in these
units has been effectively zero with the new equipment.
Figure 3. V-Style anti-sways
To prevent slack wires and wire whip in the Navajo units, a 30 Ib. single wire weight was
installed to replace the double wire weight in one-half of each precipitator. The upgraded
weight was designed with a special interface to the wire which eliminated stress breakage
at the wire-to-shroud junction. The weight was specially designed to prevent dust buildup
which in other weights has been observed to press against the wire shroud and displace
27-13
-------
the wire. The bottom of the weight was rounded to minimize spark-over to a hopper
surface. The lower frame utilized a "pigtail" retainer which was specially designed to
prevent the weights from hanging-up. All these design concepts were installed to
improve dynamic alignment.
The Stigma of Warped Collector Plates
One of the most common forms of major damage to precipitator internals is distortion of
the collector plates. This type of damage was found throughout both the Nearman Creek
and Navajo precipitators. Warped plates are usually caused by temperature excursions.
The result is close wire-to-plate clearances and reduced precipitator performance.
Warped collector plates can also disrupt gas flow, encourage sneakage into the hoppers
and around the treatment area, dampen rapping, and promote re-entrainment.
To straighten warped collector plates in these precipitators, Ladder Bar spacers (Figure
4.) were used. These spacers work by allowing the internal stresses which act to distort
one collector plate to be directed to provide the force needed to straighten another plate
via transfer points. Design considerations included the following:
• Modular units linked together to form long ladders.
• Transfer points approximately every three feet in elevation.
• Ability to quickly install at the stiffener-baffle without removal of wires.
• Top suspension to prevent any spacers from falling.
• Loose interface with stiffener-baffles for better rapping.
• 3/8" diameter smooth surface exposed to wire, with welds hidden.
• Cold rolled steel construction.
The result is to straighten and stabilize the plates, restoring proper alignment without the
need to replace the collector plates with new equipment. The straightened plates allowed
excellent static alignment to be achieved while the design concepts noted above allowed
dynamic alignment to remain almost equal to static alignment. These are the design
27-14
-------
concepts which so remarkably improved the performance and reliability of the Nearman
Creek and Navajo units.
Figure 4. Ladder Bar system for plate straightening
CONCLUSION
In the coming decade, the air pollution control industry must find more ways to improve
precipitator performance. Many of the existing precipitators can be upgraded in the same
way as the units discussed in this paper. New standards are needed to predict efficiency
gains achieved by precipitators in which the dynamic alignment problems have been
corrected. Experience with the precipitators discussed in this paper has shown that there
is more to the efficiency of a precipitator than can be measured by the conventional
formulas. Quality as well as quantity of collection area counts.
27-15
-------
ACKNOWLEDGMENTS
Credit is extended to Robert B. Candelaria, Salt River Project, and John A. Jonelis, Midwest
Power Corp./PrecipTech, Inc. for help in producing this paper.
REFERENCES
Robert E. Jonelis and John A. Jonelis. "Economical, Innovative Methods of Straightening
ESP Collector Plates So As to Achieve Alignment for the Purpose of Improving Particulate
Control and Performance" Presented at the Joint ASME/IEEE Power Generation
Conference, Milwaukee, Wisconsin, October 20-24,1985.
Allen W. Ferguson, Dale S. Lindberg, John R. Meinders. "Conversion of ESP Boosts
Reliability of Coal-Fired Power Plant" Power Engineering Magazine, December 1989.
John A. Jonelis. "Notes on the Stabilized Electrode System" Midwest Power Corp. March
1987.
Robert E. Jonelis. "Improving Precipitator Performance" Presented at the Fifth Annual
International Pittsburg Coal Conference, September 12-16,1988.
27-16
-------
CONSIDERATIONS IN REBUILDING
THE SIBLEY UNIT 1 PRECIPITATOR
D. M. Greashaber
Missouri Public Service
10700 East M 350 Highway
Kansas City, Missouri 64138
P. A. Miller
Sargent & Lundy
55 East Monroe Street
Chicago, Illinois 60603
ABSTRACT
Sibley Unit 1 is a 50-MW cyclone fired boiler which was installed in 1960. A
weighted wire precipitator was installed in 1971. The precipitator will be rebuilt
using rigid discharge electrodes and collecting electrodes on 12-inch spacing during
the first half of 1990. Items considered in the decision to rebuild the
precipitator are discussed. Included are a description of the mechanical condition
of the precipitator, characterization of the present fly ash characteristics, and an
economic analysis of rebuilding the precipitator with the collecting plates on
9-inch, 12-inch, and 16-inch centers. Performance estimates are given for the
present condition and for the rebuilt precipitator with the current medium sulfur
coal for the 9-inch and the 12-inch collecting electrode spacings. Performance is
also estimated for the rebuilt precipitator when collecting the ash from three
candidate low sulfur western coals both with and without gas conditioning.
28-1
-------
CONSIDERATIONS IN REBUILDING
THE SIBLEY UNIT 1 PRECIPITATOR
INTRODUCTION
Sibley Unit 1 of Missouri Public Service is a 50-MW unit which was installed in
1960. The boiler is a pressurized cyclone fired design as manufactured by Babcock &
Wilcox. In 1971, a Universal Oil Products weighted wire design precipitator was
installed. The precipitator had a specific collecting area (SCA) of 258 ft2/1000
acfm and the collecting plates had a 9-inch spacing. Over the years the mechanical
condition has deteriorated and the performance has degraded. Recent inspections
found severe warpage and misalignment of the collecting plates. An inspection in
March 1989 found that approximately 95% of the collecting plates had cracks at the
top of the plate on both the leading and trailing edges.
A study was performed to determine the most cost-effective method of restoring the
precipitator's performance. The poor mechanical condition was found to be the
primary cause of the poor performance and fly ash characteristics were a secondary
cause. It was concluded that the collecting plates and the transformer rectifiers
should be replaced. The study considered replacing the collecting plates
(replating) with the plates on 9-inch centers, 12-inch centers, and 16-inch
centers. Based on the study's conclusions, the Sibley Unit 1 precipitator will be
rebuilt with the collecting plates on 12-inch centers and rigid discharge electrodes
will be used. Field work is scheduled from January 29 to April 2, 1990 and Unit 1
will be returned to service on April 16, 1990.
This paper describes factors which lead to the decision to rebuild the Sibley Unit 1
precipitator with a wide plate spacing. Performance estimates are also presented
for the rebuilt precipitator when collecting the fly ash from the present coal and
from three candidate low-sulfur western coals both with and without gas
conditioning.
28-2
-------
PRESENT MECHANICAL CONDITION
Inspections performed in 1986 indicated that the Unit 1 precipitator was in poor
condition (I).
The collecting plates were severely warped.
The wire-to-plate clearance was less than 3 inches for approximately
one third of the collecting plates. The remaining plates had
wire-to-plate clearances less than 4 inches. The design clearance
was 4.5 inches.
Fly ash tended to build up in the inlet transition to a depth of
several feet. The fly ash buildup deflected the flue gas away from
the bottom making the collecting plates ineffective. A cold air flow
test with the ash in place indicated the standard deviation of the
velocity field (RMS) at the precipitator inlet was 19.7% and the RMS
at the outlet was 21.3%. An RMS of 17 to 20% is desirable for good
performance.
Gas in oil tests on the transformer/rectifier (T/R) sets showed the
possibility of sparking and the estimated remaining life was only
about 5 years (2).
The less than design wire-to-plate clearance resulted in poor performance.
Mechanical spacers were installed in 1987 in an attempt to restore the wire-to-plate
clearance and improve performance. The inlet field and the center field had spacers
installed at the second, fourth, and sixth rib of the seven rib collecting plate.
The outlet field had spacers installed on the second and the sixth rib. The
expected performance improvement did not occur. However, the mechanical spacers
made the collecting plates much harder to clean with the original rappers. Heavier
20-foot pound gravity impact rappers were installed in 1988. An inspection in March
1989 indicated that about 95% of the collecting plates had cracks in the upper
corners on both the leading and trailing edges. The cracks were a few degrees below
the horizontal and were from 6 inches to 18 inches in length (3).
PRESENT FLY ASH CHARACTERISTICS
The present fly ash was analyzed to determine if it contributed to the reduced
performance. Tests included in-situ fly ash resistivity measurements, laboratory
resistivity analysis, sulfur oxides sampling, fly ash mineral analysis, and
particulate distribution.
The various resistivity analyses were contradictory. The in-situ
resistivity analysis by the V-I method measured a resistivity of 1 or
2 x 1011 ohm-cm. The spark method measured resistivities in the
range of 2 x 109 to 7.8 x 1010 ohm-cm. The confidence level in the
results of the in-situ analysis was low because the sample layer was
thinner than desired. However, it was observed that the precipitator
performed as though the fly ash resistivity was about 5 x 1011 ohm-cm
(4).
28-3
-------
Sulfur oxides sampling indicated that the SC^ levels ranged from 2048
to 2876 ppm and the SOj levels were 3 to 10 ppm. The conversion of
S02 to SO, ranged from 0.1 to 0.4%. The S03 level increased from
3 ppm to 10 ppm following air heater sootblowing. The air heater was
retubed in 1988 with Cor-Ten tubes and it was speculated that the
cleaner air heater surfaces increased the levels of SO^ due to
catalytic action (4).
were typical for this
ohm-cm at 320°E without
Laboratory tests indicated resistivities
type of coal. The resistivity was 4 x 10
acid vapor. The measured resistivity at 320°F was 1 x 10 ohm-cm in
the presence of 5.3 ppm sulfuric acid. The ash was somewhat unusual
because breakdown occurred at 8 kV/cm rather than at a normal value
of 12 kV/cm. Calculation models predicted that the resistivity would
be about 8 x 10 ohm-cm at 320°F and 10 ppm of acid vapor. There
should be no difficulty in collecting fly ash with a resistivity of
8 x 108 ohm-cm at 320°F and 10 ppm of acid vapor (5).
Fly ash mineral analysis and particle size distribution tests were
performed using a proportionately blended sample from each of the
three hoppers in the direction of the gas flow. A similar sample
from Sibley Unit 3 was also analyzed for comparison. The Sibley Unit
3 precipitator was in good mechanical condition and the precipitator
performance has been satisfactory. Sibley Unit 3 is rated at 356 MW
and is also a Babcock & Wilcox pressurized cyclone boiler which
burned the same coal as Sibley Unit 1. Sibley Unit 3 has the same
precipitator manufacturer as Sibley Unit 1, but Sibley Unit 3 has a
smaller precipitator (178 SCA) than Sibley Unit 1 (258 SCA). The
results are listed below.
Ash Analysis, Weight Percent
Li?0
NAoO
K?0
MgO
CaO
FE203
AloO,
si623
P2°5
SO,
LOT
Particle Size Distribution
Mean Diameter, micrometers
Geometric Standard Deviation
<10 micrometers, percent
< 2 micrometers, percent
Sibley
Unit 1
0.02
0.90
2.69
1.01
3.03
10.57
17.80
60.70
0.74
0.16
1.60
3.7
6.1
2.7
68.9
12.9
Sibley
Unit 3
0.0
0.94
2.41
1.25
3.37
17.29
20.50
47.90
1.29
0.28
3.04
0.0
9.7
2.9
51.2
7.2
28-4
-------
The above ashes were chemically similar, but the smaller size of the
Sibley Unit 1 fly ash would make it more difficult to collect than the
Sibley Unit 3 fly ash. Based on past experience, if both precipitators
were in good condition, the Sibley Unit 1 precipitator should perform
about the same as the Sibley Unit 3 precipitator. The increased
difficulty in collecting the smaller Sibley Unit 1 fly ash was equal to
the difficulty in collecting the Sibley Unit 3 fly ash using the 30%
smaller Sibley Unit 3 precipitator. On this basis, the fly ash
characteristics were concluded to be a secondary cause of the observed
poor performance.
The various resistivity analyses suggested two different methods of improving the
precipitator performance by resistivity modification. First, the resistivity was
temperature sensitive and would decrease at lower flue gas temperatures. Second,
the resistivity could be lowered by flue gas conditioning. Sibley Station personnel
investigated these two methods by increasing sootblowing, removing the steam air
preheating coils from service and by injecting magnesium oxide. At the beginning of
the tests on Sibley Unit 1, the opacity was reduced from 40% to 26% by the injection
of 1 gal/hr of magnesium oxide. When the steam coil air preheater was removed from
service, the flue gas temperature decreased from 302°F to 285°F, and the opacity was
14%. At the end of the sootblowing cycle, the opacity had decreased to 10%.
Lowering the flue gas temperature was the most effective temporary method of
improving the Sibley Unit 1 precipitator performance. Injection of magnesium oxide
was less effective.
ESTIMATED PERFORMANCE WITH THE PRESENT MEDIUM SULFUR COAL
The performance of the Sibley Unit 1 precipitator was estimated using the SoRI/EPA
calculation model. The inputs to the calculation model were based on the observed
1988 values of average current density, average voltage, particle size, particle
distribution, and standard deviation of inlet velocity distribution. Other values
had to be based on assumptions. A 1978 efficiency test was used to establish the
assumed values for sneakage and for "good" values of average current density and
average voltage (5). The model was used to calculate the estimated collection
efficiency and the equivalent opacity. Equivalent opacity was used because a common
stack serves all three units at Sibley Station. The estimated collection efficiency
and predicted equivalent opacity for a variety of conditions is summarized below.
28-5
-------
Estimated Predicted
Collection Equivalent
Efficiency Opacity
Present Condition 92.85 35.9
Eliminate Inlet Transition 94.07 31.4
Fly Ash Buildup
Restore Good Alignment 97.70 13.9
Good Alignment and Eliminate
Inlet Transition Fly Ash Buildup 98.19 11.2
Ideal Alignment and Eliminate Inlet 99.20 4.7
Transition Fly Ash Buildup
The above results indicated that the poor present alignment was the main cause of
the poor performance. Elimination of the fly ash buildup on the inlet transition
had only a minor effect on precipitator performance.
The poor condition of the collecting plates and the limited remaining life of the
T/Rs lead to the decision to rebuild the Sibley Unit 1 precipitator. Methods of
improving the performance by resistivity modification had drawbacks. Lowering the
flue gas temperature by discontinuing air preheating was only feasible during the
summer because the lowered flue gas temperature was in the acid dew point during the
rest of the year and increased corrosion would result. Flue gas conditioning was a
costly annual expense. Furthermore, these methods did not address the main problems
of poor alignment and cracked collecting plates. The previous attempt to improve
performance by correcting the alignment using mechanical spacers was largely
unsuccessful. The cracks at the upper edges of the collecting plates could only be
temporarily repaired. Consequently, the only permanent solution was to rebuild the
Sibley Unit 1 precipitator.
ECONOMIC ANALYSIS
An economic analysis was conducted to determine the most economical spacing of the
collecting plates. Previous EPRI research and recent actual experience at Dale
Station of East Kentucky Power Cooperative had shown that precipitators with 9-inch
plate spacing could be rebuilt with 12-inch plate spacing at a considerable savings
without sacrificing collection efficiency. The following three options were
considered.
28-6
-------
Option 1 - Replate on 9-Inch Centers
This option was a duplicate of the original design. The precipitator
internals were demolished and new material was installed. The new
material included collecting plates and supports, weighted wire
discharge electrodes and supports, T/R sets, and automatic voltage
controls. It was assumed that the casing, casing insulation, weather
enclosure, rappers, and rapper controls would be reused.
Option 2 - Replate on 12-Inch Centers
This option was similar to Option 1. All the same components would
be replaced, but the wider spacing allowed the use of rigid discharge
electrodes which were considered more reliable than weighted wire
discharge electrodes. In addition, the T/R voltage was increased
from 45 kV to 55 kV because of the increased plate spacing.
Option 3 - Replate on 16-Inch Centers
This option was also similar to Option 1 above. All the same com-
ponents were replaced and rigid discharge electrodes were used. In
addition, the T/R voltage was increased from 45 kV to 70 kV because
of the increased plate spacing. 70 kV was considered a commercial
limit on T/R size which limited any further plate spacing increase.
The economic analysis was based on the expected changes in the present value of
revenue requirements for the capital cost of each option. The results are
summarized below.
Present Value of
Revenue Requirements
Option 1 - 9-Inch Centers $1,589,000 +$194,000
Option 2 - 12-Inch Centers $1,453,00 +$ 58,000
Option 3 - 16-Inch Centers $1,395,000 Base
The economic analysis indicated that Option 2 - Replate on 12-Inch Centers and
Option 3 - Replate on 16-Inch Centers were essentially the same cost. However, the
12-inch spacing was chosen because there was a concern about the actual performance
of a precipitator with plates on 16-inch centers when collecting the fly ash from a
cyclone-fired boiler. The calculation models were based on data from pulverized
coal-fired boilers. The effect of the finer fly ash from a cyclone-fired boiler was
not accounted for in the calculation model.
ESTIMATED PERFORMANCE WITH MEDIUM-SULFUR COAL AND CANDIDATE LOW-SULFUR WESTERN COALS
Table 1 presents the estimated performance of the rebuilt Sibley Unit 1
precipitator. Performance was estimated for both the present medium-sulfur coal and
for three candidate low-sulfur coals, which were being considered as a method of
complying with future S0£ emission limits. The estimates were made by several
pollution control equipment manufacturers (6, 7, 8).
28-7
-------
Based on estimated performance shown in Table 1, the rebuilt Sibley Unit 1
precipitator would comply with present particulate emission limits of 0.12 Ib fly
ash/106 Btu or 97.1% with either the 9-inch or 12-inch plate spacing. If any of the
candidate low-sulfur western coals were burned, the estimated collection efficiency
of the rebuilt Sibley Unit 1 precipitator would decrease to between 91 to 97%. The
rebuilt Sibley Unit 1 precipitator probably would not comply with particulate
emission limits and further measures would be required. The addition of a flue gas
conditioning system would increase the estimated collection efficiency to about
98.8% allowing compliance with the present particulate emission limits. If further
reductions in particulate emissions were necessary, the precipitator casing could be
expanded to add a fourth field. The estimated collection efficiency would increase
to about 99.7% if a fourth field was added and SO^ conditioning was used.
PRECIPITATOR REBUILD WORK
The Sibley Unit 1 precipitator rebuild was competitively bid, and the contract was
awarded to Research Cottrell on September 22, 1989. Erection work was scheduled to
begin on January 29, 1990 and complete on April 2, 1990. The total time from
contract award to scheduled completion of field work was 27 weeks.
The first 9-foot field was mechanically split into a 3-foot and a
6-foot field. Space was allocated for the installation of a future
T/R and the necessary cold roof modifications were included in the
rebuild work. It is anticipated that the sparking tendency would
increase if low-sulfur coal was burned. Having a 3-foot field and a
6-foot field on the inlet would be more electrically stable than a
single 9-foot field. The inlet field of the smaller SCA Sibley Unit
3 precipitator was split in a similar fashion and performance loss
due to sparking was significantly reduced.
The number of specified collecting plate rappers was reduced from two
rappers per five collecting plates to a maximum of one rapper per two
collecting plates. Low-sulfur coal ash has been harder to rap off
the plates than the ash from other coals. In addition, the unit fly
ash loading on the collecting plates increased. There was less
surface area to collect the same amount of fly ash.
The automatic voltage controls included provisions for intermittent
(pulse) energization. Recent EPRI work has suggested substantial
power savings by using intermittent energization without sacrificing
particulate collection efficiency.
The precipitator manufacturer found it necessary to add a perforated
plate at the precipitator outlet to correct an apparent velocity
maldistribution which was indicated by the cold air flow test. The
original design did not have a perforated plate on the precipitator
outlet.
28-8
-------
SUMMARY
The condition and the performance of the Sibley Unit 1 precipitator had deteriorated
over the last several years and corrective actions were required. Inspections in
1986 and 1987 found severe warpage and misalignment of the collecting plates. An
inspection in March 1989 found that approximately 95% of the collecting plates had
cracks at the top edge of the plate on both the leading and trailing edges.
A study was performed to determine the most cost-effective means of replating the
Sibley Unit 1 precipitator which would comply with the existing particulate emission
limits. The effect on the precipitator of a switch to low-sulfur western coal was
also addressed. The study compared replating on 9-inch, 12-inch and 16-inch
centers. It was found that the Sibley Unit 1 precipitator could be replated on
12-inch centers at a considerable saving compared to replating on 9-inch centers
without sacrificing performance. Based on the present value of revenue
requirements, the savings amounted to $136,000 (9%) compared to replating on 9-inch
centers. Replating on 16-inch centers also showed promise, but was eliminated
because of the limited experience with 16-inch centers and concern that the
calculation model did not account for the finer fly ash from a cyclone boiler. The
rebuilt Sibley Unit 1 precipitator would not comply with particulate emission limits
if low-sulfur coal was burned and some form of flue gas conditioning would probably
be required. The rebuild work is in progress and is scheduled to be completed on
April 2, 1990.
ACKNOWLEDGMENTS
The guidance and advice of Dr. Ralph Altman EPRI is gratefully acknowledged. Field
testing and the calculation of estimated performance was provided by Dr. M.
Anderson, APCO Services. Performance estimation and budgetary information were
provided by R. Kimberl, Environmental Elements; R. A. Mastropietro, Research
Cottrell; and Dr. H. V. Krigmont, Wahlco.
REFERENCES
1. Personal communication with J. Taylor, Field Service Associates.
2. Personal communication with L. Dykes, Westinghouse Electric.
3. Personal communication with C. Craig, PrecipTech.
4. Personal communication with Dr. M. Anderson, APCO Services.
5. Personal communication with Dr. R. Bickelhaupt, Clean Air Engineering.
6. Personal communication with R. Kimberl, Environmental Elements.
28-9
-------
7. Personal communication with R. A. Mastropietro, Research Cottrell.
8. Personal communication with Dr. H. V. Krigmont, Wahlco.
28-10
-------
Table 1
PREDICTED PRECIPITATOR EFFICIENCY
FOR VARIOUS CONDITIONS
SIBLEY UNIT 1
MISSOURI PUBLIC SERVICE
Manufacturer A
Current Medium-Sulfur Coal
Any of Three Candidate Low
Sulfur Western Coals
Manufacturer B
Current Medium-Sulfur Coal
Candidate A Low-Sulfur Coal
Candidate B Low-Sulfur Coal
Candidate C Low-Sulfur Coal
Manufacturer C
Worst Candidate Low-Sulfur
Coal
Worst Candidate Low-Sulfur
Coal with S03 Flue Gas
Conditioning
Predicted Precipitator
Efficiency
Percent
9" 12"
98.7 98.6
97.7 97.8
16"
98.7
97.7
99.1
93.4
91.0
93.25
93.5
98.8
96.9
93.4
28-11
-------
Table 2
Manufacturer
Design Flow
Design Temperature
Collection Area
Specific Collection
Area
Plate Description
Average Velocity at
Design
Number of Fields
Number of Chambers
T/R Rating
Number of Collecting
Plate Rappers
Number of Discharge
Electrode Rappers
COMPARISON OF THE ORIGINAL
AND THE REBUILT PRECIPITATOR
SIBLEY UNIT 1
MISSOURI PUBLIC SERVICE
Existing Design
Universal Oil Products
188,000 acfm
366°F
48,600 sq ft
258.5 sq ft/1000 acfm
(31) plates per each of
3 fields. Each plate 9 ft
x 30 ft 18 Ga 9 in.
plate spacing.
4.64 ft/sec.
1st field: 800 mA, 45 kV ave
2nd field: 800 mA, 45 kV ave
3rd field:1200 mA, 55 kV ave
(replaced)
36
12
Replacement Design
Research Cottrell
204,000 acfm
335°F
36,018 sq ft
177 sq ft/100 acfm
(25) plates per each of
3 fields. Each plate 9 ft
x 30 ft 18 Ga 11.25 in.
plate spacing.
5.04 ft/sec.
(provision to split
inlet field into
two fields)
500 mA, 55 kV ave
750 mA, 55 kV ave
1000 mA, 55 kV ave
36
16
28-12
-------
FLUE GAS FIELD STUDY, MODEL STUDY, AND POST STUDY
REVIEW TO IMPROVE THE PERFORMANCE OF A CHEVRON ESP
AT DUKE POWER'S BELEWS CREEK STATION
Scott L. Thomas
Duke Power Company
Charlotte, N. C.
Lawrence A. Zemke
CAE Diagnostic Services
Palatine, II.
ABSTRACT
The units at Belews Creek Station have always operated within the North Carolina
State environmental regulations. However, in 1986 a new regulation (Annual
Average Opacity) was introduced in conjunction with emission limit changes. The
Belews Unit #1 exceeded its Annual Average Limit in the fall of 1987 for a brief
period. It was apparent that improvements must be made to insure consistent
compliance. Poor gas flow distribution has long been the trade mark of the
'Chevron' design precipitators. The Belews Creek 'Chevron1 ESPs were suspected
to have such problems. This paper describes the air flow field test, the model
study, and the subsequent review that was performed on unit #1. Both the field
test and the model study verified the suspected gas flow problems. However, an
unexpected situation was discovered. The high flows discovered at the airheater
side of each inlet chamber in the field study were not apparent using standard
modeling techniques. It was discovered that the 2" trailing edge pipe support
acted like a turning vane and was directing the gas to the high flow side of
each chamber. The completion of the recommended improvements from the post test
review have resulted in improved opacities and continued compliance. Duke Power
has completed two more gas flow studies and in both cases field tests were
performed prior to the model study to give actual gas flow conditions to help
verify the model results.
29-1
-------
BACKGROUND
Duke Power's Belews Creek Station consists of two 1150 mw units each with two
150 SCA 'Chevron' cold side precipitators (Fig. A shows a plan view of Unit #1).
Due to low sulfur bituminous fuel required to meet state SO,, emission
regulations, SO, gas condition systems were installed to insure compliance with
state particulate regulations. Historically, the units have maintained
emissions well within state limits, however, in 1986 the state of North Carolina
implemented an annual average opacity limit (AAO). Unit #1 exceeded this limit
for a brief period in the fall of 1987. Duke personnel were already
investigating cost effective improvements which would insure continuous
compliance. One potential improvement was to modernize the gas flow
distribution and sneakage prevention through each precipitator. During unit
start ups and load changes, momentary opacity excursion occur. Also the unit
has rapper reentrainment problems. This is compounded by the use of gas
conditioning. We were hoping by improving the gas flow distribution we could
'desensitize' the unit. This paper describes the work completed in this area.
Precipitators with the 'Chevron1 designs built in the early 70's can have gas
flow problems. This is primarily due to the change in flow direction at both
the inlet and outlet faces making cross flows possible (i.e. in collection
hoppers, inspection walkways). During inspections by plant personnel evidence
of cross flows were noted in the hopper area. An additional field inspection by
Research Cottrell representatives, utilizing a smoke test, verified not only the
cross flows but also gas flow sneakage. We felt the next step was to schedule a
velocity traverse at the precipitator inlet and outlet faces. This would
provide the data necessary to determine if flow distribution problems existed
and where. The test, conducted with the unit off line, was performed by
personnel from CAE Diagnostic Services (formerly Boyle Laboratories, Inc.) in
July 1987.
FIELD INSPECTION AND TEST
An inspection of the existing condition of the gas flow path of Belews Creek
Station, Unit 1 was conducted by CAE personnel. As part of this inspection the
gas distribution devices and major structural members affecting flow in the
precipitator inlet flue between the air preheater and the first collection field
of the precipitator were identified. A similar inspection was conducted in the
outlet plenums to each I.D. fan. The precipitator inlet ductwork contained the
following (Figure A):
29-2
-------
The 'Chevron' inlet plenum was divided down the length to provide two
separate duct sections, one to each precipitator.
A set of 59 ladder type vanes were located immediately upstream of the
diverging inlet diffuser to each precipitator. Each vane was 2'-0" in
length with a nominal 2" stiffening pipe vertically on the leading and
trailing edge of each vane.
Three solid division walls extended from each precipitator chamber column
upstream across the inlet diffuser expansion to form separate flow paths to
each precipitator chamber.
Three perforated plates were located in each inlet section. These plates
were nominally 50% in free area and were on spaces of 2'-0", 3'-0", and
3'-0" from the upstream bend line of the precipitator inlet diffuser.
The third perforated plate in the direction of flow was blocked to reduce
the free area across specific sections of the chamber opening (Figure B).
This blockage was accomplished by the typical method of stitch welding flat
stock horizontally and vertically over the holes of the plate.
On each side of the 'Chevron1 duct center line and in each chamber the
lowest free area section of the third plate was closest to the air
preheater side of the precipitator. This 3'-0" wide section was 19% open.
The second section, 11'-6" wide was 25% open. The third section was 7'-0"
wide and was nominally 30% open. The fourth section of 8'-0" was 38% open.
Only the last 6'-0" of the plate remained at the unmodified free area of
50%.
Each precipitator outlet consisted of two ducts, one to each I.D. fan. Each
outlet duct contained the following (Figure A):
A set of seven large turning vanes were located immediately downstream of
the precipitator flange line.
A very large pile of ash was found in each outlet duct. The height of this
pile was estimated at ten to twelve feet at the maximum. The pile sloped
downward to both the precipitator outlet flange and down to the I.D. fan.
29-3
-------
A gas velocity traverse was conducted across the inlet and outlet of the 1-B
precipitator shortly after the inspection. The traverse was conducted with
ambient temperature air flow and was compared to the current I.G.C.I, criteria
for uniform gas velocity distribution. The results of this flow traverse
indicated very poor flow quality entering the precipitator (Figure C). Overall
76% of the readings taken were less than 1.15 times the average velocity, and
93% were less than 1.4 times the average velocity (IGCI criteria = 85% and 99%
respectively).
In each chamber high gas velocities were noted to the air preheater side of the
inlet in spite of the relatively low free area created by the blockage pattern.
The average velocity in the gas passages did not increase smoothly from side to
side across the chamber face but in a jagged profile. This high level of
turbulence was related to the presence or absence of a blockage strip in front
of the particular lanes traversed.
While flow patterns near the outlet of the precipitator showed improvement,
distribution remained below industry guidelines.
THE MODEL STUDY (Early 1989)
The second phase of the program was the construction and testing of a l/12th
scale three dimensional model of the gas flow path. The model included the air
preheater shape, the existing gas distribution devices with blockage and
structural simulations, the precipitator and ductwork to the I.D. fans.
The ladder vane detail was modeled by using thick material with a rounded
leading and trailing edge (Figure B). In this way gas flow was impacting on a
simulation of the rounded pipe that exists in the field.
The variable porosity third perforated plate in the direction of flow in the
inlet plenum was modeled by a 50% free area plate that was blocked with strips
similar to the method used in the field. The lowest free area was nearest the
air preheater side of each precipitator chamber (Figure B).
The first flow test conducted in the precipitator model simulated the conditions
of the field test. The model inlet test plane was located 2" model (24"
prototype) downstream of the leading edge of the first collecting field.
Results of this test demonstrated very poor gas uniformity in the first field
due to the blockage on the first perforated plate. A comparison of the vertical
29-4
-------
average velocity profiles between field test and model test showed similar shape
(Figure C). Higher than average flow was noted near the top of the
precipitator.
A review of the horizontal profiles for each chamber demonstrated higher flow to
the stack side of the precipitator, lower flow to the air preheater side. This
pattern was the reverse of the field date. A check was made to determine that
the blockage pattern and side to side orientation of the blockage of the third
plate was correct for both the model and the prototype. In addition, the
numbering of the gas passages side to side in both field and model were
verified.
The next step in the model test program was to remove the blockage from the
third perforated plate to reduce turbulence. Once this was done the horizontal
flow pattern in the model was uniform across the face of each chamber. This
demonstrated that the flow pattern in the model was distorted by the blockage
pattern, but did not identify the cause of the flow pattern reversal between
model and field test.
During this time frame further research provided results of field and model flow
tests from prior contracted testing at the time of construction and start up.
The initial model study for the unit did not include any blockage on the third
inlet perforated plate and showed a uniform horizontal distribution. This
confirmed the second data set taken in the model. Secondly, the first field air
flow test showed a side to side distortion with high flow to the air heater side
of each chamber of the unit. The blockage pattern of the third perforated plate
was provided to improve the noted field air flow pattern and a second flow test
was conducted which demonstrated a degree of improvement. This confirmed the
current field test data horizontal pattern.
On the basis of this analysis some element that greatly affected the inlet flow
pattern of the precipitator was missing from the model simulation. The first
choice was the ladder vane simulation used in the model. One section,
comprising one chamber of the inlet plenum of the model was modified to include
complete detailed simulation of the leading and trailing edge structural member
pipe geometry. This included the addition of round pieces installed at the
leading and trailing edge of each vane (Figure D).
The result of this modification of the model demonstrated a dramatic side to
side flow shift when compared to the uniform flow of previous testing. The flow
shift was to higher velocities to the air preheater side of the chamber
(Figure D).
29-5
-------
Testing then continued to recommend a minimal field modification while providing
support for ladder vane elements. To identify the effect of each structural
member effect in the model the trailing edge pipe was removed. A uniform
horizontal velocity profile was again generated. This limited the degree of
field modification of the ladder vanes needed to correct the field problem. A
modified form of stiffener consisting of an angle welded to the stack side
surface of the ladder vane trailing edge completed this phase of the testing
program (Figure E). Data for this stiffener configuration showed no additional
distortion of the horizontal flow profile for placement of the new member
(Figure E).
During this testing program modifications were also made to the outlet ducts.
An outlet perforated plate was designed with a variable porosity to improve
balance and flow pattern at the outlet flange. This device also was used to
further reduce flows at and below the bottom of the precipitator outlet opening
(Figure F).
To permanently eliminate the large accumulations of ash in each outlet duct a
false floor was designed to modify the flow path of the flue gas by increasing
gas velocity near the outlet duct floor. This method required a one-time
construction so as to avoid the installation of additional equipment to service
and maintain.
The end result of the model study was the design of modifications and additional
devices that provided gas velocity distribution uniformity in excess of the
current industry standards. Additional testing documented uniform flow without
the installation of the false floor in the outlet duct.
The recommended configuration of the ladder vane modification and the majority
of the recommended modifications were then incorporated in the scope of work for
the spring 1989 outage.
POST TEST REVIEW
During the implementation of the model phase, several conversations between Duke
and CAE took place. The goal was to reach cost effective methods to correct the
gas flow problems. The most important of these discussions was to find the
reason for the discrepancy between the field and model inlet flow data. As
previously discussed the pipe used to support the trailing edge of the ladder
vane was turning the flow back towards the air preheater side of each chamber.
With out having the field data available for the review, it would have been safe
to assume the inlet flows demonstrated in the model were in line with the real
29-6
-------
world. The recommended fix to eliminate the inlet flow problem came from that
discussion. After all the objectives were reached in the model phase (this
included eliminating the ash fallout in the outlet plenum), Duke personnel
prioritized the recommended fixes for implementation in Unit #1 during the
Spring 89 outage.
The primary concern as with any upgrade was cost and outage time. The
installation of the false floor was delayed for a future outage due to the work
hours required. The material was ordered for all the remaining recommendations
and labor was arranged to complete as many of recommendations as outage time
would allow. The goal was to complete the most critical items (i.e. the
modification of the ladder vane, removing the blockage strips, installing the
outlet perforated screen) and stay within a tight budget. As it turned out, the
installation of the recommendations went better than expected and everything
(with the exception of the false floor) was completed during the outage. In
addition to the planned work, several horizontal turning vanes were attached to
the outlet vertical vanes and angled towards the plenum floor. This idea
originated from work a consultant had done at another site and was discussed
with CAE as a possible (inexpensive) way to minimize the ash buildup. They
verified the vanes would not impact gas flow distribution and concurred with
their installation.
IMPACT ON EMISSION
As expected, the stack opacities following the outage were low. However, the
opacity levels increased over the next few months. This was not unexpected.
The SO., gas conditioning system was upgraded during the outage along with
boiler ignition upgrades. Some fine tuning was necessary to optimize ESP
performance. Also, the unit was cycled and was taken off line several times
during the summer. By September the stack opacity had stabilized in the low to
mid teens at full load. This was an improvement over preoutage opacities. An
annual stack test was performed September 12th, and the results showed a slight
reduction in emission as compared to the preoutage test. However, more time
will be needed to more fully evaluate the impact of the modifications on the
unit's annual average opacity.
The second and most important change noted was during load changes and air flow
adjustments. The degree of opacity change was lower than experienced prior to
the precipitation modification. To quote plant personnel, 'the precipitator is
29-7
-------
less sensitive' than before. Rapper reentrainment was also less apparent. The
reduction in reentrainment could be the result in minimizing high velocity zones
within the ESP.
FUTURE PLANS
In conjunction with the flow test and during the model phase, temperature
stratification and recommended fixes were determined. Future plans might
include implementing these recommendations. The ash fall out in the outlet
plenum is still a problem. Initially, the small outlet horizontal vanes did
minimized the fall out. Installing the false floor, however, may still be
appropriate. Lastly, our plan is to have a post field test performed so as to
evaluate what impact the gas flow modifications have made to unit #1 gas flow
distribution.
CONCLUSION
The phased approach was critical in the success of the project. Without the
availability of field date, the model would not have been as closely
representative of the field situation. This does not minimize the importance of
good modeling technique and experience. In addition, larger scale models (i.e.,
1/12 or larger) are advantageous to get the best representation of model vs.
real world.
The problem with the inlet flows was noted in the original OEM model study. A
reasonable attempt was made to minimize the impact, however, the cause of the
problem (the ladder vane support) was not identified.
Emission improvements were noted. The most important was reducing the
sensitivity of the ESP to changes (i.e., load swings, changes in gas
conditioning).
29-8
-------
ro
<£>
in
TO UL FAN
U. FAN
runs FHM
MOMMY
PRDCATER
GENERAL ARRANGEMENT PLAN
3ELEVS CREEK STATION
UNIT 1
FMI
LAMKR
ELEVATIDN DP
PRECIPITATDR INLET DUCT
FIGURE A
CAE IKAGNDST1C SCXVItZX
-------
no
-------
CAB DIAGNOSTIC SERVICES
PROJeCT 4531
TEST I OA AHD GRID 01
DOIX POKER COKFAMY
BEX IMS Caen STATION
BUTT IB
DRKMKX OF FBKC1PITA1O8 CHAMBER 2
PICLC AH FLOM TEST
100% PUS
CAE DIAG»STIC SERVICES
PROJECT 1531
TEST I «1 AHD GRID 01
DDXZ POMES COMAtPf
BELEHS CREEK STATION
DHTF IB
EHTZABCE OF PBZCIP7.TATOR CHAMBER 2
EXISTING GAS DEVICES III MODEL
100* FLO*
GA£ VELOCITY DIAGRAM
LANE HDMBCRS
2 S
370 330
360 340
320 290
210 330
320 3 SO
350 3 BO
35D 3*0
340 360
339 3 BO
330 310
320 J20
280 320
2SC 330
300 30«
320 31-0
360 340
-25 -20
* ji
400 350
311 270
351 310
400 370
340 360
«DC 320
330 360
370 2EO
330 360
380 380
340 340
310 310
360 350
330 370
3iO 350
410 350
-IS -21
14 17
510 420
450 400
140 360
190 430
450 440
430 420
120 350
460 410
400 420
410 430
380 420
340 360
440 390
420 410
430 430
460 490
1 -4
20 23
430 470
270 480
300 340
3 tO 400
380 440
430 460
90 270
290 480
400 430
440 440
450 430
360 410
310 350
300 330
280 400
410 400
-20 -5
26 29
460 180
S<0 190
440 170
390 260
460 310
5BO 390
290 1
500 460
520 340
490 490
520 340
370 370
400 240
490 210
500 310
530 350
9 -33
32 35
COO 620
560 540
420 540
460 560
480 550
570 600
210 370
510 530
600 560
590 580
510 580
390 570
360 490
440 570
510 550
590 530
13 27
-•—
-^
38 41
606 S40
650 400
550 360
550 400
620 520
600 340
5 BO 340
600 420
600 460
SBC 480
600 550
610 480
490 390
560 430
640 360
520 320
36 -1
— — v
\
^ >
44
590
530
600
640
600
640
340
610
600
560
570
570
650
640
550
360
31
f—
•t
46 IDE?
620
620
560
590
590
610
350
630
610
630
560
580
540
560
420
420
29
^ ^ „, s./ z^zrVzi-
9
D
-7
-1
5
10
-26 - : ;
5
8
10
5
-4
-7
-3
-3
-1
t
<
\
S,
„ • "
•
\
1
{
\
ft
'
-25* 0 «5»
T- +251
GRAPES SB(»
. j » DEVIATION
PEON AVC.
VELOCITY
J- -25
GAS VELOCITY DIAGRAM
LANE NUMBERS
25 27
516 537
507 526
£21 519
409 448
533 562
557 585
571 612
543 635
510 573
574 616
562 594
S34 593
463 539
50 a 564
519 572
SS3 589
22 32
— *•
29
497
447
422
390
428
460
467
458
458
484
507
464
409
490
512
544
8
^r^
31
51 S
497
474
412
508
528
533
515
500
545
537
475
452
517
549
563
18
^
33
432
430
306
313
326
272
416
443
249
324
385
336
303
276
376
425
-18
35
452
430
410
275
460
420
542
459
514
360
447
331
396
475
480
477
1
37
453
406
416
324
466
526
521
509
497
519
534
475
412
til
473
482
8
-^ —
39
460
372
369
287
341
300
397
343
340
342
356
370
232
342
449
429
-17
41
495
425
314
224
323
275
451
349
350
340
262
370
134
378
420
467
-19
43
542
488
435
399
474
427
470
461
311
372
313
439
234
427
461
513
-1
45 tDBV
451
371
251
287
268
233
301
200
229
262
217
189
169
236
343
417
13
4
-6 t>'
-20 • ; _
-i
-3 •
12
4
-4 ^
0
0
-3 ,i
-21 ' '
-2 ' •
10
16
-36 -254
: • V^ \ sf \
f
1
%
. '
V
0 «3»
t GRAPHS SBCK
FKOM AVG.
VELOCITY
H_K.S.- 26.65 *
AVERAGE VELOCITY- 430.4 F.P.N.
B.M.S.- 24.37 %
AVERAGE VELOCITY- 428.9 F.P.K.
FIGURE C
-------
DUKE POWER COMPANY
BELEWS CREEK STATION
UHIT IB
ENTRANCE OF FRECIPITATOR CHAMBER 2
INTERIM DEVICE CONFIGURATION
190% FlOH
GhB VELOCITY DIAGRAM
25 27
29
NDMBSR6
31 33
3S 37
39
41 43 45
16«
138
123
120
170
189
186
194
199
193
192
141
158
163
203
204
211
183
164
164
214
213
231
236
221
232
227
203
148
175
226
322
261
230
209
119
237
234
244
264
273
260
275
25S
178
219
269
331
281
292
I SO
844
313
319
344
358
349
346
346
318
241
287
359
405
302
311
294
288
352
361
36S
373
356
371
370
343
246
280
317
407
351
330
324
257
346
384
362
330
375
373
360
338
228
311
342
350
383
3B5
386
351
391
42 8
461
461
439
445
417
413
353
391
404
407
419
411
411
35S
422
471
477
462
459
471
433
436
344
367
435
421
442
429
422
399
447
472
516
520
467
492
500
468
388
423
464
452
432
430
422
393
443
476
510
514
516
483
482
475
395
401
425
433
48S
474
486
432
S14
540
603
606
577
575
540
535
456
418
454
451
-51 -40 -29 -10 -5
-4
16 21
30
26 45
»DEV
-4
-6
"9
-17
-1
6
11
12
10
8
1
-19
-11
1
9
«
-23»
+25%
- 0
GRAPHS SHOW
* DEVIATION
FROM AVQ.
VELOCITY
R.M.S." 31.85
-25
AVERAGE VELOCITY- 351.7 F.P.M.
TLQV FROM
MR PREHEATES
. I \ \ \ '
\ \ \ ^ "~~
MR
HEATEB
DETAIL 1
PLAN VIEW
LADDER VANE
FINAL MODEL SIMULATION DP LADDER VANE
FIGURE D
CISC DIMiHOSTIC SERVICES
29-12
-------
DUKE POWER COMPANY
BELEWS CREEK STATION
UNIT IB
ENTRANCE OF PRECIF1TATOR CHAMBER 2
RECOMMENDED GAS DISTRIBUTION DEVICES
100% FLOW
GAS VELOCITY DIAGRAM
LANE NUMBERS
25 27 29 31 33 35 37 39 41 43 4S
342 318 331 345 335 347 368 377 365 349 397
339 329 336 361 364 340 373 37ft 382 362 401
33B 316 31S 345 332 354 377 375 374 349 407
306 303 310 317 343 319 355 373 371 346 366
366 372 377 382 398 393 391 418 404 380 431
401 364 374 3SO 426 411 451 456 428 414 461
414 402 416 433 466 452 491 485 476 458 510
399 414 415 439 474 465 503 488 495 457 513
414 372 404 427 455 443 478 457 441 460 502
407 397 401 429 447 458 462 463 467 444 511
392 391 427 434 448 451 427 421 458 444 477
3 88 360 404 403 432 430 420 432 439 436 466
331 312 338 366 366 387 384 392 381 348 390
319 342 379 405 372 430 408 416 407 341 400
375 394 428 434 419 430 428 444 434 371 439
437 422 455 458 450 426 400 403 417 395 447
-8 -10 -5 -1 1 1 4 5 4 -2 10
-, - - S*
.
%DEV
-13
-11
-12
-16
-3
3
13
14
9
10
7
4
-10
-5
4
6
-2
G
= - 0 *
-. -55
«
j
/
<
s
^
\
\
\
I
r
1
r
1
}
T
5t 0 +25»
RAPHS SHOW
DEVIATION
FROM AVG.
VELOCITY
R.R.S.- 11.S6
AVERAGE VELOCITY' 403.7 P.P.M.
r
FLOW r*OM
AIR PKOCATEK
SEE 0ETML1^> ^ \ \ \ \ ^ ^
A"K\ ^ ""' "" — - ""
r
Jr"'
JT~
1
%
tTACK Stt£
nr VMC -i|-
1 1/8*
Afifi I* ANCLE
Mft I— rw SUPMIRT IT
HCATC* RC9UWEB
RECOMMENDED LADDER VANE
FIGURE E
CAE
29-13
-------
GAS TIGHT PARTITION
MODIFY BOTTOM
OF PLATE TD
REDUCE PILING
CHECK AMD
UPGRADE BAFFLES
IN PREC1PITATOU
MODIFY STRUCTURAI
OF LADDER VANES
\ \
\ CHAMBER 2 \
\ \
TO LB. FAN
INSTALL FALSE FLQOR
INSTALL DISCHARGE
PERFORATED PLATE
INSTALL FALSE FLOOR
RECOMMENDED MODIFICATIONS FDR BELEWS CREEK
FIGURE F
CAE
SERVICES
-------
EXPERIENCE WITH DUAL FLUE GAS CONDITIONING OF ELECTROSTATIC PRECIPITATORS
H. V. Krigmont
E.L. Coe, Jr.
Wahlco Inc.
3600 W. Segerstrom Ave.
Santa Ana, Ca 92704
ABSTRACT
Nearly 300 boiler units world-wide are equipped with Flue Gas Conditioning Systems
(FGC) to enhance the performance of the electrostatic precipitators (ESP's). FGC
systems using gaseous sulfur trioxide (S03) injection are predominant.
However, over the past few years a new technology, Dual Flue Gas Conditioning (Dual
FGC) has been developed. Dual FGC systems, as the name implies, use two feedstocks,
sulfur trioxide and anhydrous ammonia (NH3) injected independently in controlled
stoichiometric ratios for optimum electrostatic precipitator performance.
Applications of Dual FGC systems range from enhancing sulfur trioxide utilization
and decreasing rapping losses of precipitators with high velocity and low aspect
ratios to conditioning fly ashes with high amounts of unburned carbon.
The theory of and experience with Dual FGC systems for applications on boiler units
with various sizes is described.
30-1
-------
EXPERIENCE WITH DUAL FLUE GAS CONDITIONING OF ELECTROSTATIC PRECIPITATORS
INTRODUCTION
Introducing additives into flue gas to enhance the performance of electrostatic
precipitators is well known and has been applied in commercial installations since
the early part of this century [1].
Over the past 20 years, increasingly restrictive particulate and sulfur dioxide (S02)
emission limits have been mandated by new Governmental regulations. To reduce S02
emissions many utilities have switched to lower sulfur coals. A consequence of this
change is reduction of electrostatic precipitator performance due to back corona
effects ironically just at the time increased efficiency is needed. Moreover, many
precipitators are marginally sized with high face velocities and low aspect ratios.
Alternatives available to utilities facing these problems are to replace or rebuild
their existing precipitators, to add collecting area (units) or to overcome the
difficulties resulting from undersized precipitators using variable quality coals
by installing Flue Gas Conditioning.
Flue gas conditioning of fly ash relies on one or more of the following effects:
Modifying the surface electrical conductivity of the dust;
• Increasing the inter-electrode space charge, and/or
. Increasing dust cohesivity to reduce losses during rapping.
CONDITIONING BY SULFUR TRIOXIDE
The most common conditioning agent used for modifying high fly ash resistivity is
sulfur trioxide [1, 2]. Two major factors control ash resistivity: the sulfur
content of the coal, and the overall elemental composition of the ash.
When coal is burned, more than 95 percent of the sulfur appears in the flue gas in
the form of sulfur dioxide. The kinetics of sulfur oxidation in boiler flue gas
do not allow more than a small fraction of the oxides to appear as sulfur trioxide
[3]. When the temperature of flue gas drops to around 300 °C (about 572 °F), a
significant fraction of the sulfur trioxide gas reacts with water vapor to produce
sulfuric acid vapor. This process is essentially complete at temperatures around
150 °C (about 300 °F) where electrostatic precipitators normally operate. The
sulfuric acid vapor is adsorbed or condensed on an otherwise poorly conducting fly
ash surface and directly participates in the electrical conduction process. At very
low concentrations of sulfuric acid vapor conduction in fly ash is principally
influenced by the charge-carrying ability of the alkali metal ions as affected by
the interaction of water vapor and the ash surface. At high concentrations of
sulfuric acid vapor, the conduction process is principally controlled by the adsorbed
acid. At intermediate concentrations of acid vapor, both mechanisms contribute to
the conduction process. Thus, in a strict sense, in cold-side precipitators resist-
ivity modification results from sulfuric acid, rather than sulfur trioxide, effects.
30-2
-------
The acid may, however, react with basic constituents of the ash and thus undergo
conversion to a poorly conducting layer of sulfate salts. Calcium oxide, a fairly
common component of ash from many coals, is a probable cause of acid neutralization
and its nullification as a conductor. Thus two coals similar in sulfur content may
produce similar concentrations of sulfur trioxide, but the one containing a more
alkaline ash is likely to have a substantially higher resistivity.
To summarize, sulfur trioxide conditioning is effective at flue gas temperatures
below about 204 °C (400 °F) where surface conduction prevails. At higher tempera-
tures, effectiveness of sulfur trioxide conditioning is decreased because the ash
resistivity is partly or wholly determined by volume conduction, and also because
the fraction of the available acid which condenses on the particles is reduced due
to vapor pressure and equilibrium characteristics.
Sulfur Trioxide Generation
The most commonly used method for generating sulfur trioxide is catalytic conversion
from sulfur dioxide. This method can use liquid sulfur dioxide as a feedstock;
however, the majority of installations burn elemental sulfur to produce gaseous
sulfur dioxide which is then catalytically converted to sulfur trioxide. Heated
ambient air is normally used as a carrier gas in the injection system with the sulfur
trioxide being a small percentage of the carrier volume. The sulfur trioxide flow
is modulated to match boiler load changes, fuel variations or other variables.
Injection into the flue gas stream may be either upstream or downstream of the air
heater, depending on convenience and accessibility.
DUAL FLUE GAS CONDITIONING
A Dual Flue Gas Conditioning system simultaneously and independently injects two
conditioning agents, sulfur trioxide and anhydrous ammonia, into the flue gas stream.
These agents work together in a variety of ways. Although the exact mechanisms
responsible for improved precipitator performance are not completely defined, they
appear to include such phenomena as space-charge effects, agglomeration and increased
cohesiveness of the fly ash. Additionally, some evidence suggests that injection
of ammonia promotes improved attachment of available sulfuric acid to fly ash
particles, making it more effective for resistivity control.
Chemical Processes
Elementary explanations of Dual FGC effects have assumed that the chemical compounds
formed when ammonia is injected, either concurrently with injection of sulfur
trioxide or when the latter is formed by "natural" (combustion) processes in
sufficient quantities, are only ammonium bisulfate NH4HS04 or, if the stoichiometric
ratio is appropriate, the normal sulfate (NHJ2S04. Besides these, the existence of
intermediate chemical compounds (Table 1) is possible [4]. The melting points change
continuously according to the H^oyN^ stoichiometric ratio, and there is a
possibility that compounds with very low melting points may be formed in the process.
In addition, the ones which are stable in solid form at temperatures above 150 °C
(300 °F) are those with stoichiometric ratios 1.2 and above.
The reaction of sulfuric acid mist with ammonia in the flue gas is believed to follow
a course of absorption of ammonia into the sulfuric acid mist, evaporation of water,
formation of solids, and solidification [5].
30-3
-------
1. When ammonia is absorbed into sulfuric acid mist, ammonia ionizes in the
water solution; this is because the sulfuric acid mist in the flue gas
exists as an aqueous solution. When the reaction proceeds, the
equilibrium partial pressure of the water vapor is lowered, and the water
in the solution evaporates.
2. As the reaction and evaporation of the water progress, the viscosity of
the solution increases, and its state proceeds from the state of ionized
water solution to that of a mixture of molten salt.
3. When the reaction proceeds further (dissolving of ammonia), crystals in
solid form are deposited in the solution. In this case, the solids
deposited differ according to the gas temperature.
4. After the solids start to precipitate, crystals are deposited only in the
amount corresponding to the quantity of the absorbed ammonia; and finally,
the solution is exhausted and solidifies.
There is no way to confirm the reaction of ammonia gas with sulfuric acid gas
experimentally. But, if the theoretical tendency is considered qualitatively, the
following changes can be visualized.
•^Slgas)"1" NH3(gas)+ '"^(gas) ^ Nn4rloU4 |iquid| acidic sulfuric ammoma (ammonium bisulfate) ( *• I
NH4HS04 has a melting point of 146.9 °C (296.4 °F), and deliquescent characteristics;
it absorbs water readily. When the flue gas temperature is about 140 °C (284 °F),
NH4HS04 begins to solidify.
about 1 sec
'"'~'3(gas)+ ''l"l4j||qmd/vaporjT noU4j||ql]|d/vaporjS> ^[ (NHJ2S04 (3)
The melting point of (NHJ2S04 is 513 °C (955 °F). In the above equation (2),
(NHJ2S04 is dissociated at temperatures higher than 200 °C and decomposes into
NH4HS04.
A variety of interactions of ammonia with sulfuric acid are possible as follows:
NH3(gas)+ H2S04(gas)---> NH4HS04(gas)---> NH4HS04(solld) (4)
NH3(gas)+ NH4HS04(gas)---> (NHJ2S04(gas)--> (NH4)2S04(solld) (5)
NH4HS04(solld)+nucleus ---> NH4HS04(part,de) (6)
(NHJ.SO^+nucleus ---> (NHJ2S04(part,cle) (7)
NH3(gas)+ NH4HS04(partlde)---> (NH4)2S04(partlcle) (8)
H2S04(gas)+ (NH4)2S04(partlde)---> NH4HS04(partlcle) (9)
The reactions in (4) and (5) are collision reactions. Due to the low vapor pres-
sures of NH4HS04 and (NHJ2S04, these compounds usually solidify easily. The
quantity of (NHJ2S04 formed increases as the quantity of ammonia increases. The
reactions in (8) and (9) are subject to the effects of the diffusion of the gas
currents of NH3(gas) and H2S04( ,. The respective diffusion constants are 0.37 cm2/s
and 0.15 cm/s in 140 °C (284 °F) air, therefore, the reaction in (8) should be
somewhat faster than in (9).
30-4
-------
Modification of Resistivity
There has been much debate by many authors regarding the use of ammonia alone for
modifying the electrical resistivity of fly ash [2, 6, 7]. While successes have
been reported using ammonia as the sole conditioning agent for some low sulfur
Australian coals, U. S. experience has shown improved precipitator performance for
some low sulfur coals and not for the others. It has been theorized that ammonia
conditioning is more effective for acidic ashes in terms of the relative con-
centrations of acidic and basic oxides present (acid/base ratio) but this has not
been true in every case [8]. Ashes having high acid/base ratios with small to
moderate amounts of sulfur trioxide present in the flue gas can have high
resistivity and it may be that resistivity is reduced as a result of ammonium
bisulfate acting to facilitate sulfur trioxide attachment. If little or no sulfur
trioxide were present, ammonia alone might have little effect. This hypothesis is
somewhat speculative at present in view of the small number of observations
documented [9].
Dual FGC creates a thin conductive film on the ash surface believed to consist of
low melting point ammonium sulfate products which contributes to resistivity
modification. It has been suggested that the addition of ammonia improves upon the
efficiency of the sulfur trioxide conditioning by an additional charge carrier,
the NH4+ ions [10].
In some low sulfur coals the percentage of acidic compounds in the ash (aluminum
oxide, iron oxide, and silicon oxide) is high (above about 90 percent). Here, the
injection of sulfur trioxide alone to reduce resistivity loses some of its effect-
iveness because the acidic sulfur trioxide cannot readily attach to the highly-
acidic ash at higher temperatures. Substantial amounts of excess sulfur trioxide
must be injected while treating such ashes to produce a conductive film on the
surface. Figure 1 represents a theoretically calculated sulfur trioxide injection
rate required to bring the fly ash resistivity to optimum operating range vs. flue
gas temperature for some typical coals [11]. At temperatures from about 165 to
190 °C (329 to 374 °F), depending upon ash composition, an inflection point occurs
in the injection rate curves, with the rate increasing with temperature above this
point. The temperature at which inflection occurs is a function of ash surface
conditions, i.e. a measure of susceptibility to attachment of acid. The increased
requirement for sulfur trioxide at temperature above the inflection point
indicates the required partial pressure of sulfur trioxide needed to attach suffi-
cient sulfuric acid to the particles to obtain the desired ash resistivity. The
difference between the injection level at low temperatures and that at a high
temperature is a measure of excess sulfur trioxide which will escape from the
stack at the higher temperature. If this difference is on the order of 10 ppmv or
more, it is likely that a condensation plume will form which may be visible, de-
pending upon the concentration of fly ash in the emitted stream. Dual FGC, tends
to overcome this problem by allowing greater adhesion of the residual sulfur
trioxide to the acidic ash particle surfaces to optimize its resistivity. This
allows for proper fly ash resistivity adjustment without using excess sulfur
trioxide.
Space-Charge Effect
When sulfur trioxide and ammonia are concurrently injected into a gas stream, a
fine fume consisting of a variety of ammonium sulfate particles, all much less
than 1 micron in size, is produced [12]. These particles alter the electrical
characteristics of the flue gas between the discharge and collecting electrodes
and produce a space-charge enhancement of the electric field. This well docu-
mented effect arises when fine fume is charged in the precipitator and the
30-5
-------
electric field is thereby increased [1, 13]. Moderate, strictly controlled quan-
tities of fine particles would therefore, increase the charge level of fly ash
particles and the field near the collecting plates. The higher field increases
collection efficiency. The space charge effect is usually more pronounced in the
front fields than in those following. This happens because a large portion of the
particles are collected in the front fields and the remaining small quantity is
not sufficient to produce a significant space charge.
Cohesive Properties of Fly Ash
Dual FGC can generate low melting point substances depending on the H2SOyNH3
stoichiometric ratio; at low stoichiometric ratios low melting point substances
having high moisture absorbency are generated; this promotes cohesion. This
cohesion, applied properly and strictly controlled, improves overall precipitator
efficiency.
The cohesive properties of fly ash influence stack opacity and precipitator
efficiency by agglomerating particles and by reducing reentrainment. To simplify,
the ammonia combines with sulfur trioxide to form ammonium bisulfate. The melting
point for this material, as discussed earlier, is about 147 °C (297 °F), so it is
semi-liquid at typical flue gas temperatures and acts as a binding agent when
mixed with fly ash. Because the agglomerated ash groupings are heavier, rapping
losses are minimized.
For high resistivity ashes it has been found that variations of cohesion due to
additives are of minor importance to precipitation efficiency. For low resis-
tivity ashes where the use of an additive has little or no effect on resistivity
it has been found that changes in efficiency of precipitation are associated with
variations of cohesion. High cohesion has resulted in high efficiency due to
reduced rapping losses and reentrainment or more effective initial capture of
particles [1, 14]. The electrical forces of attraction between particles in an
ash layer on the collecting electrode in an energized precipitator are the predo-
minant component of the cohesive strength of a highly resistive ash layer and the
variation of the mechanical component due to the additive is then relatively
unimportant. With low resistivity ash the electrical forces are reduced and may
even become repulsive. Under these circumstances mechanical forces important
[1, 14].
EXPERIENCE WITH DUAL FLUE GAS CONDITIONING
Currently, a number of coal-fired utility boilers in the United States with a
total capacity of over 6,600 MW employ Dual Flue Gas Conditioning on a continuous
basis. Last year Ontario Hydro, Canada, in order to meet the requirement of the
acid gas reduction, purchased twenty (20) Dual FGC systems for boilers with a
total capacity about 8,500 MW, sixteen (16) of which will be commissioned this
year [15, 16].
Some of the first utilities to use Dual FGC initially installed sulfur trioxide
injection systems for fly ash conditioning [9, 17, 18]. In each case it was found
that the improved ESP performance was marginal in achieving compliance with new
emission requirements. Here, the precipitator sizes were characteristically small
by present day standards, and appreciable quantities of dust were lost during
rapping cycles. Previous experiments with ammonia alone had shown increased ESP
operating voltage and improved ash cohesivity which led to reduced rapping losses
[7]. It was postulated that these effects, if obtainable in systems with sulfur
trioxide conditioning, would be additive to the effects of resistivity modifica-
30-6
-------
tion. Tests of Dual FGC were made which demonstrated that the complementary
action did indeed occur [9, 17, 18, 19].
Public Service Electric & Gas of New Jersey initially installed a sulfur trioxide
FGC system on the Hudson Station, Unit 2, 660 MW, to allow firing of low sulfur
coals (Table 2), and the precipitator performance improved as intended. A subse-
quent plant improvement program (involving a new primary air heater,major ductwork
modifications, and precipitator enlargement) resulted in seriously disturbed
temperature, gas and ash flow patterns. Initial operation of the revised plant
was unsatisfactory. Precipitator power input was low and, although the three 9 by
42 foot fields (166 SCA), precipitator was capable of achieving particulate
emissions standards with low power input, excess amounts of sulfur trioxide had to
be injected and an objectionable stack plume resulted. Flue gas temperatures
ranged from 260 to 380 °F across the ductwork with corresponding resistivities
ranging from 107 to 1013 ohm-cm. Figure 2 presents the theoretically calculated
sulfur trioxide vs. temperature curves for the coals fired. As can be seen, at
temperatures from about 300 to 350 °F an inflection point occurs in the injection
rate curves, with the injection rate required increasing with temperature above
this point. As discussed earlier, the temperature at which inflection occurs is a
function of ash surface conditions, i.e. a measure of susceptibility to attachment
of acid. The difference between the injection level at low temperatures and that
at a high temperature is a measure of excess sulfur trioxide which will escape
from the stack at the higher temperature.
Dual Conditioning was considered and an ammonia injection system was later added
to the sulfur trioxide system to remove excess sulfur trioxide and to reduce the
rapping losses which contributed significantly to the stack opacity. The sulfur
trioxide in the stack was reduced, the "blue" plume disappeared, the rapping
losses were reduced, and the required sulfur trioxide injection rate was cut by a
factor of almost 2, from 15 to 8 ppmv (Table 3). That was accompanied by a major
increase in power input which was attributed to a combined effect of adjusting the
resistivity to a satisfactory operating range and a mild space charge effect. It
should be pointed out that no changes to the gas and fly ash distribution were
made when the Dual FGC system was instituted, so the resistivity effect is
logically ascribed to improved sulfuric acid attachment.
Ontario Hydro, Canada, conducted exhaustive tests on a 512 MW unit (Nanticoke
Thermal Generation Station, Unit 2) equipped with a precipitator having three 12
by 30 foot fields [16]. The design SCA was 244 ft2/1000 acfm, however typical
operating SCA was about 208 ft2/1000 acfm. In the early 1980's the original
burners were converted to a low NOX design which resulted in high carbon-in-ash
(LOI) levels. The high carbon-in-ash was attributed to the switch over the years
to coals which were lower in calorific value and grindability, with lower volatile
content. Consequently, combustion suffered as the pulverizers were required to
grind a more difficult coal at a higher throughput. The use of low NO, burners at
Nanticoke exacerbated the combustion problem.
The test program examined the use of three fuels (Table 4). The first was the
regular 50/50 blend of high sulfur US coal and low sulfur Western Canadian coal.
The resultant sulfur content was 1.2 percent "as fired", and represents the
typical fuel burned at Nanticoke in 1989. The second fuel is termed "Low Sulfur
Blend" and is a blend of lower sulfur US coal (0.8 percent sulfur) and low sulfur
Western Canadian coal (0.3 percent) the resultant blend was 0.65 percent sulfur.
-The third fuel was the pure US low sulfur coal.
For the Nanticoke test program, the sulfur trioxide and anhydrous ammonia
injection probes were installed downstream of the secondary air heaters (Figure
30-7
-------
3). Tests were conducted with no conditioning, sulfur trioxide conditioning only
and Dual FGC on the test coals. The opacity results are summarized in Table 5.
No Conditioning. Without conditioning with the regular 50/50 blend, the unit
could achieve full load with an opacity level of 11 12 percent. This increased
to 18 percent as load was increased to secondary (10 percent) overload (Figure 4).
When the unit was switched to US low sulfur coal the opacity increased with
prolonged burning such that at full load opacity was over 22 percent (Figure 5),
and operating at secondary overload was not possible due to opacity restrictions.
Tests were not conducted with the Low Sulfur Blend as it was expected to be worse
than the pure US low sulfur coal.
Sulfur Trioxide Conditioning. With about 5 ppmv sulfur trioxide injection on the
Low Sulfur Blend coal, full load could be achieved but the opacity was 19 percent
(Figure 6). At secondary overload opacity increased to 29 percent. On US low
sulfur coal full load opacity was 15 percent with about 3 ppmv sulfur trioxide
injection (Figure 7).
Opacity charts displayed excessive rapping spikes. Examination of the fly ash
showed high carbon content, an average of 14 percent for the regular 50/50 blend
and the US low sulfur coal, and 19 percent with the Low Sulfur Blend.
Dual Conditioning. With Dual FGC conditioning the precipitator performance
improved dramatically. At 17 ppmv each of anhydrous ammonia and sulfur trioxide
full load opacity on the Low Sulfur Blend was 7 percent (Figure 8), and secondary
overload was achieved with an opacity of 16 percent. On US low sulfur coal the
opacity was reduced to 6 percent with 19 ppmv of ammonia and sulfur trioxide
(Figure 9). The most notable difference with the dual flue gas conditioning was
the reduction in the rapping spikes due to the reduced re-entrainment. The
outlet dust loading was reduced by approximately four times with Dual FGC over
sulfur trioxide conditioning alone (Table 6).
Sulfur trioxide injection is known to reduce fly ash resistivity, and is useful in
reducing back corona problems associated with burning coals which produce high
resistivity fly ash. However, in the case of Nanticoke, the high carbon content
of the fly ash appears to produce an ash which is a mixture of fine ash having
high resistivity, and coarse ash which is high in carbon content. The resultant
apparently low resistivity portion of the ash is easily re-entrained into the flue
gas stream during rapping as there is little electrical force to hold particles on
the collecting plates or attached to each other. Running continuously on low
sulfur coal results in selective re-entrainment of the low resistivity particles
leaving an insulating layer of high resistivity ash on collecting plates. Thus
sulfur trioxide is useful in preventing the accumulation of the high resistivity
fly ash, which otherwise creates back corona and deterioration in precipitator
performance.
When Dual FGC is used on the same fly ash a different mechanism appears to work.
The sulfur trioxide and anhydrous ammonia in the proper mixture form ammonium
bisulfate which acts as a binding agent with the ash. Cohesive properties of ash
on the collecting plates are enhanced due to chemical reactions rather than by
electrical charge. Thus re-entrainment problems are reduced. Furthermore, some
additional electrical current conduction may have occurred through a matrix of
carbon particles encapsulated into the ash layer on collecting plates. This could
be a reason for stoichiometric ratios observed during tests on this specific
application.
30-8
-------
Fly ash samples were analyzed and no objectionable contaminant which could affect
disposal was detected. The fly ash appears to be cohesive with the dual con-
ditioning compared to sulfur trioxide injection only, but the ash removal system
was not affected.
Injection Positions. The location of the sulfur trioxide and ammonia injection
points relative to one another does not appear to have significant influence on
the interaction between two chemicals and their eventual effect on the ash
properties. This experience supports above theoretical discussions that the
chemical reaction is nearly instantaneous and the end effect depends on the sulfur
trioxide (sulfuric acid) to anhydrous ammonia stoichiometric ratio. A variety of
different combinations has been tried with similar results (Table 7) [9].
CONCLUSION
Many publications have reported results strictly in terms of changes in ESP
collection performance without determining whether the ESP improvement, if any, is
primarily due to changes in ash resistivity, reduction of rapping losses due to
changed ash cohesivity, or voltage increases as a result of space charge effects.
Since all three occur in the presence of sulfur trioxide and ammonia, it is clear
that the ESP performance improvement without a resistivity change may very well
occur. This has led some investigators to discount the necessity for resistivity
adjustment as an ESP performance optimization requirement. We would like to state
again that "the effects are additive, and the most effective and most economical
use of a given ESP structure as a dust collecting machine requires that the three
effects be simultaneously considered" [9].
Simultaneous and independent use of anhydrous ammonia and sulfur trioxide is a
novel approach to flue gas conditioning which, by providing an independent and
simultaneous sulfur trioxide and anhydrous ammonia injection in strictly con-
trolled stoichiometric ratios allows for a multitude of new applications ranging
from enhancing sulfur trioxide utilization and performance improvement of
precipitators with high velocities and low aspect ratios to improved collection of
fly ashes with high amounts of unburned carbon.
REFERENCES
1. White, H. I., Industrial Electrostatic Precipitation, Monograph:
Addison-Wesley Publishing Co., 1963, pp. 310-315.
2. White, H. J., "Resistivity Problems in Electrostatic Precipitation",
Journal APCA. April 1974, pp. 313-338.
3. Frish, N.W. "Analysis of Air Heater Fly Ash Sulfur Acid Vapor
Interactions. In Proceedings of the fifth Symposium on the Transfer
Utilization of Particulate Control Technology, Kansas City, Missouri,
August, 1984.
4. Kendal, J., et. al., Vol.IV. International Critical Tables, 1928, pp.
42.
5. Kishi, M.,et. al., "Application of the Electro-Precipitator to Oil Fired
Boilers". Mitsubishi technical paper. Date and publication unknown.
30-9
-------
6. Dismukes, E., "Conditioning of Fly Ash with Ammonia". JAPCA, Vol. 25,
No. 2, February 1975, pp. 152-156.
7. Dalmon, J. and Tidy, D., "A Comparison of Chemical Additives as Aids to
the Electrostatic Precipitation of Fly Ash". Atmospheric Environment
fPergamon Press), V.6, 1972, pp. 721-734.
8. Dismukes, E., "A Review of Flue Gas Conditioning with Ammonia and Organic
Amines." In Proceedings of the 76th Annual Meeting of the APCA, June
1983.
9. Coe, E.L., Jr. and Lagarias, J.S., "Experience in Conditioning Electros
tatic Precipitators in the United States", In Proceedings of the Jt.
ASME/IEEE Power Generation Conference, Portland, Oregon, October 1986.
10. Dahlin, R. S., et. al., "A Field Study of a Combined NH3/S03 Condition
ing System on a Cold-Side Fly Ash Precipitator at a Coal-Fired Power
Plant." Paper 84-96.3. In Proceedings of the 76th Annual Meeting of the
APCA. June 1984.
11. Coe, E.L., Jr. and Krigmont, H.V., "Prediction of S03 Injection Rates for
Fly Ash Conditioning Systems." In Proceedings of the Third International
Conference on Electrostatic Precipitation, Abano, Italy, 28 October,
1987.
12. Dismukes, E., et. al ., "ESP Conditioning with Ammonia at the Monroe Power
Plant of Detroit Edison Company".
13. Sproul, W.T. "Corona Quenching-its Significance in Electrical Precipita
tion". In Proceedings of the 56th Annual Meeting of APCA, Detroit,
Michigan, June 1963.
14. Zimon, A.D. Adhesion of Dusts. Monograph, USSR: Ximia (Chemistry), 1976.
15. Gooder, "Options Available to Meet Acid Gas Limits and Selection of
Preferred Options." In Proceedings of the Second Conference & Exhibition
for the Power Generation Industries. New Orleans, Louisiana, December,
1989.
16. Arnott, J. A., et. al., "Ontario Hydro's Evaluation of Flue Gas Con
ditioning for ESP Performance Enhancement." In Proceedings of the Gen-
Upgrade Conference in Washington, D.C., March 1990.
17. Fletcher, H. "Operating Experience at Detroit Edison with Various Flue
Gas Conditioning Systems." In Proceedings of the 75th Annual Meeting of
the APCA. June 1982.
18. Cummings, W.E., et. al., "Baltimore Gas & Electric Experience with
Combined S03/NH3 Injection for Precipitator Performance Improvement." In
Proceedings of the Sixth EPA/EPRI Symposium on the Transfer and Utiliza
tion of Particulate Control Technology. New Orleans, Louisiana, February
1986.
30-10
-------
FIGURE 1
Theoretically Calculated SOS
Requirements
200 250 300 380 400
Flu* Qai T«mp*r«tur*. D*g. F
SOS Injection Rate vs. Temperature
FIGURE 2
PSESG. Hudson St., Unit 2
Theoretical 303 Requirements
200 2SO 300 360 400
Flu* Q«* T*mp*r«tur*. Dog F.
SOS Injection Rate vs. Temperature
FIGURE 3
Ur*2 Pnc*MJKv
(B5QBO)
NANTICOKE UNIT 2
OPACITY
REGULAR 50/50
TEST 4
S03 OFF
NH3 OFF
30-11
-------
NANTICOKE UNIT 2
OPACITY
LOW SULFUR 50/50
TEST 43
S03 OFF
NH3 OFF
ESUBtfil
NANTICOKE UNIT 2
OPACITY
LOW SULFUR 50/50
TEST 35
S03 5.5 ppm
NH3 OFF
NANTICOKE UNIT 2
OPACITY
PURE US LOW SULFUR
TEST 47
S03 3.5ppm
NH3 OFF
30-12
-------
NANTICOKE UNIT 2
OPACITY
LOW SULFUR 50/50
TEST 33
S03 17 ppn
NH3 17 ppm
NANTICOKE UNIT 2
OPACITY
PURE US LOW SULFUR
TEST 53
S03 19ppm
NH3 19ppm
TABLE 1
AMMONIUM SULFATE COMPOUNDS
SUBSTANCE NAME
SULFUnC AGO
SULFURK ACD
AMMONUM
1
1
•
IV
AMMOMUM SULFATE
CtSVKAL FORMULA
H,S04
NH4H,
104
4S.O
146.9
--
--
513
PRESSURE)
NOTE
10OS H2SO4
--
--
--
--
DECCK&OSES
AT 180-290 °C
30-13
-------
TABLE 2
PSE&G - HUDSON STN.
TYPICAL COAL & ASH ANALYSIS
TABLE 3
PSE&G - HUDSON STATION
SUMMARY OF TEST RESULTS
COAL
ULTMATE ANALYSIS
CAMMM
HYDROOEN
OXYGEN
SULRJH
MTOOGEN
MOISTURE
ASH
ASH ANALYSIS
LijO
Na,0
Kjd
*>o
CaO
Fe203
AI,O3
S»J
TO,
PjOs
SO,
A
7*00
4M
4.70
097
1.53
040
7.M
001
008
1.37
O.73
208
9.OO
30.47
4633
1.87
1.10
050
B
73.02
4.93
5*0
0.87
1.50
0.00
a.oo
0.01
0.22
0.02
090
100
718
1^ rt *
,»4-U I
91.02
2.OO
0.98
O.O7
C
7330
4.53
341
110
190
740
&77
0.01
0-27
2.07
0.70
2.18
1OO5
1 1 09
J 1.O4C
48.08
1.4O
073
1.53
D
75 37 FLUE GAS FLOW, ACFM
«?. Fa TB*>. oF
J-^* EMBSWNS, LB/«*BTU
6 1O OPACtTY, X
8^9
FFFIOFNCY ' 1
001
O.O3 SCA, FT 2 1 10OO ACFM
158
070 SOj.PPMV
1.15 „
001 NHj.PPMV
A
2,383.232
310
O.O3
8
eo.oe
101
8
4
B
2 .2 68,7s 1
3D6
O.O3
8
9O.OO
107
8
4
C
2,370,814
323
O03
e
6906
102
8
4
3O.O4
94.15
1 83 -BASED ON CALCULATED NJET
0.34
080
o
I—'
-t*
TABLE 4
NANTICOKE UNIT 2 - LOW SULFUR COAL BURN
COAL ANALYSIS
COAL
ULTMATE ( AS ORED )
CARBON
HYDROGEN
SULFUR
MTROGEN
ASH
OXYGEN
ASH ANALYSIS
SiO,
AljOj
Fe,^
Cab
MBO
Na,0
KjO
sfo
TiO
p,a
SOj
REG 50/5O BLEW)
1.2* S
7373
5OO
129
1.12
988
8.B2
51.9
2248
1019
003
1.02
098
0.92
0.12
103
0.47
284
LOW SULFUR BLEND
O65X S
73.34
492
0.00
199
9.92
B57
53.90
22.41
0.03
5.07
1.11
0.04
1.27
0.11
1.05
034
2.97
US LOW SULFUR
09X S
7430
994
1.15
132
920
B.37
54.09
2878
007
105
0.83
024
1.81
0.1
153
0.20
0.85
-------
TABLE 5
NANTICOKE UNIT 2 -
LOW SULFUR COAL BURN
OPACITY SUMMARY (AVG) *
TABLE 6
NANTICOKE UNIT 2 -
LOW SULFUR COAL BURN
TEST RESULT SUMMARY
REGULAR 5O/5O BLEND 1 1.2X S)
MCR
SEC. 01.
LOW SULFUR BLEND (0.65X S)
MCR
SEC Oi_
PURE US LOW SULFUR (0.8X S)
MCR
SEC OL
NO COND
113
17.3
--
--
220
--
SO,* NH,
..
--
0.0
16 O
8.1
--
NH,
59
--
--
--
—
--
SO,
COAL SULFUR
LOAD
SO,MJECTON
10.0 NH,MJECT1ON
29.3 OPACITY
CARBON-M-ASH
15.1 SO,
(-240, ^
OUTLET DUST LOADMQ
PHECt> EFF
"4 Mi 1 nPFRATWIN
X
MW
PPM
PPM
X
X
PPM
PPM
GR/ACF
X
REGULAR
90/9O
1.2
812
--
--
11
14
1O35
470
0.010
00.2
LOW SULFUR
BLEND
0.98
612
17
17
7
10
920
400
0.027
067
US
LOW SULFUR
0.8 08
312 912
3 10
10
17 0
14
«79
940
0.10 O.O24
090 08.7
co
o
TABLE 10 - REFERENCE 0
TABLE 7
INJECTION POSITIONS
DUAL SO 3 - NH 3 SYSTEMS
STATION
A
B
C
D
E
F
NO OF
BOLER
UMTS
1
4
1
1
2
1
DST. BET.
SO,4NH3
NJL PTS.
2OM
2 M
28 M
15U
OOM
0.3 M
MATERWL
NJECTED
UPSTREAM
so,
NH,
SO,
SO,
SO,
SO,
SOjLOC.
HOT OR COLD
SDE OF APH-
HOT
COLD
HOT
COLD
COLD
COLD
•NH , NJECTED ON COLD SDE OF AR PREHEATER
-------
MODIFICATION AND CONVERSION OF THE NEBRASKA CITY UNIT 1
HOT ESP TO COLD-SIDE OPERATION
A. W. Ferguson
R. C. Wicina
B. L. Duncan
Black & Veatch Engineers-Architects
Kansas City, Missouri
K. A. Roth
R. M. Kotaii
Omaha Public Power District
Omaha, Nebraska
ABSTRACT
Omaha Public Power District, Nebraska City Unit 1, is a 585 MW net coal fueled power plant. The unit,
which burns low-sulfur Powder River Basin coal, was originally designed and furnished with a fully enclosed,
hot-side rigid frame electrostatic precipitator. Because the original precipitator was unable to maintain
reliable and continuous stack opacity and particulate emissions levels at high operating loads, the hot-
side precipitator was modified internally and converted to cold-side operation. This conversion consisted
of relocation of the unit's four regenerative air heaters, extensive ductwork modifications, and significant
internal precipitator modifications.
The precipitator modifications included replacing the first two fields of electrodes, collector plates, and
rappers with a design different from that originally installed, straightening the plates in the last two fields,
replacing the discharge electrode rapping system, and upgrading the transformer-rectifiers. A sulfur trioxide
flue gas conditioning system was also installed.
Extensive flow modelling was performed and the results incorporated to ensure equally distributed gas
flow into the precipitator chambers, in accordance with IGCI EP-3 and EP-7 recommendations, as well
as to minimize draft loss through the ductwork.
This paper describes the preconversion operational conditions, design and construction of the precipitator
modifications, and resulting performance improvements of the modifications and the conversion.
31-1
-------
MODIFICATION AND CONVERSION OF THE NEBRASKA CITY UNIT 1
HOT ESP TO COLD-SIDE OPERATION
BACKGROUND
The Nebraska City Power Plant, a 585 MW net electric generating plant, is the largest single unit in the
Omaha Public Power District (OPPD) system. The unit is located near Nebraska City, Nebraska, and burns
low-sulfur coal from the Wyoming Powder River Basin. The plant, designed in the mid-1970s, has a hot-
side electrostatic precipitator (ESP) with a specific collecting area (SCA) of approximately 320 sq ft/1000
acfm. Since initial plant startup in 1979, the hot-side ESP has failed to perform reliably especially at
higher power generator loads. To comply with air quality regulations, OPPD had to reduce boiler load
and, subsequently, generation.
Since fall 1980, repairs and alterations had been performed during several plant outages in an attempt
to improve precipitator performance. This work included realignment of the precipitator internal parts,
stiffening the collecting plates, repair of discharge and collecting electrode rapping systems, repair of
transformer-rectifiers (TRs), installation of gas distribution devices, structural repair of damaged steel
and replacement of lubrite plates to improve precipitator expansion characteristics, replacement of the
hopper level detectors, and the addition of microprocessor-based ESP control equipment.
During this time, OPPD performed a rigorous maintenance program to keep the performance level of
the precipitator as high as practical. Maintenance activities included regular sandblast cleaning of the
collecting surfaces, discharge electrodes, and gas distribution devices, as well as extensive replacement
and adjustment of rapping system components, and removal of broken discharge electrode wires from
the rigid frames.
The operating fuels for Unit 1 are both Rawhide and Caballo coals. OPPD attempted to improve precipitator
performance through the combustion of coals with high sodium ash content, such as Black Thunder coal
and Decker coal. OPPD also tested the addition of sodium carbonate and sodium sulfate directly to the
coal and in the gas stream after the economizer, as well as ammonia addition after the economizer. None
of these test programs produced long-term acceptable results.
In fact, none of the modifications proved to be particularly successful. Frequent cleaning was burden-
some and offered only short-term effectiveness. It was decided to convert the precipitator from hot-side
to cold-side operation to achieve reliable, full load unit performance. The original system design moved
the flue gas from the economizer to the ESPs and then through the air heaters to the ID fans. To convert
the unit, the duct system had to be modified and the equipment rearranged so that the flue gas moved
from the economizer to the air heaters and then through the ESPs to the induced draft (ID) fans. In addi-
tion, significant modifications were made to enhance precipitator operations and to ensure long-term per-
formance. These modifications included ensuring that any misalignment of components was eliminated
by securing any loose or broken electrode wires, installing a more effective rapper system, maintaining
the TRs, and optimizing gas distribution into the precipitators.
31-2
-------
DESIGN ENGINEERING
Conversion Alternatives
Two alternatives for the conversion to a cold-side unit were considered: relocating the air heaters, and
rerouting ductwork. Figure 1 shows the original plant arrangement.
For air heater relocation, as shown on Figure 2, the four existing air heaters would be moved from behind
and downstream of the precipitators to a new location between the boiler and precipitators, upstream
of the precipitators. Extensive ductwork modifications would reroute flue gas from the economizer outlet
by removing the ducts between the economizer outlet and precipitators and installing new ductwork to
route the gas from the economizer outlet to the air heaters located at ground level. Flue gas would then
be routed back to the precipitators by new ductwork tying into the existing precipitator inlet ducts. Where
the air heaters had been located, new spool pieces would be installed in the existing ducts. Extensive
modifications would be required to the support steel, mechanical piping, and electrical raceway. Construction
would be limited to inside the powerhouse enclosure. This alternative, which would be expected to increase
pressure loss through the draft system, might also reduce ID fan margins.
The ductwork rerouting alternative, as shown on Figure 3, meant that flue gas from the economizer would
be routed using new ductwork to bypass the precipitators. The new ductwork would wrap around the
outside of the precipitator building and tie into the existing ductwork upstream of the air heaters. Flue
gas from the air heaters would be routed by new ductwork back to the ESP inlet. New ductwork would
be added to route the gas discharging from the precipitators to the existing ductwork at the ID fans.
New columns and framing would tie into the existing structure outside the precipitator building to sup-
port the new ductwork. Significant modifications would also be required to the existing building steel.
Reinforcement of existing foundations and new foundations, including piles for the new columns, would
be required. Modifications to mechanical piping and electrical raceway would be minimal. The long horizontal
duct that runs around the precipitators would create significant potential ash dropout zones. The duct
arrangement would increase potential draft loss through the system even more than would the air heater
relocation.
Preliminary cost estimates indicated that the air heater relocation would cost less than duct rerouting.
The potential draft loss would be less for the air heater relocation plan, thus maintaining the largest ID
fan margin possible. Also, potential ash deposition in the duct was less than in the duct reroute plan.
Therefore, the air heater relocation plan was chosen.
Model Study
As part of the conversion work, the ductwork was significantly rerouted to accommodate gas flow into
the relocated air heaters. This included cutting the ductwork at the point where it entered the precipitators,
routing it to the ground level to the relocated air heaters, then routing it back up to the existing precipitator
inlets. Because the work was performed inside the building, in an already congested area, rerouting the
ductwork was not easy. The rerouted ductwork had to skirt existing equipment and utilize open areas
within the precipitator building. A flow model study was required to ensure that gas distribution to the
precipitators was uniform and that increased draft loss was minimized. This study was necessitated by
the extensiveness of the rerouting of the ductwork, stacked design of the precipitators, evidence that the
31-3
-------
The requirements of The Industrial Gas Cleaning Institute's (IGCI) standard EP-3 and EP-7 were used
to achieve uniform gas distribution into the precipitators. This ensured that the gas was uniformly divid-
ed among the four precipitators and uniformly distributed within precipitator inlets. The side-to-side gas
split between the north and south precipitators was made at the economizer outlet, a natural division
because of the dual economizer outlet configuration. Because of the large physical size and the symmetry
of the plant, a 1/12 scale model was built of only the north half of the unit. The top-to-bottom vane place-
ment and flow division was a result of the modelling effort.
In addition to a satisfactory division of flow to the top and bottom precipitators, ICGI EP-7 requires
that the precipitators have good gas distribution within each individual inlet nozzle. Each precipitator
has two inlet nozzles. Three perforated plates were originally installed in each inlet nozzle. Turning vanes
were originally installed in the 90° duct bend at the nozzle entry.
As a result of the model study, changes were made to the three perforated plates at the inlet to the
precipitators. This included changing the porosity of some of the plates and adding flow diverters to en-
sure gas distribution in accordance with IGCI EP-7.
Another potential problem area was the 90° turn in the duct at the entry to the inlet nozzle opening of
each ESP. During early operation, there had been particulate buildup on the vanes in this area. The model
showed that buildup would probably continue on these vanes, especially at low loads. The modeller recom-
mended that a vane cleaning system be used, and provisions were made to install the system at a later
date. Operation of the precipitator since conversion does not show that excessive buildup occurs on these
vanes at high loads, even though the complete cleaning system has not been installed.
The precipitator and ductwork were modified based on the modeller's recommendations for vane place-
ment and perforated plate changes required to meet IGCI EP-7 criteria. A field verification test of the
modified installation was performed. The field testing verified that the installation was performing as
predicted in the model recommendations.
In addition to the flow model, an engineering model was constructed in Black & Veatch's model shop.
The area, modelled at 3/8 inch equals 1 foot, included all details and equiment in the air heater area and
duct revisions within the plant. Construction of the engineering model began while design was in pro-
gress and before completion of the flow model. The model provided a thorough design check. Several in-
terferences were found, and the design was revised before the interferences became construction problems.
The model also enhanced review of maintenance access, especially around the air heaters where access
was reduced by the conversion. The model was frequently used in the field to review construction pro-
cedures.
PRECIPITATOR UPGRADES
The precipitator at Nebraska City Unit 1 had experienced a significant number of problems during its
operation. The original design was a two-precipitator installation with each having four fields and six
bus sections per field and requiring a total of 24 TRs. One of the most notable problems was the bending
and distortion of the leading and trailing edges of the collecting plates in the gas stream. Original collec-
ting plate spacing was 10 inches. This resulted in approximately 5 inches of center-to-center distance be-
tween the discharge electrodes and the collecting plates. With the bending and distortion of the leading
and trailing edges of the plates, the clearance in some places was as little as 1-3/4 inches between the
discharge frames and the collecting plates. This significantly hampered the collection efficiency of the
precipitator by causing sparks and reduced voltage operation. This damage appeared to occur over time
and inspections showed that the problem was most noticeable at the inlet to the first field, and its severi-
ty decreased in the subsequent fields of the precipitator. At the time of the conversion, damage in the
first and second fields was significant. The third and fourth fields had begun to exhibit bending and distor-
tion of the leading and trailing edges, but the damage was not significant.
31-4
-------
Because of this damage, OPPD investigated the possibility of either straightening the plates or replacing
them. It was determined that either should result in satisfactory operation. The question that remained
was that if the plates were bent back to their original position, how long would they stay there before
they would revert to their bent position. Proposals were evaluated for an option which included bending
the plates back to their original position and installing a bar-type device on the leading and trailing edges
of the plates to help maintain their position. The alternative was to replace the collecting plate and discharge
electrode assemblies inside the precipitator with new material.
The decision was made to replace the plates located in the inlet, or first and second, fields of the precipitators.
The design selected was a flat plate with 12-inch plate-to-plate centerline spacing. Also, new discharge
electrodes of a spiral design replaced the original fluted square wire electrodes in the first and second
fields. The original fluted square wire discharge electrodes in the third and fourth fields were left in place.
To accommodate the complete removal of the original collecting electrode and discharge electrode systems
and their replacement with new systems, the precipitator roof over the first and second fields had to be
temporarily removed. The TRs immediately above these field had to be moved temporarily, and the ex-
isting lagging, insulation, and roof plate had to be cut out between the box girders supporting the
precipitator internals. The existing overhead TR maintenance hoists were used to help lift out the old
material and to lower the new plates and electrode frames into position.
The third and fourth fields of the precipitators had some structural damage in that the leading and trail-
ing edges of the plates were bent. The curled shape of the leading and trailing edge of the collection plates
was originally cold rolled. The high operating temperature of the hot-side precipitator was allowing the
stress in these edges to relieve in an uncontrolled manner, resulting in bending and distortion. The bulk
of particulate was removed in the first and second fields. Because the third and fourth fields removed
a smaller ash load than the first two fields, the decision was made to straighten these plates rather than
replace them. Several straightening methods were investigated. One method was stress relieving, which
included heating the plates in localized areas, allowing them to straighten out. A second method was
cold bending of the plates, a brute force method. It was decided that a special tool would be made and
used to cold bend the plates back into their original position, since this was the least-cost method. It
was felt that the stress relieving and distortion of the plates during hot-side operation would not occur
with the lower operating temperatures of a cold-side precipitator. This effort was carried out in the third
and fourth fields.
In addition to the plate straightening in the third and fourth fields, each of these fields was then re-aligned
using a bracing system attached to the discharge electrode frame. This gave the frame additional strength
and allowed the frame section for an entire bus section to be centered in the gas path.
The TRs had experienced significant problems in that they had carbonized the oil and subsequently failed.
The problem probably resulted from split arching of the switch (located within the oil of the transformer).
Detailed investigations showed a space between the leaves of the switch which may have caused the oil
to carbonize when the switch arced. Another problem was the lack of a visual verification that the TR
and internals were grounded when internal access was required. Each transformer was modified to remove
the internal switch and to replace it with an external switch. This external switch is air-insulated and
located in the bus duct where the position of the switch could be verified through a window.
In addition to this modification, an intensive maintenance and calibration program was conducted on
the control system. Other changes made to the operating parameters of the control system included ad-
justing the spark rate to improve operation.
Prior to the conversion, the original rappers were a maintenance-intensive system, especially the discharge
electrode rappers. It was also questionable whether the original rapper system was capable of delivering
the intensities needed for proper cleaning. The original design used a lift drop approach. The rapper motors
31-5
-------
ran continuously and lifted the rapper hammers attached to a shaft. Each shaft contained 13 rapper ham-
mers positioned on two different levels within the precipitators. All the connections were pinned. When
a pin fell out of the rapping system, the rappers became inoperable and whole sections of the precipitators
remained "un-rapped." These pins fell out with alarming regularity. There was also a problem with ham-
mers falling off the discharge electrode rapper drives. The rapping system on the discharge electrodes
was replaced with a tumbling hammer design. The drive shafts for the tumbling hammers penetrated
the roof through the same locations that the lift rods had, so no additional holes were required in the
roofing. The tumbling hammer system spans more than one bus section, so insulators were placed bet-
ween the internal shafts. To date, these have not been a problem. The tumbling hammers have proven
to be more reliable than the lift drop approach used in the past. As a part of the new first and second
field and as a part of the work that was done on the third and fourth fields, the collecting electrode rap-
pers were also replaced, because the existing tumbling hammers had shown excessive wear and
misalignment.
Several other items were addressed in the precipitator upgrades. Because cold-side operation has a much
lower operating temperature, several steps were taken to eliminate air leakage and to avoid condensation
of moisture within the precipitator which might lead to corrosion and/or the setup of fly ash. The reliabili-
ty of the existing ash hopper heaters was doubtful; therefore, they were replaced with larger heaters to
minimize hopper pluggage. Also, the heating blankets installed on the discharge electrode support in-
sulators (which were in the box girders) were replaced with a more reliable sytem. The interior of the
precipitator casings and the interfaces between the precipitator casings and the ash hoppers and the in-
let and outlet nozzles were inspected after the ash was removed. All cracks were cleaned and repaired.
Contrary to typical experience with hot-side precipitators, which would indicate widespread cracks or tears,
especially at corners, only a few small cracks were found. Other sources of air leakage such as rapper
shaft seals and access door seals were replaced.
S03 GAS CONDITIONING SYSTEM
Since the conversion, the precipitator has an SCA of approximately 490 sq ft/1000 acfm at full load. For
the Wyoming Powder River Basin coals burned at this unit, the cold-side precipitator performance was
considered sufficient to meet both the opacity and the mass emission limitations of the plant with no
conditioning required. To ensure the reliability of the unit, however, an 803 system was installed for par-
ticulate removal enhancement. Its purpose was to increase unit reliability by providing design margin
when experiencing operating problems with items such as TRs, failure or partial failure of the control
system, shorts caused by loss of discharge wires within the precipitator, and buildup of ash on internals
which might be difficult or impossible to remove by rapping. The 803 gas conditioning system allows
condensation of the 803 on the fly ash, thus lowering resistivity of the ash and making the ash easier
to collect in the precipitator.
The second reason for installing the 803 system was that OPPD wanted the flexibility to burn other
Powder River Basin fuels, some of which could cause difficulties in meeting particulate emission compliance.
The 803 gas conditioning system includes a liquid sulfur storage tank, a burner for burning the liquid
sulfur to form SO2, a catalytic converter stage which converts SO2 to 863, and an injection system to
add 803 to the flue gas. The 803 system is sized to provide up to 30 parts per million of 803 in the
conditioned gas at the inlet of the precipitators.
CONSTRUCTION
Construction included extensive modifications to the ductwork, relocation of equipment to convert the
unit to cold-side operation, and modifications to the precipitator.
31-6
-------
The conversion construction was performed in two phases: pre-outage and outage. During the nine-month
pre-outage construction phase, work which could be completed with the unit on-line was done. This in-
cluded installing new air heater piles and foundations plus extensive reinforcement and replacement of
existing structural steel.
During the 20-week outage, conversion construction accelerated to a two-shift per day, six-day per week
operation. Ductwork modifications began immediately at the start of the outage. The closeness of the
building enclosure and the equipment arrangement greatly limited construction access. The ability to
preassemble large pieces was minimal. Individual duct panels were lifted, then assembled and welded in
the air. Approximately 900 tons of new duct material was erected in this manner. Additional modifica-
tions were made to structural steel, piping, and electrical raceway. Toward the end of the outage, OPPD
conducted extensive startup testing programs to ensure proper functioning of the many status systems
which had been affected during the course of the conversion and upgrades.
Work on the precipitator internal modifications could not begin until tho outage began. The upgrading
construction was also very intense, proceeding on a two-shift per day, si\-
-------
Air Ducts
\
^_
•*-
1
I >
71
>\
^ v
Air ID Fans
Heaters
Figure 1. Original Arrangement
Figure 2. Air Heater Relocation Arrangement
31-8
-------
wvv
Hot Gas
Ducts
Figure 3. Duct Reroute Arrangement
31-9
-------
RESULTS OF THE ROY S. NELSON UNIT 6
HOT-SIDE PRECIPITATOR
STRUCTURAL EVALUATION
C. R. REEVES
Lead Civil Engineer
Gulf States Utilities Co.
Beaumont, Texas
and
S. A. JOHNSON
Project Manager
and
R. L. SCHNEIDER
Structural Engineering Supervisor
Sargent & Lundy
Chicago, Illinois
32-1
-------
RESULTS OF THE ROY S. NELSON UNIT 6 HOT-SIDE
PRECIPITATOR STRUCTURAL EVALUATION
INTRODUCTION
During the 1970s and early 1980s, utilities building new coal-fired power plants had
two viable options to meet the S02 limitations established by the 1971 NSPS: to
install FGD equipment or burn low-sulphur coal. Most utilities opted to burn the
low-sulfur coal. Since baghouses were not yet a proven technology, hot-side
precipitator technology was developed to address the problems collecting the high
resistivity ash associated with low-sulfur coal. For various reasons, many
utilities are now experiencing unanticipated operational and structural problems
with their hot-side precipitators. There is strong evidence that the problems are
more pronounced at units with higher operating temperatures, especially above 800°F-
The hot-side precipitators at Gulf States Utilities Company's Roy S. Nelson Station
Unit 6 have had a number of operating and maintenance problems that are likely
common throughout the hot-side precipitator industry. Nelson Unit 6 is a 550-MW
pulverized-coal unit that was placed online in 1982. A Wyoming Powder River Basin
subbituminous coal is burned in a Combustion Engineering balanced draft, controlled
circulation, drum-type boiler. Two Joy western hot-side, weighted-wire preci-
pitators are used for flue gas particulate collection. The precipitators were
designed for a flue gas temperature of 800°F; however, during the first year of
operation, flue gas temperatures in excess of 900°F were recorded. In December
1982, an operational limitation of 850°F was placed on the economizer exit gas
temperature, and the unit has operated within this limitation since that time.
Soon after initial operation the following problems were identified with the
precipitators at Nelson Unit 6:
» declining precipitator performance,
• air and water in-leakage, and
« structural damage within the precipitators.
Gulf States Utilities (GSU) instituted several remedies to solve the performance
problems, but periodic water washing was the only one with some measure of success.
In-leakage and structural damage was addressed by a periodic field examination and
repair program, but this program did not prevent the same damage from reoccurring.
In 1987, GSU retained the services of Sargent & Lundy (S&L) to assess the current
condition of the precipitators and to investigate several alternate emission control
strategies for Nelson Unit 6. The following alternate strategies were considered
most feasible and were selected for detailed investgation:
• continued hot-side operation,
• cold-side conversion with the existing arrangement,
• cold-side conversion with a reversed gas flow,
• new cold-side precipitators, and
e new baghouses.
The precipitator assessment program focused on the recurring structural problems and
consisted of a series of detailed structural examinations and an abbreviated
material test program. The assessment program revealed that the structural
integrity of the precipitators was very much a concern and raised doubt about the
technical feasibility of continued use of the existing precipitators in either a hot
or cold arrangement.
32-2
-------
There were two major reasons for the structural concerns. First, differential
temperatures of up to 320°F were a normal occurrence during unit startup, as
revealed by a temporary thermocouple array placed in the precipitator inlet. The
locations of the differential temperatures closely correlated with the locations of
the most severe structural damage. The examinations also revealed that the damage
usually reoccurred in the same locations after each unit outage. Second,
significant material degradation was present. The abbreviated material test program
revealed that critical structural components in the precipitators had become
embrittled. There was also some concern that creep damage had occurred.
As a result of the assessment program findings, GSU authorized S&L to initiate a
more detailed material evaluation program. The purpose of the program was to
determine to what extent material degradation had occurred and to determine the
technical feasibility of continued use of the precipitators in either a hot- or a
cold-side configuration. If continued use was feasible, the program was to identify
the design-allowable stresses in either operating mode upon which structural
modification would be based.
The detailed material test program verified the feasibility of continued long-term
use of the precipitators in the cold-side configuration provided the conversion was
performed in the near future. Continued operation in the hot-side configuration was
possible but only on a short-term basis. This limitation was because material
degradation would continue in the hot-side configuration and the material test
program could not predict the extent of future degradation. The evaluation,
however, did recognize that continued hot-side operation was possible provided that
continued periodic material testing indicated that the material degradation did not
fall below the short-term structural modification design limits.
Cost estimates for each of the five options were prepared. The estimates for the
continued hot-side options included structural modifications based on the new design
allowables for a 7-year operating period with the intent to convert to cold-side
operation at that time. Short-term performance in the hot-side configuration would
be maintained by semiannual water washings. The assumption was that further
structural modifications for the future cold-side conversion would not be required.
The estimates for the cold-side conversion options included structural modifications
based on the new design allowables with the conversion taking place within 2 years
and S03 conditioning being implemented to address the performance issues.
Cost benefit analysis by GSU clearly indicated that the new precipitator and
baghouse options could not be justified. The reverse gas flow configuration was
found to be the preferable cold-side conversion option. However, continued hot-
side operation with plans to convert to cold-side in 7 years was the most cost-
effective option. The implementation of this option started in December 1988 and
that project is currently in progress.
This paper discusses the history of the Nelson Unit 6 precipitator structural
problems and the strategies developed to address them. Of most importance is the
implementation of a material testing and evaluation program and an innovative
structural design methodology that will allow GSU to continue operating the
precipitators in the hot-side configuration until 1995 followed by future cold-side
operation.
This paper has been divided into five sections to illustrate the problems discovered
in performing the structural evaluation and the methodology used in addressing these
problems:
• field structural examinations,
• causes of the structural and material problems,
• material testing program,
• fracture mechanics assessment, and
• structural modifications to the precipitator.
32-3
-------
FIELD STRUCTURAL EXAMINATIONS
Extensive structural field examinations of the precipitators and ducts at Nelson
Unit 6 were performed during unit outages in 1987, 1988, and 1989. Extensive
structural damage was identified internal and external to both precipitators. Minor
damage was identified in the ductwork. Slide plate degradation was also noted. The
examinations revealed severe damage to the vertical bracing, the tube columns, the
roof and hopper girders and their connections, the hopper corners, and the strut
connections on the inlet column rows of the precipitators. It was determined that
the inlet column rows had very little capacity remaining to resist lateral loads,
such as wind, and the columns had reduced gravity load carrying capacity due to the
loss of column bracing and the extensive column cracking. Although the structures
appeared to be stable, the redistribution of loads caused by the damage resulted in
an indeterminate state of stress within the various structural members.
To temporarily restore the structural integrity of the precipitators, interim
repairs were performed as the damage was identified. After each examination it
became apparent that some of the damage was repetitive, with new failures often
occurring in or near the previously made temporary repairs. On that basis, GSU
decided that a long-term permanent solution was necessary to address the structural
integrity concerns and prevent a major, unanticipated unit outage caused by a
precipitator failure.
CAUSES OF THE STRUCTURAL AND MATERIAL PROBLEMS
Based on the observations of field structural examinations, most hot-side
precipitator systems experience structural problems. Buckling of members, failed
connections, and cracking of the plates and members are the most significant
structural problems that can lead to more severe failures. More information on
these problems is discussed in Reference 1. These failures can jeopardize the
stability of the structural system. These deficiencies can also affect the
performance of the precipitator by causing or aggravating collector-plate
misalignment, shorting out of electrical fields, and air and water in-leakage. Air
and water in-leakage has been a serious problem at Nelson Unit 6.
The structural problems mentioned above can be attributed to the following four
major causes:
t a rigid structural framing arrangement that does not properly consider the
interaction of the members under thermally induced loads,
• unanticipated high differential temperatures within the precipitator,
• degraded slide bearing plates, and
• degradation of the steel due to high temperature exposure.
Structural Framing Arrangement
A redundant structure is usually desirable because then there are several load paths
so that an overall structural collapse will likely not occur. However, when a
redundant and rigid structure is exposed to large differential temperatures, very
large stresses and deformations occur to the weakest members within the structure.
Compression members will buckle and tension members will fracture. Every time that
a member fails, a load path is broken and a new, different stress profile occurs
within the structure. When a member fails, the redundancies in the structure allow
its portion of the load to be redistributed to other members. This redistribution
is acceptable as long as the redistributed loads do not result in propagating
failures that culminate in the collapse of the entire system. Progressive
32-4
-------
structural failures have been tracked over time in hot-side precipitators with
rigid, redundant framing and a multitude of external supports.
At Nelson Unit 6, the precipitators, ducts, and all support steel are braced in each
bay in the longitudinal and transverse directions. Thus, they are rigid and
indeterminate. There are 56 supports under each precipitator and the ducts are
supported every 23 feet. Theoretically, the precipitators need only 4 supports and
the ducts are large enough so that they could easily be designed to span up to 100
feet.
When the unit is offline, most of the perimeter columns on rows Lx and S, (these
locations are shown later on Figure 8) are out of contact with the support steel.
when the unit is online, thermal expansion causes these columns to come down into
contact while other, interior columns are lifted off the support steel. This change
is caused by the precipitator rigid internal bracing system reacting to the large
thermal differentials. When these uplifts occur, large redistributions of gravity
loads occur (in addition to the previously discussed thermally induced stresses),
and these can cause failures to the precipitator members and to the support steel.
Differential Temperatures Within Precipitators
Sargent & Lundy is aware that differential temperatures within hot-side
precipitators may be on the order of 300°F or higher. During unit startup, the
temperature change of the steel out of the gas flow lags that of the steel directly
in the gas flow. This temperature differential has been measured at Nelson Unit 6
to be as large as 320°F as the unit goes from ambient temperature up to a maximum
operating temperature of 850°F. These temperature differentials produce stresses in
the structural members due to the restraints imposed by the structure. The
differential temperatures are large enough to cause compression failures in the
hotter members and tension failures in the cooler members. More information on this
topic is presented further in References 1 and 2. Eventually, the temperature
differentials cause the precipitator structure to self-destruct if it is not
flexible enough to accommodate the differential growth patterns produced. In fact,
these differential temperatures are believed to be the primary cause of the member
buckling and connection failures within the Nelson Unit 6 hot-side precipitators.
These local failures can accumulate over a period of time and could potentially
result in a catastrophic failure.
Slide Bearing Plates
Slide bearing plates are installed under the support points of precipitators and
ducts to accommodate the thermal movements, which can be as great as 10 inches.
Slide bearing plates must have a sufficiently high bearing capacity for the gravity
loads and a low coefficient of friction to facilitate free movement. Typical causes
of slide plate degradation are as follows:
• exposure to temperatures larger than they were designed to accommodate;
• bearing overload due to redistribution of the loads within the precipitator
and/or ductwork;
• exposure to nonuniform or point loading due to rotation of the support stub
columns, improper initial alignment, or rotation of the support beams; and
• accumulation of debris on the slide plate surface.
Temperature degradation is common in hot-side precipitators with short stub columns
requiring the slide plate to be either partially or completely encased within the
hopper insulation. Overloading and nonuniform loading is the result of differential
temperatures.
32-5
-------
As slide plates continue to deteriorate, their effectiveness is reduced and the
coefficient of friction will increase significantly. This increase imposes
additional restraints on the thermal growth of the precipitator and/or ducts, which
will become worse with time. Further damage can result to the precipitator internal
structure and/or the support steel structure as this effect become's additive with
the aforementioned structural layout effect.
Material Degradation
A major cause of structural-related problems associated with hot-side precipitators
is accelerated degradation of the structural material within the precipitator and
ductwork. The two key temperature- and time-related phenomena that reduce the
strength and service life of the steel are creep rupture and temper embrittlement.
Of these two phenomena, creep rupture is the most well known and was typically
considered as the primary material degradation concern in hot-side precipitators.
However, temper embrittlement was not identified as an industry problem until the
mid-1980s and recent experience has shown that the majority of the hot-side
precipitator failures can be attributed to brittle fracture. Generally, creep
rupture is considered to be a gradual phenomenon with the rate of degradation
increasing with time. The behavior of temper embrittlement, however, is believed to
be different, in that the majority of degradation occurs during the initial phase of
the material's operating life. The failures at Nelson Unit 6 that were investigated
were attributed to temper embrittlement.
Temper Embrittlement. Many of the common high-strength low-alloy steels are
susceptible to temper embrittlement when exposed to temperatures in the range of
700° to 1070°F . Of all of the structural steels commonly used in hot-side
precipitators, ASTM A242 Type 1 appears to be one of the structural steels most
susceptible to temper embrittlement. This steel was used for the precipitators and
ducts at Nelson Unit 6.
One popular explanation for temper embrittlement is that the alloy and uncontrolled
embrittlement elements in the steel tend to migrate to the austenite grain
boundaries. This migration decreases the cohesive strength of the boundaries and
provides a path for easy fracture. Brittle fracture can occur when the steel is
subjected to impact loading or very high stresses at an imperfection while below the
fracture transition temperature. However, there is no apparent reduction in the
strength and ductility of the steel under slowly applied static loading. Thus, a
normal tensile test will not reveal temper embrittlement.
The major embrittlement elements are phosphorus, impurities such as antimony, tin,
and arsenic, and to a lesser degree, manganese and silicon. The rate of
embrittlement is usually quite rapid initially but decreases with the time of
temperature exposure. This embrittlement rate when plotted against time takes on a
parabolic shape where the time to embrittlement is a function of the chemistry of
the steel. Figure 1 shows a typical plot of the embrittlement rate. There is no
scale to this curve because the embrittlement (loss of toughness) versus time
relationship varies with the steel chemistry and the temperature. Test data are not
currently available to accurately provide a curve for most steels.
32-6
-------
Loss of
toughness
Maximum loss for given temperature
Time
Figure 1. Temper Embrittlement
Temper embrittlement in hot-side precipitators has important structural significance
only when the steel is below the fracture transition temperature. All steels are
brittle below their fracture transition temperature, which is somewhere around 0°F
for new steel. However, temper embrittlement can raise this to over 250°F, as shown
in Figure 2. Temper embrittlement could cause cracking of structural members and/or
plates during the unit startup and shutdown when thermally induced stresses are at
their highest. More information on temper embrittlement is presented in References
3 and 4.
Brittle
Toughness
Test
temperature i
Ductile
Brittle
Fracture —
appearance
transition
temperature
Ductile
Temperature varies
depending on degree
of embrittlement and
on steel chemistry
Figure 2. Fracture Transition Temperature Curve
Creep Rupture. Creep deformation and rupture is a time- and stress-dependent
phenomenon in which a steel member continues to deform over a period of time under a
constant load until the steel ruptures. An increase in either stress or temperature
accelerates the creep process. The rate and extent of creep continues and increases
32-7
-------
with time, as shown in Figure 3. As the creep continues, microcracks develop and
propagate, which lead to voids and eventually result in the rupture of the steel.
If structural steel members are not designed keeping their sustained stresses below
their creep rupture failure level for their design life, the service life of the
structure is in jeopardy and failures can occur.
Daformatlon
Tertiary
creep
Creep deformation
Elastic deformation
Time
Figure 3. Creep Rupture Deformation Curve
High-strength, low-alloy steels, such as A242 Type 1, the steel used in the Nelson
Unit 6 precipitators, have significantly higher strength against creep rupture than
common carbon steel, such as ASTM A36. Creep rupture does not effect the steel at
temperatures below a certain level, depending on the type of steel. Creep does not
occur in A242 Type 1 steel at temperatures below 750°F, and in A36 steel creep
degradation can be expected at temperatures in excess of 650°F.
MATERIAL TESTING PROGRAM
Preliminary material testing performed during April 1988 indicated that the
precipitator and ductwork steel had undergone substantial embrittlement. At this
time, the degree of creep damage was not known, so a more extensive material testing
program was developed for the Nelson Unit 6 precipitator steel. These tests were
conducted in September 1988.
The objective of the precipitator and ductwork material testing program was to
determine the extent of current structural degradation and to establish a design-
basis approach for the continued use of an embrittled steel with possible creep
damage. This objective was combined with a structural analysis to establish the
modifications that were required to maintain the structural integrity of the
structures. GSU decided that a design basis should be developed for both hot-side
and future cold-side operation. Of particular concern was the fracture toughness
(degree of embrittlement) of the steel, the fracture transition temperature, and the
remaining creep life.
The success of any testing program is only as good as the data collection. In this
case, in order for the test results to be considered reliable, the test samples had
to be taken from the steel that is most degraded. There was a concern that this
would not be achieved. Included in the following is the procedure that was used,
since testing every piece of steel in the precipitators and ductwork was certainly
not practical. 32-8
-------
The material testing program consisted of two distinct and separate testing
subprograms: fracture toughness testing and creep rupture strength testing. These
subprograms were applied to the precipitators and ductwork separately since the
material for each probably came from different sources.
Basis for the Sampling Approach
Since testing every piece of steel in the precipitators was not practical, a
semistatistical sampling approach was used. To enhance the confidence of the test
results representing the worst case, a relatively large population of 60 samples for
each test parameter studied was chosen in both precipitator and ductwork testing.
The sample locations were determined using the informed selection process described
below.
Since the samples for the fracture toughness testing were not necessarily the same
as those for creep, a maximum of 240 samples was possible. However, combined sample
locations were chosen wherever possible to keep the total sample population as low
as possible. As a result, 138 small plug samples were tested initially for chemical
composition. The small plug sample locations were taken from structural members
with expected high stress levels and bracing members required to provide column
stability. Consideration was given to previously known locations of embrittlenient,
chemistry, and high temperature.
To reduce the amount of further testing and still maintain a high degree of
confidence in the test results, 30 large creep samples and 30 large fracture
toughness samples were selected for additional tests based on the chemical
composition results of the small plug samples. The following sections detail the
complete large sample selection criteria and the further testing.
Description of the Fracture Toughness Testing Program
Phase 1 - Initial Toughness Testing. As explained above, 30 large samples (220 in2)
were collected for preliminary toughness testing. The most common toughness test,
the Charpy V-Notch (CVN) test, was chosen for this further testing. These 30
samples were taken from the plug sample locations found to have the chemistry most
susceptible to temper embrittlement. From each of these samples, the following
tests were conducted:
• three CVN impact tests per ASTM E23 in both the longitudinal and transverse
directions in the as-received condition and
0 one CVN impact test per ASTM E23 in both the longitudinal and transverse
directions after the specimens have been properly de-embrittled.
Deembrittlement can be performed on the steel samples by the testing laboratory.
The CVN results from the deembrittled samples give a fracture toughness
approximating that of the new, unembrittled condition of the sample. Comparing the
as-received CVN test results to the deembrittled CVN test result provides an
indication of the degree of temper embrittlement.
The CVN test is performed using an ASTM procedure with a controlled specimen that is
fractured by impact. This test is used as a qualitative screening impact test of
unembrittled steel. However, the CVN test result is not an engineering parameter
that can be directly used in fracture mechanics. Since empirical correlations are
not available to estimate the fracture toughness of embrittled steel, the results
from this test can only be used for comparison between like samples. With no code
or specification method available for determining the acceptable CVN values
associated with embrittled steel, the engineer, in collaboration with a fracture
32-9
-------
mechanics expert, must make this determination.
The test average results for the Nelson Unit 6 steel ranged from 1.0 to
48.5 fflb, with the average being 13.5 ft'lb in the longitudinal direction and 5.7
fflb in the transverse direction. From these results, the most embrittled steel
samples (those with the lowest CVN values) were tested further as stated below.
Also as part of the material testing program, a fracture transition temperature
curve, such as shown in Figure 2, was generated. This curve was created by
performing a set of CVN tests in temperature increments of 50°F. The curve for the
samples taken from the Nelson Unit 6 precipitators indicated that the current
fracture transition temperature of the steel was 175°F. This value meant that the
steel would not be brittle when the unit is at the future cold-side operating
temperature of approximately 300°F. The result of this test is important because it
defines the point where brittle fracture occurs. All stress caused by differential
temperatures occurring below 175°F must be checked against brittle fracture, and all
stress occurring above 175°F could be checked against normal allowable stresses.
Phase 2 - Slow Bend Fracture Toughness Testing. From each of the seven samples with
the lowest CVN results and, therefore, the most embrittlement, the following
additional tests were performed:
• two tension tests in both the transverse and longitudinal directions per
ASTM A370 and
• three slow-bend crack tip opening displacement (CTOD) tests in the most
brittle direction at the following temperatures: 70°F, 120°F, and 170°F.
Two samples were tested in both directions.
The purpose of conducting the CTOD tests was to develop the fracture toughness value
representative of the worst embrittled steel in the precipitator at the actual
loading rate. The CTOD-determined value of fracture toughness is the key material
property used in determining the steel's susceptibility to brittle fracture. The
CTOD tests were performed at a slow rate of loading that actually was faster than
the loading rate seen by the precipitators in operation. The governing rate of
loading was that of the thermal stress, which was calculated from the thermal
gradient data collected during unit startup.
The results from the CTOD tests ranged from 0.6 to 32.6 mils. The average values
were of no concern since the decision had been made to use the lower bound of the
results. Utilizing the results from the CTOD tests, the embrittled precipitator
steel was more realistically assessed for how it responds to the actual field
loading. The CTOD test results were then converted to a design-basis stress that
should not cause a preselected defect to propagate. This test program is explained
in detail in a following section.
Description of the Creep Rupture Testing Program
A cold-side conversion would eliminate any further creep rupture degradation.
However, the steel had undergone an unknown amount of creep degradation from 6 years
of exposure to temperatures in excess of 750°F. The remaining creep rupture life
and the creep strength was unknown before the testing.
Phase 1 - Initial Creep Rupture Investigation. As explained above, 30 large samples
(120 in"1) were collected for creep rupture investigation. These samples were taken
from the plug sample locations found to have the chemistry most susceptible to creep
rupture. Each sample was subjected to the following tests:
32-10
-------
• chemical analysis,
• scanning electron microscope (SEM) examination of the steel to determine
the stage of creep, and
t two tension tests per ASTM A370.
The microscopic examination was performed to determine whether creep was present in
the primary, secondary, or tertiary stage. The tension tests determined the current
strength and ductility of the steel.
The microscopic examination revealed no evidence of any creep damage in any sample.
In all samples there was an absence of grain boundary voids.
Phase 2 - Accelerated Creep Rupture Testing. Previous research on creep rupture of
steels at elevated temperature shows that a correlation exists between the long- and
short-term creep rupture strength of steel. In order to determine this correlation,
a stress rupture curve must be developed. For the Nelson Unit 6 precipitator steel,
18 accelerated creep rupture tests were performed on each of six samples selected
from the Phase 1 program to produce a curve for each sample. The tests were
performed in accordance with ASTM E139.
The 6 samples were selected from the existing 30 samples that were tested in
Phase 1. These samples were taken from girder bottom flanges and truss struts.
They were chosen over others because of their expected sustained high level of
tensile stress.
The tests were conducted at temperatures varying from 900° to 1100°F and test times
from 3 to 1629 hours. From the curves generated from the test results, as shown in
Figure 4, the expected creep rupture stress at 100,000 hours at 850°F was
extrapolated. This is the expected maximum sustained temperature of operation and
the anticipated longest duration until cold-side conversion will be performed. The
extrapolated creep rupture strength using these limits was determined to be 18.5
ksi.
Since this is an ultimate, or failure, stress, a factor of safety had to be used in
the design. A decision was made to base the Nelson Unit 6 structural evaluation on
a creep rupture allowable stress of 12 ksi.
Stress (ksi)
40 •
• PA 16
O PB12
• PB17
P » (T * 460) (20 + log t) 110
,-3
Figure 4. Creep Rupture Test Results for Nelson Unit 6
(Larson-Miller Plot for A242 Type 1 Steel)
32-11
-------
FRACTURE MECHANICS ASSESSMENT
The evaluation of a structure with severely embrittled steel is not a common
occurrence. There is no code or specification to guide the engineer in this
process, but there is knowledge in the field of fracture mechanics that can be used
to conduct this work. Even though most metallurgists will advise that severely
embrittled steel should not be considered serviceable, methods can be used to
develop a defendable technical approach to deal with this problem short of
abandoning the structure. The problem of continued use of embrittled structural
steel can be managed if properly addressed. However, there are risks associated
with the type of evaluation that is discussed in this paper. See Reference 5 for a
more detailed description of this evaluation procedure.
Structural reliability for an embrittled material depends on many factors, such as
the magnitude of the loading, rate of loading, number of load paths, and the nature
of the loading, and not just on fracture toughness alone. In a hot-side
precipitator, the loading rate is low, there are usually multiple load paths, and
the thermal load is self-relieving. As a result, the likelihood of a sudden
catastrophic structural failure is considered remote, although much local damage
should be anticipated. The danger is that if the local damage is left to
accumulate, it can turn into major damage that could cause the unit to be taken out
of service.
Stress Intensity Factor
The results of the CTOD tests were converted to a critical stress intensity factor,
K_, by a qualified fracture mechanics expert. This parameter is related to both the
stress level and a material flaw or crack size. In order to have crack propagation,
a flaw must be present in the material being investigated. Or rather, when a
particular combination of stress and flaw size leads to unstable crack growth, the
value of KC is defined for this material. From this principle, equations have been
defined in fracture mechanics textbooks, such as Reference 6, that establish the
relationship between stress, crack type, flaw size, and K.,. Figure 5 shows the
equations for the three common types of crack: through thickness crack, surface
crack, and edge crack.
Through-
thickness
crack
Surface
crack
Where Q = f(a'C
When c 4a
Q 1 4
Edge crack
Figure 5. KC Values for Various Crack Geometries
32-12
-------
From the CTOD test results, the lower bound value of KC equal to 43 ksi
determined for the Nelson Unit 6 preclpitator and ductwork steel in its current
state. The values of Kc determined from this testing procedure are based on the
embrittled state of the steel at the time the samples were taken. The results are a
snapshot in time of the degree of embrittlement. At this point in the embrittlement
assessment, it was necessary to approximate the level of the embrittlement expected
at the end of the service life of the hot-side precipitators. The results of the
CVN tests, the fracture appearance transition temperature curve, the CTOD test
results, and the KC values were considered when performing this determination. The
operating history and future use of the precipitator were also important parts of
this determination and were carefully considered.
The test results indicated that a KC of 40 ksiVinch could be used for the Nelson
Unit 6 precipitator and ductwork steel for up to 7 additional years of hot-side
operation, as long as a prudent factor of safety is used. This decision was based
on the fact that most of the embrittlement will have occurred during the first 6
years of operation at 800° to 850°F.
Material Flaw Size and Type
As previously discussed, the crack propagation stress is a function of both KC and
the material flaw size. Notice from the equations for each crack type in Figure 5
that KC is directly proportional to the stress but proportional to the square root
of the crack length. Therefore, for a constant value of Kc, the stress required for
crack propagation increases to the second power as the crack length decreases. By
plotting the stress versus the crack length with a constant Kc for the three types
of crack, as determined in Figure 6, there are considerable differences in the
stress levels depending on the type of crack and the crack length chosen.
Applied
stress
Through-
thickness
crack
Critical crack length
Figure 6. Applied Stress Versus Critical Crack Size
In choosing the criteria for the crack type and length, field conditions were
considered. In making this decision, each type of crack was evaluated for the
potential of existing in the precipitators. Also, the crack length that could be
discovered during a structural examination was defined. Based on that evaluation, a
decision was made to base the design on a 2-inch-long through-thickness crack. This
decision was based on the belief that exposed edges were not prevalent and any edge
cracks that did exist could be easily found and repaired, the possibility of surface
cracks was remote, and discovering a 2-inch-long or longer through-thickness crack
in the various members during a normal precipitator and ductwork structural
32-13
-------
examination was a reasonable expectation. This decision was based on the known
accessibility to view cracks within the Nelson Unit 6 precipitators and the normal
degree of ash removal performed by GSU before the field examination.
Design-Basis Stress Determination
Once the value of KC and the type and size of flaw, c, was decided, the design-
basis stress was determined using the following fracture mechanics formula for a
through-thickness crack:
a =
Kc
From this formula using a through-thickness flaw size of 2 inches, the crack
propagation stress was determined to be 22.6 ksi for a KC of 40.
The design-basis stress, or the allowable stress considering crack propagation, is
the afore mentioned crack propagation stress divided by a factor of safety. The
factor of safety is necessary because the crack propagation stress is an ultimate or
failure value. The factor of safety chosen is the decision of the designer
performing this assessment and the station owner, since there are no codes to cover
this type of design. Items that must be considered are the remaining design service
life of the structure and the future operating temperature of the steel.
For Nelson Unit 6, a factor of safety of 2 was used for the assessment of the
structure for continued hot-side operation for the next 7 years and a factor of
safety of 3 was used for the assessment for future cold-side operation after
conversion. These values were used with the crack propagation stress of 22.6 ksi.
Therefore, as a result of the fracture mechanics assessment, the following design-
basis stress values were used:
Hot-side operation until 1995: 11.3 ksi
Cold-side operation after 1995: 7.5 ksi
Limiting the member stresses to these values is expected to control brittle crack
propagation in the steel. The application of these two allowable values is
explained in the following section of this paper.
STRUCTURAL MODIFICATIONS TO THE PRECIPITATORS
Allowable Stresses
From the material testing and the fracture mechanics and creep assessments,
allowable stresses were determined for the structural steel members of the
precipitators for brittle fracture and creep rupture for both hot-side and cold-
side operating conditions. It is important to realize the limitations to these
design-basis allowable stresses.
Creep rupture is only applicable to loading combinations with long-term sustained
stress that occur at temperatures when creep degradation occurs. The operating
temperature of the Nelson Unit 6 precipitators, over 800°F, is above this level.
Once the precipitators are converted to cold-side operation, creep is no longer a
concern. Based on the Nelson Unit 6 steel testing program, an allowable stress of
12 ksi was used for loading combinations that occur while the unit is on line at
steady-state hot-side operating conditions with sustained loading.
The fracture mechanics design-basis stress is applicable only to loading
combinations that occur when the steel is at a temperature below the steel's
fracture transition temperature. This is the temperature below which the steel is
32-14
-------
brittle and above which the steel is not brittle and is ductile. For the steel in
the Nelson Unit 6 precipitators, the fracture transition temperature was determined
by testing to be about 240°F. Therefore, the fracture mechanics design-basis stress
is only applicable for loading combinations that occur when the unit is off line,
going through startup, and going through shutdown. Based on the Nelson Unit 6 steel
testing program, an allowable stress of 11.3 ksi was used for hot-side operation
until 1993 and an allowable stress of 7.5 ksi was used for cold-side operation after
1993 for loading combinations that occur while the unit is going through startup
and/or shutdown.
For the most part, normal AISC allowable stresses were used for all other loading
combinations. The normal AISC allowable stresses was based on the steel's yield
strength at the design temperature of the loading combination. The yield strength
of steel varies from its full ASTM value at 80°F to approximately 66% of its full
value at 850°F. However, for stresses generated from loading combinations with
thermal loads that were not governed by creep rupture or brittle fracture, the
allowable stresses for the Nelson Unit 6 precipitators were based on an ASME
methodology. The methodology used was taken from the ASME Boiler and Pressure
Vessel Code, Section III, Division 1 - Subsection NF, Component Supports. Member
stresses generated by differential thermal expansion were considered to be secondary
stresses and, thus, their allowable stress is much higher. In this case, secondary
stresses are defined as being self-limiting and do not produce instability of the
structure.
The design using AISC allowable stresses would indicate that extensive precipitator
column reinforcing would be required because of the large thermal stresses. This
practice is not consistent with the philosophy of trying to eliminate thermal
stresses by making the structure more flexible. Stiffening the columns would
actually result in higher thermal forces within the structure. Reinforcing the
columns on the interior of the precipitators was also inconsistent with the actual
field conditions since little column damage was found on the internal rows. The use
of the ASME methodology produced a more flexible structure, which provides less
resistance to differential thermal expansion. This change results in a structure
where behavior is more predictable under differential thermal conditions. Combined
with the new bracing arrangement, the new column thermal stresses are less than the
thermal stresses with the original bracing arrangement. Therefore, since no damage
has been found to the precipitator interior column row columns, reinforcing of these
columns was not justified.
Loads
The gravity, live, pressure, ash, wind, and friction loads imposed on the
precipitator were calculated. The only difference between the hot-side and the
cold-side loading combinations is the precipitator internal pressure, which will be
less (more negative) during cold-side operation. This larger negative pressure will
increase the stresses in the structure during cold-side operation.
The thermal loads are generated by the structure itself from the differential
thermal growth of the various structural elements of the precipitator caused by
differential temperatures. In order to properly account for the thermally induced
stresses, the temperature of various structural members was field measured.
Thermocouples were installed at various locations within the precipitator and
temperatures were collected at enough locations within the precipitator to measure
the actual temperatures within the structure. Figure 7 shows the locations for the
thermocouples. Temperature data were collected on the precipitator inlet row steel
and the outlet row steel. Experience has shown that the largest differential
temperatures occur on the inlet and outlet rows, therefore temperatures on the
interior rows were not collected.
32-15
-------
Figure 7. Temperature Probe Locations
Once the thermocouples were installed, the temperatures were recorded during unit
start up, steady-state operation, and unit shutdown. Usually the differential
temperatures are largest during unit startup; however, during unit shutdown the
temperature differentials are usually reversed, although smaller. Depending on the
structural framing arrangement, these reverse differentials may actually induce
larger stresses within the structure, which cannot be determined until the
structural analysis is performed. For hot-side operation, the maximum differential
used in the analysis was 320°F during unit startup and 300°F during unit shutdown.
Temperature differentials were also required for the cold-side analysis loading
combinations. Since recording these values was not possible at this time, they were
prorated from the hot-side readings. For cold-side operation, the maximum
differential used in the analysis was 135°F during unit startup and 105°F during
unit shutdown.
Structural Analysis
The structure was analyzed for two different operating conditions, hot-side
operation for the next 7 years and future cold-side operation for the remaining life
of the station. This computer analysis was performed using the structural analysis
computer program STAAD III.
The following loading combinations were investigated for both hot-side and cold-
side operation:
1. normal operation using operating pressure and steady-state differential
temperatures but no live load;
2. normal operation using maximum design pressure, steady-state differential
temperatures, and full live load;
3. unit startup differential temperatures with other normal loads;
4. unit shutdown differential temperatures with other normal loads; and
32-16
-------
5. normal operation with full design wind and steady-state differential
temperatures.
For loading combination 1, the creep rupture allowable stress was used for hot-side
operation and the normal allowable was used for cold-side operation. The creep
rupture allowable is used because all of the loads in this combination are sustained
loads and creep rupture is a design occurrence only for hot-side operation. For
loading combinations 2 and 5, the normal AISC allowable stresses and/or the
allowable stress adopted from ASME Subsection NF were used for both hot- and cold-
side operation. The members were designed for the stresses without the thermal
contribution using the AISC normal allowable stresses and were checked for the
stresses with the thermal contribution using the allowable stress adopted from ASME
Subsection NF. These loading combinations are based on the precipitator being at
its operating temperature, which is above the steel's fracture transition
temperature, with maximum design loads. Therefore, brittle fracture is not possible
and thus is not a design consideration. Creep rupture is not possible since the
design loads are transient, not sustained. For loading combinations 3 and 4, the
appropriate fracture mechanics design-basis stresses were used for hot- and cold-
side operation since these loads can occur while the unit is coming on- or offline
when brittle fracture is possible. Normal AISC allowable stresses were not used
because they are much higher than the brittle fracture allowables.
Since the measured differential temperatures exceed those assumed in the original
design and because of the low fracture mechanics design-basis allowable stress, many
of the members were overstressed. Because this occurred, the structure was modified
so that the stresses are below the appropriate design-basis allowable stress.
However, in order for this assessment to be successful, the stresses in each member
must be below the appropriate design-basis allowable stress for each loading
combination.
Structural Modifications
The following modifications to the structure were necessary:
• replace the precipitator inlet row and the inlet duct interior columns;
• revise the inlet duct to be cantilevered off of the precipitator;
• revise the precipitator bracing arrangement to increase flexibility;
• delete half of the precipitator external supports;
• replace the slide bearing plates under the remaining supports;
• reinforce columns, girders, and various member connections; and
• repair the discovered damage including all flaws greater than the crack
length assumed in the assessment.
Design drawings were then prepared to indicate these structural modifications to the
precipitators and ducts so that the modifications could be installed during the next
sustained unit outage. These structural modifications involve over 250 tons of new
structural steel plate and shapes and the installation of 54 new slide bearing
plates. Figure 8 shows the modifications to the inlet column row bracing. The
construction outage is scheduled to last 13 weeks starting April 3, 1990.
32-17
-------
I
Original bracing arrangement
Revised bracing arrangement
Figure 8. Precipitator Inlet Row Bracing Modification
With embrittled steel or steel susceptible to creep rupture, continuous monitoring
of the structure in the form of field examinations is imperative, even after all
structural modifications are implemented. Repairs of local cracking and other minor
damage are to be anticipated. The frequency of future maintenance examinations
should be determined based on the precipitator1s structural performance record. It
is planned to implement this activity at Nelson Unit 6. Any future fracture
surfaces discovered should be laboratory tested to determine the mode of failure as
part of the structural monitoring program evaluation. The failure phenomena must be
investigated to determine whether the material degradation is advancing.
CONCLUSION
Structural problems with hot-side precipitators depend on specific parameters
related to each installation. The operating temperature, precipitator size, bracing
layout, steel chemistry, structural condition of the precipitator, and original
structural design parameters all affect the types of modifications that would
resolve the problems. Each installation requires an individual and comprehensive
study to determine the best course of action for that unit.
Sargent & Lundy and GSU have developed a solution to address the hot-side
precipitator structural problems at Nelson Unit 6. The methods presented here for
dealing with the temper embrittlement and creep rupture problems at Nelson Unit 6
32-18
-------
may also be appropriate for other hot-side precipitators. However, there is little
practical experience available regarding the use of an embrittled structural system,
and therefore, there is an associated risk. This risk can be minimized by utilizing
a conservative material test program and design philosophy to develop lower-bound
design values. In each case, the risk associated with using this method must be
evaluated against other alternate particulate emission control strategies.
REFERENCES
1. Werhane, J. A., Schneider, R. L., and Vacek, M. G., "Hot-Side Precipitator
Structural Problems, Solutions, and Monitoring," presented at the EPRI Workshop
on Hot-Side Electrostatic Precipitator Technologies, May 1987, Birmingham,
Alabama.
2. Zaben, 0., and Fang, S. J., "High-Temperature Effects on the Structural
Performance of Hot Precipitators and Ductwork," American Power Conference,
Spring 1988, Chicago, Illinois.
3. Lamping, G. A., and Watson, P., "Mechanical Properties of A242 Structural Steel
After Exposure in a Hot-Side Precipitator," presented at the EPRI workshop on
Hot-Side Electrostatic Precipitator Technologies, May 1987, Birmingham, Alabama.
4. "Embrittlement of Steels," Metal Handbook, Vol. 1, 9th Edition, ASME, 1978, pp.
683.
5. Schneider, R. L., and Zaben, 0., "Structural Integrity Assessment of Embrittled
Structural Steel in Hot Precipitators," American Power Conference, Spring 1989,
Chicago, Illinois.
6. Barson, J. M., and Rolfe, S. T., "Fracture and Fatigue Control in Structures -
Applications of Fracture Mechanics," Prentice-Hall Inc., New Jersey, 2nd
Edition, 1987.
32-19
-------
COLUMBIA UNIT 2 PRECIPITATOR
HOT TO COLD CONVERSION
M. Vakili
Wisconsin Power and Light Company
Madison, Wisconsin
A. W. Ferguson
Black and Veatch, Engineers-Architects
Kansas City, Missouri
33-1
-------
Introduction
The Columbia Station Unit 2 of Wisconsin Power and Light Company is a 527 MW
coal fired unit, located south of Portage, Wisconsin. This unit started
commercial operation in 1978. It burns low sulfur (0.4 percent) Powder River
Basin western coal and has been in operation since 1978. The unit was
originally equipped with a hot side electrostatic precipitator for particulate
control. Due to a degradation of performance and the subsequent need for
frequent manual cleaning, WP&L decided to convert the precipitator to cold side
operation. This was accomplished by changing the gas path, so the flue gas is
passed through the air preheaters prior to entering the precipitator.
Background
The New Source Performance Standard (NSPS Subpart D) limit of 0.1 #/MBtu for
particulate and 20% for opacity is required for this unit. The unit is
equipped with two, weighted wire type electrostatic precipitators. The
precipitators are arranged in a chevron configuration. The collecting plate
centerline spacing of the precipitators is 9 inches. Each precipitator
consists of four gas tight chambers with 43 gas passages per chamber. The
electrical configuration of these precipitators, prior to conversion, was five
fields in the direction of the gas flow and four bus sections across the gas
flow. The flue gas temperature to the precipitator, prior to conversion, was a
maximum of 850 F.
Shortly after initial operation, precipitator performance degraded presenting
concerns over compliance with the opacity limit. In order to stay in opacity
compliance, the unit load had to be reduced. This performance degradation was
due to a buildup of tenacious ash on the discharge electrodes and collecting
plates. In order to increase the precipitator's particulate removal
performance, it had to be manually cleaned. The cleaning at first took place
during the planned scheduled maintenance outages, and later, during forced
outages. Since 1987, a number of upgrades, modifications, and experiments have
been tried in order to significantly improve the precipitator's performance.
The following are some of these attempts at improvement:
Tested various coal supplies
Tested off power rapping
Tested different T/R controllers
Tested a low frequency sonic horn
Tested different rappers and vibrators
Tested two different pulse energization systems
Tested sodium conditioning of coal.
Prior to converting the precipitator from hot side to cold side operation, the
need for frequent cleanup required an outage every five to seven weeks. The
outage duration required for this cleanup was about four days.
Due to acid rain legislation passed in Wisconsin, starting in 1993, the total
emission limit from all of our coal fired generating stations cannot exceed 1.2
Ib/MBtu total S02 emission. This requires a high capacity factor for the
units with lower SCL emission. Based on this requirement and the need for
more reliable plant performance, the Columbia unit 2 precipitator was converted
from hot side to cold side operation in the fall of 1988.
33-2
-------
Hot to Cold Conversion
Columbia Unit 2 precipitators have some unique characteristics that made this
conversion different than the previous conversions by other utilities. Some of
these characteristics are as follows:
• The size of the boxes. The total square footage of collecting plates
per 1000 ACFM of gas flow (SCA) of the hot side precipitators was
about 268, at a flue gas temperature of 810 F. This SCA, after the
conversion, increases to 320 at 300 F., which is very small in
comparison with the other recently converted precipitators.
0 The ductwork arrangement. The physical arrangement of the boiler, in
relation to the induced draft fans and the air preheaters, limited the
options for ductwork redesign.
• Precipitators' configuration. The precipitators are arranged in a
chevron configuration. There was a significant fly ash buildup in the
precipitators' inlet nozzle areas and the chevron section of the inlet
ducts. These areas of ductwork had to be redesigned in order to
eliminate fly ash buildup, caused by low gas velocity.
• Extent of precipitator sectionalization. In order to increase the
reliability of the precipitators, the number of electrical fields was
significantly increased.
Studies
1. Boiler and Air Preheater Study
The boiler manufacturer was contracted to conduct a boiler and air preheater
performance study, in order to improve the boiler and air preheaters'
performance, and, therefore, reduce the flue gas exit temperature. This study
evaluated the boiler and air preheaters' performance independently of the
precipitator operation.
As a result of this study, 10 retractable soot blowers were added to the
boiler, coal burners were modified, soot blowers were added to the hot side of
each Ljungstrom air preheater, and air preheater baskets were replaced with the
loose packed type. The new basket elements are six inches taller than the old
ones. These modifications were done prior to the precipitator conversion.
With these modifications, the boiler exit gas temperature, at full load, was
reduced by 60 F. These modifications also eliminated slagging and fouling at
the economizer, the superheat, and the reheat sections of the boiler. They
also reduced the reheat and superheat steam temperature mismatch.
2. Precipitator Study
In order to predict the performance of the precipitators after the hot to cold
conversion, an independent consultant was retained. This study also included a
detailed evaluation of the physical condition of the precipitators and the
ductwork.
33-3
-------
For the performance prediction, typical coal characteristics from twelve Powder
River Basin mines were used. The analysis was done for two temperatures, both
300 F. and 370 F. Precipitator performances, both with and without an
SO, fly ash conditioning system were also considered.
The SO, fly ash conditioning effect was included in this study, because the
preliminary performance prediction that was done by equipment suppliers and
A&Es indicated that due to the size of the box, the fly ash resistivity would
have to be reduced, in order to achieve successful operation of the cold side
precipitators.
A physical survey of the precipitators indicated that they were in very good
physical condition. However, a number of cracks were found in the ductwork
plates and internal trusses. The good condition of the precipitators was
credited both to maintaining the boiler exit gas temperature below 850 F. and
the use of moderate cleaning practices. Ductwork internal plate cracks and
truss damage were attributed to thermal distortion. The bulk of the ductwork
was unusable.
The cold side precipitator performance study indicated that opacity would
potentially exceed 20%. without a fly ash conditioning system and at a flue gas
temperature around 370 F, The study also indicated that, at a flue gas
temperature of about 300 F and with an SO, fly ash conditioning system, an
opacity level of 10% or less could be achieved, even when burning the worst
coal.
This study was based on the precipitators being in good physical condition.
The precipitator components were fully evaluated to determine what should be
upgraded, in order to optimize their performance. As a result of this study,
the following improvements were implemented:
• Decreased the number of plates being rapped by each plate rapper, from
9 plates per rapper to 4 and 5 plates per rapper. This was achieved
by relocating the trailing edge rappers to the leading edge of the
plates and dividing the plate support anvils.
• Replaced the discharge electrode wire vibrators with a magnetic
impulse gravity impact (MIGI) type. These rappers are identical to
the plate rappers.
• Replaced the discharge electrode rapper insulators from a bolted
connection type to a tapered shaft type.
• Added discharge electrode support insulator heaters.
• Replaced the rapper control system with a system having more
operational flexibility.
• Replaced the transformer rectifier current limiting reactors in order
to better match the TR sets.
• Increased the volume of conveying air in the fly ash handling system.
• Installed blanket type hopper heaters on the lower one third of the
fly ash hoppers.
33-4
-------
• Increased the number of electrical fields in the direction of the gas
flow from five fields to eight fields to increase precipitator
reliability. The total number of TR sets for each precipitator was
increased from 20 to 32. Figure 1 shows the electrical field
arrangement, before and after the sectionalization. These changes
were accomplished by doubling the number of supports for each of the
first three fields and splitting the discharge electrode support
angles in half.
• To avoid congestion at the precipitator roof area after the addition
of 12 more TR sets in this area, a hot penthouse was added to each
precipitator half. This modification required the removal of all the
components on the precipitator roof, the addition of a false floor
about 6' above the existing floor, and the replacement of TR sets,
rappers, the insulator heating system, and the rapper control cabinets
on this floor. With this modification, most of the insulator
compartments above the precipitator penthouse roof were eliminated.
The new floor arrangement allowed the replacement of TR sets, using
the existing monorail system. Figure 2 shows the precipitator
penthouse floor arrangement, before and after the conversion.
3. Model Study
The design of the ductwork was conducted considering another requirement of
good precipitator performance a uniform gas flow distribution at the
precipitator inlet. The average gas velocity at the precipitator inlet, after
the hot to cold conversion, was estimated to be about 5.8 feet per second.
Since this velocity was determined to be higher than desirable, it was very
important to achieve uniform velocity distribution at the precipitator inlet.
In order to eliminate fly ash dropout, the ductwork was designed for a gas
velocity of about 60 feet per second. A model study was done at 1/12 scale
using two separate scale models. The first model consisted of one of the
economizer outlet ducts to the air preheater inlet. The second model included
the two air preheater outlet ducts, the redesigned precipitator inlet ducts,
the precipitators and the precipitator outlet ducts to the two I.D. fan inlets.
Limited space due to the existing chevron arrangement of the precipitators,
resulted in the new nozzles being only ten feet in length. To eliminate any
ash buildup on the nozzle floor, the floors of the nozzles are sloped 60
degrees. The precipitator inlet expansion joints were removed and replaced by
nozzle inlet expansion joints. This was done in order to reduce the horizontal
surfaces prior to the precipitator inlets.
A number of flow distribution arrangements were tried, before a final
arrangement was selected. The final configuration of the nozzles incorporated
three rows of perforated plates with various porosities and egg crate flow
straighteners upstream of each row of perforated plates. Although the IGCI
limits were not achieved at the precipitator inlets, a good distribution of gas
flow was accomplished. Figure 3 shows the precipitator inlet and outlet
nozzles design.
33-5
-------
Electrical Fields After Conversion
Electrical Fields Before Conversion
Figure 1
33-6
-------
a n a a a
a n or a a D
Penthouse Floor Plan Before Conversion
o
o
«/
o
0
o
o
0
o
r
/
I
1
o
R
d
o
o
R
d
o
— Plate
o o
R
d
0 0
o o
R
d
o o
Rapp
o
o
o
o
o
o
0
o
o
;r
0
o
o
o
Q
0
o
o
o
o
R
b
o
o
R
d
o
1—
1
o
o o o
fl
d
o o o
o
0
o o o
R
U
o o o
D. E.
o
o
o
Q
o
0
o
o
Rapper
o
R
d
o
o
R
d
o
fl
o
o
o o o
R
0
d
o o o
o
o
o o o
R
Q
d
O O O
o
o
o
0
o
o
o
o
o
R
d
o
o
R
d
o
o
o o o
R
0
d
o o o
o
o
Q O O
fl
0
d
o o o
Gas Flow
1
Penthouse Floor Plan After Conversion
Figure 2
33-7
-------
REDESIGNED
OUTLET DUCT
REDESIGNED
INLET DUCT
I FROM
I AIR
HEATERS
FROM
AIR
HEATERS
REDESIGNED
OUTLET DUCT
INLET AND OUTLET NOZZLE DESIGN
PERFORATED PLATES
AND EGG CRATE VANES
PERFORATED PLATES
AND EGG CRATE VANES
10'
SIDE VIEW INLET NOZZLE
TOP VIEW 1/2 INLET NOZZLE
Figure 3
33-8
-------
Construction
The precipitator and ductwork modifications were accomplished during two 10
hour shifts, in a 13 week planned unit outage. The peak labor force during the
construction was 360 workers. The ductwork erection and the precipitator
sectionalization were performed by a general contractor, Research-Cottrell.
Research-Cottrell supplied the ductwork structural support steel, the ash
hopper heaters, and the precipitators' electrical equipment. Ductwork steel,
expansion joints, and automatic voltage controllers were supplied by other
contractors.
The construction management was performed by Wisconsin Power and Light. Other
contractors during the construction period installed the SO- fly ash
conditioning system, removed the asbestos expansion joints, and performed
miscellaneous electrical work. A major turbine overhaul was also done during
the same outage.
Cold Side Precipitator Performance
The converted precipitator was placed in service in December of 1988. The
opacity, at the unit maximum capacity, is about 6%. Maximum opacity, during
upset condition, is about 12%. Precipitator efficiency tests were performed in
April of 1989. The following are the average results of three test runs:
Inlet grain loading 1.7184 gr/dscf
Outlet grain loading 0.0084 gr/dscf 6
Outlet emission 0.0180 pounds/10 Btu
SO., Injection Rate 6 PPM
Removal efficiency 99.5 %
No precipitator degradation or opacity spikes have been noticed. Since the
conversion, there have been a number of broken discharge electrodes. The mode
of failure appears to be fatigue. Also, a number of discharge electrode rapper
support insulators have failed. The discharge electrode wire failure and the
rapper support insulators failures have not affected the precipitators'
performance. There have been no forced outages due to precipitator problems.
The operators are now able to ramp the load as fast as needed, without any
opacity exceedances. The S03 gas conditioning system has been operating very
well.
The unit availability in 1989 was 92.47%, compared to 76.3% in 1986 and 64.77%
in 1987. Precipitator operation had a major impact on the lower availability
numbers in 1986 and 1987. In addition the boiler efficiency has increased by
1-1.5%. This efficiency improvement is attributed to the boiler modifications,
the air preheater basket replacements and the precipitator hot to cold
conversion.
33-9
-------
Conclusion
Conversion of the Columbia Unit 2 precipitators from hot side to cold side
operation has been very successful for UP&L. Prior to the conversion, the unit
load often had to be limited, so as not to exceed the unit opacity limits.
Frequent unit outages were required to clean the precipitator. There were
forced outages of the unit, in order to clean the precipitator. Wisconsin
Power and Light is very pleased that the precipitator performance has enhanced
the unit availability and improved plant over all operation.
33-10
-------
DESIGN OF PULSE GENERATORS FOR PPCP APPLICATIONS, MASSIMO REA
(No paper provided)
34-1
-------
IMPROVED CARBON PARTICULATE CONTROL
VIA ADDITIVE INJECTION
D. Farrar - University of Toronto, CANADA
J. Reuther - Battelle Columbus, USA
W. Steiger - Battelle Frankfurt, GERMANY
R. Schmitt - Battelle Geneva, SWITZERLAND
R. van der Velde - Velino Ventures, CANADA
March 20-23, 1990
STATEMENT OF THE PROBLEM
Quantitative burnout of carbon in flames of solid (coal) and
liquid (petroleum) fuels is a continuing combustion-
engineering challenge, made more difficult by the use of
many NOx-control technologies
COMBUSTION TECHNOLOGY NEED
Chemical catalysis of combustion chemistry to promote fuel-
carbon burnout at lower flame temperatures in the presence
of less available oxygen
35-1
-------
PROJECT OBJECTIVE
Experimentally determine, with certainty, the effect of a
prototype additive, called Carbonex, on the combustion
performance of various raw or refined fuels in laboratory
and field cornbustors
CARBONEX
Organoiron in solvent carrier: xylenes or 2-ethyI hexanol
RAW/REFINED FUELS TESTED
• High-volatile bituminous and lignite coal
• Crude oil
• Heavy petroleum oil
• Light petroleum oil
• Diesel
35-2
-------
COMBUSTION DEVICES UTILIZED
• Laboratory multifuel furnace
• Laboratory diesel engine
• Field package boiler
GROUND RULES FOR CARBONEX EVALUATION
• Standardized experimental conditions
• Well defined/controlled experiments
• Extensive parametric testing
• Realistic operating conditions
• Challenging combustion situations
• Strict quality control measures
COMBUSTION PARAMETERS EVALUATED
• Firing rate: 0.5-10 x 106 Btu/hr
• Excess combustion air: 5-25-250%
• Secondary coal combustion air swirl
• Average coal size: 20-50 microns
• Additive use: premixed-injected
• Additive [Fe]: 0-0.05-1.0-5.0 ppm
• Effect of carrier: with-without Carbonex
35-3
-------
COMBUSTOR TEST FACILITIES
Laboratory muitifuei furnace
• Field package boiler
• Laboratory dlesel engine
35-4
-------
CRITICAL COMBUSTION INDICATORS
• CO2, CO, NOX, HC in flue gas
• Carbon, ash contents of participates
• Mass loading of participates
• Size distribution of participates
QUALITY CONTROL CRITERIA
• Single fuel batches
• Replicate testing
• Random test sequences
• Redundant combustion diagnostics
• Prescribed test data tolerances
• Experimental bias
• Interlaboratory corroboration
SIGNIFICANT EFFECTS OF CARBONEX
• Increased combustion efficiency
• Reduced paniculate emissions
• Reduced hydrocarbon emissions
• Reduced carbon monoxide emissions
• Reduced NOX emissions
35-5
-------
MECHANISM FOR CARBONEX ACTIVITY
Although not elucidated in this practical evaluation, a
plausible mechanism for Carbonex activity can be derived
from precedence for catalysis of combustion by iron-based
additives
PRACTICAL EFFECTS OF CARBONEX
Poorer quality fuels can be fired in less tuned combustors
at lower excess air levels without sacrificing combustion
efficiency and with lower NO emissions
SUMMARY OF INDEPENDENT EVALUATIONS
Velino Ventures Carbonex is a viable additive in stationary
and mobile combustors for the improvement of
environmental and economical performance
35-6
-------
LAB RESEARCH SUMMARY: BITUMINOUS COAL
Standardized experiments and analyses were performed by Battelle
under laboratory-simulated utility-boiler conditions to determine, with
certainty, the effect of a coal combustion catalyst called Velino Ventures
Carbonex (xylene + organoiron) on the combustion performance of high-volatile
bituminous and lignite coal. These tests employed either a pilot-scale
furnace to obtain direct data on combustion performance, without or with
Carbonex, and/or a thermogravimetric analyzer to obtain indirect data on the
same. The influence of flame-injected or pretreated Carbonex iron (5 ppm) was
evaluated at two levels of excess combustion air (10 or 26%) as a function of
secondary combustion air swirl (tuned or detuned) and coal particle size (60
or 80% minus 200 mesh). The characteristics of neat pulverized coal flames
injected with an equal volume of iron-free xylene carrier established baseline
combustion performance.
The following were the significant findings of the evaluation:
• The ability of 5 ppm Carbonex iron to enhance the
combustibility of already efficiently burning bituminous or
lignite coals was unambiguous proof-of-concept for Carbonex as
a combustion catalyst.
• The use of Carbonex in pulverized bituminous or lignite coal
flames offers a means by which to maintain high levels of
combustion efficiency (99+%) at reduced levels (10-%) of excess
combustion air.
• The use of Carbonex allows strategic technology supplements to
low-excess-air-firing, such as swirl and coal particle size
modification, to be used effectively to control NOX emissions,
without a penalty to combustion efficiency.
• Carbonex usage offers an effective means by which to achieve a
net reduction in NOX emissions to compliance levels, while
maintaining efficient pulverized coal combustion.
• Carbonex usage will probably enhance the performance of
electrostatic precipitation for particulate control.
In summary, this study confirms that Velino Ventures Carbonex is a
viable catalyst for the improved combustion of bituminous and lignite coal.
35-7
-------
LAB TEST DATA SUMMARY: BITUMINOUS COAL
SUMMARY TABLE
RESULTS OF COMBUSTION TESTS ON PULVERIZED BITUMINOUS COAL FLAMES
INJECTED WITH CARBONEX AS A FUNCTION OF SWIRL AND PARTICLE SIZE
Tut
Ho.
Ml
Al
61
M2
tt
B2
M3
A3
O
M4
A4
84
CirboMi Burner
Agtnb Iron Swift
InjKUd (pel, rt) (.rb)
Xiltnt
Xyltn*.
>jilm
Cirtaonu
Ctrfconu
Cirbtnu
Sjltot
Xgrltnt
CiraoMi
Carbonu
Carixmi
Tuwd
DoUjnW
Twwd
Twwd
Dttuiwd
Tua«l
TuftM1
Oatunad
TUM^
Turwd
Dttwntd
Tunod
Coil
Grind
(X-200 M*h)
10
N
60
M
M
60
60
60
60
10
60
60
IteM-totfl
Co»l Silt
(aicrofu)
22
22
M
22
22
M
22
22
U
22
22
U
EsctM
Coibuition
Air
m
10
10
10
10
10
10
26
26
26
26
26
26
NO,,
Eflimsionj
(pp., 01 02 Or,)
MO
400
500
660
(70
5(0
610
600
660
1000
600
720
PirticuItU
Silt
(•icront)
10
14
15
6
10
U
6
U
14
7
11
U
Particulttt
Losing
(lb/109 Btu)
(.0
(.4
7.6
4.1
4.2
(.4
4.5
J.I
4.4
J.2
2.1
3.4
Coabufttion
Eff icitftcjr
(D
99 4
95.0
95 6
99.6
96.1
96.2
99.1
•6.5
97.0
99.9
99.0
96.7
35-8
-------
LAB RESEARCH SUMMARY: LIGHT AND HEAVY OIL
Standardized tests were conducted by Battelle under simulated boiler
conditions to determine with certainty the effect of Velino Ventures Carbonex
(xylene + organoiron) on the combustion performance of surrogate light and
heavy Swiss-like oils. The influence of three levels of Carbonex iron (0.2,
1.0, and 5.0 ppm) was evaluated at two levels of excess combustion air (10 and
26 percent).
The reproducible results indicated the following significant
findings:
• Pretreatment of heavy oil with 1+ ppm Carbonex iron:
Reduced particulate carbon 90-97 percent,
Increased combustion efficiency up to 1.6 percent,
Reduced average particulate size 50 percent, and
Increased NOX emissions up to 12 percent.
• Reducing excess air from 26 to 10 percent, with the Carbonex-
treated heavy fuel, resulted in a net reduction in NOX of 11
percent while maintaining combustion efficiency at 99.90+
percent.
• Pretreatment of light oil with 0.2+ ppm Carbonex iron at 10
percent excess air:
Reduced particulate carbon by 38 percent,
Reduced particulate loading by 33 percent,
Had no influence on CO, NOX, particulate size, or the
already acceptable combustion efficiency,
effects that were equivalent to increasing the excess air to 26
percent.
• The pretreatment of the Swiss-like oils with 0.2-1 ppm Carbonex
iron allowed them to be combusted at 10 percent excess air
without reducing the combustion efficiency achievable at 26
percent excess air or compromising the lower NOX emissions
achievable at this reduced level of excess air.
35-9
-------
LAB TEST DATA SUMMARY: LIGHT AND HEAVY OIL
RESULTS OF COHBUSTION TESTS OH SURROGATE LIGHT SWISS OIL
PRETREATED WITH 0.2 TO 5 PPM VELINO VENTURES CARBONEX
IRON AT VARYING LEVELS OF EXCESS COHBUSTION AIR (10-26%)
Agent Pretreated Into Oil
Furnace Exit Temperature (F)
02 (V, volume) , .
C02 (V, volume) (A>
CO (ppm, volume)**)
NOX (ppm, volume)13'
Carbon in Particulate Ash (V, weight)
Particulate Loading (pounds/million Btu)
Average Particulate Size (microns)
Combustion Efficiency (V)
Xylene
(0 ppm Iron)
Excess Air
10% 26%
2230 2200
2.2 4.8
16.8 16.6
40 35
100 110
5.2 3.2
0.0006 0.0004
<1 <1
99.99+ 99.99+
Carbonex
(0.2-5 ppm Iron)
Excess Air
10V 26V
2220 2250
2.2 4.6
16.6 16.8
40 35
105 115
3.2 3.1
0.0004 0.0004
<1 <1
99.99+ 99.99+
(a) As-measured concentrations corrected to 0* oxygen basis.
RESULTS OF COMBUSTION TESTS ON SURROGATE HEAVY SWISS OIL
PRETREATED WITH 0.2 TO 5 PPM VELINO VENTURES CARBONEX
IRON AT VARYING LEVELS OF EXCESS COMBUSTION AIR (10-261)
Agent Pretreated Into Oil
Xylene OR Carbonex
(0 ppm Iron)
(0.2 ppm Iron)
(1
* Excess A1r »
Furnace Exit Temperature (F)
02 (%, volume)
C02 (%. volume) (»)
CO (ppm, volume) (•)
NO, (ppm, volume) (•)
502 (PP*. volume) (•)
Carbon in Paniculate Ash (%, Might)
Particulate Loading (pounds/million Btu)
Average Paniculate Size (microns)
Combustion Efficiency (%)
10%
2200
2.3
16.6
40
275
1600
16.5
0.06
10
98.32
26%
2095
4.8
16.8
30
315
1575
7.3
0.06
10
99.43
10%
2260
2.2
16.8
30
280
1550
0.68
0.06
5
99.90
Carbonex
ppm Iron)
Excess Air
26%
2150
4.6
16.8
25
355
1625
0.73
0.06
5
99.90
Carbonex
(5 PF
m Iron)
f Excess Air
10%
2270
2.6
16.9
37
300
1625
0.40
0.06
5
99.95
26%
2175
4.9
16.9
26
400
1600
0.43
0.06
5
99.95
(a) As-measured concentrations have been corrected to a 0% excess oxygen, or 'air-free', basis.
35-10
-------
FIELD RESEARCH SUMMARY: LIGHT OIL
Standardized combustion tests were conducted under actual industrial
boiler conditions in the field (Geneva, Switzerland) to determine with
certainty the effect of Velino Ventures Carbonex (xylene + organoiron) on the
performance of flames of light petroleum oil. These combustion tests were
conducted by a team of scientists and engineers from Battelle laboratories in
Frankfurt, Geneva, and Columbus. The influence of one concentration of
Carbonex iron (1 ppm) premixed with the light oil was evaluated as a function
of boiler load (firing rate) and excess air.
The measurable and reproducible results of the field evaluation
provided the following information:
• Pretreatment of light Swiss oil with 1 ppm Carbonex iron:
Reduced particulate loading by 67 percent and volatile
particulate fraction by 68 percent at full load and low
(12 percent) excess air,
Reduced particulate loading by 34 percent, particulate
carbon by 20 percent, and particulate volatile fraction by
33 percent at half load and ultralow (8.5 percent) excess
air, and
Had no direct effect on NC" , CO, or HC emissions,
particulate size, or the already acceptable combustion
efficiency.
• With 1 ppm Carbonex iron premixed into the light oil, the
industrial boiler could be fired at about 9 percent excess air
at half load with acceptably low CO and zero smoke emissions,
whereas without Carbonex, such performance was only achievable
at much higher levels of excess air (30+ percent).
• About a 2 percent overall additional reduction in all pollutant
emissions, including NOX could be realized by virtue of the
increase in thermal efficiency gained by being able to cleanly
and stably fire the boiler at reduced levels of excess air.
35-11
-------
FIELD TEST DATA SUMMARY: LIGHT OIL
RESULTS OF FIELD COMBUSTION TESTS OH LIGHT SWISS OIL
PRETREATED WITH 1 PPM VELINO VENTURES CARBONEX
IRON AT VARYING LOADS AND EXCESS AIR LEVELS
Agent Pretreated
Xylene
Load (V)
Firing Rate (kW)
Excess Air (*)
Furnace Exit Temperature (C)
02 (*. volume) , .
C02 (%, volume) (i)
CO (ppm, volume) (?'.
NCL (ppm, volume!1?'
He (ppm, volume) ^ '
Carbon in Particulate Ash (V, weight)
Volatile Particulate Carbon (V, weight)
Particulate Loading (mg/«r)
Average Particle Size (microns)
Combustion Efficiency (*)
(0 pp«
50
1500
27.1
133
5.9
10.9
43
113
4.7
13
0.39
-------
LAB RESEARCH SUMMARY: DIESEL FUEL
Standardized combustion tests were performed by Battelle Columbus
Division in a production diesel engine to determine with certainty the effect
of various levels of the additive called Velino Ventures Carbonex Diesel.
This diesel engine additive consisted of an organoiron compound diluted in a
2-ethyl hexanol carrier. The influence of two levels of Carbonex Diesel iron
(0.04 and 0.08 ppm) was evaluated in a 4-stroke, 6-cylinder, 4300 cubic-inch
diesel engine, rated at 1400 brake-horsepower and 1200 revolutions per minute
at full load. The diesel engine was operated at about 85 percent of full
load to artificially create a particulate emissions problem.
The measurable and reproducible results provided the following
information:
• The addition to conventional diesel fuel of 0.04 to 0.08 ppm
Carbonex Diesel iron:
Reduced CO emissions 7-10 percent,
Reduced HC emissions 3-9 percent,
Reduced particulate carbon 13-26 percent,
Reduced particulate emissions 29-43 percent,
Increased combustion efficiency 0.2-0.4 percent, and
Had no effect on NOX emissions.
• Use of Carbonex Diesel is superior to other diesel particulate
control technologies in that it does not result in increased
NOX emissions or diminished fuel economy.
• Carbonex Diesel is competitive with other chemical additives
for diesel particulate reduction because less of the active
metal (iron) is required to accomplish the reduction.
35-13
-------
LAB TEST DATA SUMMARY: DIESEL FUEL
RESULTS OF EVALUATION OF VELINO VENTURES CARBONEX
DIESEL AS A DIESEL ENGINE COMBUSTION ADDITIVE
~2-Ethyl Hexanol
Carrier
Agent/Diesel Fuel Ratio (Voluoe)
Iron Content of Agent (pp«, Might)
CirboMi Iron Added to Diesel Fuel (pp>, Might)
1:1500
0
0.0
Aaent Adaixed
Carbonex
Additive
1:1500
68
0.04
Into Diesel Fuel
2-Ethyl Hextnol
Carrier
1:900
0
0
Carbonex
Additive
1:900
68
0.08
Engine Exhaust Temperature (F) 843 832 843 839
0? (%, voluw, as-aeasured) 12.25 12.25 12.25 12.25
CO? (*. voluae, as-aeasured) 6.5 6.5 6.5 6.5
NO, (gram/brake horsepower-hour, air-free) 7.96 8.01 7.90 8.02
CO (gran/brake horsepower-hour, air-free) 0.82 0.76 0.80 0.72
HC (graas/brake horsepower-hour, air-free) 0.42 0.41 0.44 0.40
TO, (gram/brake horsepower-hour, air-free) 0.07 0.05 0.07 0.04
Carbon tn Participates (*, Might) 23 20 23 16
Coabustion Efficiency (%) 99.5 99.7 99.5 99.9
35-14
-------
IN-LINE PARTICLE MEASUREMENT INSTRUMENTS FOR POWER GENERATION SYSTEMS
Don J. Holve
(No paper provided)
36-1
------- |