Scribd
Upload
2K views

Applied Electrostatic Precipitation

Increased awareness of the effects of atmospheric pollution and ever tightening legislation have meant that electrostatic precipitators, which have been widely used to separate particulate matter from process gas streams, are now required to achieve collection efficiencies in excess of 99.9% for a number of applications. These changes have challenged the precipitation industry to consider how the equipment can be improved to meet the latest legislation, where control is now focusing on heavy metal and respirable size particulate discharges. In addition to achieving the increased performance at an economic cost, the emissions have to be maintained on a continuous basis, as failure to do so could have serious economic consequences to the plant operators. These demands have meant that all aspects of technology, engineering and operational concerns have had to be examined, reviewed and in some instances completely modified to meet the present criteria. In spite of commercial precipitators being used for almost 100 years, it has only been in the past two to three decades that the system has changed from essentially a 'black box' art to a scientifically-based technology. Fluid-dynamics, electro-dynamics, solid state electronics and microprocessor disciplines are now the basis for current precipitation theory, design, etc., i.e. the currently accepted 'State of the Art'. Although computational fluid dynamics and finite element analysis and other computer programs are now widely used by the industry, the sizing of a precipitator to satisfy a certain performance for a specific application is still very dependent on the supplier's knowhow and experience. The original 'black book' approach, although now taking the form of an extensive computerized data bank, is still the basis of sizing. In spite of a great deal of work being carried out using high speed computers and complex programming, there is a reluctance among the suppliers to use this approach, although it is very useful for analysing the field data. This situation is likely to change over the next few years and undoubtedly future precipitator sizing will be computer generated. Following a review of the early development of electrostatic precipitation, this volume, containing contributions by many of the world's leading experts in the field of electrostatic precipitation, covers the theory of precipitation from both fluid and electrodynamic standpoints, plus the basic practical designs and the gaseous and particulate features which impact on the precipitators' performance. XVI PREFACE To assist the non-specialist, Chapters 12, 13 and 14 cover the applications of dry, wet and mist type precipitators, including how the designs are modified to meet a specific duty and an examination of the major process factors which can affect performance. Chapter 11 gives a full description of plant commissioning and Chapter 10 plant testing, for both mass concentration measurements and elemental analysis in terms of particle sizing and chemical make-up. Finally, Chapter 15 shows how it may be possible to improve the performance of an under performing precipitator while Chapter 16 indicates where future developments in precipitation theory, design or application may lead. As Editor, I am indebted to the publishers for their backing and to the following, without whose experience and expertise, the book could not have been complied - Clive Cottingham, Ed Dismukes. Filip Knuttsen, Henry Krigmont, Leif Lind, Grady Nichols, Colin Paulson, Kjell Porle, Massimo Rea, Victor Reyes, Claus Riehle, David Styler and John Westbury, together with their respective companies for allowing permission to use the material. I would also like to thank Sheila Shepherd for correcting the final manuscript, Ken Darby, who was my mentor and a source of inspiration for some 35 years, and my wife, Maureen, for her encouragement and support in the project.

Uploaded by

tcbbss
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
2K views

Applied Electrostatic Precipitation

Increased awareness of the effects of atmospheric pollution and ever tightening legislation have meant that electrostatic precipitators, which have been widely used to separate particulate matter from process gas streams, are now required to achieve collection efficiencies in excess of 99.9% for a number of applications. These changes have challenged the precipitation industry to consider how the equipment can be improved to meet the latest legislation, where control is now focusing on heavy metal and respirable size particulate discharges. In addition to achieving the increased performance at an economic cost, the emissions have to be maintained on a continuous basis, as failure to do so could have serious economic consequences to the plant operators. These demands have meant that all aspects of technology, engineering and operational concerns have had to be examined, reviewed and in some instances completely modified to meet the present criteria. In spite of commercial precipitators being used for almost 100 years, it has only been in the past two to three decades that the system has changed from essentially a 'black box' art to a scientifically-based technology. Fluid-dynamics, electro-dynamics, solid state electronics and microprocessor disciplines are now the basis for current precipitation theory, design, etc., i.e. the currently accepted 'State of the Art'. Although computational fluid dynamics and finite element analysis and other computer programs are now widely used by the industry, the sizing of a precipitator to satisfy a certain performance for a specific application is still very dependent on the supplier's knowhow and experience. The original 'black book' approach, although now taking the form of an extensive computerized data bank, is still the basis of sizing. In spite of a great deal of work being carried out using high speed computers and complex programming, there is a reluctance among the suppliers to use this approach, although it is very useful for analysing the field data. This situation is likely to change over the next few years and undoubtedly future precipitator sizing will be computer generated. Following a review of the early development of electrostatic precipitation, this volume, containing contributions by many of the world's leading experts in the field of electrostatic precipitation, covers the theory of precipitation from both fluid and electrodynamic standpoints, plus the basic practical designs and the gaseous and particulate features which impact on the precipitators' performance. XVI PREFACE To assist the non-specialist, Chapters 12, 13 and 14 cover the applications of dry, wet and mist type precipitators, including how the designs are modified to meet a specific duty and an examination of the major process factors which can affect performance. Chapter 11 gives a full description of plant commissioning and Chapter 10 plant testing, for both mass concentration measurements and elemental analysis in terms of particle sizing and chemical make-up. Finally, Chapter 15 shows how it may be possible to improve the performance of an under performing precipitator while Chapter 16 indicates where future developments in precipitation theory, design or application may lead. As Editor, I am indebted to the publishers for their backing and to the following, without whose experience and expertise, the book could not have been complied - Clive Cottingham, Ed Dismukes. Filip Knuttsen, Henry Krigmont, Leif Lind, Grady Nichols, Colin Paulson, Kjell Porle, Massimo Rea, Victor Reyes, Claus Riehle, David Styler and John Westbury, together with their respective companies for allowing permission to use the material. I would also like to thank Sheila Shepherd for correcting the final manuscript, Ken Darby, who was my mentor and a source of inspiration for some 35 years, and my wife, Maureen, for her encouragement and support in the project.

Uploaded by

tcbbss
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 536

Applied Electrostatic Precipitation

JOIN US ON THE INTERNET VIA WWW, GOPHER, FTP OR EMAIL:


WWW:
GOPHER:
FTP:
EMAIL:

http://www.thomson.com
gopher.thomson.com
ftp.thomson.com
[email protected]

A service of

IJl

Applied Electrostatic
Precipitation
Edited by
K. R. PARKER
Consultant in Air Pollution Control
Sutton Coldfield, West Midlands, UK

BLACKIE ACADEMIC & PROFESSIONAL


An Imprint of Chapman & Hall

London Weinheim . New York Tokyo Melbourne Madras

Publisbed by Blackie Academic & Professional, an imprint of


Cbapman & Hall, 2-6 Boundary Row, London SE18HN, UK
Chapman & Hall, 2-6 Boundary Row, London SEI8HN, UK
Chapman & Hall GmhH, Pappelallee 3, 69469 Weinheim, Germany
Chapman & Hall USA, 115 Fifth Avenue, New York NY 10003, USA
Chapman & Hall Japan, ITP-Japan, Kyowa Building, 3F,
2-2-1 Hirakawacho, Chiyoda-ku, Tokyo 102, Japan
DA Book (Aust.) Pty Ltd, 648 Whitehorse Road, Mitcham 3132,
Victoria, Australia
Chapman & Hall India, R. Seshadri, 32 Second Main Road, CIT East,
Madras 600035, India
First edition 1997

1997 Chapman & Hall


Softcover reprint oftbe bardcover 1st edition 1997
Typeset in 1O/12pt Times by Doyle Graphics, Tullamore, Ireland
ISBN-13:978-94-010-7193-2

e-ISBN-13 :978-94-009-1553-4

001: 10.1 007/978-94-009-1553-4

Apart from any fair dealing for the purposes of research or private study,
or criticism or review, as permitted under the UK Copyright Designs and
Patents Act, 1988, this publication may not be reproduced, stored,
or transmitted, in any form or by any means, without the prior permission
in writing of the publishers, or in the case of reprographic reproduction
only in accordance with the terms of the licences issued by the
Copyright Licensing Agency in the UK, or in accordance with the terms
of licences issued by the appropriate Reproduction Rights Organization
outside the UK. Enquiries concerning reproduction outside the terms
stated here should be sent to the publishers at the London address
printed on this page.
The publisher makes no representation, express or implied, with regard
to the accuracy of the information contained in this book and cannot
accept any legal responsibility or liability for any errors or omissions that
may be made.
A catalogue record for this book is available from the British Library
Library of Congress Catalog Card Number: 96-83827

@) Printed on acid-free text paper, manufactured in accordance with


ANSI/NISO Z39.48-1992 (Permanence of Paper).

Contents
List of contributors
Preface

1 Why an electrostatic precipitator?

xiii
xv

K.R. PARKER
1.1
1.2
1.3

1.4

Introduction
Control system characteristics
Control operating principles
1.3.1 Inertial separation
1.3.2 Wet scrubbers
1.3.3 Fabric filter
1.3.4 Electrostatic precipitation
Summary of control system properties

2 Milestones in the history of precipitation

2
4
4
4
5
6
9

11

K.R. PARKER
Precipitator installations
2.1.1 Early investigations and developments
2.1.2 Full-scale precipitator developments
2.2 Development of electrical supplies
2.2.1 Rectifier types
2.2.2 Primary control systems
2.2.3 Automatic control systems
References

2.1

3 Basic and theoretical operation of ESPs

c. RIEHLE
3.1
3.2

3.3

3.4

General remarks
Ion production
3.2.1 Principles
3.2.2 Corona initiation field strength
3.2.3 Corona onset voltage
3.2.4 Current-voltage relationship
3.2.5 Electrical field distribution
Particle charging
3.3.1 Charging process
3.3.2 Cochefs charging model
3.3.3 Time dependence and saturation charge
Particle migration
3.4.1 Equation of motion
3.4.2 Theoretical migration velocity

11
11

13
20
20
22
22
23

25
25
29
29
31
33
36
44
52
52
52
54
55
55
57

CONTENTS

VI

3.5

Measuring and modelling particle separation


3.5.1 Grade efficiency and total efficiency
3.5.2 Laminar model
3.5.3 Deutsch model
3.5.4 Flow field and particle trajectories
3.5.5 Diffusivity models
3.6 Deposition
3.7 Removal
References

4 Mechanical design considerations for dry


precipitators

59
59
62
62
76
82
85
86
87

89

F. KNUTTSEN and K.R. PARKER


4.1 Introduction
4.2 Discharge electrodes
4.3 Discharge electrode mounting
4.4 Collectors
4.5 Casings
4.6 HT insulators
4.7 Rapping
4.8 Hoppers
4.9 Electrical clearances
References

Aerodynamic factors affecting performance

89
89
92
94
97
100
102
108
110
111

113

L. LIND
Introduction
Turbulence and secondary flow
5.2.1 Historical resume
5.2.2 Turbulence
5.2.3 Secondary flow
5.2.4 Numerical flow model
5.3 Gas velocity
5.4 Gas distribution
5.4.1 Standards
5.4.2 Residence time
5.4.3 Space charge
5.4.4 Re-entrainment
5.4.5 Erosion
5.4.6 Sneakage and sweepage
5.4.7 Optimal distribution
5.5 Model testing
5.6 Computational fluid dynamics
5.7 Field testing
5.8 Dust build-up and wear
References

5.1
5.2

The physical and chemical properties of particles


and their effect on performance

113
113
113
118
122
126
127
129
130
131
134
136
137
137
138
139
142
148
149
150

153

K. PORLE and K.R. PARKER


6.1

Particle size and shape


6.1.1 Particle sizing
6.1.2 Particle shape and structure

153
154
160

CONTENTS
6.2 Optical properties
6.3 Agglomeration
6.4 Cohesivity
6.5 Particle electrical resistivity
6.6 Chemical compositon and reactivity
References

7 Performance design considerations

c. COTTINGHAM
7.1
7.2
7.3

Introduction
What are we trying to achieve?
Assessment of the process
7.3.1 Typical assessment
7.4 Plate spacing
7.5 Configuring the ESP
7.6 Conclusions
References

Electrical operation of precipitators

VB
161
162
163
166
172

178

180
180
180
181
182
185
186
190
191

192

V. REYES
8.1
8.2

Introduction
Precipitator performance and electrical energization
8.2.1 Examples
8.3 Corona suppression and space charge effects
8.3.1 Electrical characteristics with air load
8.3.2 Characteristics with dust load
8.4 High tension sectionalization
8.5 Traditional DC energization
8.5.1 Basic principles
8.5.2 High voltage power supply ratings
8.5.3 Influence of the linear inductor
8.6 Intermittent energization
8.6.1 Basic principles
8.6.2 Comparison with traditional DC energization
8.6.3 Collection efficiency
8.7 Automatic voltage control and instrumentation
8.7.1 Introduction
8.7.2 Instrumentation
8.7.3 Basic control principles
8.7.4 Spark detection and voltage recovery
8.7.5 Back-corona detection and corona power control
8.8 Pulse energization
8.8.1 Introduction
8.8.2 Electrical configuration
8.8.3 Main features of pulse energization
8.8.4 Power consumption
8.8.5 Collection efficiency
8.8.6 Applications
8.8.7 Summary
8.9 Supervisory computer control
8.9.1 Stand-alone computer
8.9.2 Supervisory computer control via a gateway unit
8.9.3 Advanced control functions
Appendix 8A
Appendix 8B
References

192
192
194
195
196
196
199
201
202
206
208
210
211

212
214
217

217
217

220

223
226
230
230
231
235
238
239
240
241
241
243
244
245
246
247
248

CONTENTS

VlJI

Precipitator sizing methods and models of


electrostatic precipitators

e.

PAULSON and M. REA

Editor's note

250

9A Precipitator sizing methods

e.

250

252

PAULSON

9A.1

Theoretical considerations
9A.1.1 Basic dust-collection equation for gas in a duct
9A.1.2 Electrostatic precipitation
9A.1.3 Improvement of the Deutsch equation
9A.1.4 Factors affecting electrostatic precipitation
9A.2 Practical considerations
9A.2.1 Interpretation of test results
9AJ Precipitator modelling
9AJ.1 Mathematical modelling
9AJ.2 Practical testing
References to 9A

9B Models of electrostatic precipitators

252
252
254
255
258
265
265
274
274

275

278

280

M.REA
9B.l

Basic concept
9B.1.1 The Deutsch equation
98.1.2 Charging of particles and the modified Deutsch equation
9B.2 The modern approach to computer modelling
98.2.1 Early models
98.2.2 Model by Caiiadas et al. [5]
98.2J Modelling at Padova university [6]
References to 9B

10

Sampling and analysis for particles and heavy


metals in gas streams

280
281
284
285
285
286
288
291

292

G.B. NICHOLS and E.B. DISMUKES


10.1
10.2

Sampling and analysis


Heavy metals
10.2.1 General considerations
10.2.2 Sampling methods for multiple types of heavy metals
10.2.3 Sampling methods for mercury alone
10.2.4 Metal analysis in the laboratory
10.2.5 Prospects for real-time monitoring
References

11

The commissioning of electrostatic precipitators


D.A. STYLER and
11.1
11.2

J.e.

292

298
298
300
302
302
303
303

305

WESTBURY

Introduction
Mechanical commissioning

305
306

CONTENTS
11.2.1 Construction stage
11.2.2 Post construction stage
11.2.3 Cold commissioning
11.2.4 Hot commissioning
11.3 Electrical commissioning
11.3.1 An overview
11.3.2 Managers and commissioning
11.3.3 Familiarity revisited
11.3.4 The programme - who writes the programme and when?
11.3.5 Monitoring
11.4 Process commissioning
11.4.1 Hot commissioning
11.4.2 Back to the real world!

12

Dry type precipitator applications

IX

306
319
323
326
326
326
326
327
327
329
339
340
348

349

K. PORLE and K.R. PARKER


12.1 Introduction
12.2 Power generation industry
12.2.1 Bituminous coals
12.2.2 Anthracite coals
12.2.3 Subbituminous coals
12.2.4 Brown coals
12.2.5 Lignites
12.2.6 Oil-based fuels
12.3 The cement industry
12.3.1 Wet process manufacture
12.3.2 Semi-wet processing
12.3.3 Dry process production plant
12.3.4 Alkali bypass plant
12.3.5 Clinker cooler precipitators
12.3.6 Cement mill precipitators
12.4 General steam-raising plant
12.4.1 Moving grate combustors (chain grate, reciprocating
and cyclic beds)
12.4.2 Fluidised bed units
12.5 Biomass-fired steam-raising plants
12.5.1 Wood chip combustion
12.5.2 Chicken litter, etc.
12.5.3 Municipal wastes
12.6 Iron and steel works
12.6.1 Sinter plants
12.6.2 Pelletising plants
12.6.3 Steel making
12.6.4 Operations involving the casting of hot metal
12.7 Non-ferrous industries
12.7.1 Copper and nickel recovery
12.8 Aluminium smelting
12.9 Paper and pulp industry
12.9.1 Bark firing
12.9.2 Cellulose pulp production
12.9.3 Lime sludge burning
12.9.4 Magnesium sulphate burning
12.10 Conclusions
References

349
349
350
353
354
355
356
356
359
359
360
360
363
363
364
365
366
366
367
368
368
368
369
369
372
372
375
375
376
378
379
379
379
380
380
381
381

CONTENTS

13 The wet electrostatic precipitator: design and


a pplica tions

382

K.R. PARKER
13.1
13.2
13.3
13.4
13.5
13.6
13.7
13.8
13.9

Introduction
Design considerations
13.2.1 Dust deposition and removal
Discharge electrodes
HT insulators
Casing/hopper design
Water treatment
Materials of construction
Electrical energisation
Typical applications of wet precipitators
13.9.1 Applications in the iron and steel making fields
13.9.2 Applications in the chemical industries
13.9.3 Applications following acid gas scrubbers
13.9.4 Incineration-type processes
13.9.5 HAC discharges
13.9.6 Glass manufacturing
13.9.7 Other applications

14 The mist precipitator: design and applications

382
383
383
386
387
387
390
391
393
394
394
397
398
399
400
400
400

402

K.R. PARKER
14.1
14.2

14.3

Introduction
Applications of mist precipitators
14.2.1 Collection of sulphuric acid mist
14.2.2 Gas detarring
14.2.3 Collection of radioactive particles
14.2.4 Other mist precipitator applications
Conclusions

15 Upgrading of existing precipitator efficiencies

402
407
407
411
415
416
417

418

K.R. PARKER and H. KRIGMONT

15A

Modifications/changes to existing plant

418

Assessment of required performance improvement


Performance improvement modification options
Alternative solutions
15A.3.1 Electrical
15A.3.2 Mechanical changes

419
421
423
423
423

Precipitator improvements achieved by


changing the electrical resistivity of the
particulates

425

15A.l
15A.2
15A.3

15B

158.1 Change of temperature/relative humidity of the gases


158.2 Flue gas additives to improve performance
References to 15A and 15B

425
426
428

CONTENTS

15C

Theory, principles of operation, equipment and


applications of flue gas conditioning
15C.1 Introduction
15C.2 Electrical resistivity
15C.2.1 Prediction of fly ash resistivity
15C.2.2 Resistivity effects in ESPs
15C.3 Flue gas conditioning
15.C.3.1 Conditioning by sulfur trioxide
15C.3.2 Ammonia conditioning
15C.3.3 Dual flue gas conditioning
15C.3.4 Balance of plant impact
15C.4 Flue gas conditioning equipment
15C.4.1 S03 FGC systems
15C.4.2 Ammonia FGC systems - design features
15C.4.3 ESP power consumption
15C.4.4 Mixing requirements
15C.5 Application of flue gas conditioning in converting hot-side
fly ash precipitators to cold-side operation
15C.5.1 Introduction
15C.5.2 Flue gas conditioning
15C.5.3 Installation design
15C.5.4 FGC operation
15C.5.5 Economics
15C.5.6 Conclusions
15C.6 FGC systems - optimization
15C.6.1 General
15C.6.2 Theoretical approach
15C.6.3 Procedural approach
15C.6.4 Statistical approach
15C.6.5 Rapper adjustments
15C.6.6 Data collection
15C.7 Conclusions
References to 15C

16 Possible future developments in the field of


electrostatic precipitation
K.R. PARKER, c. RIEHLE and H. KRIGMONT
16A

Electrical developments
High frequency power conversion or switched
mode power supplies
16A.2 Nanosecond pulse operation and acid gas control
References to 16A
16A.1

16B

Use of natural sulphur dioxide as a feed stock


for flue gas conditioning systems: flue gas
conditioning today and tomorrow
16B.1 Background
16B.2 'Native' or 'internal' feed stock FGC technologies

xi

429
429
430
432
433
436
437
445
452
453
454
454
461
462
463
463
463
465
466
468
469
471
471
472
472
474
475
480
481
481
481

483
487
487
490
491

492
492
494

xii

CONTENTS
168.2.1 'Slip-stream' FGC systems
168.2.2 In-duct FGC systems
168.3 In-situ gas conditioning (lGC) approach
168.3.1 Variable exposed area IGC system
168.3.2 Variable catalyst temperature IGC system
168.4 Variable flow IGC system
168.5 Catalyst selection
References to 168

16C

High temperature/high pressure precipitators


for advanced power generation systems
16C.l
16C.2
16C.3
16C.4
16C.5
16C.6

Fundamentals
Voltage and current
Particle charging
Particle migration
Grade efficiency
Open questions
16C.6.l Electrical resistivity
16C.6.2 Mechanical stability of material
16C.6.3 Rapping
16C.6.4 Electrical insulation
16C.6.5 Emptying of hoppers
16C.6.6 Electrical power consumption
16C.7 Symbols
References to 16C

16D

494
495
496
496
497
498
499
500

501
503
505
509
510
513
515
515
515
515
516
516
516
516
517

Computer sizing of precipitators

518

Index

519

Contributors
C. Cottingham

Lodge Sturtevant Ltd, George Street Parade,


Birmingham, B3 1QQ, UK

E.B. Dismukes

Grady Nichols Enterprises Inc., 400 Kiowa Street,


Montevallo, AL 35115, USA

F. Knuttsen

ABB Flakt Industri AB, S-35187, Vaxjo, Sweden

H. Krigmont

Allied Environmental Technology, One Pacific


Plaza, 7755 Center Avenue, Suite 1100, Huntingdon
Beach, CA 92647, USA

L. Lind

FLS Miljo a/s, Ramsingsveg 30, DK2500, Valby,


Denmark

G. Nichols

Grady Nichols Enterprises Inc., 400 Kiowa Street,


Montevallo, AL 35115, USA

K.R. Parker

17 Somerville Road, Sutton Coldfield, West


Midlands, B73 6JD, UK

C. Paulson

CISRO, Division of Coal & Energy Technology,


PO Box 136, North Ryde, New South Wales 2113,
Australia

K.Porie

ABB Flakt Industri AB, S-351 87, Vaxjo, Sweden

M.Rea

Dipartimento Di Ingegneria Elettrica, Universita


Degli Studi Di Pavoda, Via Gradenigo 6/A,
35131 Padova, Italy

V. Reyes

FLS Miljo a/s, Ramsingsveg 30, DK2500, Valby,


Denmark

C. Riehle

Bayer AG, R&D Department, Particle Technology


& Fluid Dynamic Group, D51368 Leverkusen,
Germany

D.A. Styler

Lodge Sturtevant Ltd, George Street Parade,


Birmingham, B3 1QQ, UK

J.C. Westbury

Lodge Sturtevant Ltd, George Street Parade,


Birmingham, B3 1QQ, UK

Preface

Increased awareness of the effects of atmospheric pollution and ever


tightening legislation have meant that electrostatic precipitators, which have
been widely used to separate particulate matter from process gas streams,
are now required to achieve collection efficiencies in excess of 99.9% for a
number of applications. These changes have challenged the precipitation
industry to consider how the equipment can be improved to meet the latest
legislation, where control is now focusing on heavy metal and respirable size
particulate discharges. In addition to achieving the increased performance
at an economic cost, the emissions have to be maintained on a continuous
basis, as failure to do so could have serious economic consequences to the
plant operators.
These demands have meant that all aspects of technology, engineering
and operational concerns have had to be examined, reviewed and in some
instances completely modified to meet the present criteria. In spite of
commercial precipitators being used for almost 100 years, it has only been
in the past two to three decades that the system has changed from essentially
a 'black box' art to a scientifically-based technology. Fluid-dynamics,
electro-dynamics, solid state electronics and microprocessor disciplines are
now the basis for current precipitation theory, design, etc., i.e. the currently
accepted 'State of the Art'.
Although computational fluid dynamics and finite element analysis and
other computer programs are now widely used by the industry, the sizing of
a precipitator to satisfy a certain performance for a specific application is
still very dependent on the supplier's knowhow and experience. The original
'black book' approach, although now taking the form of an extensive
computerized data bank, is still the basis of sizing. In spite of a great deal
of work being carried out using high speed computers and complex
programming, there is a reluctance among the suppliers to use this approach, although it is very useful for analysing the field data. This situation
is likely to change over the next few years and undoubtedly future precipitator sizing will be computer generated.
Following a review of the early development of electrostatic precipitation,
this volume, containing contributions by many of the world's leading
experts in the field of electrostatic precipitation, covers the theory of
precipitation from both fluid and electrodynamic standpoints, plus the basic
practical designs and the gaseous and particulate features which impact on
the precipitators' performance.

XVI

PREFACE

To assist the non-specialist, Chapters 12, 13 and 14 cover the applications


of dry, wet and mist type precipitators, including how the designs are
modified to meet a specific duty and an examination of the major process
factors which can affect performance. Chapter 11 gives a full description of
plant commissioning and Chapter 10 plant testing, for both mass concentration measurements and elemental analysis in terms of particle sizing and
chemical make-up. Finally, Chapter 15 shows how it may be possible to
improve the performance of an under performing precipitator while Chapter
16 indicates where future developments in precipitation theory, design or
application may lead.
As Editor, I am indebted to the publishers for their backing and to the
following, without whose experience and expertise, the book could not have
been complied - Clive Cottingham, Ed Dismukes. Filip Knuttsen, Henry
Krigmont, Leif Lind, Grady Nichols, Colin Paulson, Kjell Porle, Massimo
Rea, Victor Reyes, Claus Riehle, David Styler and John Westbury, together
with their respective companies for allowing permission to use the material.
I would also like to thank Sheila Shepherd for correcting the final manuscript, Ken Darby, who was my mentor and a source of inspiration for some
35 years, and my wife, Maureen, for her encouragement and support in the
project.
K.R. Parker
November 1996

1 Why an electrostatic precipitator?


K.R. PARKER

1.1

Introduction

Stricter environmental legislation in many countries is producing evertightening regulations and standards governing the emission of fine particles
to the atmosphere from all sources. With the ease and rapid means of
international communication, many believe that the control of pollution is
a modern concept; history, however, indicates that the first recorded
measure was in the UK, when, in the 1600s, Parliament prohibited the
burning of bituminous sea coal in London, to avoid what the Clean Air Act
of 1956 referred to as 'smog'.
It was not until the nineteenth-century UK Industrial Revolution, when
water power gave way to steam produced by the burning of carbonaceous
fuels to evaporate water, that pollution became a serious threat to those
living and working in the vicinity of the new works. The problem was
exacerbated by the rapid development of the blast furnace for making iron
for the ever expanding needs of industry, and also by the large-scale
production of chemicals.
The first attempt to control emissions in the world, certainly in the UK,
was the 1863 Alkali and Works Act, aimed specifically at controlling
hydrogen chloride emissions released during the manufacture of sodium
carbonate using the Leblanc Process. The Alkali Act was revised and
extended in 1906 to cover most chemical works discharges and probably
formed the basis of most current environmental legislation.
In recent years there has been worldwide recognition of the problems of
environmental pollution and most industrialised countries have enacted
legislation covering all uncontrolled emissions. The most stringent measures
are associated with wealthy countries having high population density and
heavy industrialisation, e.g. Japan, North America and Western Europe. For
the developing countries, since much of the heavy equipment tends to be
imported from the industrialised Nations, a great deal of the plant is
supplied fitted with some form of pollution-control device, dependent on the
country of supply, proposed or existing regulations and the finance package.
There are many forms of emission which are of worldwide concern, e.g.
noise, water, gas and particulate discharges, which are the subject of control
and whose discharge or emission levels are steadily reducing with each
phase of legislation. In the UK, the allowable particulate emission rate from

WHY AN ELECTROSTATIC PRECIPITATOR?

power-station chimneys has seen, over the past 30 years, a 1O-fold decrease
and the need for minimising emissions is steadily being recognised by the
developing countries in their own right. These changes in legislation do not
however signify that the problems of pollution have been overcome and
there is still a need for the major technological countries to further improve
the 'State of the Art' for equipment designed to reduce pollution. Any
developments, in addition to improving performance efficiency, should also
increase plant availability and make any plant more cost effective, either by
changing the design of existing equipment or the application of new
concepts.
There are many aspects of pollution control; this publication, however,
will restrict itself to the control of entrained particulates, both solid and
liquid phase, in mainly gas-borne streams using electrostatic precipitation,
the principle of which is readily applicable to the collection of particles in
liquid-phase streams, provided the carrier medium has electrically insulating
properties.

1.2

Control system characteristics

There are numerous methods of separating particles from process streams


using different principles, e.g. gravity or inertial separation as used in
cyclones, impaction and diffusion as applied to fabric filtration, electrical
means as applied to electrostatic precipitation and contacting, impingement
and impaction in the case of wet scrubbers. Because of the different basic
principles used by each form of separator, they each have different properties
regarding collection efficiency, process and application suitability.
The effectiveness of the above forms of device is indicated in Figure 1.1,
a dust spectrum, which shows the particle size range over which reasonable
collection efficiencies can be achieved. While all the various approaches can
be very efficient in collecting the large particles, i.e. greater than 10 11m
(1 11m = 1 x 10 - 6 m), the current legislative emission levels mean that
effective separation of particles of 1 11m or less is now essential for a large
number of processes if they are to comply with legislation.
In theory, while the effective range of inertial separators can be extended
towards the submicron range, the required centrifugal force can only be
obtained by reducing the diameter of the device, which restricts gas
throughput, unless a large number of individual separators are operated in
parallel. This becomes impractical for many industrial processes.
Although the high energy wet scrubber, e.g. venturi type, can effectively separate submicron particles, its power consumption, typically
1000 mm w.g., to capture 111m particles at 99% efficiency not only means
high operating costs but, as the particles will be contained in a liquid

CONTROL SYSTEM CHARACTERISTICS

Micron size

100

1000

10

1.0

0.1

0.01

Settling
chambers
Cyclone
Practical
size range
covered by
typical
collectors

High efficie rcy cyclones

Wetw ~shers and to flers

High energ scrubbers

II

Fabri filters
Elect ostatic precipi ators.
Dust

Grit

Fume

T
Rain, mist
drops
~ Pulverised coal

Fog

Tobacco smoke

I
Fly ash
TYPical range
of
atmospheric
impurities

Foundry
sand

Bacteria

.J

Pollens
Heavy
industrial
dusts

Carbon black
Virus

S03 mist

Temporary
atmospheric
impurities

Permanent
atmospheric impurities
Zinc oxide_
~
fumes

~ Cement dust
, ~
Incinerator dust

Free-falling
velocity of
particles having
a specific
gravity of 2.0

I
0
0
~

-,

0
~

I
I

I
I

0
0

0
0
0

0
0
0

Velocity in air (lt/s)


based on the Stokes-Cunningham formula

Brownian
movement
exceeds
that due
to
gravity,
I.e. particles
do not
settle

Figure 1.1 Dust spectrum. The size of any particle is defined as the diameter of a sphere having
the same free-falling velocity as the particle, usually termed the Stokes diameter.
1 I'm = 1 x 1O-6 m.

effluent, this needs to be treated to avoid changing an air pollution problem


into a more difficult water pollution problem.
The choice of equipment having the capability to effectively collect
submicron particles at high efficiency from large process gas streams
(50m 3 /s upwards) tends to be limited to the fabric filter and electrostatic
precipitator as cost-effective approaches.

4
1.3

WHY AN ELECTROSTATIC PRECIPITATOR?

Control operating principles

For completeness, the operating principles of the various major types of


equipment, outlined above, can be summarised as follows.

1.3.1

Inertial separation

There are many different forms and arrangements of collectors using this
principle, usually termed cyclones or centrifugal collectors, where the
entraining stream is caused to spin or rotate rapidly within a cylindrical
vessel. Because of the much higher mass of the particulates compared with
the gas molecules, the resultant centrifugal force causes the particles to
migrate across the flow to the wall of the containing vessel, where they
become disentrained in the low flow region of the device.
The cyclone is a very simple device having a wide range of operating
temperatures, but requires a driving force, e.g. a pressure drop of some
lOOmm w.g., for effective separation. For most practical applications its use
for high efficiency collection is limited to particulates having a particle
diameter greater than 10 }.lm.

1.3.2

Wet scrubbers

All wet scrubbers operate by contacting the particles with large quantities
of liquid in a fully turbulent area of the device. The particles are allowed to
impact or impinge on the liquid droplets, normally water, such that their
size and mass increase. As larger water-wetted particles they can be removed
from the entraining gas stream in a simple impingement or cyclonic form of
separator.
As a contact/impaction device, the effectiveness of particle droplet collision is dependent on the total energy expended in the device. This can be
either as pumping power, to provide small liquid target particles, or as
pressure drop, to develop the high relative particle/droplet velocities necessary to ensure that impingement occurs. For submicron particles, venturi
scrubbers, having efficiencies in the 99% region, have total operating power
requirements equating to some 1500mm w.g.
The scrubber design is relatively simple; having a small footprint, it can
handle a full range of gas temperatures and is insensitive to sticky dusts, but
can suffer from erosion and corrosion. Its usage is normally limited, for high
efficiency collection duties, to processes having small gas flows, because of
the high power demand, large water usage, probable efHuent problems and
poor plume buoyancy unless gas reheat is practised. The wet scrubber,
because of its excellent mass transfer characteristics, however, is widely used
to control gaseous emissions, i.e. acid gases from many processes.

CONTROL OPERATING PRINCIPLES

1.3.3

Fabric filter

The principle is that the particulate-laden gas stream is passed through a


porous membrane which filters off the particulates and allows the clean gas
to pass through. As the retained particles deposit and build up to form a
layer, the efficiency of separation, which is basically, in its simplest form, an
interception and diffusion mechanism, increases, such that particles having
a much smaller diameter than the membrane pore size are collected. In the
original design of plant, the media comprised a plain woven cloth formed
into a bag or sock; the filter material is now usually felted and surface
treated to reduce particle penetration while still maintaining high gas
porosity.
In theory, such a device should be close to 100% in capturing particles
as the layer on the media builds up. In order, however, to maintain
operating pressure drop, e.g. 100-150mm w.g., it is necessary to periodically
remove the layer from the media. Assuming that the media can be satisfactorily freed of the dust layer, by means of vibration or air reverse purging
to regain a low pressure drop, on returning the filter to service there is
always a degree of slippage of the finer particles, which automatically limits
the ultimate efficiency or emission level.
From the early days, when the media were manufactured from natural
fibres in the form of a bag or sock, the operating temperature range has been
considerably extended by using synthetic fibres, woven and sintered metals,
or ceramics in various forms. In the latter, the operating temperatures can,
dependent on the media selected, approach 1000C. Whilst the actual form
of the filter can be different for the extreme temperatures, the principle of
separation remains unaltered.
Simultaneous development of special media coatings and finishes, to
simplify cleaning and general operation, has extended the expected life of the
media. This has enabled suppliers to consider firm commitments/guarantees
on the life of the media, in addition to performance. This is particularly
important for large applications where, in the case of a major power plant,
some 35000 conventional bags would be used to treat the output from each
500 MW boiler installation and, to avoid major emission excursions and
legislation violations, a strict maintenance/bag replacement programme
needs to be implemented.
Compared with an electrostatic precipitator installation handling a
'difficult fly ash', the cost of the equivalent bag filter would be somewhat
cheaper, but its operating power, the maintenance of compressors and
cleaning mechanisms, plus the cost of a complete media (bag filter) replacement, probably every 3 years or so, to ensure emission compliance, must be
borne in mind and added to the overall cost comparison.
Figure 1.2 compares the installed capital and 15 year total cost of various
arrestment systems for a 250 MW power plant designed to meet an emission

WHY AN ELECTROSTATIC PRECIPITATOR?

30r-----------------------------------~

----------RC3B-ho~se----------------------------en

co
'5 20

"0

rJJ
::J

en
c

~ 10
~

Bags - 4 year life


Pressure drop 190 mmHg
ElectriCity $O.OS/kwh

OL-------~--------~------~------~

10 10
(a)

10 11
10 12
10 13
PFA electrical resistivity (Qcm)

12r---------------------------------~

----------RC3B-ho~se----------------------------Ul

~ 9
(5
"0

~ 6
Ul

3
OL-------~--------~------~------~

10 10
(b)

10 11
10 12
10 13
PFA electrical resistivity (Qcm)

Figure 1.2 Cost comparisons for electrostatic precipitator (ESP), pulse jet fabric filter (PJFF)
and reverse gas bag (RGB) house. 1994 costs based on a 250 MW power plant designed to meet
an emission of 50mg/Nm 3 at 6% 02' (a) Total capital installed cost. (b) 15 year total cost.

of 50mgjNm 3 at 6%02 dry. The costings indicate that for either a reverse
gas bag house or pulse jet fabric filter the cost remains constant, irrespective
of the pulverised fuel ash (PF A) resistivity, whereas the electrostatic precipitator (ESP) increases proportionately to the ash resistivity. Overall the ESP
has the economic advantage taking all costs into consideration.
Another point, which needs to be carefully considered for some applications, is that if the ash becomes difficult to remove from the bag filter media
and the pressure drop continues to increase, the plant process gas throughput and hence output will drop as the induced draught (ID) fan reaches its
maximum pressure capability or limiting motor power.

1.3.4

Electrostatic precipitation

The basic theory of operation is that the gas-borne particles are passed
through a corona or charging field where they receive an electric charge,

CONTROL OPERATING PRINCIPLES

usually negative in the case of industrial precipitators, and then as charged


particles are deflected by the electric field producing the charging regime.
The charged particles then move across the gas stream from the negative
electrode to be deposited on the positive electrode, which for convenience is
normally earthed or grounded. From the collectors, the particles are
removed into receiving hoppers or troughs, either by mechanical shock
impulse rapping in a dry application or by water washing in the case of wet
precipitators, i.e. those plants operating close to or at water dew-point
temperature conditions.
Commercially, electrostatic precipitators have been used for almost a
century for the collection of dust, fume and mist particles from all types of
processes. The initial applications from mainly chemical/metallurgical based
applications were both for reducing air pollution and for the recovery of
valuable byproducts, for example the cleaning of combustible gases, or
material losses from metal smelting processes. The ability to size precipitators for a specific efficiency was ideal in this respect and economics dictated
mid-90% efficiencies for this recovery type of application, since the collected
material normally had a positive value.
More recently, the major application has been to control air pollution to
meet specific legislative emissions and, consequently, design efficiencies can
now approach, and sometimes exceed, 99.9%.
Figure 1.3 illustrates some recently measured emissions for various
process applications, together with their relevant legislative emission levels.
This indicates that a well-designed and operated electrostatic precipitator
can produce emissions well below lOmgjNm 3 for a wide range of duties and,
as such, can comply with the most stringent regulations.
With the most recent legislation demanding more stringent controls with
regard to toxics and heavy metals, many existing in the submicron fume
range, then, provided these are in a solid or liquid phase at the normal plant
operating temperature, the electrostatic precipitator is an ideal capture
vehicle.
For compounds which are in a gaseous phase at the normal process
discharge temperature, then, provided the gases can be further cooled to
condense the material, a wet precipitator can be used for very efficient
secondary cleaning, e.g. following a sulphur dioxide scrubbing system on a
power station. An alternative approach, provided the toxics or gaseous
phase material can be adsorbed by an injected target material, e.g. activated
carbon, means that the precipitator can operate above any dew-point
temperature and hence can be constructed from normal carbon steels at
much reduced cost.
The latter method is a preferred option, particularly in the UK, since
under Integrated Pollution Control (IPC) legislation, changing air pollution
to water pollution is not a viable option since the cleaning of dirty process
water can prove to be very expensive. In some instances, such as sticky dusts

WHY AN ELECTROSTATIC PRECIPITATOR?

100

60
40

20
USA

incinerators

-.s

(')

C J)

c
0
.u;
(J)

10

boilers
and
incinerators

ill

Cement

Figure 1.3 Comparison of some measured emissions with statutory regulations. Lodge Cottrell
brochure 'An Introduction to Electrostatic Precipitation'.

or dew-point conditions, there is no alternative to using a wet precipitator,


but often as this is associated with some form of scrubber or quench system,
the water clean-up is an economic part of the total process.
Although it is possible to collect these submicron toxics and heavy metals
with either a high energy venturi type scrubber or bag filter, these suffer
from very high energy costs in the case of the scrubber or possible
'bleed-through' in the case of the fabric filter, unless expensive precoating
systems are adopted; even then there is still a risk. Consequently, there are
commercial and performance advantages in using electrostatic precipitators
in order to comply with legislation.
The electrostatic precipitator can be summarised as having the following
advantages compared with other forms of device for particulate collection or
control:
1. can be sized for any efficiency;
2. can operate at temperatures up to 850C;

SUMMARY OF CONTROL SYSTEM PROPERTIES

3. can operate at any positive pressure condition but limited to suctions


that will maintain corona without the discharge developing into a
plasma;
4. can operate with fully saturated gas;
5. has a low pressure loss;
6. acceptable electrical operating costs;
7. not particularly particle-size sensitive;
8. proven long life;
9. excellent reliability;
10. low maintenance requirements.
1.4

Summary of control system properties

The purpose of this initial chapter is to list the major types of device that
are available for the removal of particulates, fumes and mists met in
industrial air pollution control applications and to attempt to identify the
advantages and disadvantages of each. While one can argue that all have

Device
Inertial
separator

Wet collector

Fabric
filtration

Electrostatic
precipitator

Advantage
Simple construction,
low cost,
low maintenance,
no temperature
limitation,
both wet and dry
applications
Simple construction,
low capital cost,
small footprint,
sticky particulates,
effective on wet gas
and dusts
Intrinsic high efficiency
99.5% plus,
reasonable footprint

Sized for any efficiency,


full particle size range,
low pressure loss,
low maintenance,
long life 20 + years,
wet and dry collection

Disadvantage
Limited effective
particle size range,
high pressure loss,
typically lOOmm w.g.,
possible erosion
High pressure loss
up to 1500mm w.g.,
large water usage,
wet em uen t,
low plume buoyancy,
possible plume odour
Pressure loss,
average 150mm w.g.,
media temperature
limitation,
fire/melting risk,
media of limited
life-span,
unsuitable for
sticky/adhesive dust
High capital cost,
efficiency sensitive to
dust resistivity,
large footprint

10

WHY AN ELECTROSTATIC PRECIPITATOR?

their own specific applications, for all-round efficient and effective collection
of all particulate forms, the electrostatic precipitator ranks probably the
highest, if one accepts the cost and ground space implications.
The properties of the devices, outlined in section 1.4, as effective high
efficiency particle collectors, are summarised in the table on the previous
page.
Please note that the foregoing table should not be used for deciding
which of the various devices should be used, but is useful in that it identifies
the advantages of each and can be used as a first cut.
As each application is specific, it will be necessary to complete a full
costing exercise, considering all the relevant factors, to determine which
collection system is most cost effective.
The subsequent chapters will cover the history of the development of the
electrostatic precipitator, from its early beginnings through to the latest and
possible future applications. In addition, the theory of precipitation and the
factors which affect the efficiency and sizing will be covered in detail,
together with practical hints and assistance to engineers faced with operating and optimising their electrostatic precipitation plant.

Milestones in the history of precipitation


K.R. PARKER

2.1
2.1.1

Precipitator installations
Early investigations and developments

While the first commercially successful precipitators were installed in the


early part of this century, historically, the first report of the electrical
attraction of smoke appears in 1600, when William Gilbert [1] describes the
phenomenon: 'Everything rushes towards electricks excepting flame and
flaming bodies and the thinnest air ... yet they entice smoke sent out by an
extinguished light.' The next report was in 1824, when Dr M. Hohlfeld at
Leipzig [2], performed an experiment of clearing fog in a jar containing an
electrified point. This work was repeated by C. F. Guitard in 1850 [3].
The first significant investigations using electrostatic precipitation were
by Professor O. J. Lodge and Mr J. W. Clark, working at Liverpool University in the early 1880s. Following publication of this work [4,5], Lodge
was approached and collaborated with Mr Alfred Walker of Parker, Walker
& Co. of Chester, in attempting to remove fumes from gases arising during
the smelting of lead at their Works in Deeside.
This was the first attempt to use electrostatic precipitation industrially for
the collection of particulates. In 1884 a UK Patent, No. 11120 [6], was filed
by Alfred Walker describing how the apparatus was arranged (Figure 2.1).
At this stage it should be appreciated that the only means of developing high
voltages was by Wimshurst or Voss electrostatic friction generators, which
stored their charge in Leyden Jar capacitors. It should be noted that unlike
later plants, Walker's drawing indicates positive energisation of the electrode system. The adoption of negative energisation of the discharge
electrodes follows Cottrell's investigations and his patent of 1908.
The plant at Bagillt in North Wales was energised from two 1524 mm
(5 ft) diameter Wimshurst machines driven from a 1 hp steam engine. This
method of energisation proved ineffective; while an electrostatic machine can
provide the necessary voltage, it cannot simultaneously deliver the required
particle charging current and, consequently, the plant failed to live up to
expectation.
It should be remembered that in the early 1880s electric motors and
generators were still being developed and were not readily available outside
the laboratory. Even had high voltage equipment been available, electrical
insulating materials, such as glass, mica, ebonite and similar materials would
not have been entirely satisfactory for precipitation duties.

12

MILESTONES IN THE HISTORY OF PRECIPITATION

Discharge system
Earthed flue

Figure 2.1 Illustration from first US patent on lectrostatic precipitation. A.O. Walker, No.
342548 (1886).

This early failure did not prevent Walker from realising the potential of
electrostatic precipitation and he took out additional Patents in Europe and
the USA for the collection of fumes and dust from all types of chemical
plant.
Almost in parallel with Lodge and Walker's work in the UK, Dr Karl
Moeller [7] in Germany rediscovered the art of precipitation and took out
a German Patent, No. 31911, 'Rohrenformiges Gas and Dampfilter', describing his findings, in 1884.
Following the installation and failure of the plant in North Wales, Lodge
appears to have temporarily lost interest in precipitation and tended to
concentrate on other experimental work, such as meteorological and atmospheric electrification for the removal of fog and creating rain artificially,
electro-magnetic radiation as a precursor to wireless telegraphy and also
X-rays for medical use. In 1903, however, he obtained a UK Patent, No.
24305 [8], covering a high voltage rectifier bridge arrangement, using
Cowper-Hewitt mercury arc vapour lamps for the deposition of smoke,
dust, mist and the like from gases. The patent describes a recognisable
precipitator using barbed wire electrodes and plate collectors. The high
tension (HT) equipment employed a gas engined driven dynamo feeding a
Ruhmkorff induction coil, then the mercury arc rectifiers in full wave
configuration.

PRECIPITATOR INSTALLA nONS

13

In 1905, another UK Patent, No. 25047 [9], was issued under the names
of Lodge, Muirhead and Robinson, covering an improved vacuum type
rectifier. These were manufactured by A. C. Cossor who, in their sales
literature, describe the device as 'Sir Oliver Lodge's High Tension Valve' for
obtaining unidirectional current for use with X-ray tubes, etc. These valves
were used extensively by Lodge in the following years for precipitation
duties.
Work in the USA, at this time, was being carried out by Dr F. G. Cottrell,
a physical chemist, who was studying various ways of controlling air
pollution from the Californian Smelters; he made the technological breakthrough for providing high voltage and charging currents simultaneously by
using a high voltage AC transformer coupled to a synchronous mechanical
switch rectifier. This device proved vastly superior to the earlier forms of
equipment and led to the successful development of an electrostatic precipitator for the collection of sulphuric acid mist.
Cottrell also recognised that negative corona had advantages over
positive and, in 1908, took out a US Patent No. 895729 [10], which
describes the early apparatus in some detail. Figure 2.2 illustrates how the
equipment was arranged. It should be noted that the discharge electrode is
described as pubescent and was made of semi-conducting fibrous material.
This proved to be important in the success of obtaining a uniform corona
over long lengths of wire at relatively low voltage. Because of insulation
limitations, the high voltage transformers available were only capable of
supplying 10-15 kV.

2.1.2

Full-scale precipitator developments

With Cottrell's success in collecting sulphuric acid fumes in the Laboratories


of the University of California, the first commercial precipitator was applied
in 1907/8 to collect fumes and dust from the Powder Works of the DuPont
de Nemours Plant at Pinhole, California [11]. This plant was quickly
followed by another installation on the Lead Smelter Facility at Selby,
which was embroiled in acute air pollution difficulties over their gaseous
discharges. This precipitator was successful in significantly reducing particulate emissions and collected some 2 gal/min (0.151/s), of sulphuric acid mist,
at a gas flow rate of some 5000 cubic feet per minute (cfm) (2.4 m 3 /s).
The next precipitator to be installed was in 1910, at the lead smelter at
Balaklala in California [12]. This plant was to remove lead fume, etc., at a
gas flow rate of 250000 cfm (118 m 3 /s). Although the plant design had
improved with the experience gained from the earlier installations, the
precipitator did not live up to expectations, and in spite of extensive
investigation, only achieved some 90% collection efficiency. At this stage, it
had not been established that to successfully handle lead dust and fumes, it
is important to optimise temperature and flue gas moisture to control

14

MILESTONES IN THE HISTORY OF PRECIPITATION

Mechanical rectifier

Insulator bushing
Clean gas
--- for bushing

_Gas
outlet

Pubescent discharge
electrode

Figure 2.2 Illustration from Cottrell's first electrostatic precipitation patent, No. 895729 (1908).

particulate electrical resistivity. This later finding, together with the problems of inadequate power supplies, probably halted the 1880s developments
by Lodge et al.
In spite of the only partial success of the Balaklala plant, further
investigations and development work was carried out in the US, since it was
recognised that electrostatic precipitation was a most effective means of
controlling and removing dust and fumes from many industrial processes.
Walter Schmidt, a student of Cottrell's, developed and patented the fine
wire discharge electrode. This breakthrough, enabled larger plants to be

PRECIPITATOR INSTALLATIONS

15

constructed, as illustrated by the precipitator installation at the Cement


Works of the Riversdale Portland Cement Works in 1912 [13], where the
precipitator was designed to handle 1000000 cfm (472 m 3 /s) of process gas
to remove cement and other dusts from their kilns. The plant was highly
successful and remained in operation for some 50 years.
Dr Cottrell had, with magnificent generosity, in 1912 decided to relinquish all rights and claims to his US patents, east of the Rockies, with the
exception of the cement industry, and gave them in trust. An endowment
fund was set up under the auspices of the Research Corporation, so that the
income could be used for the further advancement of scientific knowledge.
His remaining patents and interests covering the Western US and foreign
rights were assigned to Walter Schmidt.
The application of precipitators to the cement industry to collect kiln
dust gave rise to the American Potash Industry. This allowed the US to be
quickly self-sufficient, not only for its use as a fertiliser but also in pig iron
and ferromanganese production, where large quantities of potash are used
for fluxes. Although the quantity of potassium in the basic raw cement feed
material is small, the processing drives off the low temperature volatile
compounds, which after condensing, can be readily captured at a much
higher concentration [14]. On some installations, Feldspar and/or sodium
salts were added to the raw material to preferentially distil off the potassium
salts as halogens, thus enhancing the value of the collected material.
The first precipitator applications worldwide were mainly to combat the
worst effect of particulate air pollution. At the time there were no specific
emission levels to be met, so the size and installation cost of the electrostatic
precipitator was based on the recovery value of the collected material and
hence the collection efficiency was generally in the mid-nineties except for
easy to collect particulates such as sulphuric acid mist and the like.
Much of the development work in the smelter industry was carried out
in California, where severe air pollution, as a result of acid gas and
particulate emissions from the treatment of ores, slimes and concentrates,
was being experienced by all operating companies who were in litigation
with the fruit farmers. To counter this, precipitators were built by the
operators under licence from the Research Corporation, so different designs
were to be found, although most employed the 5 inch diameter tube, as used
by Cottrell et al. during their early work at Berkley, California.
A good example of the use of 5 inch tubes, is the precipitator installation
on the Garfield Smelter, Salt Lake City, Utah, where the builder, Mr W. H.
Howard [15], describes the precipitator, designed to handle 200000 cfm
(94.4 m 3 /s) of converter flue gas, as comprising 2520, vertical 5 inch diameter
by 10 ft long steel pipes, having No. 10 gauge iron wire as the discharge
electrodes, operating from a 25 kV rectifier arrangement. This plant operated satisfactorily and collected several tonnes of fume per day containing
some 50% lead, plus various precious metals, which had, up to then, been
lost to the atmosphere.

16

MILESTONES IN THE HISTORY OF PRECIPITATION

In 1914, because of pollution problems at the giant Anaconda Smelter,


investigations to optimise space and cost were carried out using tubes
between 12 and 36 inch diameter, which were powered by rectifiers of up to
150 kV. Reports of tubes up to 48 inch diameter, using 240 kV power
supplies, are also mentioned [11]. These large diameter tubes, however, had
their own operational problems and the final installation at Anaconda, in
1916, was a vertical flow arrangement having corrugated plate collectors.
The unit handled a gas flow rate of 700000 cfm (330 m 3 /s), and even by
today's standards was large.
Another smelter type installation was at Trail, British Columbia, where
one precipitator treated the gases from a lead blast furnace, at a design gas
flow rate of 100000 cfm (47.3 m 3 /s), using 13-inch diameter tubes and
another similar unit on a lead roasting furnace.
These installations are of historic interest in that people tend to consider
wide spacing between the electrode systems as a fairly new concept, whereas,
in reality, the early investigators recognised the cost advantages of larger
spacings and were not averse to trying them, in spite of the higher voltage
requirements.
As indicated earlier, not only were the problems of high resistivity dust,
and the use of temperature and moisture conditioning to overcome these
effects, recognised by operators in the smelting industry, but the investigators also found that by operating one precipitator at high temperature,
around 350C (650 OF), to remove the metallic elements, then after cooling
and operating a second unit at around 105 DC (220 OF), it was possible to
collect high purity arsenic trioxide commercially. This investigation gave
rise to the principle of selective material collection using precipitators.
Another area of special mention in the smelter industry is that of the
recovery of gold, silver and other bullion metals. In refining these are
distilled off as vapour, at the processing temperature, to consequently
condense on cooling as fume-size particles. These fumes were virtually
impossible to collect by conventional means, such as cyclones and scrubbers,
and were lost to the atmosphere. With the high commercial value attached
to these materials, a large number of precipitators were installed to collect
these fumes, their value being so high as to make the precipitator installation costs insignificant.
A surprising early application of the basic precipitation process can be
found in the Oil Field Industry, where Hydrofiners, or Dehydrators, for the
removal of water and the breaking of crude oil/water emulsions can be
traced back to 1915. Trials using both DC and AC energisation were found
to reduce 65% water/crude oil emulsions down to 0.5% water [16]. This
application has now been extended to cover other similar type applications
where there is a significant difference in the insulation properties of the
various phases being processed.
The earliest application in the iron and steel industry was the cleaning of
blast furnace gases for reuse. The gases from the blast furnace have a

PRECIPITATOR INSTALLATIONS

17

reasonable calorific value/heat content, but contain a fairly high dust


loading. This dust, which is rich in terms of potassium salts and hence a
valuable source of potash, produces difficulties with duct blockages, rapid
fouling of the burners, and unstable combustion. To overcome these
problems, the gases were originally cleaned at high temperature to retain
their sensible heat.
Possibly the earliest recorded successful hot dry precipitator installation
is that at the UK Skinningrove Ironworks, commissioned in 1917. This unit
is of the vertical flow, tube type and was energised by the Lodge valve form
of rectifier. Although many dry-type hot gas precipitators were installed for
cleaning blast furnace gases in the US, UK and Europe, problems with high
electrical resistivity of the particulates meant that performance was variable
and, subsequently, the wet electrostatic precipitator was found to be more
effective overall, in spite of losing the sensible heat content of the gases by
operating the plant at a much lower temperature.
Another early application was that of detarring illuminating (Town's
Gas) retort or iron works coke oven gases. During the production of town's
gas, or the manufacture of coke for use in the blast furnace, reasonably high
volatile coal is heated under a reducing atmosphere to distil off the
hydrocarbons and leave the coke behind.
The gases given off during this processing have a higher calorific value
than blast furnace gas and, in addition to a small quantity of dust, contain
a high level of tars, which condense on cooling the gases. To eliminate
downstream problems, these gases can be effectively de tarred using precipitators [17]. The earliest forms of precipitator were of the tubular vertical
flow type, but other designs, such as horizontal and vertical flow plate types,
can now be found.
An interesting development occurred in 1915, when the Hooker Electrochemical Company at Niagara Falls was under pressure from the local
authority to reduce the emissions from their bleach manufacturing plant.
The main problem arose over the discharge of chlorine gas. As the
electrostatic precipitator cannot collect gas molecules, it was decided to treat
the precipitator inlet gases with finely divided slaked lime; this reacted with
the chlorine to produce calcium chloride, which was satisfactorily collected.
This could be the first application of acid gas emission control using a dry
scrubbing technique; similar approaches are now widely used to abate many
acid gas discharges.
Although the sectors first tackled were the metallurgical smelters, followed by cement and iron and steel plants, all producing material having a
significant commercial value, it was not until the early 1920s that precipitators were applied to the power industries.
These followed the need for electric power and higher steam rate
production, which led to the development of the pulverised coal combustor.
These com busters gave serious air pollution, since up to 90% of the ash can
be carried over with the furnace gases, as compared with some 10% from

18

MILESTONES IN THE HISTORY OF PRECIPITATION

hand fired, chain grate and similar type stokers previously employed. With
pulverised coal firing, as the coal fly ash has only a limited commercial
value, the precipitator installation was primarily for air pollution control.
This is still true today, where precipitators are designed, dependent on the
fuel ash content and local legislation to collect in excess of 99.9% of the fly
ash. The collection of power station fly ash forms the largest single
application, in terms of both cost and number, of electrostatic precipitators.
While the foregoing includes some of the original installations, it is not
intended to be a complete history of events. With the formation of Research
Corporation as a worldwide licensing organisation by 1917, Linn Bradley
[18], reports: 'there are many plants in commercial operation on the
continent of North America, several on the continent of Europe, in England,
some in Japan, Africa and South America, and the work is growing very
rapidly'.
The following table lists some of the pioneer installations in, as far as
possible, chronological date order:
1884
1907
1910
1912
1912
1913
1914
1914
1914
1915
1915
1915
1915
1915
1916
1917
1918
1919

Lead fumes - roaster


Sulphuric acid mist -contact
process gases
Lead zinc - smelter
Cement kiln dust
Lead fume - con verters
Bullion recovery-slimes
Lead - Dwight Lloyd sinter
Lead - blast furnace
Cleaning roaster off gasessulphuric acid production
Lime inj. chlorine collection
Tar removal- towns gas
Carbon from calcium carbide
furnaces
Ventilation air cleaning
Dehydration crude oils
Paper pulp alkali salt
Hot BF gas cleaning
Copper smelter plants [19]
Copper smelter - central gas
cleaning. 2000000 cfm

Bagillt, Wales, UK
Pinhole, CA, USA
Balaklala, CA, USA
Riverside, CA, USA
Garfield, UT, USA
Niagara Falls, NJ, USA
Tooele, UT, USA
Trail, B.c., Canada
Germany
Niagara Falls, USA
Portland, OR, USA
Germany
New Haven, CT, USA
CA, USA
Canada
Skinningrove, UK
Japan
Anaconda, MT, USA

By 1920, the art of electrostatic precipitation had made great strides, from
the initial laboratory investigations of a scientific phenomena to becoming
an established industrial approach to collecting all forms of particulates and
fumes on many diverse applications, as indicated in the foregoing table.

PRECIPITATOR INSTALLATIONS

19

Although many of the initial plants adopted the vertical tube and axial wire
configuration, H. D. Braley, in his 'Notes on Electrostatic Precipitation' 1919
[16], indicates the following:
1. Plant designs
Gas-type treaters
(a) Tube
(b) Plate type
(c) Water film plate type
Liquid treaters
(a) Stationary electrodes
(b) Rotating electrodes.
The water film plate arrangement, where water is allowed to continually flow over
the collectors, was developed for the collection of 'difficult dusts', which had been
recognised by the early investigators, or alternatively, for use as a secondary series
low temperature unit for selective dust precipitation, i.e. following an upstream
dry unit or scrubber for removing the heavy coarser materials.
In the treatment of liquids, the idea of rotating electrodes was to break up
coalesced water chains, which on some designs shorted out the electrode system
when handling water/oil emulsions.
2. Operational temperatures
Investigations of operating practices have indicated that the following represent a
fair average of operating conditions.
Cement plant and pyrites roasters
300-500 DC
Copper convertors and metallic dusts
180-200 DC
Condensible arsenic trioxide
Below 125 DC
80-120 DC
Lead smelters and similar
Below lOODC
Water or wet film treaters
3. Gas velocity
Majority of plants operating
Lowest velocity found
Highest velocities used

5-6 fps (1.5-1.8 mm/s)


2.5 fps (0.76 m/s)
8-10 fps (2.4-3.0 m/s)

4. Direction of flow
For tube type, both up and down flow can be found, dependent on the
application and site conditions. With plate types, horizontal as well as vertical
approaches have been used. Both pressure and suction designs have been installed
to suit a particular application.
5. Capacity of rectifier equipment
Most precipitators in service are energised from rotary rectifiers and high voltage
transformers providing a maximum operating voltage of 60 kV. Figures quoted
for specific powers are interesting, in that, for cement plants the specific power is
given as 175 W per 1000 cfm (269 W per m 3 /s) or 1.25 W per ft 2 (13.5 W per m 2 )
of collector, whilst for copper and lead smelters, the specific power is 600 W per
1000 cfm (922 W per m 3 /s) and 3.0 W per ftz (32 W per m 2 ) of collector. This
would explain one of the dilemmas facing some of the early investigators and

20

MILESTONES IN THE HISTORY OF PRECIPITATION

designers, in terms of correctly sizing the rectifier equipment, particularly in


association with the very primitive control methods of handling power arcs
within the precipitator.

Most industrial precipitators at this stage were sized and designed by


'rule of thumb' approaches, based on the broad foundations established by
Dr Cottrell et al. Although the basic physics of particle charging and
migration were known, it was not until 1919/1920 that Evald Anderson, a
co-worker of Walter Schmidt, experimentally established the exponential
relationship between efficiency and gas flow, later translated to plant size
[20]. Walter Deutsch, a mathematical physicist, working for Metallgesselschaft in Germany, theoretically proved the logarithmic relationship between efficiency gas flow and collector area in 1922 [21].
The equation developed by Deutsch in 1922, later revised in 1926, was
used for almost 50 years by precipitation engineers for sizing plants having
similar dusts, process applications and design efficiencies, but different flow
rates. In the 1960s, when new and more stringent legislation called for
emissions to be reduced by an order of magnitude, the Deutsch Equation
was modified by Matts and Ohnfeldt [22] amongst others, into a more
operational form. All present day precipitator equipment suppliers have
adopted some version of the modified Deutsch Equation for plant sizing
duties and the success of these modern plants can be traced back to the
findings of the early workers in the precipitation field.

2.2.

2.2.1

Development of electrical supplies

Rectifier types

The ultimate success of electrostatic precipitators for industrial applications


can be attributed and paralleled to the development of suitable electrical
supplies. Mention has already been made of the need to simultaneously
provide high voltage and corona current for particle charging and the
reason for the failure of the initial plants using electrostatic-type generators.
The contribution of Dr Cottrell, in developing the synchronous switch
rectifier, was possibly the greatest technological breakthrough and allowed
the precipitator to be taken from the laboratory into the field.
Although Lodge's high tension valve rectifier was widely used in the UK,
by the then recently formed Lodge Fume Deposit Co., the system, however,
was not very efficient as a rectifier, had power limitations and, being fragile,
was not readily acceptable by most users.
Several DC generators have been investigated and developed over the
early part of the century, the earliest probably being that from the Girvin
Electric Co. of Philadelphia, where the machine was a vertical belt driven

DEVELOPMENT OF ELECTRICAL SUPPLIES

21

device, having a rotating field and provided with intermediate commutating


poles. The armature coils were oil immersed and each connected to the
commutator and wired in such a manner as to develop the required HT
voltage. The machines were some 33 inch in diameter and 62 inch in height,
and could deliver voltages of up to 75 kV at 10 kW.
In the late 1950s, Prof. Fellici proposed a DC generator working on the
Van cler Graff principle, and although successful in small field trials, the
rapid development of silicon technology supplanted its use industrially.
In order to progress the art of precipitation worldwide, a strong friendly
relationship developed between the main investigators, Sir Oliver Lodge in
the UK, Dr Cottrell in the US and Dr Moeller in Germany. Archive
evidence shows that there were biannual meetings between the parties to
review their latest developments, patents and findings. This close academic
and free interchange of ideas between what was to become Lodge Cottrell,
Research Corporation and Metallbank und Metallurgische Ges. (Lurgi),
lasted until the US Sherman Anti-Trust Laws made such Associations
illegal.
One result of these discussions was that Dr Cottrell allowed Sir Oliver
Lodge to use the rotary switch rectifier for precipitator duties. In recognition
Lodge, in 1921/1922, renamed his company Lodge Cottrell Ltd. In Germany, Lurgi also used the rotary switch rectifier during this period of major
and very varied plant installation.

V1.cl
Electrically 180 0 shift

Figure 2.3 Basic principle of mechanical switch rectifier. C, edge connections on insulating disc;
E, earth or ground connection (positive); M. synchronous motor (half mains frequency); p.
precipitator HT connection (negative); T. transformer secondary winding (AC).

22

MILESTONES IN THE HISTORY OF PRECIPITATION

While the rotary rectifier was mechanically robust and reasonably


efficient, it was noisy and needed a large room because of insulation
requirements and ventilation since the arcing produced both NOx and
Ozone. The operation of the rectifier is shown schematically in Figure 2.3
and, although dating from 1908, this form of rectifier was still being supplied
in an updated configuration by Lodge Cottrell Ltd in the 1960s for Utility
Plant applications.
To reduce some of these difficulties there was a trend, particularly in the
US, to use thermionic high voltage diodes or Kenetrons as the valve was
called. In Europe development in the selenium rectifier field in the 1940s and
early 1950s, tended to replace the rotary rectifier for new installations.
The advent of silicon technology in the late 1950s and early 1960s
changed the whole of the precipitator electrics scenario. Virtually all new
installations from the mid-1960s onwards were fitted with silicon rectifiers
because of their advantages over other forms of rectifier. Not only are they
vastly superior in terms of efficiency and are much smaller in size, but the
control of the primary input and automatic optimisation systems now
incorporate the latest silicon technology.
2.2.2

Primary control systems

For many years the only method of controlling the voltage, and hence
corona current, was by means of tap changing on the transformer input.
This simple tap changing was later modified to an autotransformer or
moving coil regulator, where the incoming voltage was modulated, initially
by hand wheel, then by a motor drive, as voltage optimisation systems were
introduced.
The next type of control was by the use of magnetic saturable reactors
or transductors located in the primary circuit, where the output from the
saturable reactor is varied by altering the impedance of the device. This is
achieved by a DC current passing through a separate winding on the core
of the reactor.
Although reasonably successful, these devices were rapidly supplanted by
the silicon controlled rectifier or thyristor in the early 1960s. These devices
control the power into the transformer by modulating the firing angle of the
incoming supply voltage and are superior to previous methods of control.
2.2.3

Automatic control systems

With the need to optimise precipitator performance at all times, the


development of the automatic voltage control system dates from the 1950s.
It was recognised, by this stage, that the optimum precipitator collection
efficiency was obtained at the maximum operating voltage as predicted from
theory. Once the maximum operating voltage had been achieved, any

REFERENCES

23

further increase in primary supply caused the secondary voltage to fall and
the current to rise as a result of increased arcing within the precipitator.
The early automatic voltage controllers were electro-mechanical systems,
where either the kV or current was monitored and action taken if a certain
value was reached or exceeded. These systems were fairly basic in concept
and led to discussions as to whether voltage or current control was the best
approach. Modern systems monitor both current and voltage to fully
optimise performance.
Several forms of wholly electrical device were developed using thermionic
valve or magnetic amplifier approaches, but with the rapid development of
silicon technology in the 1960s, these were quickly superseded. Initially
simple analogue designs were used, then digital and finally microprocessor
based, silicon high speed switching systems, which operate in conjunction
with the silicon controlled rectifiers, or thyristors, used for primary power
modulation.
The latest controller systems are of the stand-alone type, using sophisticated microprocessor units, having programmes which can make the
precipitator completely automatic in terms of operation. Facilities can be
included to provide complete start-up and shut-down of the precipitator,
rapping control, data and fault logging and some provide even a fault
finding menu aid, as well as back ionisation detection, pulse modulation and
an interruptable power supply to suit a particular set of site conditions.
Generally these controllers interface with the plant DeS system for overall
simplicity of control.

References
It is not intended for this chapter to be a complete review of the development of electrostatic precipitation, but only to indicate the major milestones
in the development of the present high performance plant widely used in
industry for the collection of all types of particulate. For those interested in
obtaining further insight into the early development of electrostatic precipitators, the following reference works should be examined.
1. Gilbert, W. (1900) De Magnete. (English Edit.) Thompson S.P., pp. 24-5, London, UK.
2. Hohlfeld, M. (1824) Das Niederschlagen des Rauches durch Elektricitat. Kastner Arch.
Gesarnrnte Naturl., 2, 205-6.
3. Guitard, C.F. (1850) Condensation by electicity. Mech. Mag. (London), 53, 346.
4. Lodge, 0.1. (1884) Dust Free Spaces. Lecture to the Royal Dublin Society, April 2nd (see
Transactions of Society).

5. Lodge, 0.1. (1886) The electrical deposition of dust and smoke with special reference to the
collection of metallic fume and to a possible purification of the atmosphere. J. Soc. Chern.
Ind., 5, 572-6.
6. Walker, A.O. (1884) British Patent No. 11120.
7. Moeller, K. (1884) German Patent No. 31911.
8. Lodge, 0.1. (1884) British Patent No. 24305.

24

MILESTONES IN THE HISTORY OF PRECIPITATION

9. Lodge, 0.1., Muirhead, A. and Robinson, E.E. (1905) British Patent No. 25047.
10. Cottrell, F.G. (1908) US Patent No. 895729.
11. Cottrell, F.G. (1916) Recent progress in electrical smoke precipitation. Paper Presented at
the Second Pan-American Scientific Congress, Washington, Dec. 1915-Jan. 1916. Eng.
Min. J., 101, 385-92.
12. Cottrell, F.G. (1911) The electrical precipitation of suspended particles. J. Ind. Eng. Chern.,
3, 542-50.
13. Schmidt, W.A. (1912) The Control of Dust in Portland Cement Manufacture by the
Cottrell Precipitation Processes. Proc. Eighth International Congress. Appl. Chern., 5,
117-24.
14. The Times Engineering Supplement (1917) October 27th.
15. Howard, W.H. (1914) Electrical precipitation at Garfield. Trans. Am. Inst. Min. Eng., 49,
540-60.
16. Braley, HD. (1919) Notes on electrostatic precipitation. 25th General Meeting American
Electro-Chemical Society New York. April 3-5 1919, pp. 13-43.
17. White, A.H. et al. (1914) The electrical separation of tar from coal gas. Am. Gas Light J.,
101, 177-80.
18. Bradley, L. (1917) The Cottrell process in practice. Abstract of Paper Presented at
American Institute of Mining Engineers, and American Electro-Chemical Society,
January 26 1917. Electr. Rev., 80 (2066).
19. Hirota, R. and Shiga, K. (1920) Electrical precipitation in Japanese smelters. Chern. Metall.
Eng., 22 (6), 276- 7.
20. Anderson, E. (1925) Some factors and principles involved in the separation and collection
of dust, mist and fume from bases. Trans. Am. Inst. Chern. Eng., 16, 69.
21. Deutsch, W. (1922) Bewegung und Ledung der Elektricitatstrager in Zylinder Kondensator. Ann. Physik, 68, 335.
22. Matts, S. and Ohnfeldt, O.P. (1963-1964) Efficient gas cleaning with the SF electrostatic
precipitator. Flakt Ret"., 6, 7, 105-22.

3 Basic and theoretical operation of ESPs


C. RIEHLE

3.1

General remarks

In electrostatic precipitators (ESPs) the separation of particles from flue


gases is based on electrical means. The separation process can be divided
into five essential steps which are schematically illustrated in Figure 3.1:
1. generation of charge carriers (section 3.2);

2.
3.
4.
5.

charging of the particles (section 3.3);


deflection and separation of the particles (sections 3.4, 3.5);
dust deposition (section 3.6);
dust removal (section 3.7).

In most applications the steps are not as isolated as represented here, but
the steps can happen all in one operational section. These are the so-called
single-stage precipitators, which are the most important configurations for
industrial processes (Figure 3.2a). In so-called two-stage precipitators steps
1/2 are executed separately from steps 3/4/5, i.e. in the first stage the ions
are produced and the particles are charged while in the second stage the
particles are collected and removed (Figure 3.2b). The two-stage configurations are only used in small-scale applications as, for example, clean room
technology. Since single-stage precipitators have a far greater industrial
significance than two-stage ESPs, the discussions will be confined to the
first. However, most of the presented subjects and equations can be equally
applied to two-stage configurations.
In single-stage electrostatic precipitation, two principal different designs
exist: tube-type (Figure 3.3a) and plate-type precipitators (Figure 3.3b). In
plate-type precipitators a row of discharge wires - orientated along the
gravitational axis - are positioned between parallel collecting plates forming
a duct. Dust is deposited on the collecting plates due to electrical and Van
der Vaals forces. The plates are cleaned by mechanical impact and the dust
layer or large agglomerates fall down into the hoppers. An alternative to
removing the dust layer by mechanical impact occurs in wet precipitators,
where the dust is removed by a flushing liquid flowing over the plates.
In industrial precipitators the total length of the collecting plate is
divided into series and parallel fields. A 'field' is characterized by an
independent power supply unit. Therefore, the fields of a precipitator can be

26

BASIC AND THEORETICAL OPERATION OF ESPs

flue gas

generation of
charge carriers

charging and
deflection of
particles

E
rapping

dust hopper

Figure 3.1 Schematic illustration of the separation process in ESPs.

operated with different electrical conditions leading to different transport


situations for the particles. Thus the operating conditions can be optimized
to different dust loadings and different particle sizes as the gas passes
through the precipitator.
In tube-type precipitators the discharge wire and the tube are orientated
along the gravitational axis and, in most current applications, the collecting
plate is cleaned by liquid films running down into a sump at the precipitator
base. For this reason tube-type ESPs are often applied to processes with
self-draining liquid particles, e.g. acid mists or tar. In the past, tube-type
precipitators were used for dry applications, where the dust was removed by
mechanical rapping as in the plate-type approach previously mentioned.
In contrast to the plate-type, in the tube-type, the gas flow and the
discharge wires are aligned. Apart from the precipitator arrangement in
question, the gas flow conditions should be carefully designed, since reentrainment of particles from high gas velocities and sneakage are often
severe problems preventing high efficiencies being obtained. Therefore,
special care should be taken in obtaining a homogeneous gas flow distribution at the precipitator inlet.

27

GENERAL REMARKS

clean gas

(a)

1. stage

charging

2. stage

collecting

clean gas

raw gas

~d",'
(b)

Figure 3.2 (a) Single-stage precipitator; (b) two-stage precipitator.

Besides the flow field, the precipitator's state of operation is determined


by the electrical conditions and the dust which is to be collected. Figure 3.4
overviews the influencing parameters and their interactions with respect to
particle transport. The interaction between the electrical state and the flow
field is called electrohydrodynamic. For example the turbulence intensities
of the flow field change when switching on the high voltage [1- 3]. The
interaction between the dust particles and the electrical state governs
charging and discharging mechanisms and space charge effects. The interaction between the dust particles and the flow field governs the particle

28

BASIC AND THEORETICAL OPERATION OF ESPs

clean gas
discharge
wire with
high voltage

high voltage frame with


discharge electrodes

earthed
collecting
electrode

gas

(a)

18

LNE - - - 1 - - -

rawgas

(b)

earthed collecting electrode

Figure 3.3 (aJ Tube-type electrostatic precipitator; (b) plate-type electrostatic precipitator.

Figure 3.4 Overview of influencing parameters and their interactions with respect to particle
transport in a precipitator.

dispersion and the existence of particle concentration profiles. Figure 3.4


illustrates that particle transport is governed more by the interactions than
by the influencing parameters themselves. Furthermore, apart from the dust,
all influencing parameters and interactions are a function of the precipita-

ION PRODUCTION

29

tor's geometry. Therefore, similar transport conditions for particles on


different scales of precipitator size provide at least geometrical similarity.
The problem of electrical similarity in ESP operation was investigated
recently [4, 5].
3.2

3.2.1

Ion production

Principles

To produce large numbers of charge carriers in a gas phase, a critical


electrical field strength has to be overcome. To create this high electrical
field strength, high voltage is applied to a special electrode design. When the
applied voltage exceeds a distinct value, an electrical current between the
two electrodes can be measured indicating a corona discharge. This is the
corona onset voltage as illustrated in Figure 3.5. A further increase in
voltage will lead to a progressively increasing current until spark-over
occurs. The spark-over marks the electrical breakdown of the gas. Figure
3.6a shows a photograph of corona discharges along five parallel smooth
wires, while Figure 3.6b shows a successful photograph of a spark-over in a
laboratory ESP taken by J. Miller [6]. This shot also visualizes the corona
discharges at the ends of the emitters, in this case, spiked-type electrodes.
What is the characteristic of a corona discharge compared with a
spark-over and how does it start? Usually a gas contains in the order of 10 19
neutral molecules per cm 3 . Because of natural radioactivity and cosmic
radiation some molecules become ionized and recombine immediately
afterwards. If an electrical field is present during ionization, the electron will
be accelerated and separated quickly from the remaining positive ion
(Figure 3.7). After a short distance the electron will hit another neutral gas

spark
over

-...
c:

::l
U

voltage-

Figure 3.5 Typical current-voltage relationship.

30

(a)

BASIC AND THEORETICAL OPERATION OF ESPs

(b)

Figure 3.6 (a) Corona discharges along five parallel smooth wires [24]; (b) spark-over in a
laboratory ESP (photograph taken by J. Miller [6]).

molecule and an additional, second electron will be produced, provided the


kinetic energy has been high enough for ionization. Thus, an avalanche
effect is able to start in regions where the electrical field strength is high
enough to bring the electrons to ionization energy levels. Recombining ions
and highly activated molecules emit photons and the weak blue glow is
therefore characteristic for a corona discharge in addition to a crackling
noise. An inhomogeneous electrical field occurs at curved surfaces, e.g. wires,
which means that the high field values are restricted to regions close to the
wire. This region, where ionization processes occur, is called the active zone.
When the moving electrons come into the region of lower electrical field
strength, the passive zone, they are not able to ionize any further molecules,
but they will attach themselves to an electronegative gas molecule, such as
02' SOz, Cl z, thereby forming negative gas ions. The spatial extension of the
active zone is only a few percent of the passive zone, i.e. particle separation
occurs almost completely in the passive zone.
The drifting charge carriers, which are continously produced by the
so-called 'trichel pulses' of the corona discharge, represent the electrical

ION PRODUCTION

: D

: passive

31

:~~:~;,?(; ~

:~1~g'f,~~
o

00

active zone: electron + molecule ---> 2 electrons + positive ion

passive zone: electron + molecule ---> negative ion

Figure 3.7 Principle of a corona discharge.

discharge current. If the applied voltage, however, exceeds a certain level, a


discharge begins to pass straight through the gas as a 'streamer'. Suddenly
electrical breakdown of the gas sets in and spark-over occurs.
In general, the discharge electrode can be operated on negative or
positive polarity. For a given geometry, however, the corona initiation voltage and the electrical breakdown of the gas occur at higher voltages for
negative energization than for positive. Because of these higher electrical
field strengths, most industrial applications prefer negative corona, i.e. discharge electrodes are energized with negative high voltage and the collecting
plates (positive) are grounded. Readers who are interested in more details
of corona discharge with respect to the ESP application may refer to [7,8].
3.2.2

Corona initiation field strength

From section 3.2.1, it should be clear that the precipitator configuration, the
gas state and especially the gas composition, determine an essential part of
the ESP's electrical properties. The electrical field required to start a
contino us ionization process will obviously depend on the ionization
energies of the gas species present and the mean free path between collisions.
Since the mean free path is related to the gas state, the corona initiation field
strength will be a function of gas density. Furthermore, the curvature of the
wire determines the inhomogeneity of the electrical field. A pure theoretical
description, however, is still missing because the analysis of a complex

32

BASIC AND THEORETICAL OPERATION OF ESPs

mixture of gases is difficult. Peek, however, has proposed a semi-empirical


relationship (equation 3.1) where the initiation field strength Eo is expressed
as a function of the relative gas density b (equation 3.2), the discharge wire
radius r SE and two empirical constants, A and B, characterizing the gas and
corona polarity [9].

Eo

Ab

+ BJ b/rSE

(3.1)

t5 = P2 = P2 . Tl
Pl Pl T2

(3.2)

Usually temperature and pressure are related to normal conditions, i.e.


Tl = To = 273 K and Pl = Po = 1 bar. For typical ESP conditions (negative
corona), Robinson [1OJ recommends as empirical constants:

Figure 3.8 shows the corona initiation field strength Eo as a function of


wire radius. The thick line represents the relationship for the recommended
values; the thin lines were calculated with the constants A, B for pure S02
and COb respectively, in order to show the sensitivity to gas composition.
In general an increase of wire radius leads to a decrease in electrical field
strength, this trend reducing for radii> 1 mm.

200

...>
o

.5

--0---

ESP recommendation
Air

---{'S)--

SO 2

--D--

CO 2

...g'

..c

150

!!!

-;

'C

Qj

:;:::

0.1

0.3

0.5

0.7

0.9

1.1

1.3

1.5

wire radius in mm
Figure 3.8 Corona initiation field strength Eo as a function of wire radius.

33

ION PRODUCTION

3.2.3

Corona onset voltage

The conditions for the corona initiation field strength are essentially set by
the design of the discharge electrode. The voltage necessary to overcome this
critical field strength, however, is set by the complete configuration of
discharge and collecting electrodes. For this reason the corona onset voltage
has to be additionally a function of the collecting electrode's design.
Therefore, a distinction between the two main ESP designs must be
considered (see Table 3.1 for typical characteristic values of ESP operation).
3.2.3.1 Tube-type. The collecting electrode of a tube-type ESP is simply
described by the tube's radius r NE . From electrostatic field theory the
following relationship can be deduced for coaxial electrode configurations:

(3.3)
In Figure 3.9 the corona onset voltage is plotted as a function of the
collecting tube radius for three different wire radii. The curves hold for a
relative gas density of 1.0 (this dependence will be discussed in chap. 16).
Generally, increasing tube size needs higher voltages for corona onset.
Alternatively, the onset voltage can be lowered by reducing the radius of the
discharge wire.

Table 3.1 Typical characteristic values of ESP operation

Va (mjs)
2s(mm)
E(kVjcm)
j(mAjm2)

Particle concentration
in raw gas (gjN m 3)
Efficiencies (%)
Applications

Tube

Plate

1.5-2.5
150-250
5-5.5
-1.0
Up to 10

1.0-2.0
250-400
3.5-4
-0.5
Up to 1000

99-99.5
Mist collection, e.g. S03'
hydrocarbons, tars

Flow rates (m2js)


tJ.p
Energy consumption

Up to 20
20

99.5-99.9+
Dry dusts, e.g. P.F., cement, iron
ore; wet dusts e.g. BF gas, BOS
gas, DESOX gas
Up to 1000
10-15

Efficiency (%)

DC power (Wjm 3 js)

DC power" (W jm 3 js)

99.0
99.5
99.75
99.90

500
1000

170
300
450
1000

"Depends on type of discharge electrode.

34

BASIC AND THEORETICAL OPERATION OF ESPs

45
-----cr- rSE = 0.5 mm

>
.II:

.=
CD
Cl

ca
"0
>

li
III

40

--0-- rSE

= 1.0 mm

--!Sf-- rSE

= 1.5 mm

35
30

c:
0

ca

25

.0

.,'"

c:

...
0
0

C)

.0

-7

20

iii'
<{l

0'
::::J

15

50

100

150

200

radius of collecting tube rN

250

300

in mm

Figure 3.9 Plot of corona onset voltage as a function of the collecting tube radius for three

different wire radii.

3.2.3.2 Plate-type. Beside the wire radius rSE and the wire-to-plate distance s for the determination of plate-type configurations another parameter
is necessary, i.e. the distance between the discharge wires, 2e (Figure 3.10).
When describing plate-type configurations, a characteristic length d is
introduced, which is a function of the wire-plate distance s and the
wire-to-wire distance 2e. The corona onset voltage for plate-type ESPs is
given by equation (3.4) while the characteristic length ratio is determined by
equations (3.Sa,b,c).

35

ION PRODUCTION

14

12

;.

10
3

d/2c

2.5

:1:

,'.2<.

2
0.8

0.85

0.9

0.95

1.05

1.1

4
2
0

0.2

0.4

0.6

0.8

1.2

1.4

s/2c
Figure 3.11 Plot of values for d/2c, dependent on the s/2c ratio.

des, 2c)
Vo = Eo(6, rSE )' rSE 'In - - rSE
d

2c

n 2c'

(3.4)

for

s/2c < 0,3

(3.Sa)

~
= 18 . exp (2 96 --:)
2c'
'2c '

for

0,3 < s/2c < 1,0

(3.5b)

~
= ~ . exp (n . ~ ).
2c 2n
2c '

for

1,0 < s/2c

(3.Sc)

Dependent on the s/2c ratio, the values for d/2c can be calculated
according to the different relationships (3.5a,b,c) and are plotted in Figure
3.11, from which it will be seen that wire-plate spacings larger than the
wire-to-wire distance progressively dominate the characteristic length. Higher characteristic lengths mean that higher onset voltages are required. In
other words, a very short wire-to-wire distance interferes with the corona
discharge, while wide wire-to-wire distances need lower corona onset
voltages (the electrical field in the space between them and the collecting
plate, however, will decrease, resulting in deteriorating particle collection).
Table 3.2 gives examples for the value range of the characteristic length
d for typical wire-plate distances and some variations in wire-to-wire
distance. Figure 3.12 illustrates the corona onset voltage for plate-type

36

BASIC AND THEORETICAL OPERATION OF ESPs

Table 3.2 Examples of typical size values in plate-type design with characteristic length d

200

ISO

s,/mm

250

5/2c

d/2c

2c/mm

d/mm

2c/mm

d/mm

2c/mm

d/mm

0.5

0.79
3.57
17.72

300
150
100

237
535.5
1772

400
200
133

316
714
2357

500
250
166

395
892.5
2941

1.0
1.5

configurations as a function of the characteristic length d for different


electrode radii. The behaviour is analogous for tube-type ESPs, where an
increasing d means higher corona onset voltages. Again the onset voltage
can be lowered by reducing the radius of the discharge wire.

3.2.4

Current-wltage relationship

The electrical current is formed by gas ions passing through the passive zone
as mentioned above (section 3.2.l). The drift velocity of these gas ions is
proportional to the electrical field applied and the constant of propor-

70

--D---- rSE

= 0.5 mm

---0-- rSE

= 1.0 mm

---{5J--

>

60

~~~~~~~~--~~~~~~"

rSE = 1.5 mm

.=

GI

01

50

:l

'0
>

3ic:
o

.oo

ca
c:

CJ

40

30

~
~
a.

20

dJ
'{2

0"

10~~~~~~~~~~~~~~~~~~~~~~~~ ~

500

1000

1500

2000

2500

3000

3500

characteristic length d in mm
Figure 3.12 Corona onset voltage for plate-type configurations as a function of the characteristic length d for different electrode radii.

37

ION PRODUCTION
Table 3.3 Mobilities of single-charged gas ions at 0 'C and 1.0 bar [8]

Gas
Air (dry)
Air (very dry)
COz (dry)
HzO (lOOT)
SOz

Negative ions

Positive ions

2.1
2.5
0.98
0.95
0.41

1.36
1.8
0.84

1.1
0.41

tionality is called the electrical mobility, hi' of gas ions. Since theoretical
expressions for mobilities, derived by kinetic theory of gases, are still
questionable, experimentally determined values are recommended whenever
these are available. Table 3.3 gives a few mobilities for positive and negative
gas ions [8]. It has been found experimentally that ion mobility is almost
inversely proportional to the relative gas density (5 (equation 3.2) over a
wide temperature-pressure range, thus equation (3.6) holds. Since most
ESPs operate with negative corona and a gas composition similar to air, the
following mobility value for normal conditions is assumed (equation 3.7).
h(

p,

T) = h(p 0' To)


(5

(3.6)
(3.7)

F or typical electrical field strengths e.g. 1 - 5 X 10 5 V1m, the gas ion's


velocity ranges between 20 and 100 m/s.
The operating condition of an ESP is mainly governed by its electrical
state. The relationship between the applied voltage and the resulting current
is called the current -voltage characteristic. This curve describes all possible
electrical operating states of a given ESP configuration. Usually the total
electrical current i tot is related to the total collecting area ANE,tot. Thus the
electrical current density jNE results (equation 3.8), having the unit of current
per unit area. Sometimes the total electrical current is related to the total
length of discharge wires L SE , tot. Thus an electrical current per unit length
results (equation 3.9).
.

i tot

JNE=~~

(3.8)

ANE,tot

.
i tot
JSE = - LSE,tot

(3.9)

In practice, however, the electrode designs are not always as simple as


illustrated in Figures 3.3a and 3.3b. Especially the discharge electrodes,

38

BASIC AND THEORETICAL OPERATION OF ESPs

L
~SP;"
~wlre

IT ""oat'"
M
strip

Star

Figure 3.13 Discharge electrode elements.

which are rarely round wires. All sorts of sharp edges, points, barbs, etc., are
common and Figure 3.13 gives only a few examples. For most of these
discharge electrodes, a theoretical description for the current-voltage relationship is impossible. Thus the characteristic of a certain electrode has to
be measured in the laboratory. A great variety of collecting plate designs
also exist and some examples are shown in Figure 3.14. The background of
collecting plate profile design is directed more towards the improvement of
particle collection and retention and less by electrical properties. Most
collector profiles can be considered flat in relation to the electrode/collector
spacmg.
When designing industrial precipitators an estimation of the electrical
power is needed and the dependence on geometry might be helpful.
Therefore, the next sections introduce the theoretical approaches to
current - voltage relationships for idealized geometries.

3.2.4.1 Tube-type. In tube-type ESPs, the number of tubes Nt equals the


number of discharge elements NSE and the tube length LNE is approximately
equal to the wire length LSE (see Figure 3.3a). Therefore, equation 3.10 holds
for the current per discharge wire jSE' When the total current is related to
the collecting area jNE' equation (3.11) results, wherein, it represents the
current per tube and i tot the total current.
(3.10)

39

ION PRODUCTION

~~-~~~~~~~~~~~

J~--~J---J~~J

~~-~~~~~~~~~~~

J~--~J---J~~J

Duct section NTS electrodes omitted

Gas flow

Opzel baffle

Collector bottom tube


Opzel design (Research Cottrell)

Figure 3.14 Collecting plate designs: Opzel Design (Research Cottrell); Catch Space (Lodge
Cottrell).

(3.11)

Considering the last two equations a general transformation rule for both
kinds of current density in tube-type ESPs can be given:
.

jSE

(3.12)

JNE=~r

NE

With the assumption of low current flow, a quite simple relationship can be
deduced from the theory of electrostatic fields (equation 3.13). This approximation was originally published by Townsend in 1915 [11]. It is applicable as long as equation 3.14 holds. The permittivity of vaccuum 8 0 is
8.86e12 As/Vm.

c5

.
jSE
48 0 . b(c5) . U (U - U O(rNE' r SE '
JNE = - - - = - " - - - - - - - - - - ' - - - ' - " = - - . = : . . 2nrNE

rNE )3 . InrrNE
-

(3.13)

SE

(3.14)

40

BASIC AND THEORETICAL OPERATION OF ESPs

~E
,5
UJ

-0--- rNE

= 100 mm

---0- rNE

=150 mm

-----Q-- rNE

= 200 mm

,~

-...
>-

rSE = 1.0 mm

'iii
c::

CD

'C

T=423K

=3.1*1 04 m2Ns

c::

;:,

()

40

20

60

80

100

applied voltage U in kV
Figure 3.15 Dependency of current density, j"E' on the collecting tube radius.

Figure 3.15 illustrates the dependence of current density jNE' according to


equation (3.13), on the collecting tube radius. The wider the duct, the
progressively lower the current density becomes for a given voltage. Compared with the collecting tube radius, the influence of the wire radius is
negligibly small.
In the case of higher currents, a simple calculation becomes impossible,
The corresponding expression is given in equation (3.15), according to
Townsend [IIJ, where 'P t and <1\ are defined by equations (3.16) and (3.17),
respectively. Equation (3.15) is plotted in Figure 3.16, therefore for practical
purposes only equations (3.16) and 3.17) have to be used.

~-

'P t

'P

= U - U0
t

U0

<D _ (

rNE

t-

EO'r SE

1 -In 1 +

In rNE
rSE

)2, r

NE -jNE

Bo'b(<5)

(3.15)
(3.16)
(3.17)

3.2.4.2 Plate-type, Plate-type ESPs consist of a number of parallel ducts


Nd (see Figure 3.3b). The number of discharge electrodes per duct is N SE '
The current per unit wire length jSE is given by equations (3.18), where ito\, id
and iSE represent the total current, the current per duct and the current per

41

ION PRODUCTION

350
300
250
200
<Ilt

150
100
50
0
0

12

15

'l't
Figure 3.16 Plot of equation (3.15).

discharge wire, respectively. For the current density per unit collecting area
equation (3.19) holds.

jNE'

.
JSE

.
JNE

i tot
T tot

'-'SE

Ndid

=N N h
d

SE SE

N did

i tot

NSEiSE
NSEhSE
id

= A~~ = NdANE = 2LNEhNE

iSE

hSE

(3.18)

(3.19)

Combining the last two equations, a general transformation for both types
of current density can be deduced (equation 3.20). This can be approximated
to equation (3.21) (where 2c = discharge element spacing, Figure 3.10)
because the relationships (3.22) and (3.23) hold for most plate-type designs.
.
JNE

.
= JSE

NSEhSE
2L h

(3.20)

NE NE

(3.21)
(3.22)
(3.23)

42

BASIC AND THEORETICAL OPERATION OF ESPs

--0-- 5

0.8

.~

0.6

0.4

E
>'iii

--!Sf--

-......

P = 1 bar

b = 3.110-4m 2Ns
.0

r:::

41

:::s

(.)

5 = 2c = 250 mm

T = 150C

r:::

41

= 2c = 150 mm

~s=2c=200mm

0.2

'"

rsE =1 mm

-"

Eo = 43S10 5V/m
Uo = 28.5 kV

2-

10

20

c;i.

w
f:

W
~

30

40

50

60

70

80

Applied Voltage in kV
Figure 3.17 Plot of current-voltage relationship for a range of parameters according to
equation (3.24).

The wire-plate arrangement is much more difficult to handle theoretically


than the wire-tube configuration. However, P. Cooperman derived an
analogous solution for the current density (equation 3.24) by assuming low
currents, which is fulfilled as long as equation (3.25) holds [12].
.

jSE

JNE = 4c =

nco' b(6) . U' (U - U o(d, rSE ' 6


d
c 52 'In

( __5_)2 . _1_.
EO'r sE

In ~

c'iNE 1
b(6)

(3.24)

(3.25)

n' Co'

rSE

The current-voltage relationship according to equation (3.24) is plotted in


Figure 3.17, for a range of parameters.
For high currents, G. Cooperman proposed a different approach [13]. He
applied an approximative solution for high currents given by White [7J to
certain parts of the precipitator duct. The result and definitions are presented in equations (3.26) to (3.28). Herein ENE.j=O characterizes the highest
electric field strength at the collecting plate achievable in the electrostatic
case.

43

ION PRODUCTION

.E

--0-- 5

;;(

E
,!:
>-

-...

= 2c = 250 mm

p = 1 bar
T = 150C

'iii
c:
CD
C

s=2c=150mm
= 2c = 200 mm

--0- 5

b = 3.1*104 m2Ns
2

.c
~

rsE =1 mm

c:

CD

...

::J

<;i
W
f:
S

Eo = 43.5*1 05V1m
U o = 28.5 kV

;y;
~

0
20

60

40

80

100

Applied Voltage in kV
Figure 3.18 Plot of current-voltage relationship for the same parameter settings as in Figure
3.17, according to equation (3.26).

+J

JNE

Bob
16s 3

(X

9(V - V 0

ENE.j=O =

[(X

nVo
d
2cln-

2
(X

+ 192(sE NE j =o) 3 (V

+ SE NE j =O)2

- V o)]

12(sE NE ,j=o)2

(3.26)
(3.27)
(3.28)

r SE

The current-voltage relationship according to equation (3.26) is plotted in


Figure 3.18 for the same parameter settings as Figure 3.17, allowing
comparison of both.
Since, in the case of higher currents, a simple calculation becomes
impossible for plate-type designs, an analogous graphical procedure can be
carried out, as for the tube configuration. The corresponding expression is
given in equation (3.29) according to Robinson [10] while 'P p and <l>p are
defined by equations (3.30) and (3.31), respectively. Equation (3.29) is
plotted in Figure 3.19; therefore, for practical purposes only equations (3.30)
and (3.31) have to be used.
(3.29a)
(3.29b)

44

BASIC AND THEORETICAL OPERATION OF ESPs

200

30
25

150

20
~

15

V
/'

10

cl>p 100 I-O~


08
0

./

1.8

/'

24

50

o ~
o

'/

'I'p

/
12

15

Figure 3.19 Plot of equations (3.29a), (3.29b) and (3.29c).

b=A+i
'I'

=
p

<D

=
p

(3.29c)

U - U 0 . In ~
U0
nrSE

(_S_)2.
EO'rsE

16s 'jNE
n 3 0b(c5)

(3.30)
(3.31 )

When looking at common plate-type designs the ratio sl2e, i.e. the half
duct width to the wire-to-wire spacing, is normally around 1. This does not
happen accidently, but the current density reaches a maximum value for that
configuration. This can be demonstrated by taking equation (3.24) and
calculating current densities as a function of wire-to-wire spacing for given
duct widths and given wire radius. This relationship is shown in Figure 3.20,
for s = 200 mm and r SE = 1.0 mm, an applied voltage of 40 kV and an ion
mobility of 3.1 x 10 - 4 m 2 IV s. If the wire-to-wire distance is very small, the
corona will be completely suppressed.
3.2.5

Electrical field distribution

Ions are generated as soon as the electrical field strength overcomes the
critical value Eo. However, the distribution of the electrical field in the tube,

45

ION PRODUCTION

0.16

('II

s = 200 mm
rSE = 1.0 mm

0.14

E
.5
w

T = 423 K
0.12

Eo = 43.5'105V/m
U = 40 kV
b = 3.1'104 m 2 Ns

0.1

...?~

iii
r::

GI
'C

-......
r::

GI

0.08
0.06

'"<3

0.04

""'"E'"

0.02

E.

:I

<;>

0
0

200

400

600

800

1000

wire-to-wire distance 2c in mm
Figure 3.20 Current density as a function of wire-to-wire spacing for given duct widths and

given wire radius.

or in the duct, governs the distribution of the current on the collecting


plates, the charging of the particles and their migration. Electrical field
strength is therefore a local function and the field distribution depends on
the geometry and the operating state of a precipitator. Quite often the
electrical state of a precipitator is simply described by an electrical field
strength according to equation (3.32). However, the value Epg> which is U Is
for plate-type and Ulr NE for the tube-type designs, represents only a rough
approximation of the real state, because it characterizes the electrical field
only for parallel plate electrodes. For that reason Eps is called the pseudohomogeneous electrical field strength.

Eps

applied voltage
distance wire-collecting plate

(3.32)

The field distribution is obviously affected by the distribution of electrical


charge. Therefore, in order to get the electrical field distribution, the
distribution of the gas ions has to be known and vice versa. This is why a
correct solution of the problem needs, for example, an iterative procedure
to solve Poisson's equation and the equation of continuity of electrical
charge as outlined in Figure 3.21 [5]. However, such detailed computations
are restricted to simple electrode geometries. For practical applications,

46

BASIC AND THEORETICAL OPERATION OF ESPs

computation of
electrostatic field

t;.

= _

C{'.'

Ei

(x. Y)

CQ. i1
E

'P i (x,y)

computation of
space charge distribution

c~

-~

+ EigradcQ.i = 0

CQ. i (x. y)

Figure 3.21 Outline of procedure to solve Poisson's equation and the equation of continuity
of electrical charge.

simpler approaches, such as neglecting space charge or assummg a


homogenously distributed charge, are often preferred.

3.2.5.1 Tube-type. In a coaxial geometry the radial position is described


by r. Taking the electrical field equation, the distribution for the electrostatic
case can easily be calculated (equation 3.33a). Considering a homogenously
distributed ionic space charge, Robinson derived the distribution as given
by equation (3.34a) [10]. In Figure 3.22 both relationships are plotted in
dimensionless form. As can be seen, the space charge, resulting from a
voltage higher than corona onset, leads to an increasing electrical field
strength near the collecting plate.
U
E(r) = - - -

r.ln

E(r) =

(3.33a)

r NE

r SE

(3.34a)

The dimensionless forms, equations (3.33b), (3.34b), result from introducing


dimensionless quantities according to equations (3.35) to (3.39). The dimensionless current density (equation 3.36) and the dimensionless voltage
(equation 3.37) have proved to be helpful when comparing current-voltage

47

ION PRODUCTION

~C>

s:.
~

!
Cii

-0-

0.8

U = U b = 21 kV; U' = 5.01 ;

u::

u
';:
u

CD

oj

1\

0.4

.~

11/
11/

C
o
'iii

I:
CD

E
is

..

..

rNE = 150 mm
rSE = 1 mm

b = 3.1*1 0-4m2Ns

Eo = 43.5*105V/m

I---

I---

t"----.

~.

..

= 0.411 -

0.2

; i'NE

p = 1 bar
T = 150C

0.6

jjj
CD

=0

-U=45kV; U' = 10.35 ; iNE = 0.68 mAlm 2

"C

Gi

iNE

..

,
,.
.,

0.2

0.4

-.
.

0.6

0.8

Dimensionless Radial Position r'


Figure 3,22 Plot of radial position and electric field strength in dimensionless form.

relationships for precipitators with different absolute size [4].


E'(r') = ,
r

(3.33b)

1
'Inr~E
.,

E'(r') =

JNE

+f

, E(r)
E(r)
E =-=-Eps
U /rNE

(3.34b)
(3.35)
(3.36)

(3.37)
r'

(3.38)
(3.39)

48

BASIC AND THEORETICAL OPERATION OF ESPs

3.2.5.2 Plate-type. For wire-plate geometries an analytical solution can


be derived for the electrostatic case [14,15]. The vector components
Ex.j=o, Ey,j=o are given in equations (3.40) and (3.41), respectively, In order
to allow simple programming on a computer, some abbreviations have been
used according to the equations (3.42) to (3.47). The electrostatic solution
is a superposition of the electrical field of m wires between two parallel
plates. The coordinates are used as introduced in Figure 3.10. The coordinates and all length scales occurring are related to the half duct width s
(equations 3.48 to 3.51) in order to derive relative quantities.

Ex j=o(x', y')
,

U
= -'
S

sinhf3

n'cos?( .

N2

(3.40)

(3.41)
?(

ny'/2

(3.42)

f3

n(x' - 2me')/2

(3.43)

cosh 'I

+ cos

2 -

~ In { cosh y -

}' =

nme'

N -

J}

cos J

(3.44)
(3.45)
(3.46)
(3.47)

x' = xis

(3.48)

y' = y/s

(3.49)

2e'

2els

(3.50)
(3.51 )

In Figure 3.23a, a three-dimensional plot of the spatial distribution of the


Ey-component in the electrostatic case is plotted in dimensionless form
(equation 3.52) as a function of the relative coordinates for a precipitator
duct with five discharge wires (the Ey-component is responsible for the
transport towards the collecting plates). As can be seen, Ey is very high only

49

ION PRODUCTION

(a)

(b)

Figure 3.23 Three-dimensional plot of the spatial distribution of the Ey-component (dimensionless form) in the electrostatic case (a), in the case of a homogeneously distributed ionic space
charge (b), as a function of the relative coordinates for a precipitator duct with five discharge
wires.

close to the wires and reaches a constant level for y' > 0.5. Between the wires
it decreases down to zero.
E~(' ')
,X,Y

Ei(X', y') = E;(x', y')


Eps

Vis

(3.52)

In considering the ionic space charge, Oglesby and Nichols [8] proposed to
take the constant space charge deduced by P. Cooperman [12]. Thereby,
the current density is assumed to be only a function of the y-coordinate. The
resulting field is a superposition of the electrostatic field plus the field of a
homogenously distributed ionic space charge. The dimensionless electrical
field strength is given in equations (3.53) and (3.54), while Figure 3.23b
demonstrates the increase of the electrical field near the collecting plate

50

BASIC AND THEORETICAL OPERATION OF ESPs

10rr---.-----.----.-----.----.
U = 40 kV; iNE = 1.48 mA/m2
E
- 0 - U = 36 kV; iNE = 1.03 mA/m2
:; 8
--<>- U = 30 kV; iNE = 0.47 mA/m2
UZ:,
~ U = 20 kV; jNE = 0.0 mA/m2
I

B, 6

s = 100 mm -+-----1

Ui
~ 4 1--1\---l---+---+----::b_==:...j

-=

OL-_~

__

__

__

_ L_ _~

0.2
0.4
0.6
0.8
Dimensionless coordinate y'

1.0

Figure 3.24 Numerical results for a plate-type ESP with a duct spacing of 2s = 200. The
increase in the electrical field strength with increasing y' depends on the current density.

caused by considering the space charge for the same conditions as in Figure
3.23a, but a voltage higher than corona onset.
(3.53)

E~(x', y') = E~,j=o(x', y')

E'Ax', y') =

E~,j=o(x', y') + j~E' 2c' In

(!) .~
r SE

(3.54)
(3.55)

As already mentioned, a correct solution to the problem needs an


immense computational effort. An iterative solution for electrical field and
charge distributions has been calculated by various investigators, for
example by [5,8,16]. Figures 3.24 and 3.25 show such numerical results for
a plate-type ESP with a duct spacing of 2s = 200 mm according to [5].
Obviously the increase of the electrical field strength with increasing y'
depends on the current density (Figure 3.24). At the collector surface, this
increase in field strength can be 200 or 300% compared with the electrostatic case. Of further concern is the local concentration of ionic space charge
ciQ ' which is continuously decreasing with increasing y' (Figure 3.25).
In an electrical field, the space charge concentration is distributed in the
interelectrode space. An example of a calculated distribution of ionic space
charge according to Miller [17J is depicted in Figure 3.26. This shows that

51

ION PRODUCTION

5.0 x 1014 r - - - . . , - - - - - , - - - - - , - - - - - - r - - - - - ,

(]

cen
Q"'E

4.0x 10 14

en~

~~ 3.0x10141-~-+-~-+O~

c'

0 0
.~ ';;

!::rD
c .

2.0 x 10 14 I---+-----'''!::I.+----''''ood-----+-----l

Q)~

u~

1 .0 x 10 14i--i--'-i---r;;;;::::=+==::J
O~--~-----L----~----L---~

0.4
0.6
0.8
Dimensionless coordinate y'

1.0

0.2

Figure 3.25 Numerical results for a plate-type ESP with a duct spacing of 2s = 200. The local
concentration of ionic space charge c'Q is continuously decreasing with increasing y'.

Concentration of gas ions Co


(1.6 x 10- 19 As/m 3 )

5.0x1014
4.0x1014
3.0x1014
2.0x1014
1.0x1014

~~

",\0~y?>

~0'ii, ~

1.0 0

o~

0~",~~'().'0
Q'~ o~o~

;..

c,o

Figure 3.26 An example of a calculated distribution of ionic space charge according to Miller
[17].

52

BASIC AND THEORETICAL OPERATION OF ESPs

the highest ion concentrations occur close to the discharge wire and
decrease towards the collecting plates. The highest value occurs at the
collecting plate just opposite the wire.
Both distributions of electrical field and space charge will influence the
particle charging process and the particle transport towards the collecting
plates.

3.3
3.3.1

Particle charging

Charging process

Usually the charging process is divided into a field charging region, for
particles > 111m, and a diffusion charsing region for particles < 0.1 11m
[7,8,10]. While field charging requires the presence of an electrical field,
which drives the free movable charge carriers, the diffusion process is based
on randomly moving gas ions caused by temperature and described by the
kinetic gas theory, i.e. Brownian motion.
Obviously, in an ESP, particles of all sizes experience both situations
simultanously, as particles < 0.1 11m are also driven by the electrical field. A
simple addition of charges resulting from both mechanisms is sometimes
made [18]; however, Oglesby and Nichols argue that both electric currents
towards the particles have to be superimposed [8]. Initial approaches
considering this were made by Murphy et al. [19] but they didn't succeed
with a complete solution. Based on their work, Liu and Yeh [20] published
a simplified theory which led to reasonable agreement with measured
results. Smith and McDonald [21] extended this model to all particle sizes,
thus including field and diffusion charging as limiting cases. The model of
Liu and Kapadia [22], which was published later, gives better approximations in some cases. All these models however, cannot be solved analytically, i.e. they need numerical effort.
For practical work it seems to be reasonable, therefore, to look for
charging theories describing charging processes continuously from small to
larger particle sizes. A reasonable alternative to models based on numerical
solution is Cochet's analytic equation [23]. This allows an easy calculation
and the correlation to actual ESP conditions is quite reasonable in the
critical size range, as Figure 3.27 demonstrates.

3.3.2

Cochet's charging model

The particle saturation charge (i.e. after infinite time) according to Cochet
is given by equation (3.56). For calculations the electrical field strength E

53

PARTICLE CHARGING

Eps =3.6105 V/m


CO.t = 1013 (1.6 .10.19 As)

.-

I~

'0

.5:

8e.

CI)

...IIICl

.c::
U

I:

;;

...::I
III

I II

en

101

Hf+---1

-<>-

(j

:e
III

CI.

- - 0 - LIU I YEH (1968)

_ _ SMITH I McDONALD (1976)

CI)

DIFFUSION CHARGING

--<>- FIELD CHARGING


f---+---1

- 6 - COCHET (1961)

0,2

0,4

EXPERIMENT (HEWITT 1957)

0,6

0,8

1,0

1,2

1,4

Particle Diameter dp in 11m


Figure 3.27 Plot of particle diameter versus particle saturation charge based on Cachet's
analytic equation [23].

and the electrical permittivity of the particle material Gr have to be specified.


(3.56)
In Figure 3.28 calculations were made for typical electrical field strengths
and a temperature of 150 o e, which changes the mean free path of the
molecules from 0.065 Jim to 0.1 Jim. The electrical permittivity of the
particles was assumed to be Gr = 10. As can be seen from the figure, a 0.1 f1.m
particle carries 5 elementary charge units in an electrical field of
3.0 x 10 5 V/m, while a 9.0 f1.m particle reaches 10000 electrons under the
same conditions. For larger particle sizes the saturation charge is proportional to particle surface and increases linearly with the electrical field
strength.
Generally, the electrical field strength is a local function in a precipitator
(see Figures 3.22/3.24) and therefore this influences the particle saturation
charge. However, these local functions have not, to date, been considered in
common ESP models. Such a solution only becomes possible when particle
trajectories are calculated (as demonstrated in 3.5.5).

54

BASIC AND THEORETICAL OPERATION OF ESPs

108
Ii)

en

'I0

....x
~

E,. 1.0 x 105 VI m ..,......,.........,..,.,...!--..."....,.;-,-.,."..."""'.....,,~-J,...m!

107

--0-

106

~ E = 3.0 x 105 V1m ~-t-+"1..,.,.j-+-;+~h~+-t-'rtTt.l


~ E=5.0x10 5 V/m

105

- ~

--~~!--~~~~-'-ri~j~il':~f

; : ~ n:.

~
Q.

104

.,',

CD

~ 1000
0

CD
(3

'E

100

b---+-.......,.............,.j....~~...,...,,<f.Ak,,04--.........+-rt'....t--j
i

nl

a.

10

; ::::.;

..

, i

p =1 bar
T = 150C
lambda = 0.101 11m
epsr = 10

: r: :,'!:!,Hi
,

r:ii';

:.; "'!

, , ' ; !::=:;i:!

,:!
". :

....
"'1

-:-

~.,-

L-~~~~~~~~L-~~~~~~~~L-~~~W

0.01

0.1

10
Particle diameter dp (11m)

100

1000

Figure 3.28 Plot of particle diameter versus particle charge: calculations for typical electrical
field strengths and a temperature of 150 cC, which changes the mean free path of the molecules
from 0.065/lm to 0.1 /lm.

3.3.3

Time dependence and saturation charge

The time dependence of the charging process is described by equation (3.57),


using the time constant 'Q for the charging process (equation 3.58) [10]. The
dynamic charging behaviour is independent of particle size if a homogenous
electrical field is considered.
(3.57)

(3.58)
Before the time dependence can be studied, the time constant 'Q has to be
specified from the electrical conditions. Referring to Ohm's law, the concentration of gas ions cQ can easily be expressed in terms of current density,
mobility and electrical field (approximation in equation 3.58). For typical
electrical states, 'Q has values ::::;; 10 ms. The dynamic behaviour of the
charging process was calculated and is presented in Figure 3.29 for time
constants 1, 10 and 100 ms. As can be seen, the particles will reach about

55

PARTICLE MIGRATION

.....

--"'"
---'"
8

:::::..

0.8

"

0.6

CD

DI

"-

~~

______-r~______~-r__~~taUa=lms
- - 0 - lau a =

as

z:.

---!Sf--

(.)

CD

+=
"-

10 ms

lau a = 100 ms

0.4

as

0.

CD

>

+=
as

0.2

Gi

a:

10

100

1000

time tin ms
Figure 3.29 Plot of relative particle charge versus time under typical electrical conditions.

90% of their saturation charge within some 10 ms under typical electrical


conditions (i.e. 'Q ::::; 10 ms).

3.4

3.4.1

Particle migration

Equation of motion

In order to characterize the state of particle motion the balance over all
forces acting upon the particle is needed. For a particle in an ESP, these
forces are the momentum force FT (equation 3.59), the electrical force Fel
(equation 3.60) and the drag force F w (equation 3.61).
(3.59)
(3.60)
(3.61)
The sum over all acting forces has to be zero (equation 3.62).
(3.62)

56

BASIC AND THEORETICAL OPERATION OF ESPs

100

::I

p =1 bar
T= 150C

lambda

=0.101

IJ.m

!...
0

10

c
s:.

aI
C

.;.

'cC

..... ~ ~

::I

1
0.01

0.1

10

100

Particle Size dp in 11m


Figure 3.30 Cunningham correction factor applied to Stokes' law when particle size reaches the
region where the fluid loses its continuum characteristic.

Before solving this differential equation, however, the drag force must be
specified. In the case of low Reynolds numbers i.e. Re 1 (equation 3.63),
the drag coefficient Cw is given by equation (3.64), which is fulfilled in typical
ESP conditions for particle sizes less than 20 Jim. The drag relationship in
this regime for spheres is given by Stokes' law (equation 3.65).
(3.63)
(3.64)
(3.65)
If the particle size reaches the region where the fluid loses its continuum
characteristic (mean free path of the molecules ),), then Stokes' law needs
correction by the Cunningham factor Cu, given by equation (3.66) and
plotted in Figure 3.30 for 150C.

Cu = 1 + 1.246 2),
dp

+ 0.42

2),

d )

dp exp -0.87' 2~

(3.66)

PARTICLE MIGRATION

57

Assuming the fluid to have no component towards the collecting plate and
that the particles reach their saturation charge, equation (3.67) results as a
differential equation, characterizing the motion of a charged sphere in an
electrical field E.

dw
dt

3.4.2

3nl1dp
m(Cu)

W=

Q: E
m

(3.67)

Theoretical migration velocity

Taking wet = 0) = 0 as the initial condition, the solution of equation (3.67)


can easily be found. This describes the time dependence of the particle
migration velocity (equation 3.68). Herein, the relaxation time Tp characterizes the dynamic behaviour of the particle (equation 3.69) and W th is the
theoretical migration velocity, which is the steady state velocity of the
particle (equation 3.70) [24].
(3.68)
= mpC Cu) =

3nl1dp

W th

QooE
-p3
d CCu)
nl1

d; CCu)

pp .
1811

(3.69)
(3.70)

The theoretical migration velocity is plotted in Figure 3.31 as a function of


particle size for three different electric field strengths, a temperature of
150C and an electrical permittivity of Cr = 10. The theoretical migration
velocity shows a minimum located at 0.35,um. For larger particles the
increase of W th proceeds linearly with particle size (equation 3.70 and
equation 3.56). For smaller particles the increase in wth is more pronounced.
From Figure 3.31 the strong influence of the electrical field becomes obvious
(wth increasing with E2).
The transient behaviour of particle motion is characterized by the
relaxation time T p ' which is plotted in Figure 3.32 as a function of particle
size. It is worth noting that the relaxation time does not depend on the
electrical conditions. The particle velocity at different time intervals, in
relation to the steady state value, can easily be calculated with equation
(3.68). In order to illustrate the acceleration phase, this velocity ratio is
plotted in Figure 3.33 for different time steps. Thus, after 1 ms, particles
smaller than about 6,um have already reached steady state velocity, while a
10,um particle has reached only 80% and a 20.um particle only about 30%.
When the migration velocities of all particles in an ESP are known, then
the separation process can be calculated.

58

BASIC AND THEORETICAL OPERATION OF ESPs

10
~ E = 1.0'105V/m

- 0 - - E = 3.0'10 5V/m

~I,.-+--+I

>-

'ou

-0-

=5.0'1 05V/m

'i

>
c::
o
:;

0.1

10
CI

:i
iii

0.D1

p = 1 bar
T = 150C

:;

lambda = 0.101 I!m


epsr= 10

CII

oCII

s:.

I-

0.001
0.01

0.1

10

100

Particle Size dp in Jlm


Figure 3.31 Plot of theoretical migration velocity as a function of particle size for three different
electric field strengths, a temperature of 150 "C and an electrical permittivity 0, = 10.

1 04
(I)

.5

1000

Go

:I
III

CII

100

p = 1 bar
T = 150C
lambda = 0.101 I!m

i=

c::

0
:;

10

III

><

III

'i
~

,,
": ; ;

,
,

:.:: !

CII

:e

0.1

::i
:::J

III

Q,

0.01
10

100

1000

Particle Size dp in Jlm


Figure 3.32 Plot of particle relaxation time as a function of particle size.

MEASURING AND MODELLING PARTICLE SEPARATION

--

--

--0--

.c

--~-

::::..

0.8

~---Y----~----~----~----~--~~

.c

~-

>-

'u0

59

1.0 ms
10 ms
100 ms
1000 ms

p = 1 bar
T = 150C

0.6

lambda = 0.101 I'm

Gi

>

0.4

r::
0

:;::

...1'0t7I

:i

0.2

Gi

a::

0
10

100

1000

Particle Size d p in 11m


Figure 3.33 Plot of particle velocity, at different time steps, in relation to the steady state value
as an illustration of the acceleration phase.

3.5
3,5,1

Measuring and modelling particle separation


Grade efficiency and total efficiency

If a particle collection process is to be characterized, a method for measuring


particle 'size' is necessary, Although particles of all sizes might occur in a
disperse system, a measuring method always produces 'size classes'. The
borders of the size classes (let's say i-I) are termed by Xi' the width of the
size class (interval) by ~Xi' while the median of a size class is termed by dp , i'
An illustration of the nomenclature is given in Figure 3,34.
Let's assume we analyse the particles collected in a hopper with a sieving
tower consisting of six sieves plus the bottom. Thus, we have six given
borders X z to X 7 and an estimate of the smallest particle size Xl' After sieving
and weighing the sieves and the bottom, the relative mass parts are
determined according to equation (3,71); these, of course, refer to the median
of the size interval. In particle size analysis, a cumulative mass distribution
is often used, where Xi is related to the upper end of a size class (see
definitions 3.72, 3.73 and Figure 3.34).

~Q(d p.I.)

M(dp.J
M tot

(3.71)

60

BASIC AND THEORETICAL OPERATION OF ESPs

Particle size (j.lm)

100

200

300
~e 400
500
1.0 ,.--r-r-r-r-r-r-r-r-rr..,.,.rr,,--.-r-r""-<!h-""'T""1-'-'
t.O(dpe)

~S

=-x 0.8

AQ(dpS)
~4

~0.6"""""""""""""""-""""""""

.0
.;::

7ii

g;

0.4 ........... -- ........ _.

.~

~3

--4

....

AQ(d p3)

0.2 ........

::.r;(~~~;

. . . ~~~A .

O~~,,--'-~~~~~-T~~~-r~~~

X11 X2 I X3 I
X4
I
I
I
d P 1 dp2 dp3

I
I

Xs

dp4

I
I

Xe

I
I

X7

d p6

dps

Size step s, set by method

Median size

Figure 3.34 Nomenclature of particle 'size' measurement.

(3.72)
j- 1

Q(x)

(3.73)

for 2 c" j c" N

.ilQ(dp.J;

i= 1

An example with numbers is given in Table 3.4, together with the corresponding density distribution as given according to equation (3.74).
d .)

q(

p,'

t1Q(d p,;}

(3.74)

LlX i

We now have a formal method to characterize the particle sizing of dust, at


the precipitator's inlet, the fine fraction at the outlet or the coarse fraction
in the hopper.
Table 3.4 Example of particle size distribution and derived quantities

1
2
3
4
5
6
7

x,/J1.m

dp,/J1.m

M(dp,)/g

L1Q(dp,)

Q(x,)

L1(x')/J1.m

40
80
120
180
250
320
400

60
100
150
215
285
360

25.927
32.998
40.069
58.925
44.783
32.998

0.11
0.14
0.17
0.25
0.19
0.14

0
0.11
0.25
0,42
0.67
0.86
1.00

40
40
60
70
70
80

235.7

q(d p' )/1/J1.m

0.00275
0.00350
0.00283
0.00357
0.00271
0.00175

MEASURING AND MODELLING PARTICLE SEPARATION

61

Grade efficiency or fractional efficiency means we evaluate the efficiency


for each particle size class. For size class dp i for example, when the collected
coarse amount is related to the inlet amount, the result is a fractional
efficiency (equation 3.75); when the emitted, fine fraction is related to the
inlet amount the result is called fractional penetration (equation 3.76).
Obviously equation (3.77) has to hold.
(3.75)
(3.76)
(3.77)
If an integral balance is made for total particle collection, the result is a
total efficiency, or a total penetration P, holding for the quantity balanced,
e.g. mass generally (number or surface can also be balanced). The total mass
efficiency can also be expressed in mass concentrations as demonstrated in
equation (3.78). Again equation (3.79) has to hold.

E=

M e tot
M e tot

E+P=l

Me. tot -

M f tot

M e tot

1 _

C f tot

(3.78)

Ce tot

(3.79)

Generally, a fractional efficiency curve gives much more information


about a separator than the overall efficiency E. In order to derive a
fractional efficiency curve, particle size distribution measurements are
needed. Combining equations (3.75) and (3.71) leads to a simple relationship
for calculating fractional efficiency (equation 3.80) from measured collected
particles and inlet distribution.
(3.80)
From equation (3.80) the calculation procedure for the total (mass) efficiency can be deduced (equation 3.81).
(3.81 )
When total mass efficiency is the only interest, then, by using an isokinetic
sampling device before and after the ESP, an alternative calculation can be
done. If the mean velocity in the ESP and the cross-sections at both

62

BASIC AND THEORETICAL OPERA nON OF ESPs

measuring locations is identical equation (3.82) holds, which considers the


different fluxes (measuring time and cross-section) for sampling.
E= 1_

fl.M r . Me'A e
Mr' Ar fl.M e

(3.82)

Measuring total mass efficiencies may often be the simplest way to characterize a separation process. However, when considering modelling the
separation processes, fractional efficiency is certainly the base and this is
why selective particle size measurements are essential.
Needless to say, all standard models refer to the ideal precipitation
process, i.e. where all negative influences are known and ignored, e.g.
back-corona of the dust layer, re-entrainment of precipitated dust and field
bypassing, called sneak age. Correctly speaking, this means that these
idealized models need correspondingly idealized experiments for confirmation or failure - a problem when taking plant results, an advantage with
lab-scale experiments.

3.5.2

Laminar model

The word 'laminar' in this case means that there is no driving force for the
particles in the opposite direction to the electrical field. In other words, there
is no remixing force arising, e.g. from turbulence, so that each particle
follows a trajectory only determined by gas flow and electrical conditions.
It can be shown by different methods [24,25] that, in the case of a fluid
velocity profile, fully charged particles and a homogeneous electrical field,
grade efficiency is determined by straight particle trajectories. This situation
is illustrated in Figure 3.35, which makes clear that grade efficiency is given
by equation (3.83). This is the ratio of the mean gas residence time in the
precipitation zone to the time a particle of size dp needs to travel the distance
s.
(3.83)

3.5.3

Deutsch model

In contrast to the laminar model, the Deutsch model assumes an infinite


remixing force, e.g. caused by 'turbulence', which redistributes the remaining
particles at each downstream location homogeneously over the precipitator
duct. Apart from the boundary layer, the particles in the downstream
direction have the mean velocity of the fluid, are fully charged and migrate
in a homogeneous electrical field. The situation is illustrated in Figure 3.36
and was firstly published in 1922 by W. Deutsch [26].

3.5.3.1 Grade efficiency. Taking a balance over the control volume


sketched in Figure 3.36, the differential equation (3.84) results (written for

MEASURING AND MODELLING PARTICLE SEPARATION

!
I
I

63

Wtht

limiting
particle
trajectory

L-e------

plane of discharge wires

Figure 3.35 Illustration of the laminar model. Grade efficiency determined by straight particle
trajectories in the case of a fluid velocity profile, fully charged particles and a homogeneous
electrical field.

I ;-., """-'' '

J.:[1':::--:":: _
dL

plane of discharge wires

Figure 3.36 Illustration of the Deutsch model.

plate-type), which can be integrated quite easily in order to get the


concentration of particles with size dp ' as a function of downstream location
L (equation 3.85).
de(d p)
eeL, dp )

w1h(d p ) dL
VoS

(3.84)
(3.85)

Remembering equation 3.75 for the definition of grade efficiency, the final
formula is equation (3.86) for plate-type ESP:
T(d )
p

1 - ex p {- Wth(dp,E)'LNE}
vo's

(3.86)

and equation (3.87) for tube-type ESP:


T(d )
p

1- ex p {- 2'Wth(dp,E)'LNE}
vO'r NE

(3.87)

When the total collecting area and the total flow rate is used, a general

64

BASIC AND THEORETICAL OPERATION OF ESPs

T(De).xls

o
0.9

0.99

c:;'
CD'

0.999

::::J

()

'<

0.9999

10

0.99999
12

Deutsch Number De
Figure 3.37 Grade efficiency plotted as a function of the Deutsch number.

formulation is possible (equation 3.88).


T(d p ) = 1 - exp { - wth(d p , E) .

ANE}
V

(3.88)

Some years ago, Self defined the exponent in equations (3.86) to (3.88) as
the 'Deutsch number' [25]. In Figure 3.37, the grade efficiency is plotted as
a function of the Deutsch number. Obviously the exponent can also be
interpreted as dimensionless residence time (equation 3.90) or as the product
of a particle drift parameter n, which includes operational conditions (va' E),
and a dimensionless length describing geometry (L NE , s) [4,24].
T(d p )

1 - exp{ - De(d p ; E, vo)}

(3.89)

(3.90)

In Figure 3.38, showing the grade efficiency for a plate-type precipitator


zone of 5 m length, the strong influence of electric field strength on
performance is demonstrated. The typical minimum efficiency at around
0.3-0.4 {lm is shifted from 5% up to 70%, by increasing the field from 1.0
to 5.0 X 10 5 Vim.

65

MEASURING AND MODELLING PARTICLE SEPARATION

.!Ii

\lUi'

0,8

.......

~------~-+~--~~~+-----~r--+----~r---~~
f-

Q.

I-

>cCI)

0,6

;;:

0,4

'u

p = 1 bar
T = 150C
lambda = 0.101 !lm
epsr = 10
Vo = 2.0 mls
s = 0.200 m
LNE = 5.0 m

CI)

'C

...
CJ
1"11

0,2

i <'"

0
0,01

0,1

100

10

Particle Size dp in Ilm


Figure 3.38 The grade efficiency for a plate-type precipitator zone of 5 m length demonstrates
the strong influence of electric field strength on performance.

Figure 3.39 compares measured and calculated grade efficiencies and


illustrates the influence of mean gas velocity. The measurements were
performed in a lab-scale ESP with ambient air under ideal precipitation
conditions (no sneakage, no rapping, no back-corona) by using an optical
particle counter, which is able to analyse the particle size distributions 'in
situ' very quickly, thus guaranteeing constant conditions [24]. From Figure
3.39 it can be seen that around 10 11m the compliance between measurements and calculations is quite good. For smaller particle sizes 2 !lm),
however, all measured efficiency values are far higher than the calculated
ones. A quantitative explanation for this behaviour is still missing. Maybe
agglomeration happens or the fine particles can be charged more efficiently
than assumed.
Furthermore, Figure 3.39 demonstrates that grade efficiency increases in
wider ducts as long as they are operated with the same pseudo-homogeneous electrical field strength. This increase can be explained with an increasing space-averaged electrical field strength (see Table 3.5). A detailed
Table 3.5 Space-averaged electrical field strength

E (kV/cm)

s (mm)

100

71

50

E p , kVjcm

5.49
2.9
5.49
5.0

5.14
3.4
4.68
4.8

4.83
4.5
3.86
4.4

mA/m2
jNE mA/m2
E p , kVjcm
jNE

66

BASIC AND THEORETICAL OPERATION OF ESPs

el
vo =0.5rn1s
--Deutsch

"tI

0,8

>c

0,6

Q.

vo =1.0rnls

=3'105 VIm

Vo

LN/S

= 2.0 rnIs

= 5.0

i="
(.)

Gl

.(j

:E
W

.......! ..

0,4

Gl
"tI

....... .. ... .. _..

<0

(;

0,2
....

..

1
Q.

"tI

i="
>(.)
c
Gl

0,6

.(j

:E
w

0,4

Gl
"tI

f!

Cl

.... --......... -... T


... i

0,8~--~-~~~~~~~~---~-b~~~+++--~

:~

... -............ +
0,4~~~--~~~~H-

__~--~~~~~--~

~-

0,1

10
Particle Size dp in Jlm

Figure 3.39 Comparison of measured and calculated grade efficiencies and illustration of
influence of mean gas velocity.

67

MEASURING AND MODELLING PARTICLE SEPARATION

discussion concerning the comparison of electrical operating conditions in


precipitators of different size is given in [4,24].
3.5.3.2 Effective migration velocity. The ESP performance is often characterized by the 'effective migration velocities' Werr or W k , which are defined
by equations of the Deutschian type (equations 3.91a/b and 3.92a/b). While
the Deutsch model predicts the grade efficiency T(d p ) by means of an
exponential function (see equation 3.88), wherein the drift velocity wth(d p )
of particles in the electrical field occur, equations (3.91) and 3.92) define a
velocity value by taking the exponential function to describe the total mass
efficiency (E3). Therefore, effective migration velocities represent overall
averages, characterizing performance as total efficiencies do.
Werr == - -

(3.91a)

In(1 - E 3 )

ANE

E3 = 1 - ex p (-

Werr~)

(3.91b)

ANE

(3.92a)
(3.92b)

>u
c

0,9999

0,0001

0,999

0,001

CD

u
;:
W

1/1

III
I'll

!.

3:

DI

III
1/1

0,99

0,01

-"tI
CD

:::J

...

CD

:::E
iii

I-

-I

DI

0,9

0,1

20

40

60

80

100

120

Specific Collecting Area ANEN in

140

:::J

1 60

m2 /m 3 /s

Figure 3.40 Dependence of total mass efficiency on specific collecting area for dilTerent
values.

Wen

68

>-

()

BASIC AND THEORETICAL OPERATION OF ESPs

0,9999

0,0001

0,999

0,001

-I

GI

.(3

==
w
(II

s:::

III

(II
(II

0,99

0,01

III
I'll

..

."
CD

........

::l
CD

==

III

ia

0
I-

2.

!.

0,9

0,1

20

40

60

80

1 00

120

140

::l

1 60

Specific Collecting Area ANEN in m2 /m 3 /s


Figure 3.41 Dependence of total mass efficiency on specific collecting area for different modified
effective migration velocity (w k ) values.

Figure 3.40 shows how total mass efficiency depends on specific collecting
area for different Weff values according to equation (3.91). The plotted
straight lines overpredict performance, when for an existing ESP the
emission needs to be further improved. The reason for this overprediction
is that mass transfer along the collecting plates is not constant. It decreases
with locations further downstream, because smaller, and therefore lighter
particles, have lower transport velocities, i.e. they need more time to be
collected.
In order to overcome this overprediction caused by neglecting particle
size distribution effects, Matts and Ohnfeldt [27] introduced a modified
effective migration velocity Wk' Their equation (3.93) is plotted in Figure 3.41
for different W k values. Obviously, for the same specific collecting area and
the same total efficiency, the W k value is much higher than the corresponding
Weff value. Furthermore it is demonstrated for W k = 30 cmls that the
empirical parameter k has a strong influence on the slope of the curve. A
typical value for k lies within 0.4-0.6 depending on the actual problem.
These Werr and W k values do not represent particle velocities (but they
characterize mass transfer; see next section). This is why it is not surprising
that effective migration velocities (typically a few cm/s) are often one order
of magnitude smaller than theoretical migration velocities, i.e. different
characteristics should not be compared. There is, of course, only one
unrealistic case, where werr is identical with a particle velocity, and that is
in the case of a monodispersion, i.e. all particles have exactly the same size.
Finally, Figure 3.42 illustrates the relationship between the total mass
efficiency and the effective migration velocity, and in particular, their

69

MEASURING AND MODELLING PARTICLE SEPARATION

-----.-

-----

E3
1.00

s= 50mm
s= 71 mm
S

= 100 mm

- -6- -

- -0- -

-{]-

1.2
iNE

..,

We"

= 2,9 mAlm2

::
CD

_0

0.95

()

.;:-

>u
c

CD

0.90

_0

CII

'u
:;::

0.6

0-

1/1
1/1

...-

s::

..

ce'
I

0.85

__ - 6
/

IV

::E
iii

0.8

[J

0.80

&/

/
/
/

.;. - - -

<
(1)
0.4

cr

0.75

0.2

0"

()

;:;:
'<

:;'
3

1/1

0.70 0

0.5

1.0

1.5

2.0

2.5 0

Gas Velocity Vo in m/s


Figure 3.42 Relationship between the total mass efficiency and the effective migration velocity
with particular reference to gas velocity and duct spacing.

dependence on gas velocity and duct spacing. The total mass efficiency (left
y-axis) stems from gravitional efficiency measurements in three geometric
similar lab-scale precipitators of different absolute size (s = 50/71/100mm
and LNE/S = 5.0). Figure 3.42 holds for a constant current density
jNE = 2.9 mA/m2 at the collecting plates, i.e. the DC voltages have been
adjusted according to the different duct widths.
It is worth noting that the effective migration velocities (calculated
according equation 3.91a) in Figure 3.42 are much larger (0.25 mls to
1.0 m/s) than those usually observed in large-scale precipitators (0.04 m/s to
0.20 m/s). There are two main reasons for this behaviour: firstly, the longer
a precipitator becomes, the lower the effective migration will be in total,
because the local mass flux continously decreases with the length coordinate
(see next section); secondly, all secondary influences, usually occurring in
practice, e.g. re-entrainment, sneakage and back-corona, all causing deteriorating efficiency and thereby reducing the effective migration velocity, were
excluded from the investigations by the design and the performance of the
experiments.
What are the trends of the measured results plotted in Figure 3.42 where
an increase in the mean gas velocity leads to a progressive decrease in the
total efficiency? From equation (3.91a) it follows that the observed decrease

70

BASIC AND THEORETICAL OPERA TION OF ESPs

in total efficiency correlates with an increase in the effective migration


velocity. This contradicts the interpretation of the effective migration velocity as a quantitative measure for the particle transport velocity, since the
transport conditions perpendicular to the main flow direction, determined
by the electrical state, have not changed. To be more precise, the local
transport velocities of individual particles have not changed, but the mass
flux towards the plates will increase because of a higher volume flow. The
increase of the effective migration velocity with increasing gas velocity
points to a correlation of ""eff and mass flux.
When comparing the precipitators with respect to constant specific
collecting areas (where v = const. since LNE/s = 5.0), the wider duct shows
higher total efficiencies. Here, a decrease in total efficiency correlates with a
decrease in the effective migration velocity. This means that an increase in
the effective migration velocity corresponds to both a decrease and an
increase in the total efficiency. This observation expresses the fact that the
mass flux towards the collecting plates can be increased in two ways: first,
by increasing the individual drift velocity of the particles in the electrical
field where the number of individual particles transported per unit time
remains constant. However, because of an increased drift velocity the
particles are deposited earlier on the plates or the same number of particles
are collected in a shorter precipitator. On the other hand, an increase in the
volume flow leads to an increase in the number of particles transported per
unit time towards the plates, but without a change in the individual drift
velocity. Obviously only the first case can represent a real precipitator
improvement.

3.5.3.3 Mass flux. The total collected mass flow (particles per unit time)
or the total collected mass flux (particles per time and area unit) can simply
be derived from fundamental considerations enabling a clearer representation of particle transport in precipitators [28]. If i m e represents the total
particle mass flow at the precipitator entrance and im.r the fine particle mass
flow at the outlet, then the total mass flow i m c of collected particles is
described by equation (3.93). Additionally, the definition of the total mass
efficiency was considered (equation 3.78).
(3.93)
The incoming particle mass flow im e can be calculated by using the particle
mass concentration at the inlet em c and the volume flow Ii according to
equation (3.94). When relating the total mass flow of collected particles imp
to the collecting area ANE , a particle mass flux results, which is called 'total
mass flux of collected particles', characterizing the total precipitation capability of the apparatus (equation 3.95). Total' means here a two-fold
integration indicated by the double bracket: the first bracket indicates the

MEASURING AND MODELLING PARTICLE SEPARATION

71

integration over particle size distribution, the second bracket indicates the
integration over the precipitator length.
(3.94)
(3.95)

vs

=c m.e E
3 L
NE
With the normal measurements of total efficiencies, this total flux of
collected mass <Gm.c can be calculated as simply as the widely used
effective migration velocity weff . However, the total flux of collected mass
has two advantages. On one hand this quantity is based on a proper
physical background. Further, it is linguistically clearly distinguished from
the steady state theoretical drift velocity of the particles W th in the electrical
field, thus avoiding misleading interpretations of Werr as a particle velocity.
On the other hand <Gm.c is an explicit function of particle mass
concentration in the raw gas: an influence which is well known in practice
but not explicitly considered in Weff.
When evaluating total efficiencies, the mass concentrations in the raw and
clean gas have to be calculated; therefore, the total flux of collected mass
can readily be calculated with known quantities according to equation
(3.95). This procedure was carried out with the efficiencies in Figure 3.43
where the resulting total flux of collected mass are plotted as open dots
(y-axis on the right-hand side). The mass flux values range from 10 to
40 mg/m 2 s and, as is to be expected, they increase with increasing gas
velocity.
A limiting straight line exists for the total flux of collected mass, since
<Gm.G has to be zero at gas velocity zero. If a precipitator is operating
with total efficiencies close to 1.0, the slope of the limiting straight line is
obviously solely a function of the ratio, precipitator length to half duct width
4E/S (see equation 3.95). Because of geometrical similarity, the precipitators
investigated belong to a configuration characterized by LNE/S = 5.0, leading
to the following slope of the limiting straight line (see Figure 3.45):
50 mg

~-20mg
3
2.5m
s

Obviously, precipitators with different LNE/s-values have different limiting


straight lines.
Furthermore, Figures 3.42 and 3.43 show differences with respect to
precipitator size. These wide-spacing effects can be explained by higher

72

BASIC AND THEORETICAL OPERATION OF ESPs

--...-

----

1.00

s= 50mm

- -6- -

s= 71 mm

-()-

s = 100 mm

-{]-

i m G> >
60
::!!
c:

..,

w
>u
c:

CD

0.95

50

0.90

40

0.85

30

'u

!E
w
en
en
III
::E
iii

...

)C

9o
2-

c.
3:

DI

: - - ---, ,

0.80

.....

0.75
0.70 0

0.5

...
...

.-..,

, ,

....

20

:.--

'

10

1.0

1.5

2.0

en
en
S

3
~
3

..

en

2.5 0

Gas Velocity Vo in m/s


Figure 3.43 Plot of total flux of collected mass with corresponding total mass efficiencies as a
function of increasing gas velocity_

space-averaged electrical field values as before (see Table 3.5, jNE =


2.9 mA/m2).
It should be pointed out that in the deduction of equation (3.95) no
model for particle transport was required, only a calculation. A model
becomes necessary when the total mass efficiency E3 or the grade efficiency
T(d p ) is to be predicted.
The relationship between effective migration velocity and the total flux of
collected mass results when combining equation (3.95) with equation (3.91).
Thus, equation (3.96) allows an easy transformation from effective migration
velocities into total fluxes of collected mass and vice versa. The only things
required to be known are the total efficiency and the total particle mass
concentration in the raw gas, both quantities required to be measured when
W eff is to be found.
(3.96)
Generally, a flux can be represented as a product of a transport velocity
and a concentration. Similarly, the total flux of collected mass can be split

MEASURING AND MODELLING PARTICLE SEPARATION

73

up into two factors, a 'velocity' Werr and a ratio representing a 'concentration'. Obviously, the ratio represents a fictitious concentration value, since
the mass concentration in the raw gas is connected in a physically uninterpre table way with the total efficiency - therefore Werr is a fictitious velocity.
However, there might be a practical argument for a representation with
effective migration velocities: when comparing Figures 3.42 and 3.43, a
different sensitivity of Weff and jm,c to changes in duct width and
changes in gas velocity becomes obvious, For the werr-representation
changes of gas velocity and changes of duct width are of the same order, but
in a <Gm,c-representation, changes in duct width are far smaller compared with changes in gas velocity,
Besides the total collected mass flux, the mass flux concept can be applied
locally, i,e. locally collected mass flow of particles of size dp , towards the
collecting plate at location L, is the product of the mass concentration of
these particles cm(L, dp ) at location L and the theoretical migration velocity
wth(d p) (steady state transport velocity) of particles of size d p (equation 3,97).
The mass concentration of particles of a given size cm(d p ) in relation to the
total mass concentration is equivalent to ~Q3(dp) and can therefore be
represented by means of the corresponding particle size distribution function (equation 3.98).
jm,c(L, dp )

cm(d p) = cm . ~Q3(dp)
T(d )
p

=
-

(3,97)

cm(L, dp ) ' wth(dp)


=

cm 'q3(d p)' d d p

1 _ cm(L, d p )
cm.e(d p)

(3.98)
(3.99)

If the validity of a certain particle transport model is assumed, e.g, that


of Deutsch, then both the particle size dependent mass concentration and
the change of particle size distribution (which is caused by the classifying
effect when passing through the electrical field) can be calculated. Figure
3.44 shows, for example, how the average theoretical migration velocity
changes when passing through a precipitation zone assuming a log-normal
number distribution for the dust (for details refer to [29]). Analogously, a
local mass flux of collected particles Gm,c(L) can be calculated by using
the local number concentration cN(L) and the number density distribution
qo(d p , L) at that location (equation 3.100),

fp,m" cN(L) . mp(d p)' wth(d p) . qo(d


dp,mm

p,

L) . d d p

(3.100)

74

BASIC AND THEORETICAL OPERATION OF ESPs

Precipitator: s = 100 mm
Vo = 1.0mfs
Eps = 5.0 kV fcm

Dust: log.norm. distribution


d p,50.0 = 5.0 IJ.m
Er = 10

1.0

(Wth) )(!"Wth2)

---

0.8

--c--

1.5

--0--

0.6

--4--

0.1

......... +....................... -1 0.8


(i)

~......,..........;............................."............................;.............................; ........................... -1 0.61

0.6

~
~

crln

r---~~-------r----~+---~--~~~~

0.4

~,,,,,,,.,,

.. ,,.,,,, ..,,+ ... ,,.,,,,,,::::',,"-!-=:.=,,,,.=,,.,,,,,,.~.,,,,,,,,,,.,,,, ..

J
,,,,,,.-1 0.4
~

....

0.2 ... ............

~.:.'.~..~...~.~.=?.J.:~.;~~.o1}..~-.""'-2'

....--5-'J
.. t--.':".,,--t-"=w;:::::::j
-=--:..-

j
:
- - - - -t.: - - - - -t.

+- - - -t. - ~

0.2

- - -t. - - - - -t. - -

OL.......L-'--'-L-..I.-'--'-~-'--'-~-'-...l-L.......L-'--'-L.......L-'--'-'-'O

Dimensionless length LNEfs

Figure 3.44 Plot to show how average theoretical migration velocity changes when passing
through a precipitation zone, assuming a log-normal number distribution for the dust.

Integrating this locally collected flux over the complete precipitator length
leads to the total mass flux
already introduced (equation
3.101).

LNE

jm,c

(3.101)
This total mass flux, calculated by means of a transport model, can be
compared with the total mass flux derived from measurements according to
equation 3.95. Depending on the level of correspondence, the transport
model could be validated.
The locally collected mass flux can also be simulated by means of particle
tracking in a computed flow field (the method is described in 3.5.4). For a
precipitator with half duct width s = 200 mm, a length of 4 m and a mean
gas velocity (air, temperature 150C) of 1 mls the flow field was calculated
by means of a commercial CFD (computational fluid dynamic) package
[30]. The turbulent flow was modelled by using the k - I-: model with an
inlet turbulence intensity of 10%. For the simulation of locally collected
mass fluxes, the particle size distribution represented in Figure 3.45 was

MEASURING AND MODELLING PARTICLE SEPARATION

75

If
...
cCD

:;:,

1/1

..
.!

0.6

r-------+----<rO--r--+-----~__t--__j

...
c:c
cr

:;:,

::::J

--

,g

C>

iii'

'3
E
:;:,

....

0.1

Particle Size In 11m


Figure 3.45 The particle size distribution represented here was assumed, as at the precipitator
inlet, for the simulation of locally collected mass fluxes.

assumed. In order to consider the stochastic walk of particles in turbulent


flows for particle tracking, a random walk model was used [30]. The
particles were homogeneously distributed across the inlet and the number
of calculated particle tracks corresponds to the relative frequency of size dp
in Figure 3.45. In total, 10 5 particle tracks were calculated and the locations
of collection were registered, or alternatively the particles escaped.
Figure 3.46 shows, for each particle size, how the precipitated mass is
distributed along the collecting electrode. The locally precipitated mass of a
size class is related to the total mass of the size class at the inlet. As expected
the large particles with high drift velocities wth are precipitated early in the
ESP. Each size class shows a curve with a maximum, which forms in the
entrance region (L' < 5), then the curve decreases with increasing precipitator length. For particles < 2.5 ,um almost no profile for the collected mass is
readily observed (presumably, the little increase at L' > 15 is an artefact
caused by the flow field conditions and the particle tracking algorithm).
When the results for the size classes are summated at each location, the
total, locally collected mass or the collected number of particles can be
plotted. In Figure 3.47 these masses or numbers are related to their total at
the ESP inlet. This simulation of collected particles shows a strong maximum when represented as mass within L' < 5. This situation is analogous
to the real behaviour where most of the dust can be found in the first
hopper. The sharp maximum is a result of the mass representation compared with the number representation. The number curve of collected

76

BASIC AND THEORETICAL OPERATION OF ESPs


Q.

..J

0.25

Q.

'I

0.2

ca(/l
ca
"0-

0.45
0.55
---- 0.65
----+- 0.75
- 0 - 0.85
~ 0.95

E(/I

0.15

Cl)U

-CI)

UN
CI).-

=(/1

OCl)

O1

UU

.-

=t:

caca
uQ.
0 ...

GiQ.
>
.!!!
.';=

...
CI)

pm

----+- 0.35 pm

(/I
(/I

-CI)

I~0.;5

f
~

-fr-

pm

--8-

pm
pm
pm
pm
pm

~2.5 pm ~J
3.5 pm
------ 4.5 pm
---Q--- 5.5 pm
l
---fB- 6.5 pm
j
~
----.- 7.5 pm 1
----{}-- 9.0 pm ~
15 pm

-0-

'-_______.p.a.rt.i.c.le_si.z.e.s... JJ ~

11
l

0.05 L

(f)

r~~-~~~~~~~

o ~
o 2.5

7.5
10
12.5
15
5
dimensionless precipitator length L'

17.5

20

= LN Ei S

Figure 3.46 Diagram showing how the precipitated mass is distributed along the collecting
electrode for each particle size.

particles shows a less pronounced maximum which is to be found slightly


further downstream in the ESP.
Although particle tracking models (PTM) are in an early stage of
development, they could have a practical application as shown by the results
presented in Figures 3.46 and 3.47, which are quite promising for future
analytical work. If, for example, for a given ESP configuration and a given
dust, the curves in Figure 3.47 can be simulated, then an approximation
function can be found, which should allow excellent scale-up in the case of
precipitation improvement by adding further collection zones. Variations in
operating conditions, such as changing gas velocity or electrical field
strength, can also be easily investigated.
3.5.4

Flow field and particle trajectories

Observing particle transport in an ESP reveals (Figure 3.48) that, on the one
hand, turbulent flow structures exist, while on the other hand, the particles
are not homogeneously distributed over the duct. As Figure 3.48 and video
tapes could show [31J, dust free zones are clearly observed in the duct under
idealized transport conditions. Obviously, this observation is more congru-

77

MEASURING AND MODELLING PARTICLE SEPARATION

0.15
/II

"0'"

(1)(1)

-0--

_.a

uE

.!!:::l

'0

c:

eo
eou"o

particle numbers: N(L')/Ntotal

- - particle mass: M(L')/M tot

0.1

~o

oc:
-co
ai/ll
.~ ~

--

..!!!E
(1)_

.n

0.05

c.:CfJ-'"'"
UJ

"'co

E
X
:::>

2.5

7.5

10

12.5

15

17.5

20

dimensionless precipitator length L' = LN EI S


Figure 3.47 The total, locally collected mass or the collected number of particles plotted in
relation to their total at the ESP inlet.

ent with the assumptions of the laminar model than with those of the
Deutsch model.
For that reason, this section considers an alternative way of modelling
particle transport in ESPs. Particle tracks can be calculated by solving the
equation of motion provided the flow field and the electrical field are
known. A simplified version of the equation of motion is given by equation
(3.102), where the drag force is assumed to be Stokesian, which holds for
particles < 20 tlm under typical ESP conditions [24].
(3.102)
In turbulent flows, each local velocity component is described by its
time-averaged value and a time-dependent fluctuation according to equation (3.103).
(3.103)

The simplest approach for the fluid flow assumes a time constant turbulent
velocity profile and a corresponding profile for the turbulence intensity,

78

BASIC AND THEORETICAL OPERATION OF ESPs

Figure 3.48 Particle transport in an ESP showing that, on the one hand, turbulent flow
structures exist, while, on the other hand, the particles are not homogeneously distributed over
the duct [4].

both only a function of y' as depicted in Figure 3.49 (in the near wall region
a laminar sublayer is assumed). The velocity profile holds for a volume flux
of 1.0 m/s and the profile of turbulence intensity is rather arbitrary; however,
values of 10% in the middle of the channel, increasing in near wall regions,
are typical for turbulent flows (for details see [24]).
When looking at the time averaged process, the mean particle tracks can
easily be calculated by common numerical methods (e.g. Runge-KuttaFehlberg). Examples of typical particle tracks are given in Figure 3.50 for
particles of sizes 1.0 and 10,um. A lab-scale ESP (s = 100 mm, v = 1.0 m/s,
U = 30 kV) has five discharge electrodes and an inhomogeneous distribution of the electrical field strength, together with the space charge according
to 3.2.5, has been considered in the simulations if the particles enter the ESP
with zero charge.
Figure 3.50 demonstrates how particles of different size are deflected in
different ways. Particles entering the ESP on the symmetry line of the
discharge electrodes get a kick by the first electrode towards the collecting
wall. For electrode numbers > 2, however, this kick and the electrodes'
influence, because of inhomogeneity, has already disappeared. Furthermore,
the particle charging process, in general, cannot be neglected, especially for

79

MEASURING AND MODELLING PARTICLE SEPARATION

..

1 ,2

'(3
0

Gi
>
iii

'x<II

1 ,0

....................................

0,8

.. 1

'E

0,4

'iii

c::
CI)

0,2

:c

-...

...........:.....:

--+- vx Iv 0

0,6

I/)
I/)
CI)

0,3

,'":

>-

,....

.......... ; ........
........ .............

.......................

c::

0,2

0-

c::
CD

.. 0-:

0
CD

..1.. 0 .. Tu

1,0

:i'
CD

1/1

::.

'<

0,25

1,00

0,75

0,50

dimensionless coordinate y'

=yls

Figure 3.49 A time constant turbulent velocity profile and a corresponding profile for the
turbulence intensity, both a function of y', comprise the simplest approach for the fluid flow.

fine particles. The 1.0/tm particles entering the ESP away from the plane of
discharge wires will not reach 90% of their saturation charge within the
considered precipitation zone.
When the stochastic character of turbulent flow is to be considered, the
time-dependent part of the velocity is expressed in terms of an rms value of
the x-component (isotropic turbulence) according to equations (3.104) and
(3.105). The rms value is expressed by means of the assumed velocity and

s= 100 mm,
iNE

Ua = 30 kV,

= 0.47 mA/m2,

Vo

t " : '

o Op,rel

1.0 m/s

= 50%

Op,rel

,~

i~~~ b;=1=.0~Il=m= =_=,= -= :=.=_


~

100

= 90%

=====~~

. =========:==========='=,='
=:

I~;:~~:~~:--,-.-:
~

0.25

-~,.

.... ... .. .

_.

..

. ..... -

....... .

O~~U--L-L-J__L-L-~-L-L~~__L-~~-L~~__~

0.5

1.0
1.5
2.0
2.5
3.0
3.5
4.0
Dimensionless length coordinate x' = LIs

4.5

Figure 3.50 Examples of typical particle tracks for particles of sizes 1.0 and 10 11m.

5,0

80

BASIC AND THEORETICAL OPERATION OF ESPs

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

Dimensionless length coordinate x' = LIs

4.5

5.0

Figure 3.51 Top: plot to show how the structure of calculated particle tracks depend on eddy
size (turbulent structure). Bottom: Five stochastic particle tracks with square starting location
and an eddy size of 5 mm.

turbulence profiles (Figure 3.49) according to equation (3.106). The average


y-component is zero.
V~(y'; t) = Xvx.rms(y')

(3.104)

v~(y'; t) = Xvx.rms(y')

(3.105)

vx.rmb')

vx(y') . Tu(y')

(3.106)

The actual, time-dependent velocity value is simulated by means of a


random number X, which has to be revised from time to time. This time step
can be interpreted as the life time of turbulent eddies or, when expressed in
a distance, as a scale characteristic for the turbulent structure (eddy size Lw)'
Figure 3.51 (top) shows how the structure of calculated particle tracks
depends on this eddy size. Compared with the observations depicted in
Figure 3.48, an eddy size of 5 mm leads to particle paths structured realistically. For this eddy size, Figure 3.51 (bottom) shows five particle tracks
started at the same locations and same conditions. As in real turbulent
flows, the simulated randomness leads to a spread in the particle paths.
With modern computational fluid dynamics (CFD) tools, however, the
calculation of the flow field and even the tracking of particles have become
more comfortable. Although an inhomogeneous electric field cannot be
considered in standard versions, an homogeneous field can be taken into
account quite easily. An example is given in Figure 3.52.

MEASURING AND MODELLING PARTICLE SEPARATION

=200 mm; v =1.0m/s; E =3.0kV/cm

81

Collecting wall

Dimensionless length coordinate x' = LIs

Figure 3.52 Example of how the homogeneous field can be taken into account in the
calculation of the flow field with modern CFD.

The method of particle tracking can be used to simulate grade efficiencies


of separators. For the lab-scale ESP discussed (s = 100 mm, L = 5 s,
v = 1.0 mis, U = 50 k V), grade efficiencies of different models together with
experimental values are plotted in Figure 3.53.
Obviously, for particle sizes around 10l1m the Deutsch model and the
measured grade efficiencies are in good compliance. For decreasing particle
size, however, measured particle precipitation is much better than the
Deutschian values. While for sizes < 111m, the performance is even better
than the values predicted by the laminar model. The results of the particle
tracking models introduced above are also plotted in Figure 3.53. It is
interesting to note that both results are rather different. Thus, the PTM
model based on a simple flow profile, but considering the inhomogeneous
structure of the electrical field with a homogeneously distributed space
charge, leads to an enormous upshift in efficiency for particle sizes < 111m although the values are less than the measurements. This might indicate
that fine particles are much better transported than assumed up to now,
originally caused by the commonly neglected inhomogeneous distribution
of the electrical field and space charge effects. This explanation is further
supported by the results of another PTM model which uses a simulated
turbulent flow field and a simple homogeneous electrical field strength.
Herein, without electrical inhomogeneities, the particles in the fine particle

82

.......g,

BASIC AND THEORETICAL OPERATION OF ESPs

0.8

'tJ

~
>0-

cCI)

0.6

'u

:;:
CI)

- -. - - lab-scale experiments
particle tracking model
IPTM) with simulated
flow field and homogeneous electrical field

0.4

- -0 - -

CI)

...

'tJ

III
C)

0.2

0000 _- - -D'
0
0.01

- -0 - -

_0

0.1

PTM with flow profile


and inhomogeneous
electrical field

10

particle size in Jlm


Figure 3.53 Plot of grade efficiencies of different models together with experimental values.

region are collected even lower than the predictions according to Deutsch.
For particle sizes> Illm the simulated efficiencies lie in between the laminar
and the Deutschian curve.
Looking at these results, there seems to be promise in combining the
inhomogeneous electrical conditions with the simulated fluid flow; even so,
such a simulation technique could still show a deviation from the measurements. It is presumed that the agglomeration effects, which occur under ESP
operation [32], will have to be considered further to obtain compliance
3.5.5

Diffusivity models

Sometimes particle transport in ESPs is described as a combination of


convective and diffusive processes [15,25]. Assuming that the average flow
field and the electrical field are of vector type as given in equation (3.107),
then the so-called convective diffusion equation looks as equation (3.108).
=.
v=

(vo)
o '
.

(0)
Eps

(3.107)

(3.108)

83

MEASURING AND MODELLING PARTICLE SEPARATION

Dedimensionalizing equation (3.109) with half plate-to-plate distance s, gas


velocity Vo and particle concentration at the entrance c(x' = 0) reveals that
the convective diffusion equation (equation 3.109), for clarity the bars have
been omitted) is governed by two dimensionless numbers: the ratio of the
transverse particle velocities n (equation 3.110), which is called the transport parameter (compare [4]), and the ratio of two counter-current transport effects, namely electrical transport against eddy diffusivity, the Peclet
number, Pe (equation 3.111).

ac' _ ~ . (a 2c' a2c') =


Pe ax,2 + ay'2

~ . ac'

n ax' + ay'

(3.109)
(3.110)

Pe

s
Dp

W th

(3.111)

The method of determining the analytical solution was comprehensively


described by Leonard et al. and a detailed solution is published [33]. The
particle concentration is given by a sum of m eigen functions (equation
3.112) and, unfortunately, the complex relations for the eigen values 8 m and
the constant em are difficult to handle. When solving equation (3.112) on a
computer, the effect of different eddy diffusivity values is easily shown as

1,0
0

C3

0,8

:;:::;

III
~

cQ)

0,6

<

laminar
Dp

<

00

Leonard et al.

Dp

00

Deutsch

c
0

u
Q)

0,4

>
:;:::;
III

a;

I:

0,2

Dp= 0 cm 2/s

Dimensionless Coordinate y' =y/s

1~

Figure 3.54 When using a computer to solve equation 3.112 the effect of different eddy
diffusivity values is easily shown.

84

BASIC AND THEORETICAL OPERATION OF ESPs

Figure 3.54 demonstrates. For low diffusivity values Dp --+ 0 the model tends
to the laminar case, while for high diffusivity values Dp --+ CfJ the Deutsch
model emerges.
c(x', y') = c(x' = 0)

f Cm exp( -0 Fm) exp (pe.2 Y')


x'

m=l

(3.112)
with:
exp ( Cz =

(3.113)

2pe)

.
Pe 2
pe2.{-pe.cosceml+(em-4e.)SIDceml+

o m +4
1 + -Pe
2 40 m2

= -

Pe
(pe)}
exp - 2

1 ( 1 - -pe 2 )sm(20
.
Pe
Pe 2 + _.
+ -80
) - ------..-cos(20
)
2
40 m
40 m2
40'"'m
m
m
m
(3.114)

(3.115)

An analytical integration of the concentration profile at precipitator's exit


will readily lead to the grade efficiency (equation 3.116).
T(d p ) = 1 -

L
00

m=l

(4)

C m exp -0

_E.

Fm

sin0
(pe)
. ~exp
0

(3.116)

For practical aspects the grade efficiency curves are important to know, but
the particle size is hidden behind the transport parameter, 0, and the
Pe-number. This is why the values for both parameters cannot be varied
independently. In Figure 3.55 grade efficiencies for a precipitator of
2s = 200 mm, LNE = 500 mm, Vo = 1.0 m/s and U = 50 kV is illustrated,
calculated according to equation 3.116.
The efficiency minimum in all curves is caused by the Cochet equation
(see section 3.3) considering diffusional effects in the field charging process
of particles. It may be surprising that the difference in efficiency for particles
below 1 pm is rather small, while the largest difference is to be observed

85

DESPOSITION

1,0

..-Q.

-'C

Models:
0,8

Dp =0 laminar

0< Dp <00 Leonard et al.

>u
c

Dp = 00 Deutsch

CII

'u

!E
w

0,6

CII

'C

...tV
C!J

0,4

0,2

Particle Size d p in J'm


Figure 3.55 Grade efficiencies for a precipitator of 2s = 200 mm, 4.E = 500 mm,
and U = 50 kV, calculated according to equation 3.116.

Vo

1.0 mls

around 2 j!m. Of course the absolute trends depend on the operating


conditions used for calculation.
Since the differences range between the laminar and the Deutschian limit,
neither of which match reality (see Figure 3.53), and because of the
complexity of the equations, the finite diffusivity models are not recommended for practical work. (When the local character of the electrical field is
considered, which requires a numerical iterative solution process, there still
remains some contradictions when simulating grade efficiencies [34].)
3.6 Deposition

When the drifting, electrically charged particles touch down on the collecting electrode, the electric charge has to flow from the particles into the
ground electrode. Since the electrical charges have to pass through the dust
layer, dust resistivity plays an important role in ESP operation (Figure
3.56). The transport of electrical charge occurs by two mechanisms, surface
and volume conductivity, which are discussed in detail elsewhere.
Because the numbers of electrical charge carriers created by the corona
are usually orders of magnitude more than needed for charging all particles
to saturation [24], not only the charges on the particles have to pass
through the dust layer but all charges created by the corona. Therefore,

86

BASIC AND THEORETICAL OPERATION OF ESPs

'*!

charge transport
through
the gas

b~particles
0

by

i~ns

through
the dust
layer

through the electrode

--

Figure 3.56 Illustration of the important role played by dust resistivity in ESP operation since
the electrical charges have to pass through the dust layer.

operating with high currents can have advantages when considering particle
charging. On the other hand, it can be critical with respect to the initiation
of back-corona in the dust layer, which dramatically deteriorates particle
precipitation. For this reason pulsing high voltage systems, resulting in
lower total charge carrier flow, can be very effective when applied to high
resistivity dusts. These lower currents lead, in addition, to less energy
consumption [35-37].

3.7

Removal

The dry removal of the dust layer is made by mechanical impact which is
achieved by hammers in certain time steps. In most common applications
these cleaning cycles are made on-line, i.e. the gas flow and electrics are still
in operation, while the dust layer breaks up into large agglomerates which
settle into the hoppers. Because of the ongoing gas flow, finer and already
collected particles can be readily re-entrained into the gas stream. This
re-entrainment, which depends for example on turbulence levels, can severely reduce the ESP efficiency.
On the other hand, the electrical current, which is still flowing through
the dust layer, holds the dust layer like an electrical glue. This electrical
force, caused by the charge transfer between touching particles, increases the
common adhesive force and is called an electric clamping force. This force
plays an important role in the cleaning process of the collecting electrode,
but has been poorly investigated to date.

REFERENCES

87

References
1. Riehle, C. and Loffier, F. (1993) Particle dynamics in an electrohydrodynamic flow field
investigated with a two-component laser-Doppler velocimeter. Part Part. Syst. Charact. 10,
41-7.
2. Shaugnessy, E.J. and Davidson, 1.H. (1986) Turbulence generation by electric body forces.
Exp. Fluids, 4, 17-26.
3. Shaugnessy, E.J. and Davidson 1.H. (1985) The fluid dynamics of electrostatic precipitators.
Aerosol Sci. Technol., 4, 471-6.
4. Riehle, C. and L6ffier, F. (1993) Reflections on similarity laws concerning particle transport
in electrostatic precipitators. Powder Techno/., 77, 201-8.
5. Miller, 1., Riehle, c., Schwab, A.J. and L6ffier F. (1994) Numerische Feldberechnungen in
Elektrofiltern im Hinblick auf elektrisch iihnliche Betriebszustiinde. J. Electrostatics, 33,
213-28.
6. Miller, 1. (1994) Private communications.
7. White, H.J. (1963) Industrial Electrostatic Precipitation, Addison-Wesley, Pergamon Press.
8. Oglesby, S. and Nichols, G.B. (1978) Electrostatic Precipitation, Pollution Engineering and
Technology, Vol. 8, Marcel Dekker, New York, Basel.
9. Peek, F.W. (1929) Dielectric Phenomena in High Voltage Engineering, 3rd ed., McGraw Hill,
New York.
10. Robinson, M. (1971) Electrostatic Precipitation, Ed. E. StrauB, Air Pollution Control, Part
1, 10hn Wiley, New York.
11. Townsend, 1.S. (1915) Electricity in Gases. Oxford University Press, pp. 375-6.
12. Cooperman, P. (1960) A theory for space charge limited currents with application to EPs.
Trans. AlEE, 79, 47-50.
13. Cooperman, G. (1981) New current-voltage relation for duct precipitators valid for low and
high current densities. IEEE Trans. Ind. Appl., A-17(2), 236-9.
14. Spiegel, M.R. (1974) Complex Variables (Schaum Outline Series), McGraw-Hill, New York,
pp. 252-3.
15. Self, SA, Kihm, K.D. and Mitchner, M. (1987) Comparison of wire-plate and plate-plate
EPs in turbulent flow. J. Electrostatics, 19,21-32.
16. Leutert, G. and Bohlen, B. (1972) Der riiumliche Verlauf von elektrischer Feldstiirke und
Raumladungsdichte im Plattenelektrofilter. Staub-Reinhalt. Luft, 32, 297-301.
17. Miller, 1. (1995) Private communications.
18. Sato, T. (1987) Charging process of fine particles in unipolar ion flow. Trans. I.E.E. Jpn.,
107, 155-61.
19. Murphy, A.T., Adler, F.T., and Penney, G.W. (1959) A theoretical analysis of the effects of
an electrical field on the charging of fine particles. Trans. AlEE, 78, 318-26.
20. Liu, B.Y.H. and Yeh, H.C. (1968) On the theory of charging aerosol particles in an
electrical field. J. Appl. Phys., 39, 1396-402.
21. Smith, W.B. and McDonald, 1. R. (1976) Development of theory for the charging of
particles by unipolar ions. J. Aerosol Sci., 7, 151-66.
22. Liu, B.Y.H. and Kapadia, A. (1978) Combined field and diffusion charging of aerosols in
the continuum regime. J. Aerosol Sci., 9, 227-42.
23. Cochet, R. (1961) Lois Charge des Fines Particules (Submicroniques) Etudes TheoriquesControles Recents Spectre de Particules, Coll. Int. la Physique des Forces Electrostatiques
et Leurs Application, Centre National de la Recherche Scientifique, Paris, 102, 331-8.
24. Riehle, C. (1992) Bewegung und Abscheidung von Partikeln im Elektrofilter. Dr.-Ing.
Thesis, University of Karlsruhe.
25. Self, SA, Kihm, K.D. and Mitchner, M. (1987) Comparison of wire-plate and plate-plate
EPs in laminar flow. J. Electrostatics, 19, 21-32.
26. Deutsch, W. (1922) Bewegung und Ladung der Elektrizitiitstriiger im Zylinderkondensator.
Ann. Phys., 68, 335-44.
27. Matts, S. and Ohnfeldt, P.O. (1963/64) Efficient gas cleaning with SF electrostatic
precipitators. Fliikten, 1-12,93-110.
28. Riehle, C. (1995) Mass flux and effective migration velocity in electrostatic precipitators.
Powder Techno!. (December issue).

88

BASIC AND THEORETICAL OPERATION OF ESPs

29. Riehle, C. (1993) Zur Messung und Beschreibung des korngrei3enabhiingigen Stofftransports im Elektrofilter. Staub-Reinhalt. Luft, 53, 383-8.
30. FLUENT Inc. (1995), User guide.
31. Riehle, C. and Leffler, F. (1991) Revision of the Deutsch model based on video tapes of
particle transport in electrical wire-plate precipitators, EPRI 9th Particulate Control Symp.,
Oct. 15-18, Williamsburg/USA EPRI TR 100471,2, Palo Alba, CA, USA.
32. Riehle, C. and Wadenpohl, C. (1995) Electrically stimulated agglomeration at an earthed
surface, Powder Technol. (December issue).
33. Leonard, G., Mitchner, M. and Self, S.A. (1980) Particle transport in electrostatic
precipitators. Atmosph. Environ., 14, 1289-99.
34. Riehle, C. and Leffler, F. (1995) Grade efficiency and eddy diffusivity models. J. Electrostatics, 34, 401-13.
35. Peterson, H.H. (1981) New trends in electrostatic precipitation: wide duct spacing,
precharging, pulse energization. IEEE Trans. Ind. Appl., IA-17, 496-501.
36. Darby, K. (1984) Pulse energization - an alternative to conditioning for highly resistivity
dusts. Proc. 2nd Int. Con! ESP, Kyoto, Japan, pp. 495-501, APCA, Pittsburgh, USA.
37. Mayer-Schwinning, G. and Neulinger, F. (1985) Sind Pulsgeneratoren tatsiichlich die
Lesung aller Entstaubungsprobleme?, Zem. Kalk-Gips, 38, 405-11.

Mechanical design considerations for dry


precipita tors
F. KNUTTSEN AND K. R. PARKER

4.1

Introduction

The major operational components of any precipitator, as will be appreciated from chapter 3, are the discharge electrodes, which provide the
necessary corona current for charging the particles, and the collectors or
receiving electrodes, on which the charged particles are deposited. Once the
particles have been deposited they will need to be removed, in order that
precipitation will continue satisfactorily, either by mechanical rapping in the
case of a dry precipitator, or by liquor washing if the application is wet.
These components are housed in a completely stable and gas tight casing,
usually thermally insulated, the lower end of which is formed into hoppers
or troughs for receiving the particulates removed from the collectors. In
order that the discharge electrodes can be electrically energized, to the
required voltage to produce the corona, they are isolated from the casing by
some form of electrical lead through insulator arrangement, which is
coupled to the transformer rectifier equipment providing the precipitator's
voltage and current needs.
While the foregoing identifies the necessary components for precipitation,
the design of the individual items have special requirements if the precipitator is to not only perform completely satisfactorily, but give long-term
consistent performance with the minimum maintenance needs. This chapter
will review the mechanical items of equipment in some detail, in order to
meet these objectives.

4.2 Discharge electrodes

The basic requirement of any discharge element used in an electrostatic


precipitator, is, when electrically energized, the field intensity adjacent to the
element surface is such that the molecules in the vicinity have their work
function raised to promote ionization of the gas, together with the release
of some electrons. The production of these ions and electrons is covered in
detail in chapters 3 and 8, so will not be discussed further in this section.
To produce the necessary field intensity at the surface of the electrode, it
is essential that the radius of curvature of the element is small compared

90

MECHANICAL DESIGN CONSIDERATIONS FOR DRY PRECIPITATORS

18

lA

lA'

lE

lE'

lD

lC

0
Discharge
edges

,','-

o
Spiral

Ribbon
peak

Ribbon
square teeth

Nail
star

Square
twisted element

Plain strip

Serrated
strip

Figure 4.1 Discharge electrode profiles.

with that of the positive collector electrode. In most practical applications


this does not present any problem, since the collector, particularly for a plate
precipitator, can be considered as having an infinite radius of curvature.
Over the years, the discharge electrode has taken many different forms,
from a simple round, square or barbed wire, through to so-called controlled
emission electrodes for specific duties, some units being claimed as 'unbreakable'. As a modern power plant precipitator can contain the equivalent of
50 km of wire electrode, the cost, in a competitive bid situation, means that
a simple readily available shaped material can offer an economic advantage,
so a round wire formed into a spiral or twisted square section wires can
often be found in many installations.
Some typical electrode profiles are illustrated in Figure 4.1; these can be
split into two different categories, uncontrolled and controlled corona forms.
In the controlled corona type, these have specific positions or areas where
the emission develops, such as the spikes on electrodes lA, or in the case of
the spiral electrode lB, the position closest to the collector, i.e. the point of
maximum field intensity. For the square or round wire format, the corona
develops somewhat randomly and, if observed, the highest or brightest
corona points move around the electrode to coincide with the maximum
field intensity which changes with dust conditions, either on the electrode
system or in the gas phase itself.

91

DISCHARGE ELECTRODES

10

<..

High emission
serrated electrode
(25 mm wide x 1.6 mm)

...
...
..

:::l
0
Q)

<G

~/

1: 6
~
.c

&l

is

.........:/

2
.,_##

00
Figure 4.2 Laboratory emission characteristics. Square us. serrated electrodes, collectors at
305 mm spacing.

Each precipitator supplier will emphasize the advantages of their specific


electrode form but, in the absence of space charge or corona suppression
effects, all electrodes can provide acceptable performance figures. One of the
disadvantages of the controlled emission electrode is that without corona
suppression or space charge effects, the emission is considerably greater than
for a simple wire electrode, as indicated in Figure 4.2 [1]. These data were
obtained under laboratory conditions in air, for a square wire electrode 1E
and a controlled emission electrode 1e. In actual operation a similar
characteristic is obtained, but the shape of the curves is modified to reflect
the space charge effect resulting from the dust particles.
This increase in power consumption, under free running operation, can
have a considerable effect on the operating costs of a large precipitator
installation. Where, however, 'power saving' measures are practised, using
modern microprocessor control systems (see chapter 8, section 8.9), this
increased power consumption can be minimized without significantly increasing the emission to any great extent.
Where severe space charge or corona suppression is to be expected, then
a high emission discharge electrode is essential to allow the particles to be
effectively charged. Under these conditions the corona emission characteristic of the high emission electrode reduces to equate roughly with that of a
simple wire profile, hence the power consumption is limited automatically
[2]. There is, however, a significant increase in precipitator performance, as
the space charge or corona suppression effects are negated by the higher
corona current flow, resulting in a higher 'drift' velocity.

92

MECHANICAL DESIGN CONSIDERATIONS FOR DRY PRECIPITATORS

Theoretically, only the first field needs to have high emission electrodes,
since once this field performs efficiently, the particles causing the corona
suppression or space charge effects, are largely removed and the downstream
fields operate normally. In many installations, it is normal to find the same
form of electrode fitted throughout the unit unless the plant is very large
and a change in electrode profile can be economically justified.

4.3 Discharge electrode mounting

In general there are two main approaches, dependent on the design


philosophy of the supplier, as to how the electrode systems are mounted
within the precipitator. The first system can be considered as the outboard
of the field proper, i.e. the discharge electrode support system is carried from
an insulated frame located in the interfield area, and the second inboard,
where the suspension is carried from an insulated frame located above the
collectors.
For the outboard suspension, individual electrode elements are normally
mounted in a rigid frame, often referred to as a 'bedstead', where the
electrodes are fixed between horizontal members forming the frame. The
total discharge system itself is typically carried from four support insulators

Figure 4.3 Schematic arrangement of complete precipitator. Wide interfield spacing, bedstead
frame (spiral electrodes), rolled channel collectors.

DISCHARGE ELECTRODE MOUNTING

93

located above the interfield area, and the individual frames are normally
positioned within the height and length of the collector duct, as illustrated
in Figure 4.3.
The inboard type of suspension, being above the field, enables the
electrodes to be hung from a top support frame to pass between the
collectors. The early installations were fitted with weighted wire electrodes,
where the tensioning weight was located beneath the bottom of the
collectors. More modern designs using the inboard approach have the
electrode elements carried in a mast, or use the 'unbreakable' full-height
electrode, both with the extremities being located above and below the
collector, as indicated in Figure 4.4.
The outboard approach, where the insulators are positioned over the
interbank area, means that the casing is longer than the inboard system, but
with this, the casing is taller to cater for the top and bottom electrode
support systems.

Support
insulator

~-=_

-----

f~~~~~~fi~r1r:[;f

Lead through
insulator

Top
frame
for discharge
electrode

Discharge
electrode at
negative polarity
~Gas

flow
Earthed collector
electrode plate

Bottom frame
for discharge
electrode

Figure 4.4 Schematic arrangement of complete precipitator field. Narrow interfield spacing,
mast-type discharge electrodes (twisted wire), catch space collectors (courtesy Lodge Sturtevant
Ltd).

94

MECHANICAL DESIGN CONSIDERATIONS FOR DRY PRECIPITATORS

Obviously one of the main criteria for plant availability is the freedom of
electrode failure, as a broken or swinging electrode can electrically short out
the field and hence impact on performance. To minimize this, it is important
that the choice of electrode design must be such that swinging must be
prevented. Many designs are arranged to have only short unsupported
lengths of electrode elements, so that the amplitude of element vibration,
arising either through electrical energization or mechanical rapping, is
reduced.
To further increase the potential electrode life, large section elements are
used instead of smaller round wires, but as the corona emission is reliant on
a small radius of curvature, the electrode edge must be as sharp as possible.
For example, under a given situation, a sharp 4 mm square wire has the
same emission characteristics as a round wire of 2.5 mm diameter, but has
some four times the cross-section and hence is mechanically superior.

4.4 Collectors
The prerequisite for any collector is that it provides a relatively flat profile
to the discharge electrode which gives a uniform electric field, and is
mechanically stable under all operating conditions of temperature, gas flow
and rapping.
Over the years, the collector like the discharge electrode has taken many
different forms, from an initial simple flat or corrugated plate to the present
large fabricated designs.
During the 1950s and 1960s, there was a vogue for so-called 'catch
pocket' profiles where, in theory, the dust, upon being dislodged by the
rapping, passed through the pocket into a vertical quiescent zone, such that
re-entrainment was minimized [3]. In practice, as the pockets caused local
electric field disturbances at the surface, and the maximum field was at the
pocket, this resulted in the greatest dust deposition adjacent the pocket, so
that when rapped, the dislodged dust tended to fall outside the protected
area. Their use has been discontinued; not only was the theory unsound, but
mainly because of cost, since to provide the quiescent zone, the collector
plate comprised two separate sheets.
Most installations today, which can have collector plates up to 5 m by 15
or 16 m high, employ either full-height cold-rolled strips having inbuilt
stiffening ribs, joined together to form the collector, or alternatively, some
type of fabricated collector, employing pressed or rolled sheets connected
between vertical stiffening members, e.g. channels. These approaches (Figures 4.5 and 4.6) have been found to provide the requisite mechanical
strength and stability to maintain the degree of electrode/collector alignment necessary to meet the low emission levels presently demanded.
Particularly, so as to minimize costs, collector metal thicknesses are typi-

COLLECTORS

~I"--- 500

95

----+11

~--------------750----------------+I

Figure 4.5 Collector design - cold-rolled strip profiles.

cally around 1.5 mm, which without the stiffeners, would be totally unsuitable.
In either approach, the elements are carried from a suitable top support
member sometimes rapped, carrying the main load, plus a lighter frame for
bottom end location purposes, or a heavier bottom frame, which can be
alternatively mechanically rapped. Again, like the discharge electrode system, the format depends on the suppliers' design concepts.
The actual profile of the collector depends on the availability of fabricators to produce a suitable profile in the lengths and tolerances required.
Each supplier will have a preferred profile for his particular cold-rolled
sheet, but essentially all forms have raised upstands at each edge and various
stiffening rib arrangements to balance out rolling stresses, such that twisting
or 'banana-ing' is eliminated. Some have slots and fingers built into the edge
stiffeners, which interlock, to provide additional stability and alignment.
Some rolled channel profiles are illustrated in Figure 4.5.
There are a number of different fabricated forms of collector having flat
sheets connected between vertical stiffening members. The so-called 'catch
space' collector is illustrated in Figure 4.6; this comprises a hollow top beam
carrying full-height rolled channels, 75 mm x 50 mm, having 1.5 mm pressed
or rolled sheets connected to them forming the collector proper.

96

MECHANICAL DESIGN CONSIDERATIONS FOR DRY PRECIPITATORS


Motor

Collector rapping assembly


Roof deck

Upper top girder

Drop rod

Collector plate

Figure 4.6 Collector assembly - catch space design (Lodge Sturtevant Ltd).

CASINGS

97

On some smaller wet precipitators, e.g. with a maximum height of 8 m, a


simple thick 5 or 6 mm hot-rolled plate collector can be found; again this
would be supported from a heavy top member and located at the bottom
by a lighter frame. Other forms of collector can be found in practice,
particularly on mist precipitators, where tubular metal or fabric reinforced
plastic (FRP) thick collectors can be found. These alternative forms of
collector are detailed in chapter 14.

4.5 Casings
The casing, contammg all the preCIpItator internals, can weigh several
hundred tonnes and is an important part of the overall design. Not only
must it be completely gas tight, but must also withstand all the various loads
imposed on it without distortion, in order to maintain electrode collector
alignment. The loads would include wind, snow and possibly seismic
conditions, in addition to the dead load of the internals, plus dust deposits
on the internals and hoppers.
Casings are generally designed to international standards utilizing the
'allowable stress' in the material, for load, gas pressure and temperature, etc.;
alternatively, some casing designs are now based on 'deflection under load'.
As any precipitator structure must be rigid enough to maintain electrode
alignment and to contain the gas under all conditions, the design based on
'deflection', rather than 'allowable stress', can result in an optimum weight/
cost arrangement.
Typically each precipitation field would be separated by a vertical
column member and portal, which carries the collector and discharge
electrode (DE) loads. For the outboard insulator suspension arrangement,
this portal comprises a double roof girder forming a hollow beam or tunnel
in which the discharge electrode suspension insulators can be located. This
hollow beam/tunnel tends to produce wider interbank access areas than the
inboard arrangement, but in this case, the discharge electrode load must be
carried by structural cross-beams connected back to the collector support
bridge girder spanning the unit (see Figures 4.3 and 4.4), showing typical
side elevations.
Most large precipitators incorporate a main lower ring beam system
upon which the precipitator casing is built; this ring beam is, in turn,
supported from a substructure having sliding bearings situated between the
substructure and ring beam stubs. This arrangement (Figure 4.7) allows the
complete casing to expand with temperature changes, which not only
eliminates mechanical stresses, but ensures that alignment is maintained.
Figure 4.8 illustrates how, on a large casing, one point on the substructure
is fixed and all other locations can move.

98

MECHANICAL DESIGN CONSIDERATIONS FOR DRY PRECIPITATORS

Typical ring beam

-N..-+--.,<'- Restraining guides parallel

to direction of expansion

,-'

~-, .---'-'---~
I
I

I
I

I
I

Sliding bearing surface

'
I

- - - - .. ,

------~-,-----------

- ------1\

, ,
,..... ---- .... _-

~~==----------------------==-~

"
"I I

"
"
""
"

""
"" ,
,

,
i

,,
,,

,,
,,,
,,,
,,
,,
,,,
,,

II

"

II

'':

,,
,
,,

1:.::_ .......................................... _-_-__ ,

Figure 4.7 Bottom ring beam and sliding bearing details (courtesy Lodge Sturtevant Ltd).

99

CASINGS

III

~+.---

---~

III
Casing expansion
Figure 4.8 Substructure/casing expansion conditions.

The lower end of the casing terminates in a hopper or trough mounted


directly onto the ring beam; sometimes the side walls are extended vertically
downwards to provide additional storage capacity for the dust. The specified
design condition for the hopper connection is usually the hopper dead load
plus the hopper completely full of dust, calculated with the bulk density of
the appropriate dust in question.
In the past, casings have been constructed from reinforced concrete and
other building materials, but the present trend, because of speed of erection
and cost, is to employ steel fabrications, usually partly prebuilt off-site, for
example field side walls, hoppers, ring beams and possibly mouthpieces,
depending on the size of installation. FRP and similar corrosion-resistant
materials can also be found on wet or mist precipitators.
For large installations, e.g. power plants, to reduce costs and to help
maintain complete rigidity, interfield support props and cross-ties can often
be employed in the design of the precipitator casing.
For smaller precipitators, some designs use a monolithic approach where,
instead of sliding bearings, the casing expansion is reflected directly into the
column legs, which are arranged as an extension to the side wall columns.
Substructures both in steel and reinforced concrete can be found; with
the latter, the precipitator design would always incorporate sliding bearings
between the ring beam column stubs and substructure, whereas, with a
monolithic approach, the substructure must be of steel and would form part
of the casing design. The foundations of the substructure must be carefully
designed to carryall the loads reflected back from the precipitator itself.
To ensure that the casing is gas tight, any access doors and cut-outs
used for rapping drives/insulators, etc., or instrumentation, must be carefully

100

MECHANICAL DESIGN CONSIDERATIONS FOR DRY PRECIPITATORS

considered and designed such that they can be sealed effectively. Any air
ingress into an operating precipitator will not only result in electrical
disturbances but invariably causes local corrosion from the gas being cooled
below dew-point. Any door used for man access, in addition to having
effective sealing, should also be electrically key interlocked for safety
reasons, as the operating voltage on any precipitator is lethal.
On the majority of dry precipitators the casing is thermally insulated to
reduce corrosion and give a long life. The thermal insulation, or, at least its
fixing to the steel work, needs to avoid all thermal bridges; otherwise
corrosion can occur on the inside of the casing, which could ultimately
affect its mechanical integrity and possibly, in the short-term, electrode
alignment.

4.6

HT insulators

Possibly one of the limiting items affecting plant availability and performance could be the HT insulators used for electrically isolating the discharge
electrode system from the casing. Not only must the insulator material be
capable of withstanding the maximum voltages imposed upon it during all
operating conditions but, on some precipitator designs, the insulator must
withstand the full weight of the complete discharge system and possible
deflection loads.
To minimize potential problems of electrical tracking, because of deposited materials, e.g. dust, carbon, moisture or sulphuric acid, it is usual to find
the insulators heated and/or purged. This is essential to comply with some
recent legislative measures demanding that the precipitators are electrically
energized immediately the process plant is switched on from cold so as to
minimize all possible emissions. Under this condition, the internals, particularly the insulators, which are usually mounted partly outside the gas
stream, can, unless preheated, encourage water/acid deposition from the
gases with potential resultant tracking. If tracking occurs, not only will the
precipitator efficiency be reduced, but the insulator will fail electrically, due
to uneven heating or arcing causing thermal stresses. In this situation the
insulator would require replacing completely to return the precipitator to
optimum performance.
Generally the insulator material best suited for this duty is high alumina
porcelain, although special high temperature applications have used 99.9%
alumina, which, from Figure 4.9, offers better high temperature insulation/tracking properties [4].
Although many designs mechanically support the discharge electrode
system directly on a cone-shaped (flower pot) insulator, others use post
insulators outside the gas stream and use the HT insulator as an unloaded

101

HT INSULATORS

Q)

<.>

c:

.l!!
(j)
'iii
~

109

Glazed
porcelain

(ij
<.>

t5
Q)
10 8
iii

10 7

100

200

300

Temperature (0C)

400

500

Figure 4.9 Insulator materials, temperature [s. resistance curve (courtesy Lodge Sturtevant
Ltd).

gas seal. Until the recent legislation measures, the post insulator support
design was used with dirty gas insulators having a tracking length of 1 m or
so. With this approach it was necessary, before energizing the precipitator
electrically, to allow sufficient time for the gas temperature to heat the
system up to give dry conditions which, with the long tracking length, has
been found sufficient on most installations to overcome dust tracking
problems.
In addition to the hollow portal beam spanning the precipitator, in which
the insulators can be mounted, two further insulator arrangements can be
considered: firstly, the use of thermally insulated and heated boxes mounted
on the roof and, secondly, a complete top housing covering the whole roof
area. Figure 4.10 shows typical insulator box and top housing arrangements.
The latter approach is normally found where the discharge electrodes are
carried from post insulators and the rapping is externally mounted, as will
be described later. The advantage of this approach is that the top housing
forms a weatherproof self-heated room over the casing, which enables both
electrical and mechanical components to be visually inspected during plant
operation and any required maintenance can be carried out, after isolating
the electrics, in warm dry conditions.

102

MECHANICAL DESIGN CONSIDERATIONS FOR DRY PRECIPITATORS

Figure 4.10 Insulator box and top housing designs.

4.7

Rapping

So that a precipitator always functions at optimum efficiency, it is necessary


for any deposits on the collector and discharge system to be periodically
removed, so that any build-up has minimal impact on the electrical
operating conditions. This removal is usually achieved by mechanical
impact rapping in the case of dry precipitators and by liquor washing for
wet applications. (The different washing systems employed are fully described in chapter 13.)

RAPPING

103

In modern dry installations, dependent on the supplier, the rapping is


achieved either by a rotating tumble hammer or by dropping a heavy rod
onto an anvil connected to the component being rapped. The tumble
hammer system is normally operated through a motor-driven shaft and the
drop rod by either a motor-driven cam shaft or a magnetic lifting mechanism. Older systems, employing vibrators or mechanical lifting and dropping
of the collectors, have been superseded by the foregoing, but can sometimes
be found.
Although, theoretically, cleaning to the metal surface would be ideal, the
energy required to dislodge a 10 J-lm particle from a metal surface is around
1000g [5J, which, while possible, would result in not only complete
disintegration of the deposited material, causing severe re-entrainment and
a very much reduced efficiency, but such impact intensity would also
invariably result in fatigue and mechanical damage to the components being
rapped.
In order that the particles reach the hopper, the size of the agglomerate,
when dislodged, must be some 1000 J-lm in effective diameter to overcome
the horizontal component of gas flow [6]. This means that a compromise
has to be established on any operating plant, such that the rapping provides
sufficient energy to dislodge the deposited layer from the surface, without
causing the agglomerated layer to shatter and become re-entrained, and also
limits mechanical damage to the components being rapped. Because the rate
of deposition varies along the precipitator, the rate or frequency of rapping
is reduced from the inlet to the outlet fields; this is necessary to enable the
deposited layer to achieve sufficient thickness to produce a 1000 J-lm agglomerate when sheared from the surface by the rapping.
Although the frequency of collector rapping is relatively slow, it nevertheless is mechanically searching, at a rate of 12 raps per hour; the
components undergo 1 million impact cycles in approximately 10 years, and
to eliminate premature failure most suppliers carry out full-scale laboratory
investigations. These are not only to ensure that the system is capable of
long-term reliability, by equivalent 10 year life testing, but also that the
energy distribution is sufficient and reasonably uniform across the collector
or discharge system. An energy distribution across a collector, rapped at the
lower end by a 5 kg tumble hammer, is presented in Figure 4.11; this shows
a typical format which has proven satisfactory in the field. The results also
indicate the potentially damaging energy levels imparted around the anvil
area, which have to be suitably designed to ensure long trouble-free life.
Figures 4.12 and 4.13 indicate typical arrangements for a tumble hammer
bottom and collector rapping system and a cam lifted drop rod arrangement
at the top of the collector. While the tumble hammer system is invariably
mounted within the casing, as shown, the drop rod can be either external,
as shown, or internal. An advantage of the external arrangement is that it
operates in a dust-free environment, but requires a number of roof penetra-

104

MECHANICAL DESIGN CONSIDERATIONS FOR DRY PRECIPITATORS

Figure 4.11 'g' distribution over a tall collector.

tions. On some installations additional rows of tumble hammer rapping can


be found in the collector height, to further improve the distribution profile,
when handling sticky or 'difficult' dusts [7].
While the greatest rate of dust deposition occurs on the collectors, even
small accumulations on the discharge electrodes can have a deleterious effect
on the electrode emission characteristics, since the radius of curvature is
increased (see chapter 3), and it is equally, if not more important, to ensure
that the discharge electrodes are effectively rapped. As the system contains
very much less material than the collectors, transference of energy through

105

RAPPING

Figure 4.12 Tumble hammer collector arrangement (Flakt).

the discharge system is more difficult. Some 50% of the total applied energy
is lost through every 90 change of direction; nevertheless, both internal
tumble hammer and external drop rod rapping systems are employed. The
internal tumble hammer raps either an individual frame or a pair of frames,
whereas the external drop rod form of rapping can only impart energy to a
specific area of the support frame. By limiting the size of individual support
frames, very effective rapping can be achieved, albeit at the cost of additional
insulators and support frames.
It is generally accepted that, if rapping intensities of 100 g (0 to peak) plus
can be achieved on the collectors and a similar value on the discharge
electrodes, then the resultant cleanliness will be satisfactory in terms of
maintaining performance. While 'g', the acceleration due to gravity, is used
as a measure of rapping effectiveness, this is very much an oversimplification, since the value determined will be dependent on the size and
thickness of the element being rapped as shown by the foIIowing formula
[6]:
0

acceleration

4 n 2 . f2 . d

where f is the fundamental frequency of vibration (Hz) and d


displacement (m), and

IS

the

106

MECHANICAL DESIGN CONSIDERATIONS FOR DRY PRECIPITATORS

Vertical
channel
member

Figure 4.13 Drop rod rapping arrangement (courtesy Lodge Sturtevant Ltd).

where K is a constant, D = Et 3 (1 - V 2 ), t is plate thickness, V is Poisson's


ratio, E is modulus of elasticity, w is weight per unit area and A is short edge
dimension.
Since the measurement of g to determine the effectiveness of rapping is
not the complete story, various manufacturers often quote lower values than
the above 100 g, for their particular system as being adequate to enable their
plant to perform satisfactorily. In this instance, one must consider the
reputation and track record of the supplier, prior to accepting the statement
covering lower g values, which might be correct for the vast majority of
particulates met in practice.
Arguments have also arisen as to how the g value should be measured
on the collector or discharge element, e.g. transverse to or in the plane of
the rapping; many specifications overcome this by stating that the acceleration will be normal to the surface, and whether the reported value is to be
peak to peak or zero to peak. Another factor to be borne in mind when
measuring g values is that the size and weight of the accelerometer and the

107

RAPPING

90
80

70
-0)

~60
o

:;0

10

j! 50

co 40

~::J

1II30
G)
:::i:
20

10
O~----------~----------~----------~

0.1

1
10
Mounted accelerometer mass (1<J3kg)

100

Figure 4.14 ElTect of accelerometer mass on measured 'g' value.

frequency cut-off range can affect the measured value, thus further complicating the situation. This is illustrated in Figure 4.14, which shows how
the measured value for a 5 m x 15 m catch space collector top rapped by a
15kg rod dropped through l00mm, is influenced by the mass of the
accelerometer [7]. Ideally, since the collector is relatively thin, the measuring device should have zero mass to eliminate local disturbances and
changes in the mode of plate vibration.
In practice, a lightweight accelerometer of 0.2 g, is possibly the nearest
approach to the ideal and, to avoid 'ringing', and hence false high readings,
some suppliers advocate that a practical approach is to use a measuring
cut-off frequency of 10kHz, which is well below the collector's natural
frequency of vibration, but there is no adequate theory that has been
adopted worldwide.
Rapping re-entrainment, which is inevitable on mechanically rapped
dry units, is minimized on most modern installations by using microprocessor control systems to ensure that collectors in successive fields are
not simultaneously rapped. Some installations have the kV and power
slightly reduced on the field immediately before the blow is delivered, to
eliminate any flashover induced re-entrainment which may arise because
of the dust disturbance. In other plants, power-off rapping is employed,
i.e. all electrostatic clamping forces are removed to assist in dust dislodgement.

108

MECHANICAL DESIGN CONSIDERATIONS FOR DRY PRECIPITATORS

Other installations minimize re-entrainment by completely isolating the


flow during rapping by means of inlet and outlet isolating dampers,
so-called off-flow rapping; all fields are then rapped simultaneously, usually
with the power off. The isolated flow is then left for some period of time to
allow the dust to reach the hoppers prior to reconnecting the flow back into
service. An alternative to complete flow isolation is to only blank-off the
duct being rapped with an air curtain or damper system [8]. The economics
of these approaches need careful evaluation, since the cost of the additional
dampers and duct seals may outweigh the gain from a potentially reduced
plate area precipitator.

4.8

Hoppers

Examination of data log records indicates that at least 70% of the reported
dry precipitator problems are associated with hopper dedusting difficulties
of some sort, so one cannot overemphasize the importance of satisfactory
hopper and dedusting system design in any precipitator installation.
The particulates dislodged from the collector system have to be initially
contained in a receptacle in order that they may be removed from the
process gas stream. These receptacles usually take the form of pyramidal
hoppers or troughs located beneath the fields of the precipitator. To ensure
that the particulates can be extracted from these, it is essential that the
discharge opening in the base, leading to the final evacuation system, should
be no smaller than 300 mm diameter, regardless of the amount of material
in the hopper. The corner valley angle depends on the characteristics of the
dust but is not normally less than 55 to the vertical.
The bottom third or so of the hopper must be heated to maintain the
dust above the gas dew-point temperature; heating can be accomplished by
thermostatically controlled electrical tapes or pads, rated at 1.5 k W 1m2, or
by low pressure steam coils or jackets. Other appendages, which are often
incorporated, are some form of manual poke rod and a hopper level
indicator.
To assist in the emptying of particularly cohesive materials, some hoppers
have vibrators or hammers fitted to the side walls of the hopper. The use of
vibrators can, however, sometimes worsen the situation by compacting the
dust, so their use must be carefully considered. For potentially adhesive
particulates, such as those from paper mills operating the older Kraft
process, the base of the precipitator is a continuation of the vertical side
walls to form a full-width trough, the particulates being extracted by a
full-width scraper chain conveyor feeding into a narrower end conveyer
fitted with a sealing device.
To optimize the performance of any precipitator, it is essential to
eliminate any gas bypass of the field, as the gas expands over the hopper
region. This can be effectively achieved by full-height vertical interfield

HOPPERS

109

baffles in the case of trough hoppers, while for pyramid-type hoppers


adequate deflection plates located so as to divert the flow upwards will be
fitted. With designs using a frame-type discharge electrode carrier, centre
hopper baffles extending up to the bottom of the collectors can be found;
for plants having the lower end of the discharge electrodes terminating
beneath the collectors, the height of any potential centre baffle would not
assist in controlling bypass and, in this instance, it is imperative that the
small inlet and outlet hopper deflector baffles are fully effective. This is
normally checked on large, even full-scale models of the area.
There are numerous devices on the market for hopper level indication/detection, employing one of the following principles: capacitance
change, change in a tuning fork frequency, change in pressure/suction,
change in rotational speed of a driven disc, ultrasonic and radioactive
sensors. As any device will only determine a local dust situation, overfull
hoppers can still arise, in spite of an apparent 'empty' hopper indication.
Should the hopper overfill with dust, not only will the dust short-out the
electrics, but could also result in severe mechanical damage if the level
should reach high enough into the field area. If 'rat holing' occurs during
the emptying from a very overfull hopper situation, since the dust is fluid, if
only one duct empties, the hydrostatic pressure exerted by the dust in the
adjacent filled ducts can result in local distortion of the lower part of the
collector and sometimes the discharge system itself, with a deleterious effect
on precipitator performance.
To reduce this risk, or at least give further early warning, in addition to
the hopper level devices, chains are sometimes hung from the four corners
of the discharge frame. The dust on reaching the chains usually causes the
electrics to trip on reduced kV output, thereby preventing further dust being
precipitated. One should remember, however, that although dust will not be
precipitated, it will still tend to be deposited under gravity, particularly at
the inlet, so that action to remedy the overfull hopper situation is still a high
priority if damage to the internals is to be avoided.
As many precipitators operate under suction, it is imperative that air
inleakage into the hopper is prevented to eliminate possible re-entrainment
of collected particulates. Even an unmeasurable inleakage flow can reentrain sufficient dust in passing through the hopper area to significantly
affect the overall emission from the plant, particularly the outlet hoppers.
For example, a precipitator operating with an emission of 50mg/Mm 3 , will,
if mixed with an outlet hopper in leakage, equating to only 0.25% of the total
flow but carrying 10 g/Nm 3 of hopper dust, result in a final emission
approaching 75 mg/Nm 3 , i.e. a 50% increase.
Although hopper evacuation systems fall outside the scope of this
publication, the precipitator designer/supplier must be fully conversant with
the proposed evacuation system capacity and design, since this will not only
impact on the choice of hopper, but also on the method selected to prevent
air inleakage during emptying sequences.

110

MECHANICAL DESIGN CONSIDERATIONS FOR DRY PRECIPITATORS

For volumetric systems, e.g. conveyors of various types, rotary valves are
often used sited below the hopper take-off point to meter the dust flow into
the conveyor. As the rotary valve acts as a pump on the return empty
half-cycle, its position in the down leg and operation must be such as to
provide a positive head of dust to eliminate inleakage. For either dense or
lean phase pneumatic conveying systems, both rely on slide-type gate valves
for isolation and sealing, so the above 'pumping' problem will not arise;
however, it is most important that these gate or slide valves seal, particularly
when positive pressure systems are used.

4.9

Electrical clearances

The electrical clearances that must be carefully considered, if the precipitator


is to operate efficiently and as designed, fall into two main categories; the
first are those associated with the discharge electrodes and collectors and
the second, the peripheral clearances outside the field itself. Any electrical
clearance which is too small will limit the operating voltage of the unit and
so affect the efficiency of precipitation.
In general, the electrodes are suspended so that they hang exactly
mid-way between adjacent collectors. In practice, although the alignment is
critical, with erection and fabrication tolerances, one must accept a misalignment of up to 10 mm on a field height of 10 m and above, but a tighter
tolerance for smaller plants.
The position of the discharge element within the collector and the
distance between elements is also important to maximize efficiency; generally, a minimum electrode element~electrode element distance of half the
collector/collector spacing is used to give satisfactory corona formation and
current distribution over the collector surface (see chapter 3). For collectors,
the clearance between the stiffening channel edges and any emitting point
should be greater than the collector half-spacing by some 25% to provide
satisfactory emission and voltage conditions. This distance applies in both
the horizontal and vertical planes, since the electrodes or collectors may not
hang exactly plumb.
Another area of concern with the discharge electrodes is the position
where any support or carrier members pass through the collectors. In this
case, not only should the member be low emitting, e.g. have a large radius
of curvature, but must also pass centrally between adjacent collectors.
For clearances outside the actual field area, the positioning of any live
discharge electrode support members must be well clear of any earthed
structural components. For general safety, the following 'rule of thumb'
minimum clearance distances should be adopted for any design. As positively energized components have a lower breakdown strength than negatively energized elements, any edges on the earthed collector, casing or

REFERENCES

111

support members can result in premature breakdown. For example, for


components having sharp edges opposing one another, a separation of three
times the discharge electrode/collector distance should be used, even with
reasonably curved surfaces, such as those on rolled angles or joists; the
minimum separation should be 1.5 times the discharge electrode/collector
distance, with intermediate separations being adopted for differing profiles
and edge conditions.
A special area of concern is where the discharge electrode support tubes
pass through the casing and insulator. To minimize the quantity of heated
purge air used with the cone or flower pot insulator, the open area at the
base is significantly reduced by a restrictor, which would take the form of a
profiled electrical stress cone or bus ring. This, with a large diameter
suspension tube, e.g. 50mm or so, effectively minimizes the risk of breakdown in this area, in spite of the actual clearance approaching, or being
sometimes less than, the discharge electrode/collector separation. The location of the suspension tube must be truly central within the bus ring to be
fully effective.
Where bus ducts, rather than HT cable, are used to feed power to the
electrodes, their diameter and profile must be such as to eliminate the risk
of breakdown within the duct itself. If insulators are employed to centralize,
or carry the bus connection, these should have adequate tracking length to
prevent breakdown because of atmospheric dust deposition, etc.
On large precipitators having discharge electrode frames spanning two or
more hoppers, the actual distance between the frame and any cross-hopper
apex member, should take into account any potential dust deposition on the
member to minimize the risk of breakdown.
Similar comments would apply to internal tumbling hammer discharge
electrode rapping systems, where the 'live' shaft and hammers could give rise
to suspect areas, which would be difficult and expensive to overcome.
Other components, which may be built into the system, should adopt the
above 'rule of thumb' minimum clearance distances in order that the
maximum applied voltage is across the discharge electrode/collector system
and not elsewhere in the plant.

References
1. Unpublished Lodge Cottrell Ltd Research Report.
2. Parker, K.R. (1980) The precipitation of difficult dust. J. Electrostatics, 8, 355-67.
3. Forrest, 1.S. and Lowe, H.J. (1957) Present performance and scope for improvement in
power station electrostatic precipitators. Proc. Conference on the Mechanical Engineers'
Contribution to Clean Air, pp. 42-59, Inst. Mech. Engs, London, UK
4. Unpublished Lodge Cottrell Ltd Research Report.
5. Lowe, H.J. and Lucas D.H. (1953) The physics of electrostatic precipitation. Br. J. Appl.
Phys., Suppl. 2, 840.

112

MECHANICAL DESIGN CONSIDERATIONS FOR DRY PRECIPITATORS

6. Baylis A.P. and Russell-Jones A. (1982) Collecting electrode rapping designed for high
efficiency electric utility boiler electrostatic precipitators. Proc. 4th EPA/EPRI Symposium
on the Transfer and Utilization of Pasticulate Control Technology. Houston, Texas, Oct
11-14th - EPRI, Palo Alto, CA.
7. Darby, K. et al. (1990) The rapping systems for cleaning the discharge electrodes and
collectors of electrostatic precipitators. Proc. 4th International Conference on Electrostatic
Precipitators, Beijing, China, Chap 7, pp. 498-513, International Academic Publishers,
Beijing, 1992.
8. Falaki, H.R. (1993) Experimental study of flow diversion during the rapping of collector
plates inside an ESP. Proc. 10th Particulate Control Symposium and the 5th International
Conference on ESPs, 5-8 April, Washington, USA, pp. 37.1-15, EPRI TR LOS048 V2, Palo
Alto, CA, USA.

Aerodynamic factors affecting performance


L. LIND

5.1

Introduction

In any precipitator the gas flow carrying the particles which have to be
precipitated plays an important role. Gas volume rates between parallel
connected chambers should not deviate too much from the average value
used for sizing, and the gas flow distribution across the face of each
individual chamber should be within certain limits. Furthermore, it is
important to include the turbulence and the secondary flow, generated from
internal structures, and from the corona wind itself. It is the intention of this
chapter to explain these matters in detail.

5.2

5.2.1

Turbulence and secondary flow

Historical resume

The flow between the electrodes has been known to be turbulent right from
the very early days of precipitation but, due to the complexity of the flow,
a more quantitative analysis was not performed until the appearance of the
computer, half a century later. The flow which is influenced by gravity,
electric field and viscosity controls the movement of the electrons, positive
and negative ions, neutral gas molecules and particles. This is indeed a very
complex flow, difficult to conceive, to measure and very difficult to calculate
theoretically.
Going back more than 70 years to 1922, Deutsch put forward his
expression for the electric particle migration velocity [1]:
(5.1)
the nomenclature being VE the electric migration velocity, q the particle
electrical charge, Ep the precipitating field strength vector, dp the particle
diameter, /l the dynamic viscosity of the gas.
Deutsch took the forces acting on spherical particles to be the electrical
Coulomb and the Stokes' fluid drag forces, presuming a quiescent gas. The
determination of the charge was his prime concern, and his charging theory
was extended down to particles of less than 0.1 /lm diameter.

114

AERODYNAMIC FACTORS AFFECTING PERFORMANCE

The diffusion charge and the field charge were later treated by, e.g.
Cochet [2] and McDonald et al. [3], by introducing the concept of particle
mobility
mp

= q/(3nd p {l)

(5.2)

Substituting in equation (5.1) gives:


(5.3)

This is the Coulomb-Stokes value, which Deutsch introduced in his


expression for the efficiency
'1

= 1 - exp[( -vdv)(L/d)]

(5.4)

where v is average axial gas velocity, L is precipitator length and d is half


the duct width. (vE/v)(L/d) is the so-calkd Deutsch number D.
In the derivation of equation (5.4), Deutsch presumed the particle
concentration profile to be flat over the electrode space due to 'whirling
motion, electric wind etc.' In other words, he assumed full turbulent mixing
and thereby the diffusion or degree of turbulence to be infinite.
According to Mayer-Schwinning [4], the influence of the electric wind
was discussed as early as 1924, and in 1931, Deutsch stated [5] that the ion
wind had a positive as well as a negative influence on precipitation. He
meant that 'the gas flow was only responsible for the uniform distribution
of the particles within the precipitator' (quote from [4]). The physical
mechanism Deutsch looked upon was a core of flow with infinite turbulence
intensity, and a boundary layer close to the collecting wall, where the
Coulomb force overcomes the Stokes' force.
In his 1963 book, [6] White demonstrates how precipitator efficiency
with laminar flow becomes 100% after a length LLam = vd/vE. The efficiency
expression is:
'1Lam

= 1- D

(5.6)

In the late 1970s Professor White expressed his great interest in achieving
realistic values of the turbulence level inside the precipitator, being well
aware of the fact that equation (5.4) might be pessimistic.
Since Deutsch, many specialists have tried to 'crack the nut' by including
a finite turbulence intensity in an appropriate manner. In 1968, Myron
Robinson [7] published a review of articles dealing with the influence of
turbulence. His own interest was the 'electric wind' and its contribution to
turbulence and the resultant secondary flow and its influence on precipitator
performance. From his review the essence from a few contributors follows.
(a) In 1959, Friedlander [8] introduced a particle vorticity diffusion and
a migration velocity in the two-dimensional differential equation for the
particle concentration in both axial and transverse directions.

TURBULENCE AND SECONDARY FLOW

115

0.008 , - - - , - - - - - - , - - - - . - - - - . - - - - - - - - - ,

---.
en

.........

"IE
.........

0.006

+------1f------+---+---t-----l

.0
:~

en

::l

'+-

0.004 - f - - - - I f - - - - - + - - - j - - - + - - - - - - i

:-!:::

"D

tu

0.002 - f - - - - I f - - - - + - - - j - - - t - - - - - - i

0...
0.000

+------1'----+---+---t---I
a
2
4
6
8
10
Distance from wall (cm)

Figure 5.1 Effective particle diffusivity measured with LOA at 2.5 mls and - 38 kV. Duct width
is 250 mm. Redrawn from [13].

(b) P. Cooperman used a similar model in the years from 1960 to 1966
[9] adding re-entrainment to advance the problem.
(c) In 1967, Myron Robinson published a paper [10] with a modified
Deutsch equation inspired by Inuyshkin and Averbukh and P. Cooperman.
(d) Crowe and Stock published, in 1974, a paper which included a
calculation on the electrodynamic flow in a tube precipitator including an
effective viscosity for a pipe flow without an electrical field [11].
(e) In 1979, Bernstein and Crowe [12] described observations using a
laser light sheet in a wire plate precipitator. The same paper deals with
numerical calculations of electric field and gas flow, the latter described by
a k-e turbulence model. They concluded that the interaction between electric
field and gas flow increases the migration in certain regions, but decreases
migration and augments re-entrainment in other regions.
(f) The same year, Harald Gross of Stuttgart published results from
measurements in a laboratory precipitator [13]. He measured the effective
particle diffusivity r, using a laser doppler anemometer, and found a peak
close to the collector wall (Figure 5.1). Furthermore, the particle migration
velocity was recorded and he considered that the particles must overcome a
diffusion barrier at the wall before precipitation. As this barrier is independent of duct width, this is a fluid mechanical argument for wider ducts
having higher migration velocities than corresponding narrower ones.
(g) In the late 1970s a group of scientists at the High Temperature Gas
Dynamic Laboratory at Stanford University took up the topic, continuing
through the 1980s [14,15]. They contributed powerfully to the understanding and the solution of the influence of turbulence on performance. The

116

AERODYNAMIC FACTORS AFFECTING PERFORMANCE

0.4 -,---r------r------r----,--------,

..
0.3

0"

+---h~~___1--_+--_+-_____4

""'
UJ

"E 0.2 -t--i-t--___1--_+--_+-_____4


~
:::J

0.1

0.0

+---+-__rl--_+---+----1

..

+---t----f---+---+---l

0.0

0.5

1.0

1.5

2.0

2.5

Axial velocity (m/s)


Figure 5.2

u;m, measured in laboratory precIpitator using LOA between electrodes. Circles and
squares are from two different horizontal levels. Redrawn from [20].

work concluded in a theoretical and practical demonstration of the expectation of a higher efficiency electrofilter with a finite level of turbulence. This
they illustrated in a graph showing penetration P in dependence of the
Deutsch number, D = vEL/(v . d) with Peclet number as a parameter, P E =
vEd/r. Suffix E stands for 'electrical', VE being determined from equation
(5.1). P E is Cf) for laminar and 0 for fully turbulent flow. The diffusivity
r ~ 0.2' U~ms' LEU' where u~ms is the rms value of the turbulent fluctuating
axial velocity and LEU is an Eulerian length scale. Typical turbulent velocity

Figure 5.3 Vortex structure topology at zero bulk velocity. Emitter point at E. From [21].

117

TURBULENCE AND SECONDARY FLOW

0.1

".....

..!..
c

:8
~C1l
C

C1l
Q.

./

Y-

...

10

100

/
o

/'
./

Deutsch Number

Figure 5.4 Penetration versus Deutsch number. Curve is laminar theory, straight line is
Deutsch's theory, while circles are from laboratory precipitator measurements. Redrawn from
[22].

fluctuations are shown in Figure 5.2 from [20]. Calculated and measured
concentration profiles, performed at Stanford, clearly demonstrate the
claims by former researchers of increasing concentration towards the
collector wall.
Y. Yamamoto et al. in 1980 published a numerical study of secondary
flow interaction [16], followed in 1981 by an experimental and theoretical
study (2 - D) of the secondary flow in a positive corona single stage
precipitator [17]. In 1982 Thomsen et al. repeated the experiment but with
negative tuft-corona [18].
Later, in 1984, Shaugnessy et al. from Duke University, North Carolina,
published their version of the influence of the electrical body forces on the
flow creating a secondary flow pattern [19].
In 1986. Larsen and Sorensen on the basis of measurements in a
laboratory precipitator, described an analytical procedure for simulating the
secondary rolls generated by the current [20]. Larsen and his group
published papers on secondary flow and turbulence in precipitators through
the 1980s. They demonstrated that the vortex picture begins having a
complicated course (Figure 5.3) at zero axial velocity, changing to vortices
being stretched by the bulk flow at higher axial velocities [21]. One of the
findings was that the performance might decrease below the Deutsch limit
at higher Deutsch numbers, because particles trapped in the middle of the
vortices escape the electrode zone without reaching the collecting walls. This
might happen with laminar flow as well. The phenomenon appears to be
confirmed from tests on a small laboratory precipitator (see Figure 5.4, from
[22J). In the figure the laminar curve and the Deutsch line are seen together
with the measuring points shown as circles. Above a Deutsch number of
approximately 4, the points drop below the Deutsch line.

118

5.2.2

AERODYNAMIC FACTORS AFFECTING PERFORMANCE

Turbulence

A description of the influence of turbulence upon precipitator efficiency,


based primarily on theories and measurements from Stanford University,
California, and the Technical University of Denmark, is given below.
Turbulence is generated from the inlet screens, the frame members and
structural members inside the casing. The discharge electrodes themselves
produce eddies or vortices, which degenerate to turbulence and finally into
heat. The boundary layer, extending from the leading edge of the first
collecting plate down through the duct of the given field, creates turbulence
due to wall friction, being enhanced from any protrusions on the collecting
plates, the so-called antire-entrainment baffles. Normally the field length is
too small to make the boundary layers of opposite walls meet, hence real
'duct flow' with fully developed velocity profiles is unlikely to occur.
The electrons of the corona become attached to gas molecules converting
them to gas ions, which move with high velocity due to the electric field. The
ions and the neutral gas molecules, thus carried away, form the so-called
'ion wind', which is a major source of turbulence.
There is a clear difference between positive and negative corona for
smooth wires; the positive corona is a rather stable glow along the wire,
while the negative corona forms corona tufts. The tufts are randomly
distributed and move with time, and the number increases with current
density. For discharge electrodes with corona spikes (high emitters) the
difference in appearance between positive and negative polarity is thought
to be smaller. As positive corona is hardly used today in industrial

60.---~---.--------.--------.

~40+---~~~--------+_-------4

~
o
::J

)
~ 20+-------~r-------+_-------4

r-lIr_

o+--------+~~----+_~----~

Bulk velocity Uo (m/s)

Figure 5.5 Turbulence intensity versus bulk velocity. Black squares without tension and circles
with -50kV and O.65mA/m2. Redrawn from [18].

119

TURBULENCE AND SECONDARY FLOW


200
lBO
r--

I 160
.........
U 140
f-

:r:

120

~.-

.. ....... .""". .....


.
'..... . ". ./" . ."......

gj

100

c::
o BO
iii

a5

60

40

"'09 "J'.

(fl

20

10

20

30

40

50

60

70

BO

90

100

Length coordinate (-)


Figure 5.6 Heat transfer coefficients from heated plate to air. Triangles from flat plate, circles
from plate with transversal baffles. Redrawn from [24].

precipitators, it will not be treated here, but the potential of positive corona
with respect to high efficiency precipitation due to low turbulence might be
worth exploiting in the future. Therefore, this chapter deals only with
negative corona.
From simple theory, we know that turbulence has a negative influence
on precipitator efficiency, as laminar flow should give 100% efficiency for a
given precipitator length, while the Deutsch theory, assuming 100% turbulence, should produce an upper limit with respect to precipitator length.
In Figure 5.5 we see the turbulence intensity as a function of axial gas
velocity [18]. This reveals the fact that too Iowan axial velocity would be
expected to have an unwanted effect on efficiency, due to turbulence, and it
should be avoided.
Often, in the literature, it has been claimed that low turbulence is created
by the shielding of collecting plate protrusions or baffles. This is not the case;
in [23] it is shown, by laboratory measurements and theory, that baffles
have a negative effect on efficiency, and [24] demonstrates the negative effect
of baffles using similarity between high heat transfer coefficients downstream
of a baffle and high turbulence (Figure 5.6). The superiority of baffle-free
plates was demonstrated by field measurements on collectors formed from
horizontal plate strips without baffles.
The laminar molecular kinematic viscosity of air, at standard conditions,
is of the order 15 x 10 - 6 m 2 /s, whereas, the turbulence particle diffusivity r,
predicted from concentration profile measurements in a parallel plate
laboratory precipitator with precharged particles, is of the order 60 to

120

AERODYNAMIC FACTORS AFFECTING PERFORMANCE

120 X 10- 6 m 2/s, i.e. four to eight times higher. With batfJ.es an even higher
level was found, 30 x 10- 4 m 2/s, i.e. about 20 times higher [14].
An equation describing the concentration balance, assuming constant
turbulent particle diffusivity, is:
(U

+ uE)on/ox + (v + vE)on/oy + (w + wE)on/oz


= qo2n/ox 2 + o2n/oy2 + o2n/oz 2)

(5.7)

where u(x, y, z) is the time-averaged velocity, n(x, y, z) is the time-averaged


particle concentration, U E , VE and WE are the electrical migration velocities,
and x and yare coordinates in the axial direction and perpendicular to the
collecting wall, respectively; z is the third Cartesian coordinate.
Initial conditions and boundary conditions are:
at x = 0
on/oy = (v
on/oy = 0
on/oz

+ vE)n/r

at y = 0
at y = d
at z = 0 and L

d is half the duct width, and L the precipitator length.


Equation (5.7) can be solved numerically, as in [25] for two dimensions,
x and y. Figure 5.7 shows a typical measured concentration profile at
0.4 mA/m2, redrawn from [25]. This profile differs from the flat 'Deutsch'
profile only in the core region, i.e. 0.20 times half-duct width.

1.0 - r - - - , - - - - - , - - - . . . . , - - - - r - - -

;;::
o
....

0..0.6

-I---l------+---+---+-----j

:;::;

b 0.4 - I - - - - , f - - - 4 - - + - - t - - - j
C
<Il

80.2

-I---f---4---l---t---j

0.0 -I--.--f-..,--4-,---l---,--t-....,--j
1.0
-1.0
-0.6
-0.2
0.2
0.6

Y coordinate

Figure 5.7 Measured concentration profile in a wind tunnel precipitator at approx.


0.4 mA/m2 Redrawn from [25].

121

TURBULENCE AND SECONDARY FLOW

D.l

/5

20

,,-....
~
.........

:;:;
0

.....

V
./

L...

Q)

Q)

10

0...

100

f/

./V

./

345

Deutsch Number

Figure 5.8 Penetration versus Deutsch number for various electrical Peclet numbers.
L - PE = X! and D - PE = 0, curves are for PE = 5 and 20. Redrawn from [26].

Integrating the concentration profiles across the duct width gives the
mean concentration nm(x) and the efficiency along the precipitator l1(x) =
1 - nm(x)/nm(O). Instead of x, (vE/v)(x/d), the dimensionless length or the
Deutsch number, is used in Figure 5.8 as abscissa with penetration,
P = 1 - 11 as ordinate. In solving equation (5.7) PE' the electrical Peelet
number, appears as a parameter: P E = vEd/r.
Figure 5.8 shows that a finite turbulence level, i.e. P E greater than 0, gives
less penetration as compared with the penetration found by the Deutsch

0.12 - r - - - , - - - , - - - - - , - - - - - - , - - - ,

D.1D

+---+----'1----+---+---1

D.D8

-1----+---1-----1----+----1

,,-....
(/J

..........
E D.D6 +-----,~-_+--+--+--_1

.........

0.04 +---+-----1---+---+---1

:::J

0.02

O.DD

+---+-----1---+---+---1

+--,--+--.----II---.--+-.--+-r--I
0.4
0.8
1.2
1.6
2.0

0.0

Bulk velocity (m/s)


Figure 5.9

U,m,

versus bulk velocity. Black squares with DC and triangles with pulse. Current
density 0.05 mA/m2. Redrawn from [27].

122

AERODYNAMIC FACTORS AFFECTING PERFORMANCE

equation. Notice that a disadvantage of the graph is the fact that VE is part
of both the abscissa and the parameter. This has been modified [26] by
introducing a dimensionless number equal to the P6c1et number divided by
the Deutsch number, i.e. vd 2 /(rL).
In [25] the reason for the measured migration velocity enhancement in
the entrance region to the precipitator is believed to be caused by increased
particle charging due to turbulent motion. This leads the authors to
conclude that relatively high turbulence ahead of the first field and moderate
and low turbulence further down through the precipitator is desirable.
In the case of pulse operation, with approx. 150 flS pulse width, it has
been shown on a wind tunnel laboratory precipitator that the turbulence
level is more moderate than for the case of DC operation. This is due to the
fact that the impulsive electrical body forces deliver a smaller momentum to
the flow than the continuous body forces of direct current (Figure 5.9, from
[27]).
5.2.3

Secondary flow

Similar to the turbulence phenomena described, the effect of secondary flow


on the precipitation rate is limited to a description of the effect of the
non-uniform electric field created by negative corona, which is dominant in
industrial precipitation.
Many commercial discharge electrodes are fitted with sharp points or
pins acting as specific or 'controlled' corona emitting points. The ion current
emitted from these entrains surrounding gas, creating an ion wind, blowing
towards the plate at the level of the pin and back again between pins in
order to maintain continuity. Even for a discharge electrode of the wire type,
negative corona tend to be generated at discrete points, where they make
tufts which move up and down along the wire. Basically the interaction
between electrical field and gas flow creates tufts that are positioned at
almost constant pitch.
This non-uniformity in electrical field and current gives rise to body
forces acting on the fluid, creating vorticity, which drives the secondary flow,
and the stretching of the vortices is a source of turbulence.
In [17], Yamamoto and Velkoff revealed secondary flow with mainly
vertical axes using a smoke visualizing technique. In [21], the structure for
low bulk velocity is interpreted from calculations of the flow field, suggesting
a vortex topology structure as seen from Figure 5.3. For low velocities,
0.1-0.2 mis, the flow field is reported to fluctuate at a frequency corresponding to the bulk velocity divided by axial electrode pitch. As bulk velocity
increases for a given current density, the secondary flow pattern, vortices
with almost vertical axes, changes and at realistic velocities their axes tend
to turn horizontal (provided the discharge electrodes are vertical). Thus, we
end up with secondary vortices like those indicated in Figure 5.10 from [20].

TURBULENCE AND SECONDARY FLOW

123

00

IJfB
a>

C\I

10 <0
C\I ~

+-,........--+----1....-==-550
300

150

Figure 5.10 Secondary rolls generated by negative tuft corona. Velocity profiles indicated, too.
From [20].

See also [28] for other suggested vortex structures. At even higher velocities,
the vortices break down to turbulence, which is dissipated fairly quickly into
heat.
Larsen [29], introduced analytical expressions for the secondary flow
with velocity components satisfying mass conservation:

v = Vo sin(ny/d) cos(nz/d)

(5.8)

w = -(L/d)vo cos(ny/d) sin(nz/d)

(5.9)

y being the coordinate perpendicular to the collecting surface, z the


coordinate along the discharge electrode body, d half the duct width, L half
the pin pitch, and Vo the maximal roll velocity or 'roll strength'. The 'roll
strength' is defined as the maximum transversal mean velocity at the level
of pins, normally found half-way between the emission electrode and the
collector wall.
If the v and w expressions are inserted in the two-dimensional particle
concentration balance equation (5.7) we get:

124

AERODYNAMIC FACTORS AFFECTING PERFORMANCE

0.1,-------"'"'"T"----y-'I

~
~
c
o

.~

a;
c

If.

10

0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6


Dimensionless length

Figure 5.11 Penetration versus dimensionless length (= Deutsch number) for roll strengths
volvE = O. 0.5, 1 and 2 for infinite Peelet number (laminar). Deutsch's theory gives the straight
line marked D. Redrawn from [30].

By assuming a constant axial velocity U0' constant drift velocities VE and


putting WE equal to zero, and ignoring axial diffusion, then using the
proper boundary conditions from equation (5.7), the efficiency can be found
by integration:

UE ,

(5.11)

1.0

O.B

~0.6
C

(J)

'u
WO.4

;;::

0.2

0.0

17

/'
V

1I

0.0

0.2

0.4

vEivo

0.6

O.B

1.0

Figure 5.12 Maximum efficiency versus inverse roll strength for infinite Peelet number
(laminar). Redrawn from [30].

TURBULENCE AND SECONDARY FLOW

125

X being a dimensionless axial position, X = (vEIU o)(xld), P E the electrical


Peclet number, P E = vEdlr and VOIVE the dimensionless 'roll strength'.
Figure 5.11 shows penetration (1 - '1) versus X for P E -+ 00, corresponding to zero diffusion. It is obvious that the expected laminar efficiency is
markedly reduced with VOIVE > 0 and for a dimensionless roll strength of 2,
the Deutsch limit is intersected at X;;::: 1.1. The achievable maximum
efficiency for laminar flow versus inverse roll strength is shown in Figure
5.12.
Roll strength has been found from measurements in a wind tunnel
precipitator to increase with increasing velocity up to a maximum, dependent on current density, and to decrease again at higher bulk velocities,
possibly due to the fact that the rolls are incompletely developed at any
given axial position.
For a given bulk velocity, the roll strength increases with DC current
density almost linearly as seen from Figure 5.13. Experiments also show that
the roll strength increases as the distance between corona points over
half-duct width approaches unity.
In reference [29], a thorough physical and mathematical treatment of the
secondary flow is given. The non-uniformity in electrical field and current
gives rise to body forces Ie acting on the fluid. The vorticity drives the
secondary flow and the stretching of the vortices causes turbulence. Mathematically the time-averaged Navier-Stokes equations including Ie can be
expressed as:
V'Vw = w'VV + vV 2 w

+V

Ie/p

(5.12)

0.5 ...,..-,----,.----,--....--,----,.----,--..,

0.4

-t---I----+--+-+--+---+--+--j

---.0.3

+--t--t--t--+--t------l:.--+--i

en

'-

-S

:;;,0.2

+--t---t-"--'1---.+--t---l--+--I

0.1

+--t---t--+-+--t---l--+--I

..

0.0 +--1----+--+-+--+---+--+--1
0.00
0.04
0.08
O. \f,
0.16

Current density (mAim)

Figure 5.13 Roll strength versus current density. Circles are with DC and squares with pulse
operation. Redrawn from [27].

126

AERODYNAMIC FACTORS AFFECTING PERFORMANCE

where V is the velocity vector u(x), v(y), w(z), P is gas density and v is gas
kinematic viscosity. w is the vorticity vector w x , w y ' W z The body force
ie = Pi E with ion density Pi and electrical field E. Assuming the ion space
charge Pi the dust space charge Pd' which is normally justified, the
electrodynamic equations and the flow equations can be solved separately.
A unidirectional and periodic field ie = (0, i y , 0) is introduced in equation
(5.12), which is solved numerically. The author presents a perturbation
analysis using a periodic, unidirectional field to establish the effect of current
density and electrode geometry on the fully developed roll strength, including an estimate of the roll development length.
The roll strength is found as:

where Re = V od/v and Pi E = im/m, the rotational part of the electrical force
field driving the ionized gas motion, m is ion mobility, C( = L z /(2d) the roll
aspect ratio and L z is the vertical distance between corona peaks.
The question, how pulse operation influences the secondary flow, in cases
where the pulse frequency exceeds the characteristic frequency of rolls, has
to some extent been answered by investigations performed by the same
group of researchers. Experiments referenced in [27J, using approx. 150-,us
pulses, show rolls of almost the same strength as for DC up to about
0.08 mA/m 2 , at bulk flow velocities of 1 m/s (Figure 5.13). At higher current
densities, pulse roll strength decreases while DC roll strength increases.
Though the exact current limit is specific to a given arrangement, one might
conclude from Figure 5.13 that there is no real difference between pulse and
DC energization with respect to roll intensity, particularly as pulse operation is normally used with high resistivity dust, demanding low current
densities.

5.2.4

Numerical flow model

A full model consists initially, in drawing up the Maxwell equations for


electric field and space charge in three dimensions and implementing a
proper particle charging theory; secondly, the full Navier-Stokes equations
in three dimensions for the flow field including the electric body forces and
the necessary turbulence model with the corresponding equations; and
finally the three-dimensional particle transport equations. The boundary
conditions, which form the solution of the partial differential equations; are
generally straightforward; only the corona region stilI needs some empiricism. It is a demanding task working out the proper codes, finding an
effective and fast solution procedure and having access to sufficient computer power. The difference equations are coupled and need to be solved
iteratively demanding significant computer space and time.

GAS VELOCITY

127

lamshid Zamany [30] tried, but had to limit the problem. He worked
with the electrical equations in three dimensions combined with a twodimensional turbulent flow model. He used a constant dilTusivity in the
calculation domain proportional to the square of axial velocity times duct
width, r - U 2 d. In order to determine the electrical operation point with
dusty gas, only the corona starting voltage is needed for a given clean gas
state. The discharge electrodes can have any form, including fully threedimensional spike types, and the collecting plate surface can be corrugated
or provided with baffies. The space charge distribution is three-dimensional
and the current density pattern on the collector surface is predicted. Corona
quenching is calculated, together with axial current density averaged in the
vertical coordinate direction. Efficiency as a function of axial length can be
predicted for each class of particle size.
Further improvement of the model includes three-dimensional flow field,
local particle dilTusivity, a turbulence model reflecting the physics inside the
precipitator and re-entrainment at the near wall region. In a later improved
version the local field strength and the local current density on the dust
surface and across the dust layer are calculated and used for prediction of
back corona.

5.3

Gas velocity

The Deutsch migration velocity, presuming a quiescent gas, only depends


on the molecular gas viscosity. Through the years a large number of
efficiency measurements has been performed showing that there is a variation in the migration velocity parameter with bulk velocity.
Among the papers dealing with this subject, the contribution from
Dalmon and Lowe from the 1950s has been chosen [31]. They used a
pilot precipitator where dust burden, temperature and velocity could be
controlled. Figure 5.14 shows the calculated migration velocities for 3.5, 11
and 51 11m particles, increasing with increasing velocity up to about 2 mls
and a decrease at higher velocities, presumably due to re-entrainment from
scounng.
By means of the theoretical code from [30], the efficiency has been
determined and the migration velocity parameter calculated using the
Deutsch expression. The result is seen in Figure 5.15 revealing an increase
with velocity, as measured by Dalmon and Lowe. The code does not take
re-entrainment into consideration, so the constant level above approximately 2 mls does not reflect reality. The data scatter at high velocities is
due to numerical instability, not to physics.
The conclusion is that an optimally designed precipitator should operate
in the range where the migration velocity is almost independent of axial
velocity. If a precipitator is to operate at a reduced velocity, the variation
should be included in the sizing. It should be remembered that the reduced

128

AERODYNAMIC FACTORS AFFECTING PERFORMANCE


25,-------,------,------,-------,

cn20+-------~--~_+~----~----~

"E
u

'-"

~15+-------~~--~~--~~----~
-+-'

o
Q)

>10+-----~f_f_--_+------~~--__1

51

:;:;
11
~ 5+------+r-----_+------~--i-__1
(Jl

3.5
o+---.---r--.--_+---r--~--.___1

Axial velocity (m/s)

Figure 5.14 Migration velocity parameter versus axial velocity for 3.5-, 11- and 51-J.lm particles.
Redrawn from [31].

migration velocity using low velocity is the result of the higher turbulence
level.
Precipitators operating at elevated axial gas velocities must have a perfect
gas distribution, as velocities above average might cause scouring and
re-entrainment, thereby limiting the achievable precipitator efficiency. A plus
for higher gas velocities is the relatively low level effect of turbulence
improving efficiency.

40

o
:;:;10

'--

11

(Jl

234

Axial velocity (m/s)


Figure 5.15 Migration velocity versus axial velocity. Calculated using the code of [30].
Re-entrainment not included.

GAS DISTRIBUTION

5.4

129

Gas distribution

Dust-laden gases are led through ducts where the velocity is chosen
according to various criteria. Dust concentration might vary between a
couple of g/m3 and several hundred g/m3 and the conveying of the dust
right up to the precipitator without particle deposition demands that the gas
velocity in horizontal or almost horizontal ducts is not too low. Depending
on particle size distribution, dust density, concentration and duct length, the
gas velocity is normally chosen between 13 and 20 mls at nominal plant
output.
The need for an adequate conveying velocity is counteracted by the
request for a reasonable pressure drop and restricted wear on walls and
internal structures. Such considerations demand velocities below about
20 m/s. Some plants operate with a variable loading factor and if dust is
deposited on the duct bottom at reduced production there is a risk that the
bottom wall and the lower part of the side walls are cooled and will corrode.
The cooling might also mean that the deposit becomes sticky and forms a
more compact layer which cannot be redispersed when production rate
increases. Loosely precipitated dust, deposited during periods of reduced
production, will normally be re-entrained during run-up, which causes an
increased dust loading to the precipitator. If the precipitator has not been
sized to cope with such peaks in inlet concentration, the result might be
increased emission, maybe even violation of the guarantee or legislative
limit.
The optimal gas velocity in the space between electrodes varies from
process to process, but for the sake of simplicity assume it is set at 1 m/s.
The transition piece between raw gas duct and precipitator housing is, fluid
dynamically, a diffuser, where kinetic energy is converted to potential energy
or pressure. The transition piece between the precipitator housing and the
clean gas duct is a contraction, which, fluid dynamically, is relatively
unproblematic, having only a small effect on the upstream flow. The design
of diffusers with an expansion ratio of up to 20, or more, is critical to
provide an even velocity profile at the outlet, both in space and time.
Normally, the demand for an even profile, means an apex angle not greater
than 8 to 10 for a 'straight' diffuser. This is obviously an impossible
approach for economic reasons; firstly there would be a risk that dust would
be deposited on the bottom and, furthermore, small angle diffusers demand
horizontal space, which is costly and rarely available, particularly if an
existing precipitator is to be upgraded. (Most plant arrangements have close
coupling with minimal duct lengths to optimize capital cost and space.)
This leads to the use of angle diffusers combined with the transition piece.
Short length diffusers, or wide angle diffusers are fitted with perforated
screens, a technique adopted from wind tunnel diffusers using nets or grids.
If a diffuser has too wide an apex angle, the flow will separate from the walls
and a centre jet with a relatively high centre velocity and recirculating flow

130

AERODYNAMIC FACTORS AFFECTING PERFORMANCE

close to the walls is developed. Separation is caused by the negative pressure


gradient created by the diffuser itself. From time-averaged measurements it
is known that the characteristic velocity profile has a peak at the centre line
and reverse flow at the walls. The instantaneous profile for wide angle
diffusers adheres to one wall at a time and separates completely from the
other wall. In the next time interval the walls change their roles. In popular
terms, screens create a pressure drop, depending on the screen porosity,
which acts against the pressure rise of the diffuser. With multiscreens there
is a tendency to have velocity peaks close to the diffuser walls instead of a
at the centre axis. See reference [32J for information about methods for
producing uniform gas flow.
It is obvious that the ability of the inlet transition piece to reduce velocity
and create an even velocity profile depends on the velocity profile in the raw
gas duct. This must not be too skew and its form at the inlet flange should
at least be known to the precipitator designer. Not only the flange velocity
distribution has to be known, but also the particle distribution, any
temperature stratification, and the distribution of, say, O 2 and S02' all of
which are important to the precipitator elftciency. The gas mixing is
influenced by molecular and turbulent diffusion, and as for the temperature,
by heat transfer from convection and radiation.
5.4.1

Standards

The US 'International Oas Cleaning Institute', lOCI, published EP-7 in


November 1980. It deals with 'standards' for gas distribution and model
testing [33]. In this publication, it is clearly pointed out that the gas
distribution demands are subordinate to the emission guarantee and need
not be corrected, if it does not comply with lOCI EP-7, as long as the
emission guarantee is met. (The lOCI has since changed its name to ICAC,
'Institute of Clean Air Companies'.)
lOCI EP-7 lays down requirements as to the number of measuring
points, their distribution and the subsequent evaluation and statistics, e.g. it
demands that 85% of the velocities in a cross-section, 3 ft (0.9 m) downstream of the leading edge of the first field, are less than or equal to 1.15
times the average velocity. Furthermore 99% must be less than or equal to
1.40 times average. The same needs apply to a cross-section 3 ft (0.9 m) from
the outlet of the last field. Specific demands with 1.10 instead of 1.15, and
90% instead of 85% are not motivated by precipitator physics, but merely
in an attempt to demonstrate effective project management.
In the case of a presumed normal velocity distribution (which is not
necessarily the case) the [1.15185%J claim corresponds to a certain coefficient of variation, (J/v rn , (J being the standard deviation and Vrn the mean
velocity. See Figure 5.16, where the abscissa is normalized velocity, v/v rn , and
the ordinate is the cumulative normal distribution. The tangent to the curve

131

GAS DISTRIBUTION

100,-------------~--~~~~--,

85 -----------------------------

------,1

r--..

~
<ll

>

:;:; 50

..2

:J

E
:J

O+---~--------~---L------~

1.15

Normalized gas velocity. v/v m

Figure 5_16 Cumulative normal distribution versus normalized velocity. Distance between
crosses is 1.253o"/v m .

in point [1150%] intersects the 100% horizontal line at 1.0 + l.253a/vrn ,


and if the curve includes the point [1.15185%], a/vrn is approximately 0.145
or 14.5%.
Experience seems to indicate that a is almost independent of mean
velocity. In fact, the lower the average velocity, the more difficult it will be
to obtain a good gas distribution, and a/vrn increases with decreasing mean
velocity.

5.4.2

Residence time

Provided that the precipitation occurs evenly along the collecting plates it
can be concluded that the efficiency is dependent on the residence time.
However, this is not the case for the bigger particles, which are precipitated
after a relatively short axial length, while the full length is exploited for the
smaller particles. This argument does not take re-entrainment, sneakage,
sweep age or contra-emission into account.
The Deutsch equation, for the penetration P, had been derived with the
implication that concentration profiles are even across the duct giving only
axial changes in 'amplitude', presuming an infinite particle diffusivity or an
infinite turbulence intensity level. In spite of such a strong presumption, the
Deutsch equation (5.4) has given meaningful results and is still used
worldwide. Implied in equation (5.1), for a given field strength, gas state and
particle size distribution, is a migration velocity independent of axial gas
velocity. With this constant migration velocity, White [6] used 'calculus of

132

AERODYNAMIC FACTORS AFFECTING PERFORMANCE


25.0

-----.

r-

~20.0

f--

:;::;
:::l

-D
'i::

15.0

+J

.~
-0 10.0

t--

>.
U

OJ

6-

5.0

OJ

L.

LL

0.0
0.4

0.6

O.B

1.0

1.2

Gas velocity (m/s)

1.4

Figure 5.17 Frequency distribution of measured gas velocities. Readings are grouped in
intervals of 0.1 m/s.

variations' to demonstrate that mInImUm penetration is achieved for a


perfectly even velocity profile across the duct. In order not to confuse the
matter with mathematical demonstration, P is calculated below for a
precipitator with (1) constant gas velocity V m , and with (2) 0.99v m in one half
and l.Olvm in the other half. vdvm is set at 0.15 and Lid at 50:
P = 553.1 ppm

v = 0.99v m and l.Olv m

554.7 ppm

the latter penetration being 0.3% higher than the penetration with constant
gas velocity. Thus, it is made probable that the non-linear Deutsch expression gives minimum penetration for constant axial velocity.
If a velocity distribution function f(v) is introduced, equation (5.4) can
be modified. Figure 5.17 shows a measured discrete distribution, and Figure
5.18 a continuous approximation of the cumulative curve found from the
summation of the step curve. Often f(v) is assumed to be a Gaussian
distribution with average velocity Vm and 'standard deviation' 0':
(5.14)
and hence:
(5.15)

133

GAS DISTRIBUTION

100.-------------~------~~_,

80
~

~
~

60

Q)

>
:;:;
o
:::J

40

:::J

20

0.6

O.B

1.0

Gas velocity (m/s)

1.2

1.4

Figure 5.18 Cumulative distribution corresponding to the frequency distribution of Figure 5.17.
Median velocity reads 0.94 m/s.

Idel'chik and Aleksandrov [34] evaluate P


way:

P(v) in a slightly different


(5.16)

where
(5.17)
Ac being the cross-sectional area. See Figure 5.19 where penetration P is
plotted against parameter M [34].
Gooch et al. [35] use a velocity weighted version of (5.15):

L~~~' P(vrn)f(v)v/vrndv

(5.18)

In Figure 5.20 the expressions (5.15), (5.16) and (5.18) are reproduced as
graphs independent of a/vrn. The actual gas distribution function f(v) is
considered as being a normal distribution. The expressions of White and
Idel'chik and Aleksandrov correspond very well, giving a weaker dependence of the gas distribution than is found with (5.18).
Among precipitator vendors, theoretical or empirical curves based on
a/vrn' much like those of Figure 5.20, are used. In Figure 5.21 a correction
curve based on measurements by FHikt [36] in a pilot precipitator is given
together with a measured curve from [37]. Points are calculated using the
Wb theory [38] and measured data from [39]. Maybe the truth lies
somewhere between the full curve and the dotted curve.

134

AERODYNAMIC FACTORS AFFECTING PERFORMANCE


12

10

:;:; 6

l-

T'
Q)

C 4

,./

Q)

1.0

1.1

1.2

1.3

1.4

1.5

1.6

Parameter M
Figure 5.19 Penetration versus parameter M. Redrawn from [35].

5.4.3

Space charge

In the end it is the distribution of the dust particles, not that of the gases,
that determines the precipitator efficiency. Especially the distribution of the
fine particles, which are the hardest to catch, further down in the precipitator must be smooth. If the particle distribution is skew, then the space

3.0

c
/

c 2.5

:;:;

....
IQ)

~2.0

0..

,1/

Q)

>

:;:;

I,l

~1.5
0::

1.0

-~

. .. ~,,:::

..:., ....

10

20

I
30

!/
/

b
I
I

~l
40

Coeff. of variation (l6)

50

60

Figure 5.20 Relative penetration versus coefficient of variation. (a) equation (5.15). (b) equation
(5.16) and (c) equation (5.18).

135

GAS DISTRIBUTION

140

130
/""',

~
Q)

'iii

120

Q)

>
:;:;
o

~110

100

~....
k::::::' l:::::::
10

..........

...............

~~

...

..............

20

30

Coeff. of variation (~)

40

Figure 5.21 Relative precipitator size versus coefficient of variation. '1 = 99%. Full line redrawn
from [36], dotted line from [37] and squares from [38].

charge tends to disturb the current distribution. The term 'current capsizing'
is often used for this phenomenon. Differences in dust mass concentration
varying from top to bottom will always occur, but there is less difference in
the number of small particles, which have the greatest surface compared
with mass. The specific surface (m2jkg) is, for a given dust mass density,
proportional to Ijd p , dp being the particle 'diameter' or characteristic
transverse measurement. As charges adhere to the surface of the particles,
the charge per unit mass will also be inversely proportional to dp .
The difference in concentration between inlet and outlet planes is larger,
e.g. if we consider a field efficiency of 90%, the outlet mass concentration is
only one-tenth of that of the inlet. Yet, taking into account the fine particles
at the outlet having the greatest surface area, the variation in space charge
is less than a factor of 10. In principle, a balance will take place between
inlet and outlet current densities, as long as current quenching, due to space
charge saturation, does not occur: if the precipitation tends to decrease at
the inlet, more particles are conveyed towards the outlet causing the current
density there to decrease.
Crosswise the conditions are quite different. An individual duct hardly
knows about the existence of the neighbouring ducts and not at all about
the ducts on the far side of the precipitator. They are all exposed to the same
voltage and a small dust concentration in a specific duct increases the local
current density, due to a low space charge, thus creating a growing
imbalance between ducts. For this reason it is better with a couple of ducts
having high loading than a couple with low loading. This is why a gas
distribution with high velocity in the outer duct, often seen in precipitators

136

AERODYNAMIC FACTORS AFFECTING PERFORMANCE

fitted with central inlet and low porosity screens, is less severe than one with
much too Iowa velocity in the extreme ducts, which is likely to exist in
similar configurations with high porosity screens.
Skew crosswise gas distribution was seen on earlier precipitators with
transverse coupling of individual bus bar sections and a common rectifier.
Recording the current of each section often revealed that one of the sections
took the lion's share and left the other(s) with too small a share. Such
arrangements result in reduced efficiency and, today, vendors normally
provide each section with individual high voltage power supplies.
Current capsizing might, in cases with high resistivity dust, reduce the
efficiency drastically, because high current density in a local area due to low
space charge provokes contra-emission. The flow of electric current with
reverse polarity disturbs the local performance and undersupplies other
regions with current. As the total current is low, under high resistivity
operation, the efficiency is expected to reduce rather significantly. Strange as
it may seem, laboratory measurements with widely distributed contraemission have demonstrated that there is less influence of gas distribution
on efficiency for high resistivities. Yet, this is not the case for incipient or
moderate contra-emission situations [40].

5.4.4

Re-entrainment

Dust once precipitated on the surface of the more permanent, 'old', dust
layer is not necessarily caught once and for all. There has been some
indication of re-entrainment taking place under normal operating conditions, even though the gas distribution is even, showing that precipitation
efficiency is the net result of dust being deposited on the collecting surface
and dust leaving the surface. Furthermore, it seems that the time constant
involved can be in the order from a few seconds up to several days.
Unburned coal or particles, conductive for some other reason, present in
the gas stream might be discharged upon reaching the collecting surface and
re-entrained due to the lack of compressive or holding force from the electric
field. They might also be only partly discharged making it easier for the gas
to scour them off and redisperse them. Processes with conductive particles
demand a correct gas distribution and a moderate axial velocity.
In cases where high resistive and conductive dust particles are present at
the same time, it is important to increase the current density, above the
contra-emission limit, in order to keep the conductive dust partially clamped
to the collector. In such cases the demand for even gas distribution and
moderate gas velocity is even more strict. During collecting plate rapping,
lumps of dust are loosened and fall under the influence of gravity. Some are
caught again and others fall into the hopper. Where the lumps slide along
the dust layer on the plate, particles are loosened and redispersed. Likewise
falling lumps will tend to disintegrate due to gas erosion or hitting

GAS DISTRIBUTION

137

obstructions, e.g. frame tubes, rapping bars, walkways and the hopper walls
themselves. In all cases dust is re-entrained and redispersed reducing the
collection efficiency.
Only a few collector plates are vibrated or rapped at one and same time
in order to avoid too massive a re-entrainment factor. Some designs use
systems where the neighbouring ducts are closed, or the gas velocity in the
ducts is reduced, during rapping, using flaps or air curtains synchronized
with the rappers; other designs close off the compartment flow during
rapping. For the same reason synchronizing of rappers between serial fields
is used in order to reduce the effect of rapping spikes, especially in cases
demanding very low average emission.
One should be aware of the effect of falling lumps inducing strong
downwash. Neglecting gas friction the terminal velocity can be calculated as
Vf = --/(2 g. h), g being the gravity acceleration and h the fall height. Putting
g = 9.8 m/s2 and h = 10 m, results in Vf ~ 14 mis, a very high value, compared with the bulk velocity. The axial flow will be disturbed during
rapping, and the gas distribution will be temporarily skew, and there is a
risk of re-entrainment from the hopper. The phenomenon is predominant at
the inlet end, where dust is most abundant.
5.4.5

Erosion

If the bulk velocity is increased above a certain level, say above 2 mis, and

the gas distribution is uneven, erosion will most likely occur; again an
argument for ensuring a perfect even distribution and a not too high bulk
velocity (see section 5.3).
5.4.6

Sneakage and sweepage

Sneakage is the term describing dust-laden gas not passing through the
active electrode system, including areas where the discharge system has no
corona points, e.g. at frame tubes, even though turbulence in the ducts tends
to improve charging and thereby precipitation of the dust inside the active
electrode system.
Gases flowing between the very outmost collector plates and the wall of
the housing, and gases flowing above and below the electrode systems, carry
dust and give rise to sneakage. If 0.1 % of the dust passes untreated from the
inlet to the outlet, it is easily seen that efficiency can never exceed 99.9%,
however efficient the precipitating system might be. With 10 g/m3 dust
loading the emission will never be less than 10 mglm 3 .
In order to minimize sneakage, extra screens, so-called baffles, are put
between the fields and below the collectors. These screens should be
preferably perforated, porosity ~ 10%, in order to prevent turbulence and
recirculation behind the screen.

138

AERODYNAMIC FACTORS AFFECTING PERFORMANCE

The greater the space above and below the electrode system and the
smaller the collector height, the more the sneakage in terms of volume flow
rate and hence the reason why low precipitators must be fitted with effective
baffles.
Sneak age is estimated from velocity measurements in the regions above
and below electrode systems. Some model laboratories judge the sneakage
from the time it takes to empty the hoppers of smoke, injected before
start-up of the fan.
5.4.7

Optimal distribution

As seen from the preceding sections the interaction between gas distribution
and precipitator physics is complex, and the demand that the gas distribution should be very even is mostly based on the modified Deutsch expression such as equations (5.15), (5.16) and (5.18).
In the first field, where most coarse particles are found, there is no sense
in trying to raise the heavy dust fraction to the upper region, as it finally
has to be accumulated in the hoppers. This supports a distribution with
velocities above average at the lower part of the field. From the second and
further fields, it is the finer particles which are to be caught, suggesting the
gas distribution should be fairly even until the exit from the last field.
Particles re-entrained in the lower part of the last field might be swept out
in the clean gas duct, and it is recommended that the bottom outlet
velocities are below average. Outlet transitions fitted with gas distribution
screens make it possible to adjust the vertical velocity profile, which is
especially important in cases where the precipitator is designed with a
'bottom-type' outlet.
Figure 5.22 shows field-recorded vertical velocity profiles with an air load
on a two field precipitator. The precipitator had a penetration about 0.07%
and almost no rapping peaks for an inlet loading of some 40 glm 3, so even
though the distribution is far from even, it is believed that this distribution
profile has a positive effect on the efficiency [40].
Arthur G. Rein [41J uses a two-dimensional model, dividing the length
and height of a precipitator into rectangles, and looks at the net mass
balance of each rectangle. Dust enters the rectangular cell from the upstream
element, partly as re-entrained dust from upstream cells at a higher level,
and leaves the cell, either because it has not been precipitated or because it
has been re-entrained. The mass balance for all cells is collated and the
system of equations solved using a computer. Rein states that 'controlled
non-uniform gas distribution at the inlet and outlet faces can be used to
improve the performance of the precipitator', and recommends that more
than half the total gas flow should be supplied to the lower half of the
treatment zone and should be exhausted from the upper half of the
treatment zone.

139

MODEL TESTING

100%
0 100% 0
>--------<
34%
~

a/V 37%

45%

100%
0
0---------0
69%
!

Figure 5.22 Vertical velocity profiles from air load field testing. From [40].

5.5 Model testing


The reason for the demonstration of a proper gas distribution in a scale
model stems from the 1960s, where precipitators often did not fulfil their
guarantee. Thus modelling became a demand, and stricter claims of gas
distribution quality were introduced, e.g. IGCI EP-7.
Model studies are performed by vendors or by specialized laboratories,
either when new transition pieces are to be designed or on request by the
client or specification. In the latter case the whole duct system will normally
be included. Such studies include first of all the gas distribution and system
pressure drop, but also an assessment of the risk of dust drop-out in the raw
gas ducting.
The model specification, measurement methods and evaluation of results
are normally part of the client's tender document. Sometimes, the clients'
consultant tightens the demands on the distribution beyond that reasonable,
with respect to precipitator efficiency and cost, and sometimes beyond the
physically possible. An extremely even distribution needs extended laboratory measurements and time plus a large number of internal flow baffie
correcting devices.

140

AERODYNAMIC FACTORS AFFECTING PERFORMANCE

The fact that it is possible to predetermine the gas distribution in a


full-scale precipitator by performing a test on a model of scale 1: 16 to 1: 8
is explained by physical-mathematical dimensional analysis. Model testing
demands geometrical, kinematic and dynamic similarity. In other words, the
ratio between any linear dimension in the model and full-scale must be the
same and the ratio between velocities in the model and full-scale at
corresponding points must be constant. Finally the ratio between forces
acting on the fluid in the model and full-scale must be identical; this is
normally automatically fulfilled if there is geometrical and kinematic similarity. The physics of the flow is described by three Navier-Stokes equations, one per coordinate in space and the equation of continuity. Making
the equations dimensionless, using characteristic mass, time and length and
combinations thereof, it can be proven that the solution is universal,
provided the dimensionless numbers appearing in the equations are equal
in the model and full-scale. This includes the Reynolds number, Re, the
Euler number, Eu, and the Mach number, Ma. The Euler number is
automatically fulfilled as it depends on Re.
Re is the ratio between flow inertia and friction, while Ma is the ratio
between the gas velocity and the speed of sound, i.e.
Re

= vdlv

and

Ma

= via = vIJ(KRT)

where v is velocity (m/s), d is characteristic length (m), v is kinematic


viscosity (m 2 /s), a is speed of sound (m/s), K is adiabatic power, R is gas
constant (J/kg/K) and T is temperature (K).
If the model is ten times smaller than full-scale and V FS = 2v M , identical
Re gives vM = 5v FS ' For V FS = 1 mis, the model velocity becomes VM = 5 m/s.
In the duct, 20 mls full-scale velocity gives 100 mls model velocity, or
Ma = 0.29, very close to the compressibility limit.
The net power, necessary to drive the air, is W = dpV '" tjJl/2pv 3d2,
where V is volume flow rate. Provided the pressure coefficient \f is the same
in model and full-scale (fully turbulent flow) the powers are found to be:
(5.19)
(5.20)
The ratio between WM and His is:
(5.21)
With a density ratio of 1.4, a velocity ratio of 5 and scale factor of 0.10, the
fan power needed in the model is 1.75 times full-scale power. This high
power is prohibitive, hence praxis is to use the same axial velocity in the
model as full-scale. The Reynolds number (Re) in the ducting of the model
will still be in the fully turbulent range, while Re in the precipitator model

MODEL TESTING

141

will be in the low turbulent range, about 104 . This reduces the Mach number
to approx. 0.05 and the model fan power to 1.4% of that of full-scale.
IGCI EP-7 recommends to double the model duct width in order to
double the Reynolds number, but this is hardly the proper solution at the
inlet of the first field, because the straightening effect of the collector plates
is of greater importance to the crosswise velocity distribution than is the
correct Re based upon duct width.
The Reynolds number based upon lengths from the leading edge of
internals, such as guide vanes, dividing walls and kicker plates in the model,
are often in the laminar regime and their effect on the flow will therefore be
underestimated.
Velocities in the model and full-scale are normally recorded with hotwire, hot-film or vane-type anemometers. None of these probes, in their
standard design, can indicate the flow direction. As the end result should be
an even distribution, this is not of great importance but, where necessary,
flow direction can be visualized using smoke, provided the smoke is injected
with the same velocity as the local flow, or visualized using wool tufts.
Knowledge of the velocity vector direction can help the troubleshooter find
the best remedy for correcting the gas distribution.
The vane anemometer has the disadvantage that velocities below 0.30.4 mls cannot be recorded, and a jet only partly filling the vane crosssection is registered with its full value. Apart from this, the vane probe is a
reliable and rugged instrument.

Figure 5.23 Photograph of model in the laboratory of Lodge Sturtevant.

142

AERODYNAMIC FACTORS AFFECTING PERFORMANCE

Some laboratories use one person to position the probe in the measurement cross-section and one to read the instrument. Other laboratories use
slides with electric motors and electronic control and only one person to
control the experiments. Data recording is done by pen or paper or by
computerized data acquisition, the latter being capable of calculating and
plotting the statistics and graphics simultaneously.
Figure 5.23 shows a photograph of a model in the Lodge Sturtevant
laboratory. IGCI EP-7 requires the gas distribution to be mapped only at
the inlet of the first field and the outlet of the last field, which does not reveal
possible maldistributions in other internal cross-sections. Therefore, it is
recommended to map at least one extra cross-section, e.g. the inlet of the
second field, where the hopper tends to draw the gases downwards creating
a bottom peak on the vertical velocity profile. This tendency is less
pronounced, with precipitators having a trough-type hopper such as those
used in lime reburning kiln precipitators.
Results are normally presented in a report which should include sketches
of the inlet and outlet transitions, plus inlet and outlet screen geometries and
guide vane arrangements. Velocity profiles, plane and isometric, iso-velocity
plots, velocity vector plots and statistics are also normally included.

5.6

Computational fluid dynamics

With the appearance of powerful and fast computers, new possibilities for
replacing time-consuming model testing and field testing have arisen. This
involves solving the differential equations describing fluid motion, using
either a finite volume or sometimes, but more rarely, a finite element
method. The methods are named computational fluid dynamics or simply
CFD.
Physically, the conservation of mass and Newton's second law are
applied on the fluid, mathematically expressed as the equation of continuity
and the Navier-Stokes equations in two or three dimensions. Even with
today's fastest supercomputers, it would be impossible to divide the calculation domain in parts small enough to describe all the details in the flow
field. Solving the equations directly is a discipline called direct numerical
simulation, which is still restricted to very limited Reynolds numbers and
small geometries. With the smallest eddies (of size, as per, the Kolmogorov
length scale, (v 3 . b/U 3 )o.25, v being the kinematic viscosity ~ 40 10- 6 m 2 /s,
b a typical shear layer thickness ~ 10- 1 m and U the bulk velocity ~ 1 m/s)
the length scale is of order 0.3 mm. This means that a precipitator of
dimensions 15 x 15 x 20 m 3 should have a mesh number of order of one
hundred million millions, far beyond the capacity of computers in the 1990s
to tackle direct numerical simulation.

COMPUTATIONAL FLUID DYNAMICS

143

Instead the turbulent variables are taken as average values plus fluctuating parts, e.g. U = U + u', u being the instantaneous x-velocity, U being the
time average and u' the time-dependent fluctuation (time average of u' == 0).
Introducing these variables into the Navier-Stokes equation and time
averaging leads to:
pDUaviDt

where the stress tensor


Lij

Lij

= pg - VPav + V' Lij

(5.22)

is:

J-l(auJax j

+ au)axJ

- p(u;u~)av

(5.23)

The first term is the laminar stress, and the second term the turbulent stress.
The turbulent part of the stress tensor is either found by solving the
so-called Reynolds' stress equations or simply by expressing it using average
flow values and the so-called Boussinesq assumption (J. Boussinesq, Paris,
1877): L t = J-ltaUlay, where J-lt is the eddy (or turbulent) viscosity, U is an
average velocity and y is a coordinate perpendicular to vector U.
J-l t is determined using either Prandtl's mixing length theory, J-lt ==
pI2Iau/ayl, I being the so-called mixing length, or using the k-s model: J-lt ==
CIlP/s, CIl being a constant or coefficient, k being the turbulent kinetic
energy, k = i(U'2 + V'2 + W,2), and s being the turbulent dissipation, s =
k 3 / 2 / L d; where Ld is a length scale for the dissipating eddies.
The more complex the turbulence model is, the more differential equations must be introduced, which can only be solved by closing the system
of equations by introducing algebraic expressions based upon empiricism.
After the break-through of IBM's computers and software in the 1950s,
a great impetus was given to the development of physical and numerical
methods for solving flow dynamic problems. At the end of the 1960s more
research groups contributed to the development, among others researchers
from the Mechanical Engineering Department at Imperial College in
London [42,43] looked at the solution of internal flow problems, whereas
the aerospace companies emphasized external flows for calculation of lift
and drag.
The most common methods of today are based on transformation of
the differential equations for conservation of energy, mass and momentum
to difference equations, solved by integration after applying the given
boundary conditions. Originally in two dimensions, the equations can be
transformed and solved using streamfunction 'I' (pu = a'l'/ay, pv =
- a'l' lax) and vorticity w (w z = av/ax - au/ax), thus eliminating pressure as
a variable. Later on, with more efficient computers, the equations were
formulated and solved in the primitive variables, pressure and velocity, in
three dimensions, thus avoiding the problem of defining proper boundary
conditions for a vorticity having steep gradients close to the walls.
As velocity gradients are also steep at the walls, special attention is paid
to the description of analytical near-wall variations, the so-called wall

144

AERODYNAMIC FACTORS AFFECTING PERFORMANCE

functions. The difference equations are often solved in Cartesian or orthogonal grids, equidistant or almost equidistant, using 'upwind differences'
taking into consideration the local direction of flow. One problem with the
grids is the appearance of 'numerical diffusion', which is maximal if the flow
vector has a 45 angle with the grid axes. In recent years advanced grids
have been developed, e.g. 'adaptive grids', where the mesh size changes
according to the gradients of the variables, decreasing in size where
gradients are steep. Another approach is the use of 'multigrids', where
calculation shifts between a fine and a coarse mesh, thus smoothing out
short- and long-wave variations.
The difference equations are solved indirectly (integrated) by iteration,
sometimes several hundreds, and numerical stability is achieved by proper
successive under- or overrelaxation. The criterion for having found the final
solution normally is the mass continuity, and the calculation stops when the
maximum residue is less than, say, 10- 4 .
The 'solver procedure' is a chapter in itself using various numerical
principles, continually updating the values in the domain as soon as new
values have been found. A number of fast and efficient solvers are commercially available today. (See Patankar [44] for more details on the numerical
principles.)
With more complicated geometries, such as precipitators including fancy
transition pieces, guide vanes and screens, grid designing is difficult. A
simple method is to use a parallelepipedal domain and block out all the
outer elements until an approximate contour is achieved, leaving the
geometric domain boundary as a step surface. This means that there is a
limit to how precise the solution close to the walls can be, even though the
internal flow is only slightly influenced. In fact, it is a grid generator,
especially easy to use, which is the main issue for the operator. The easier
the grid generation, the more different configurations can be calculated in a
reasonable time, presuming, of course, that the solver is effective and fast.
Less than 2 h per new contour and less than 1 h per modification would be
ideal.
The real progress in mesh generation is a mesh that can be fitted without
any consideration of the interfacing, at least as seen from the point of view
of the user. Up to now mesh structures have had to fit where ducts and
transition pieces, or transition pieces and precipitator housing meet, but new
grid systems are appearing. Thus, it is possible to have an essentially polar
mesh of a circular configuration corresponding with a rectangular mesh of
a box, without any concern about mesh fitting at the plane where they meet.
This block type is well suited for duct and precipitator analysis with respect
to both gas distribution and pressure drop determination. Figure 5.24 shows
this type of mesh from the code Star-CD.
The more advanced the mesh, the more variables to be treated and the
stricter the convergence criteria, the more calculation time is needed to find

COMPUTATIONAL FLUID DYNAMICS

145

Figure 5.24 Example of arbitrary interfacing of the mesh of a cylinder and the mesh of a box.
Star-CD.

a correct solution. The need for space in the memory and on the disc is
another 'eye of a needle', the cell number increasing roughly with the
product of the number of variables and the reciprocal of the mesh size to
the third power. While details such as guide vanes, kicker plates and ladder
vanes have to be modelled separately, there is a possibility to simulate
perforated screens using some sort of porosity model such as a Darcy
porosity. Screens with evenly distributed guide vanes might be modelled as
a whole, but this calls for special routines, not normally commercially
available. In the equations the so-called 'source' terms can be modified in
order to reflect the influence of the gas distribution screens. (See Patankar
[44].) Screens can also be modelled by 'blocking' out cells, but this
procedure demands a very fine mesh with many nodes. The flow before a
perforated screen is often recirculating, unless small screen guide vanes are
used. Such flows can be identified using CFD, revealing velocity vectors
parallel to the screen, making it impossible to improve the distribution by
modifying the screen by increasing or decreasing the open area alone.
Programs which cannot treat small thin oblique surfaces in an effective and
easy way should therefore not be used. Like the use of porosity for
simulating a perforated screen, it is possible to design a subroutine with a
screen combined with distributed guide vanes and use it as a black box when
the calculation is done in the screen domain. This would reduce the need for
a complex mesh with a large number of grid points.

146

AERODYNAMIC FACTORS AFFECTING PERFORMANCE

Next to the mesh generation, postprocessing, i.e. the presentation of the


results, is the most important factor for the user. Commercial systems apply
velocity vector plots and iso-curves for pressure and velocity, normally
supported by colours. Isometrical presentations of results are often used,
these being more fanciful than informative and less applicable than, e.g., 2-D
velocity profiles. Figure 5.25 shows part of a precipitator with guide vanes
in a manifold, and Figure 5.26 shows the streak lines of zero-mass particles.
Unfortunately, it is not possible here to show vector plots or velocity
profiles, because they are presented with relatively faint colours, difficult to
reproduce in black and white. For further details see references [45J and
[46].
A serious user of CFD must have access to the necessary hardware and
software, invest in training and have at least one person engaged full time
in calculations. Likewise, a company having its own CFD facilities is
necessary in cases where external specialist firms are engaged to do some or
all of the computational work, because it will make it more efficient to
arrange test programs and to interpret the results.

Sij}
PROSTAR 2.21
3 Apr 95
VIEW
-1.000
1.000
-1.000
ANGLE
0.000
DISTANCE
22189.44
CENTRE
16275.00
-3350.00
-5025.00
EHIDDEN PLOT
LIGHT SOURCE
-1.000
1.200
-1.000
y

Case 1: With porosity


FLS Miljoe
Demonstration precipitator geometry

x~z

Preliminary results

Figure 5.25 Modelling of guide vanes in the manifold above a precipitator inlet. Star-CD.

147

COMPUTATIONAL FLUID DYNAMICS

SD
PROSTAR 2.21
3 Apr 95
VIEW
-1.000
1.000
-1.000
ANGLE

0.000
DISTANCE

22189.44

CENTRE
16275.00
-3350.00
-5025.00
EHIDDEN PLOT

case t Wi1h POroSIty

FLS Miljoe
Demonstration Praci1ll1ator Geometry

~z

PrelimInary Results

Figure 5.26 Streak lines of massless particles in manifold and inlet to a precipitator. Star-CD.

Suitable hardware work stations are IBM, HP, Digital, Vax, etc., whose
prices depend on facilities. The system also requires a proper colour printer
or plotter for graphical output. If the geometry is taken from a CAD system,
it should be possible to transfer the geometry files to the mesh generator of
the CFD system.
Software can be developed by the users themselves, using physical,
mathematical and numerical backgrounds available from books. Most
organizations, however, prefer buying a commercial package, because program development is extremely expensive and upgrading and introduction
of new theories and facilities can hardly be maintained or justified by a small
group of engineers occupied with other disciplines outside CFD analysis.
Commercial software packages are numerous and without favouring a
particular code, names such as Flow3D, Fluent, Kameleon, Phoenics, Fidap,
Star-CD and Viscous exist. Some systems can be supplied as user property,
others can be hired on a yearly basis; some codes are open, i.e. there is access
to the source code, others are 'closed' delivered as executable programs only.
Upgrading, troubleshooting and hot line assistance are normally offered by
the supplier.

148

AERODYNAMIC FACTORS AFFECTING PERFORMANCE

Companies specializing in CFD offer their services, sometimes both CFD


and model testing, and often institutes, e.g. aeronautical laboratories,
technical universities and technological centres worldwide, perform calculations if required.
The advantages of numerical modelling are that (1) there is no need for
violating the model laws, (2) the model will not have to be physically
constructed and (3) the result is presented in a relatively short time. Yet, it
must not be forgotten that the flair and know-how of the skillful fluid
mechanics engineer cannot be replaced by a code, however fanciful it might
be. While using the program for solving flow problems, the changing of the
outer contour, changing design, position of guide vanes and screen porosity
will give the operator the opportunity of gaining experience in a relatively
short time. But practical experience and know-how can hardly be replaced
by keyboard and screen alone. The solution from a CFD run must not be
accepted uncritically: first of all because the solution to the difference
equations is not necessarily the correct solution to the differential equations;
secondly, because of possible limited applicability of the turbulence model
and thirdly because of imprecise boundary conditions. Finally, the numerical solver has a limited precision and the calculation might not converge
properly. Sometimes fluctuating residuals from the solution of the timeaveraged equations indicate that there is no stationary solution to the
problem. In order to gain confidence in a given CFD code, results should
be field tested on existing installations.
5.7

Field testing

The full-scale demonstration of the velocity distribution, whether model


tested, CFD, or not, can be performed at clients' or vendors' request before
start-up. Clients' interest is confirmation of a proper distribution while
vendors interest is in building-up and maintaining experience.
In most European designs of precipitators there is normally adequate
access to the interspace between fields and before the first field and after the
last field, so it is possible either to climb the emission frame with the velocity
probe or to fit some sort of 'monkey' to the collecting plate edges to carry
out the necessary velocity traverses.
In precipitators of the US type, access is normally more difficult and
sometimes the probe is lowered into the precipitator from the top of the
electrode system or from the roof.
As with model testing, hot-wire, hot-film or vane anemometers are used.
In cases where the fan power is restricted to operating with cold air, the
average velocity might be low. This speaks for using hot-wire or hot-film
anemometers which have higher sensitivity at low velocity, but generally one
should beware of low velocities as precision falls.

DUST BUILD-UP AND WEAR

149

Two operators are needed inside the precipitator and at least one outside.
In the duct a velocity probe, vane anemometer or pitot-static tube, is fitted
connected to a recorder for verification of stable fan operation. This velocity
signal can, in the case of fan or changing flow problems, to some extent be
used for correcting the measured distribution.
Inside the precipitator sufficient light must be available and an intercom,
for communication makes the job easier. Special inpection doors fitted with
acrylic glass, because it is thus possible to find the way out in the event of
a power failure, and to have radio communication with external personnel
and the control room, adds to safety.
The time needed for traversing a cross-section is of the order of 1 h;
however, large cross-sections, mapped by use of a climbing 'monkey', might
need 2 h or more. To this, one must add the time for rigging up and
dismantling. If time is scarce, simultaneous mapping in more than one
cross-section is possible. If modifications are found necessary and the
hardware has been prepared, e.g. different screen open areas and additional
guide vanes, the measurement procedure after modification must be repeated until a satisfactory result is achieved.
In some cases it is necessary to return to the laboratory and resume the
model testing or further numerical calculations in order to find a solution.
This is a normal procedure if time in the field is restricted. Later on, if
measurements have to be taken after the plant has been running, and as the
precipitator will be dirty, special precautions should be taken to protect the
personnel by using respiratory, eye and ear protection.
5.8

Dust build-up and wear

Internal gas distribution devices are subject to dust build-up and wear.
Guide vanes should be placed and designed correctly, so that build-up is
minimized, and if the risk cannot be eliminated then rapping, vibration or
acoustic horns for cleaning should be considered. This includes vanes on, or
close to, screens. Horizontal vane surfaces should, in principle, be avoided;
a better solution is oblique surfaces, if necessary double-angled, e.g. 45
upwards and 45 downwards. Furthermore, screen vanes should be integrated with the screen in order to avoid dust build-up on the vane, due to
dust impacting on solid screen areas; such screens are often supplied with
vibrators or hammers. Acoustic horns have also demonstrated their capability of keeping screens clean on many precipitators.
It must be emphasized that the designer should ensure that gas distribution demands are fulfilled, not only in the model or during field testing
under air load, but also during normal operation.
Fly ash dust from cement clinker coolers and dust from some metallurgical processes consist normally of very sharp, abrasive grains, so if the gas

150

AERODYNAMIC FACTORS AFFECTING PERFORMANCE

velocity in the raw gas duct and at the entrance to the inlet transition piece
is too high, there is a risk that wear of guide vanes and of the first
distribution screen will occur. It must be remembered that coarse particles
have much higher inertia than the gas; therefore, deceleration of the gas
velocity right before an inlet flange does not necessarily reduce the velocity
of the sharp particles. In some cases, wear is so severe that vanes and screens
disintegrate so rapidly that extra plant outages, between regular overhauls,
are needed in order to meet emission demands or in order to remove eroded
pieces blocking the dust conveying system. If there are no alternative layout
arrangements to reduce the risk of severe wear, then vanes and screens must
be made out of thick or wear-resistant steel.

References
1. Deutsch, W. (1922) Bewegung und Ladung der Elektrizitatstrager im Zylinderkondensator.
Ann. Phys. (Leipzig), (4)68, 335-44.
2. Cochet, R. (1961) Lois de charge des fines particules (submicroniques). Etudes
theoriques-Controles recents spectre de particules. Colloque International-La Physique
des Forces Electrostatiques et Leurs Applications. Centre National de la Recherche
Scientifique, Paris.
3. McDonald, J.R., Smith W.B. and Spencer III, H.W. (1977) A mathematical model for
calculating electrical conditions in wire-duct electrostatic precipitation devices. J. Appl.
Phys., 48(6), 2231-43.
4. Mayer-Schwinning, G. (1987) Walther Deutsch, a pioneer in electrostatic precipitation. 3rd
ICESP, Abano-Padova, Padova University, Italy, pp. 3-11.
5. Deutsch, W. (1931) Ann. Phys. (Leipzig), 5,251-63.
6. White, H.J. (1963) Industrial Electrostatic Precipitation. Addison-Wesley, MA, pp. 238-93.
7. Robinson, M. (1968) Turbulence in electrostatic precipitators ... a review of the research
literature. Miner. Process., May, pp. 13-7.
8. Friedlander, S.K. (1959) Principles of gas-solid separation in dry systems. Chem. Eng.
Progr. Symp., Ser. 25, 55, 135-49.
9. Cooperman, P. (1965) Eddy diffusion and particle erosion in electrostatic precipitation.
Toronto, APCA 58th Annual Meeting, Paper 65-132, APCA, Pittsburgh, USA.
lO. Robinson, M. (1967) A modified Deutsch-equation for electrostatic precipitation. Atmos.
Environ., 1, 193-204.
11. Crowe, C.T. and Stock, D.E. (1974) The effect of electrodynamic secondary flow on the
performance of an electrostatic precipitator. Heat Transfer Fluid Mech. Inst., 12-14 June
1974, 24th Proceedings, pp. 254-65.
12. Bernstein, S. and Crowe, C.T. (1979) Interaction between electrostatics and fluid dynamics
in electrostatic precipitators. 2nd Symposium on the Transfer and Utilization of Particulate
Control Technology, July 23-27, Denver, Colorado, EPA 600j9-80-039b, Vol. III, pp. 12545.
13. Gross, H. (1979) Zur wirkung der Turbulenz in Elektroabscheidern. Staub 39,197-202.
14. Leonard, G.L., Mitchner, M. and Self, S.A. (1980) Particle transport in electrostatic
precipitators. Atmos. Environ., 14, 1289-99.
15. Leonard, G.L., Mitchner, M. and Self, S. (1981) Precipitation from turbulent flows. 1st
ICESP, Monterey, October, pp. 208-56, EPA, Pittsburgh, USA.
16. Yamamoto, Y., Nakamura, S. and Velkoff, H.R. (1980) Numerical study of secondary flow
interaction in an electrostatic precipitator. Innovative Numerical Analysis for Engineering
Science, University Press of Virginia, pp.3-12.
17. Yamamoto, T. and Velkoff, H.R. (1981) Electrohydrodynamics in an electrostatic precipitator. J. Fluid Mech., 108,1-18.

REFERENCES

151

18. Thomsen, H.P., Larsen, P.S., Christensen, E.M. and Christiansen, 1.V. (1982) Velocity and
turbulence fields in a negative corona wire-plate precipitator. 4th Symposium on the
Transfer and Utilization of Particulate Control Technology, October 1982, Houston, Texas.
(Vol. II, EPA-600/9-025b, Nov. 1984.)
19. Shaugnessy, E.J., Davidson, 1.H. and Hay, J.e. (1984) The fluid dynamics of electrostatic
precipitators: effects of electrode geometry. 5th Symposium on the Transfer and Utilization
of Particulate Control Technology, August 1984, Kansas City, Missouri, Vol. 2, paper No.
28 (EPRI CS 4404, Feb. 1986).
20. Larsen, P.S. and S0rensen, S.K. (1984) Effect of secondary flows and turbulence on
electrostatic precipitator efficiency. Atmos. Environ., 18(10), 1963-7.
21. S0rensen, J.N., Larsen, P.S. and Zamany, J. (1991) Experimental and numerical analysis
for flows in negative corona precipitator. 8th Symposium on Turbulent Shear Flows, Sept.
9-11, Techn. Univ. of Munich, Germany.
22. Efficiency Measurements on a Laboratory Precipitator. FLS milj0 a/s, Internal Report in
Danish.
23. Leonard, G.L., Mitchner, M. and Self, S. (1981) Precipitation from turbulent flows. 1st
ICESP, Monterey, October, EPA, Pittsburgh, USA.
24. Lind, L. and Bojsen, E. (1993) Experience with baffle-free collecting plates in an electrostatic precipitator. 10th EP RI Symposium on the Transfer and Utilization of Particulate
Control Technology, and 5th ICESP, Washington, April 3-8, pp. 36.1-14, EPRI TR
103048, Vol. 2.
25. Self, S.A., Kihm, K.D. and Mitchner, M. (1987) Precipitator performance improvement
through turbulence control. 3rd ICESP, Abano-Padova, Italy, pp.443-80.
26. Self, S.A., Mitchner, M., Choi, D.-H., Kihm, K.-D. and Leach, R. (1986) Finite-diffusivity
effects in single-stage precipitators-theory and experiments. 6th Symposium on the Transfer
and Utilization of Particulate Control Technology, February, New Orleans, Mississippi,
EPRI CS 4918, Vol. 2, pp. 28.1-30.
27. Larsen, P.S., Christiansen, J.V. and Christensen, E.M. (1987) Secondary flows and turbulence in a pulsed, negative-corona, barbed-wire precipitator. 3rd ICESP, Abano-Padova,
Padova University, Italy, pp.481-93.
28. Davidson, 1.H. and McKinney, Peter J. (1989) EHD flow visualization in the wire-plate
and barbed plate electrostatic precipitator. Proc. IEEE, pp.2118-25.
29. Larsen, P.S. (1986) Secondary Flows in Negative Corona Precipitator. DCAMM Report No.
337, October, Technical University of Denmark.
30. Zamany,1. (1992) Modeling of Particle Transport in Commercial Electrostatic Precipitators.
Ph.D. Thesis, ATV EF316, Copenhagen, September, ISBN 87-984457-0-7.
31. Dalmon, J. and Lowe, H.J. (1961) Experimental Investigations into the Performance of
Electrostatic Precipitators for P.F. Power Stations. Colloque International- La Physique
des Forces Electrostatiques et Leurs Applications. Centre National de la Recherche
Scientifique, Paris.
32. Methods for producing uniform gas flow in processing equipment. Br. Chern. Eng., July,
359-63.
33. Gas Flow Model Studies (1981) Publication No. EP-7, International Gas Cleaning Institute
Inc., Revision 4, October or Institute of Clean Air Companies, Pub. No. EP-7, 1993.
34. Idel'chik, I.E. and Aleksandrov, V.P. (1974) The effect of the nonuniformity of the gas flow
on the efficiency of electrostatic precipitators. Teploenergetika, 21(8), 60-2, 85- 7.
35. Gooch, J.P., McDonald, J.R. and Oglesby Jr, S. (1975) A Mathematical Model of
Electrostatic Precipitation. EPA-650/2-75-037, April.
36. Rosby, S.-O. (1974) Precipitator Model Studies and Scaling-Up Experiences. Fliikt, AB
Svenska Fliiktfabriken, Issue No.2, Telub AB Teknikinformation, Viixjo, 12 pp.
37. Gas Distribution Influence on Precipitator Efficiency. FLS milj0 a/s, internal report in
Danish.
38. H0egh Petersen, H. (1990) A precipitator sizing formula 4th ICESP, Beijing, September.
39. H. H0egh Petersen and L. Lind, private communication, 1986.
40. Lind, L. (1986) Influence of gas distribution on precipitator performance. 6th Symposium
on the Transfer and Utilization of Particulate Control Technology, New Orleans, February,
EPRI CS 4918, Nov. 1986, Vol. 2, pp. 33.1-15.
41. Hein, A.G. (1989) Dust reentrainment, gas distribution and electrostatic precipitator
performance. JAPe A, 39(5).

152

AERODYNAMIC FACTORS AFFECTING PERFORMANCE

42. Gosman, A.D., Pun, W.M., Runchal, A.K., Spalding, D.B. and Wolfshtein, M. (1969) Heat
and Mass Transfer in Recirculating Flows. Academic Press, 338 pp.
43. Wolfhstein, M. (1969) Convection Processes in Steady Two-Dimensional Separated Flows.
Imperial College of Science and Technology, Report EF/R/G/l, Thesis, January.
44. Patankar, S.V. (1980) Numerical Heat Transfer and Fluid Flow. Hemisher Publishing
Corporation, New York.
45. Ettema, R. (1994) Electrostatic precipitator performance improvement through numerical
simulation. World Cement, April, pp.63-6.
46. Schwab, M.J. and Johnson, R.W. (1994) Numerical design method for improving gas
distribution within electrostatic precipitators. American Power Conference, 56th Annual
Meeting, Chicago, April 25-27, 7 pp.

The physical and chemical properties of particles


and their effect on performance
K. PORLE AND K. R. PARKER

6.1

Particle size and shape

In the air pollution control industry, particles to be collected fall into one
or more of the following general categories, i.e. dust, fume, ultrafine fume or
mist.
Dust particles usually arise as a result of mechanical disintegration or
communition of large lumps into smaller particles by grinding, crushing, etc.
The particles are irregular in shape and their size refers to some average
dimension and spans the range from above 200 Jl.m down to about 1 Jl.m
(1 Jl.m = 1 x 1O-6 m). For inertial type collector technology (see chapter 1)
the particle size is usually referred to the equivalent Stoke's diameter or free
falling velocity. Any particles produced during communition having a size
greater than 200 Jl.m normally settle out so quickly that there is no real
difficulty in their collection.
Pulverised fuel ash (PF A) is worth special mention, as worldwide, the
largest application of electrostatic precipitators is for the collection of fly ash
from coal fired utility type power generation facilities. The ash (PF A)
carried over from the combustion zone comprises a complex mixture of
material, predominantly silica/alumina based compounds, having a median
particle size of around 15 Jl.m. Using the above categories, this would be
considered as 'dust', but during combustion, the coal particles are exposed
to high temperatures e.g. 1200 C and this results in some ash material being
volatilised. This subsequently condenses in the cooler regions of the boiler
to produce submicron particles, i.e. less than 1 Jl.m, amounting to some 2%
of the total ash. Hence PF A is a mixture of both dust and submicron size
fume particles.
Fume is typically conferred to solid particles formed by either sublimation of solid phase material or condensation of a vapour phase condition,
usually as a result of exposure to high temperature such as in combustion
or smelting processes. As most of these particles have been, at one stage, in
a liquid/molten phase, surface tension effects tend to produce spherical
particles, usually in a small size band. A typical size range would be 0.1 to
1.0 Jl.m. These particles do not have a measurable settling velocity, but
exhibit strong Brownian motion.

154

PHYSICAL AND CHEMICAL PROPERTIES OF PARTICLES

Ultrafine fume is reserved for the smallest possible sizes, usually less than
0.1 ,um; the particles are normally spherical in shape, e.g. tobacco smoke, or
sometimes irregular, such as carbon black. With sizes less than O.I,um these
particles are approaching molecular dimensions and exhibit very strong
Brownian motion.
This Brownian motion is very necessary in the diffusion charging process
for electrostatic precipitators, as discussed in chapter 3. Generally their
spherical shape enables optical or laser-type measurements to be carried out
to evaluate their size.
Mists or fogs are formed by the condensation of a vapour phase material
on suitable nuclei to produce a suspension of small liquid droplets. Although atmospheric fogs can have fairly large droplet sizing, for the purpose
of precipitation duties, only those mists having a particle size less than l,um
will be considered. Being in a liquid phase, a mist particle is spherical and
the smallest particles exhibit strong Brownian motion, which is essential to
their effective charging.
In the pollution control field, particle suspensions are further characterised by their physical and chemical properties; for example, their size,
structure, surface area, electrical resistivity, chemical reactivity and composition, cohesivity, ability to absorb electric charge and propensity for
erosion, are some of the properties which have an effect on electrostatic
precipitator design and/or performance. Those characteristics having the
greatest effect on precipitation will be described in some detail, particularly
as to how they impact on performance.
6.1.1

Particle sizing

A precipitator would see at its inlet a dispersoid of particulates being carried


by a gas stream. This stream behaves like a fluid, splashes like a liquid,
supports wave motion and can be pumped through conduits. One of the
critical aspects of the dispersoid make-up is that of particle size. Particles of
dust are generally made up of a myriad of different shapes and sizes and,
unlike fume, are rarely spherical in shape, although some combustion
processes can produce cenospheres, which are hollow and result from the
outgassing of small pieces of carbon suspended in molten ash.
One of the difficulties facing the air pollution control engineer is to define
what is meant by particle size and, then, how the size will impact on a
specific method of collection. If the particle is truly spherical the size
measurement is simply the diameter, but if the shape is irregular, then the
definition and measurement becomes more complex.
The majority of dusts met in practice consist of a whole spectrum of sizes,
rather than a single size, as are sometimes laboratory produced for a specific
mono-dispersed aerosol. The spectrum comprises a continuous distribution
of sizes, usually on a log/normal form of distribution. This enables the sizing

155

PARTICLE SIZE AND SHAPE

7
6
5

'"0
)(

g4
~

>u

5i

::J
0-

!!?

II..

c
al
'6
CD

,
C'
al:

CD,
~:

10
15
Particle size (microns)

20

25

Figure 6.1 Typical particle size distribution of fly ash.

to be treated on a statistical analytical basis; for this, the size frequency


distribution function yo(x) and the cumulative distribution function y(x) are:
Yo(x) dx = the proportion of particles in the size interval x to x

and
y(x)

f:

y(x) dx

+ dx.

the proportion of particles in the size range 0 to x.

A typical example of a frequency distribution function is illustrated in


Figure 6.1. In this, three useful parameters are indicated, i.e. the mean, the
median and mode points.
The mean particle size is defined as the size where 50% of the mass lies
above and below that size, usually referred to as the d so size.
The mode point or size is that with the highest frequency or concentration. This is useful in the case of electrostatic precipitation, where one
considers whether or not corona suppression or space charge effects from
submicron particles are likely to be significant.
The median size is where 50% of the number of particles lie above and
below that value. This is important where one has to consider the fraction
of particles below a certain value, such as in the proposed US PM 10
regulations (PMlO particles < 10 11m).

156

PHYSICAL AND CHEMICAL PROPERTIES OF PARTICLES

~
.9 30
)(

Q)

i:D 20
N

'iii

.E
III

~ 10

'E

8.

eft
5

10

15
20
25
Particle" size (microns)

30

Figure 6.2 Typical particle size histogram.

Methods of determining the particle size of a particular dust do not


produce a continuous distribution curve as indicated in Figure 6.1. Most
evaluate the number or mass of particles within a specific size band or range,
for example in a sedimentation analysis. This produces a histogram-type
curve as shown in Figure 6.2. The mid-points of each specific size range are
then taken and a cumulative curve is produced as in Figure 6.3. If this is
plotted as a log probability curve, a straight line is usually derived (Figure
6.4); this however can deviate somewhat at both upper and lower ends. This
deviation is not altogether surprising, since in practice a single large particle

100

Q)

80
60

~
E
i3 40
:g
>-

20

10
15
20
Particle size (microns)

25

Figure 6.3 Cumulative particle sizing.

30

PARTICLE SIZE AND SHAPE

157

100

10
50
90
99
Cumulative percent by weight

Figure 6.4 Log-probability curve for particle distribution.

has a big impact on the mass distribution at the coarse end and fine particles
may arise not from communition of the feed material, but because of
volatilisation/condensation, and so affect the fine end.
One of the problems with many analytical sizing methods is that of
obtaining a truly representative sample for analysis, i.e. as the precipitator
or downstream device really sees the dispersoid. Even if an isokinetic sample
is obtained from sampling, most analytical methods redisperse the sample,
so the sizing could measure agglomerates arising from the sampling which
are not broken during redispersion, or true agglomerates in the flue which
are. Materials which are volatilised and recondense in the system can be
found adhering to larger particles upon which they have impacted or used
as condensation nuclei; hence the question, does one consider the agglomerate mass as a whole or a big particle plus a number of separate small
ones? Figures 6.5 and 6.6 [1] show both a large particle with small
fume-size particles attached and an agglomerate of ultrafine particles. For
mass collection, the weight of the small fine particles has no significant effect
on the overall efficiency, but, where a specific efficiency for a particular
material or size is required, the need to know the true distribution is
important.
Recent investigations [1,2] carried out on high efficiency power plant
precipitators, 99.85% + , using the latest sizing techniques, e.g. Berner low
pressure impactors, differential mobility analysers (DMAs) and optical
particle counters, have indicated that PF A, although predominantly monomodal at around 8 j1.m, has a smaller but distinct secondary mode at around
0.2 j1.m. This second mode is almost coincident with the minimum efficiency
(maximum penetration) in the ESP fractional efficiency curve (see Figure
3.34 in chapter 3). Where specific emission requirements are to be met it is
important to be aware of this bimodal distribution.

158

PHYSICAL AND CHEMICAL PROPERTIES OF PARTICLES

Figure 6.5 SEM micrograph of fly ash particles from the combustion of South African coal.
Photograph of large particle with smaller attachments. Magnification 15000 x .

Another interesting fact arising during this work was that, with the lower
combustion temperatures associated with fluid bed combustion (FBC), the
number of particles below 0.1 J.lm was an order of magnitude lower than
from conventional pulverised fuel (PC) fired units.
There are many different methods of measuring a parameter which can
be assigned as a particle size. This parameter is defined in various ways,
depending on the method or apparatus used; generally for irregularly
shaped particles, the size is normally defined in terms of an equivalent
diameter, which will depend on the physical or geometric properties of the
particle.
The following lists the main types of equipment available and the
principle associated with each method.
Sieving. This depends on passing the particles through a range of defined
size apertures, usually in the form of a mesh, in descending stacked order.
The particle retained by a specific mesh is normally allocated the mesh
size, i.e. sizing is based on a linear dimension. The method is only
acceptable for particles in excess of about 50 J.lm because of agglomeration
difficulties. Some systems use wet sieving, which extends the range slightly
downwards.

PARTICLE SIZE AND SHAPE

159

Figure 6.6 High-resolution SEM micrograph of fly ash particles from the combustion of South
African coal. Photograph of agglomerated small particles. Magnification 15000 x .

Inertial systems. These depend on suspending the particles in a gas stream


and modifying the gas flow such that the particle's momentum allows it
to leave the stream and be captured. The parameter in this case is the
equivalent Stokes' diameter, i.e. the diameter of a sphere having the same
density and settling velocity as the particle.
Liquid sedimentation systems. These produce a measure of the Stokes'
diameter, but instead of a moving air stream, a liquid column is used
through which the particles fall. At specific times, samples are taken and
the particulate concentration measured and back-calculated as the equivalent free falling diameter.
Volume measurement. In this approach, the particles are suspended in an
electrolyte and passed through a small electrolytic cell. The change in cell
current is then converted to a resistance effect and assuming only one
particle is present in the cell, its volume can be calculated.
Area measurement. The particles suspended in a gas stream are irradiated
by a laser beam and the change in beam intensity measured as a result of
absorption/scattering by the particle.
Diameter measurement. For the smaller particles, such as those classified as
fume, the electron microscope enables the diameter to be calculated. As
these particles are usually spherical the equivalent size is that as meas-

160

PHYSICAL AND CHEMICAL PROPERTIES OF PARTICLES

ured. Larger irregular particles, using an optical microscope, are more


difficult to assess, since they could be flat platelets.
From the foregoing, it is obvious that there is no single universal method
of sizing and each differs in the parameter measured. For general gas
cleaning duties, the determination of the equivalent Stoke's diameter offers
advantages but, except for the cascade impactor, most methods rely on
redispersing collected particulate samples, which because of agglomeration
may introduce errors.

6.1.2

Particle shape and structure

While the above sizing methods are imporant in determining one of the
particles' physical properties, they do not give a complete picture as far as
electrostatic precipitation is concerned. Another parameter, the shape of the
individual particles, needs to be assessed, since very few dusts are truly
spherical and this means that their shape can play an important role in the
overall precipitator efficiency.
Particles arising during the incineration of paper tend to retain their
original platelet form, so have a large surface area in two planes but are very
thin in the third. Another form for concern is that of partially combusted
coal, which produces a voided coke-like particle having a large surface area
but low mass. Both paper char and coke particles, comprised mainly of
carbon, have very low electrical resistivities, which, as detailed in section 6.4,
can be readily re-entrained by the gas flow after reaching the collectors.
The precipitation engineer is also concerned with processes giving rise to
particles which are small in cross-section but are very long, e.g. hair or rod
form. These particles tend to align themselves with the field lines and can
join or 'chain-up' causing electrical breakdown.
In the case of fume and mists, because they usually arise through
sublimation, volatilisation and condensation processes, surface tension effects during their transition phase tend to give rise to spherical particles
usually 1 11m or less in diameter. Elutriation or sieving procedures are not
practical in this size range and equivalent volume, area or actual measurement of diameter is normally carried out, typically giving results on a
frequency basis before conversion to mass. If actual mass distribution is
required, the cascade impactor, used as an in-situ device, will produce a true
sizing and will determine any naturally agglomerated particles as an
equivalent diameter sphere.
Although particles of carbon black have been found to 'chain-up', the
major problem facing the precipitation engineer from small submicron
particles is one of potential space charge and corona suppression effects on
the electrical operation of the unit.

OPTICAL PROPERTIES

6.2

161

Optical properties

The optical properties of aerosols are of great importance to the airpollution control engineer, because the degree of pollution is often judged
by the appearance of the stack discharge. The colour of emitted smoke or
particulates is determined for the smaller particles by their shape, size, and
refractive index and for the larger particles, by their colour and surface area.
Raleigh [3] carried out theoretical calculations on the scattering of light by
spherical particles having diameters of 0.1 ,urn and less and proposed that
light scattering is proportional to the sixth power of particle diameter and
inversely proportional to the light's wavelength. This is mainly of interest as
it explains the redness of the morning and night sky and the intense blue of
outer space.
For particles greater than O.1,um Raleigh's Law does not hold and
changes to a second power law of diameter. In the case of fresh tobacco
smoke, the particles of tar are around 0.25 ,urn; this is below the wavelength
of normal light, 0.4 to 0.8 ,urn, so the scattered light appears blue since this
is the wavelength scattered most. When the smoke has been exhaled the
particle size is much larger, because of condensed water vapour, and the
smoke appears white, as the light is scattered more uniformly.
For even larger particles, scattering is mainly by reflection, i.e. in
proportion to their surface area and is the basis of optical extinction type
meters. The physics of the extinction meter is based on Lambert's Law,
which states that, if a beam of light intensity lois passed through a column
of particles of length x, then the amount of light reaching the far side will
diminish exponentially depending on the properties of the aerosol. This can
be expressed mathematically as:
1 = 10 ' e- kx

where k is a function of particle size and mass concentration, I.e. total


surface area, or
1= 10 ' e- Smx / 4

where S is the specific surface area of the particles (m 2 /g), m is the mass
concentration (gjN m 3) and x is the path length (m).
In practice, the emission from a specific source, after a precipitator for
instance, tends, even if the mass emission should vary, to maintain a roughly
similar particle sizing and hence specific surface area. This means that any
change in the opacity for a given path length is, as a first-order approximation, directly proportional to the mass concentration.
The following example illustrates how a shift of 20% in the legislative
emission limit for a power station (utility plant), affects the opacity for a
10 m path length. If we assume the dust concentration to be 50 mg/Nm 3 and
the dust has a specific surface area of 9000 cm 2 /g, then from Lamberts' Law

162

PHYSICAL AND CHEMICAL PROPERTIES OF PARTICLES

we have:
Emission.
(mg/Nm 3 )

Srnx/4.

Transmission
factor (I/lo)

50
40
60

0.1125
0.09
0.135

0.894
0.914
0.874

For comparison, normal window glass reduces the amount of transmitted


light by around 8%, so in the table, all conditions would give 'fairly clean'
chimney conditions.
The above calculations indicate that, for a 20% shift in emission, the
intensity of light received on the far side of the column changes by 2%, and
although small, can be used for control purposes. In chapter 8, covering the
electrics, this approach is in fact detailed for both precipitator performance
optimisation and power saving by means of a feedback loop between the
opacity meter and TR electrical control system.

6.3

Agglomeration

Particle agglomeration is attributable in the main to the collision and


impaction of the smaller finer particles. These are in continuous motion, as
the result of bombardment of the particles by gas molecules, to exhibit what
is termed Brownian motion. The effect of the bombardment is maximised
for the finer particles and, as they agglomerate, Brownian motion decreases,
so the possibility of further collision is reduced.
In precipitation, the effect of agglomeration is to shift the particle sizing
upwards; this will impact on their free-falling velocity as per the StokesCunningham relationship and may reduce the possibility of space charge
and corona suppression effects.
Another aspect, which is gaining prominence in the air-pollution field, is
that of the need to efficiently collect toxic and heavy metals; these materials
in most processes are very small in terms of mass concentration, but recent
toxicology investigations have indicated their importance to health.
Heavy metals usually exist in the downstream area of the process as
condensed submicron particles after passing through a volatilization stage
in their life cycle. During condensation, some of the particles use larger
particles as condensation nuclei, so it is not unusual to find small metalrich fume adhering to larger inert particles, as indicated in Figure 6.5.
This means that these fine heavy metal particles are removed along with
the easier to collect coarser particles. This would help explain why solid
phase toxic and heavy metal mass balance measurements, in spite of their

COHESIVITY

163

smaller particle size and potentially lower performance, compared with the
coarser particles, produce almost the same order of efficiency as the bulk
materials [4].
The effects of agglomeration and cohesion, although producing particle
flocculation, are the result of different mechanisms and should not be
confused. Agglomeration, as stated above, arises wholly in the gas phase
because of Brownian motion and, except for impaction on some larger
particles, predominantly applies to small submicron particles forming larger
but still small units (see Figure 6.6). Cohesion, on the other hand, applies to
the collected precipitated dust, as will be covered in section 6.4 below.
Cohesion, is a measure of the binding mechanisms holding together all size
particles, as a result of electrical and mechanical forces acting on the layer
rather than individual particles.
6.4

Cohesivity

In electrostatic precipitation, particle cohesion plays an important role in


the plant's performance, firstly in respect of how the particles are held to the
collector and secondly how the particles hold together in their transference
to the hopper, following rapping.
Particles after reaching the collector plate are held initially by electric
forces and after losing their charge would, without the mechanical binding
force, be readily released and possibly re-entrained by the gas stream. From
Stokes' Law, with a precipitator gas velocity of 1.5 mis, only bound particles
greater than 500 f.1.m equivalent diameter stand any chance of reaching the
hopper. Without this particle cohesion the dust would be subject to massive
re-entrainment, which would have a deleterious impact on efficiency.
The forces holding the particles to the collectors and themselves are a
combination of electrical and mechanical Van-Der-Waal forces. Both are
linked to the surface chemistry of the particle, the nature of which depends
not only on the composition of the dust but also on the gas constituents to
which the particle has been exposed. For particles having electrical resistivities above 1011 Q-cm. the electrical holding force tends to predominate over
the mechanical force, but for low resistivity dusts the mechanical force
becomes more dominant, depending on the dust layer properties.
In general, light fluffy dusts adhere poorly, while dense or sticky dusts
adhere well. Particles such as carbon and ionic salts, which can form 'snow
flakes', have low packing densities and poor cohesion, so the particles are
only loosely held together. An exception is where the ionic salt is deliquescent/hygroscopic and absorbs moisture from the flue gas which cements or
binds the particles together. If the temperature should approach the gas
dew-point the dust becomes sticky and may be difficult to remove from the
internal components of the precipitator.

164

PHYSICAL AND CHEMICAL PROPERTIES OF PARTICLES

The importance of cohesivity and precipitator performance has been


discussed in general terms by a number of investigators; for example, Lowe
and Lucas [5] calculated the forces to which a dust particle would be
subjected on the collector; Penney and Klinger [6] measured the cohesion
of an electrostatically precipitated dust layer after the power had been
switched off and compared the result with that of a mechanically formed
layer with fairly good agreement.
Dalmon and Tidy [7], determined the relative importance of cohesion of
both mechanical and electrical origin and then related the cohesive properties to precipitator performance. This investigation was carried out with two
very different power station fly ashes, one relatively fine having high
resistivity and another coarse containing 33% carbon and of low resistivity.
Both dusts gave rise to poor precipitator efficiency.
The collected dust samples were initially washed to remove all surface
conditioning effects/agents and injected into an oil-fired combustor rig fitted
with a 250 mm hexagonal tube precipitator. Precipitator efficiency measurements were made with the dust being self-conditioned by the flue gas and
then with various conditioning agents, known to reduce resistivity, injected
into the rig, upstream of the precipitator with about a half-second exposuretime.
At the end of each run, a detachable section of the collector was removed
and the bulk density and the minimum compressive load, i.e. the load to just
observe further compression of the dust precipitated on the sheet, were
determined. Dust from the collector plates at the end of each run was placed
in a powder tensiometer [8] and the ultimate tensile strength (UTS) of the
medium was determined for each sample over a range of bulk densities.
(Although the cohesivity of the powder sample was not directly measured,
Farley and Valentin [9] have shown that for powder beds the UTS is
directly proportional to the cohesivity.)
The results from these measurements showed that each dust had a
distinct but separate UTS vs. compressive load relationship, which was
independent of conditioning agent, and was basically a function of particle
size, shape and hardness. The force required to fracture the bed is proportional to the product of the mean strength of the individual forces and their
number density in the fracture surface. For a specific ash, the tensile strength
at a fixed bulk density can, therefore, be used as a measure of the
particle/particle force, or cohesion, and thus indicates the relative ease of
re-entrainment.
The precipitator performance, when treating the variously conditioned
high resistivity fly ash, showed little dependence on the UTS value measured
at the precipitated bulk density or the UTS at constant bulk density. There
was, however, as expected, a distinct effect of resistivity on collection
efficiency due to the conditioning agents and hence the conclusion, for high

165

COHESIVITY

100

Ash + 503

Carbon + 503

)I( Ash + (NH 4)2 504


)I( Carbon + (NH4)2 504

50

100

150

200

250

300

350

400

450

Increase in additive ion concentration in ash (mg mol)

Figure 6.7 Effect of additives on precipitator efficiency for ash and carbon [7]. 0, ash + SO,;
., carbon + SO,; @, ash + (NH4)2S04; ~, carbon + (NH4)2S04'

resistivity dusts, that the improvement in performance is primarily by


reduction in electrical resistivity.
With the coarser low resistivity ash (10 7 Q-cm), the conditioning agents
were not expected to produce any further reduction in resistivity; however,
the addition of conditioning agents was found to significantly improve the
performance. The collection efficiencies of the fly ash and carbon were
separately evaluated for a range of injection rates for S03 and ammonium
sulphate; these are reproduced in Figure 6.7. This clearly demonstrates that
it is the carbon efficiency which is improved by the conditioning agent by
preventing re-entrainment. When the UTS was measured at constant bulk
density, the emv vs. UTS relationship was found to be linear (Figure 6.8),
with the higher emvs being associated with the higher UTS. This increase in
the individual particle/particle force is responsible for the improved collection efficiency of the carbon and the reduced re-entrainment potential.
Dalmon and Tidy [7] concluded that tensile strength measurements
made on mechanically formed beds of highly resistive dusts gave results
which closely align with those measured on a precipitated bed after the field
was removed. Cohesivity will be present in any deposited layer and will be
augmented by an equal force due to the electrical field when the ash
resistivity is high. As the ash resistivity reduces, the electric field effect
reduces and may even become repulsive, so particles are readily re-entrainable. Increasing the tensile strength of low resistivity particles by the

166

PHYSICAL AND CHEMICAL PROPERTIES OF PARTICLES

Figure 6.8 Relationship between effective migration velocity (EMV) and UTS for variously
treated fly ash samples [7].

injection of conditioning agents produced increased precipitation efficiency,


mainly from the capture of very low resistivity carbon particles, as a direct
result of the higher cohesive forces prevailing, leading to reduced reentrainment.

6.5

Particle electrical resistivity

Perhaps as regards precipitator engineering, the electrical resistivity of the


particles is of paramount importance to performance and hence the size of
the precipitator itself. As most particulates from power stations are complex
compounds of basically silica and alumina, both having excellent insulating
properties, one would anticipate resistivities in the 10 14 Q-cm and higher
range as a bulk resistivity. The electrical resistivity of the particles can be
modified, however, by the presence of impurities in the ash or where the
surface of the particle is naturally conditioned by components present or
added to the entraining gases. The nett resistance of a surface conditioned
particle, even for a thin monatomic conductive coating, is significantly
reduced. (Resistances in parallel.)
The electrical resistivity is determined by two mechanisms, the bulk
volume conductivity, which is a function of the particle matrix constituents,

PARTICLE ELECTRICAL RESISTIVITY

167

and surface conduction. The latter is governed by the adsorbed surface layer,
which is related to the surface reactivity and gas components.
The measurement of resistivity, because of this surface conditioning effect,
makes laboratory determination difficult. There are, however, standard
techniques and apparatus [10], where the sample is placed in a cell and the
surrounding environment simulates the original gas condition in terms of
temperature, water dew-point and, if necessary, acid gas concentration. To
obtain comparative laboratory data, the accepted procedure uses a standard
cell configuration and the dust is compacted to a set pressure. Once the
apparatus has reached the correct 'equilibrium gas' condition, the current
passing through the cell is determined for a range of applied voltages. In
practice, resistivity evaluations are made for both increasing and decreasing
temperatures to enable the peak and general resistivity profile to be found.
Laboratory measurements tend only to be comparative, since the packing
density will not necessarily be the same as on the collector plates and the
sample itself may have aged and so affected its surface reactivity. In addition,
measurements by Goard and Potter [11], have shown that the resistivity,
not unexpectedly because of the ionic nature of the conduction mechanism,
is strongly field dependent.
For actual in-situ field determinations the US Southern Research Institute (SoRI) developed a point/plane apparatus [12], which is inserted into
the flue. Dust is electrostatically precipitated onto the plate by energising
the point electrode negatively, with respect to the plane or plate electrode.
After a certain time, depending on the dust concentration, a circular plate
attached to the point is carefully lowered onto the dust surface and the
current, for a preset range of voltages, is measured. From this, and with
known cell dimensions, the resistivity of the precipitated dust can be
evaluated.
An alternative form of apparatus for both in-situ and laboratory measurements is a cell mounted in the base of a small 25 mm diameter sampling
cyclone [13]. Gas is drawn through the cyclone, which collects all but the
smaller submicron particles, and after rapping the cell to produce a constant
packing density, the current through the cell is measured in the usual way,
thereby enabling the resistivity to be calculated.
With either type of in-situ apparatus, while the environmental conditions
are reproduced exactly, the actual ash sample is from a single point and may
not be representative of all the ash and there is the possibility that the fines
escape the apparatus completely.
A typical laboratory derived temperature/resistivity curve for a power
station fly ash, is illustrated in Figure 6.9. This shows that at temperatures
higher than 200C, where the surface conditioning effects have been destroyed, the resistivity follows classical theory of volume conduction, as a
result of increased thermal motion of the molecules, producing a linear
relationship with the inverse of absolute temperature. Below 200C, the

168

PHYSICAL AND CHEMICAL PROPERTIES OF PARTICLES

....
..
Measured in
vacuo-surface
Films removed
by outgassing
.0-0.

...

:\

"0.

ti

6 Measured in
moist air

(40C O.Pt)

(laboratory)

:' Measured by site


resistivity apparatus
108 '--_--L_ _\~--L~----L-

100
200
Temperature (Oe)

300

Figure 6.9 Effect of flue dust resistivity of surface film condition. Willington power station flue
dust.

curve shows a decreasing resistivity as the surface conditioning, due in this


case to moisture, becomes more significant as the temperature falls, because
of the increasing vapour pressure. The result of vacuum outgassing the
sample produces a continuation of the classical theory line confirming that
the decreasing resistivity below 200C is the result of surface conditioning.
Figure 6.10 has been developed for a dry process cement kiln dust which
illustrates the effect of changing gas moisture content on particulate resistivity. From this, one can appreciate the need for water conditioning of this
type of dust in order to avoid the problems of reverse ionisation and to limit
the size of the precipitator.
While many engineers consider the electrical resistivity of the particulates
to be all important, the problems of accurately determining the true value,
as indicated above, really means that resistivity is only one of the tools used
in the precipitation industry and one must not overemphasise its importance
in the design and sizing of precipitators. The effect of resistivity on
precipitator operation and performance can be summarised as follows.
For particles having resistivity in the range 10 10 to 1011 Q-cm the particle
charging and discharging regime, when the particle arrives at the collector,
proceeds normally and so has minimum impact on performance.

PARTICLE ELECTRICAL RESISTIVITY

169

Figure 6.10 Effect of moisture content of entraining gas and temperature on cement dust
resistivity.

As the resistivity increases, although the charging occurs normally, the


particle on arriving at the collector only slowly loses its charge and a voltage
begins to develop across the dust layer according to Ohm's Law. Dependent
on the resistivity and layer thickness the voltage, for resistivities in the
1013 n -cm range, can reach a point where positive ions begin to be emitted
from the surface of the dust. These positive ions cross the interelectrode space and collide with and neutralise negative ions and charged
particles, which significantly reduces precipitator performance. This condition is termed reverse ionisation, or back-corona, and the operating electrics
exhibit a much reduced average voltage but an extremely high current. The
average indicated voltage actually falls with increasing current, which is the
opposite to a 'normal dust' operating condition.
Recent investigations of the applied voltage and current (A VC) waveforms have shown that, under reverse ionisation conditions, the peak applied

170

PHYSICAL AND CHEMICAL PROPERTIES OF PARTICLES

voltage tends to be maintained, but with the high current flow, the ripple
voltage significantly increases and the main feature is that the minimum
ripple voltage level dramatically falls. To control/minimise the deleterious
effects, modern Ave units monitor this minimum voltage condition and
take appropriate corrective action to prevent the positive current 'runaway'
(see chapter 8).
For slightly lower resistivities, or dusts having poor packing density,
although a voltage develops across the dust layer, interstitial breakdown
occurs through the layer and gives rise to a 'streamer' which results in field
breakdown. The symptoms of this condition are a slightly reduced but
normal average operating voltage, but a very low discharge current, any
attempt to raise the current immediately resulting in increased flashover.
Examination of the voltage waveform shows a minimum ripple voltage and
any attempt to increase the current raises the threshold voltage resulting in
breakdown. While this condition does not have such an impact on performance, as does the classical reverse ionisation or back-corona scenario, it
nevertheless causes an efficiency limitation.
At the lower end of the resistivity range, for very conductive particles,
such as metallics or carbon, the charge on arriving at the collector is lost so
quickly that the particle sits on the surface as a neutral particle and can be
entrained by the gas stream. While there are no specific characteristics
identifying this phenomena, the potential re-entrainment results in poor
collecting efficiency. Examination of the outlet dust samples shows an
increasing quantity of these conductive particles (see section 6.6). The extent
of the re-entrainment depends on a number of factors, such as particle
cohesion, gas stream velocity and turbulence and the electrostatic forces
acting on the particle.
Durham [14J working with low resistivity spray drier particles shows
that, by theoretically examining the conditions existing at the boundary of
the dust layer (Figure 6.11), it is possible for a repulsive electric force to
develop which expels the particles back into the gas stream. From Figure
6.11 an electric field Eg exists at the surface of the layer as a result of the
corona electrode voltage, which also creates the ion density N; at the
surface. The resultant current density J g is in the positive x direction,
although the ions move toward the collector.
We consider:
J g = N; x e x B; x Eg A/m2

where N; is the ion density, e the charge on an electron and h; the ion
mobility.
Assuming homogeneous and steady-state conditions we have:
J g = J 1 = J p A/m 2

where J 1

IS

the current density through the dust and J p

IS

the specific

171

PARTICLE ELECTRICAL RESISTIVITY

+00+0
+0 P10 +0 NI
+0 0 +0
---~,

,EoE,

I
I
I
I

Plate

Gas

Layer-!

Gas-

Figure 6.11 Boundary conditions for a layer of particles on the collector.

collector current. So
E 1 = J 1 X P1 V1m
where E1 is the electric field within the dust layer and P1 the dust resistivity
(Om).

The surface charge density is given by:


(j

= eo{E g -

e1 x P1 x J p )

clm

where eo is the dielectric for free space and e1 the dust layer dielectric. From
this, the surface charge density can be positive or negative, depending on the
resistivity and layer conditions. The force acting on the surface charge
density, ix, is given by:

ix = {E + E 1 )/2 N/m2
g

If the surface charge density is positive, the force tends to pull the particles
from the surface, and if negative, the force holds the particles to the surface.
Figure 6.12 has been taken from [14] showing the effect of these forces
for a current density of 0.06 mA/m2 and for different resistivities. For the
higher resistivities, the attractive force holding the particles to the plate
becomes predominant, whereas for the low resistivities there is an apparent
electric repulsive force promoting re-entrainment.
While theory would suggest that low resistivity dusts, such as those
arising from spray driers, should be subject to electrical re-entrainment,
there are a number of full-scale plants fitted with precipitators, following
spray drier desulphurisers, where re-entrainment does not appear to be a
problem, [15]. This does not necessarily mean that the proposed theory is

172

PHYSICAL AND CHEMICAL PROPERTIES OF PARTICLES

2
Current density (J
60 nA/cm 2

t
c
o

'w

:;

0.

Q)

a::

-2~---------L----------~

10 7

108

109

________ __ ____
~

10 10

Particle resistivity (n/cm)

Figure 6.12 Force acting on particle surface for different electrical resistivities and field
strengths [14].

incorrect but, as the properties of the particulates are controlled by a large


number of factors, it is likely that a slight difference in the approach to
dew-point temperature could make the dust more adhesive in spite of the
lower resistivity.
Fortunately for the precipitation engineer, knowledge of the process or
previous experience enables the correct choice/size of unit to be made to give
a certain efficiency. If, however, an error of judgement occurs then there are
various means of overcoming the worst effects, either by modifying the
surface condition, by the injection of various chemicals, as described more
fully in chapter 15, or by applying modern electrical energisation techniques,
as covered in chapter 8.

6.6

Chemical composition and reactivity

Virtually all dusts met industrially consist of a multitude of chemical


compounds in differing particle sizes, very few having identical composition
because of the presence of impurities, either in the raw feed material or the
fuel used in processing.
In the case of steam raising plants, the ash at the precipitator consists
mainly of the fuel's ash components, mainly silica and alumina, which have
been exposed to high temperature in an oxidising atmosphere, Some of the
latest DeNOX technologies, however, employ an initial reducing atmosphere and a somewhat lower flame temperature which can affect ash

CHEMICAL COMPOSITION AND REACTIVITY

173

chemistry. There is invariably a higher carbon carryover which has to be


considered as it will effect precipitator performance.
In the power industry it is fairly usual to consider carbon carryover as
approximately equal to the loss on ignition at 800C. In other industries the
loss on ignition can be the result of the loss of water, the release of carbon
dioxide from carbonates, the evaporation of alkali salts and some metallic
components, so for industries outside power, the carbon needs to be
determined in some other way.
At the precipitator inlet, the ash chemistry and reactivity is important in
that it determines the particle resistivity. Raw coal and ash analyses in fact
provide the precipitator design engineer with invaluable information on the
probable fly ash resistivity and hence the potential size of precipitator for a
specific efficiency. Chapter 15 provides a fuller explanation of Bickelhaupt's
relationship [16J between coal and ash chemistry on fly ash resistivity.
In general, for steam raising units, the coal and ash components which
produce high resistivity are silica and alumina, both excellent insulators in
their own right, whereas those which reduce resistivity in the bulk media are
coal sulphur and ash sodium.
Although during combustion most of the coal's pyritic sulphur forms
sulphur dioxide, a small percentage converts to sulphur trioxide which is the
important substance in significantly effecting resistivity. In a boiler flue gas
there is always moisture present, either from the moisture or converted
hydrogen in the coal or from atmospheric moisture; the gaseous S03 with
this moisture reacts to produce condensed phase sulphuric acid. This then
uses the surface of the particles as condensation nuclei, thereby reducing
their resistivity, as conduction is now able to proceed through this absorbed
or adsorbed layer of acid. The effect of surface conditioning is dependent on
the amount of sulphur trioxide present, the gas temperature and hence
vapour pressure as indicated in Figure 6.13 [17]. For coals containing in
excess of 1.5% sulphur, there is sufficient natural surface conditioning at
normal operating temperatures of 130C to give acceptable values of
resistivity for effective precipitation.
Although the foregoing is generally true, there are incidents where the
fuel, particularly the hydrocarbon content, is not totally combusted, but is
volatilised off to coat the particles with an insulating-type material which
can promote severe reverse ionisation-type operation. The phenomenal
occurrence is probably more prevalent during plant start-up, when the
furnace is cold and the oil start-up burners are not ideally set up, but can
arise at other times, so one must be aware of this possible situation.
The presence of sodium in the ash behaves very differently, where, instead
of surface conditioning, the sodium ions act as charge carriers, so high
resistivity effects are negated to a certain extent. Measurements on a
Japanese installation firing Australian low sulphur (0.5%) fuels, and therefore expected to give 'difficult' precipitation, showed that the precipitator

174

PHYSICAL AND CHEMICAL PROPERTIES OF PARTICLES

200

6'

e.... 180
~

:J

'
Q)

~ 160
.l!l
C

'0

a.
~
Q)

140

"0

gj
Ol

"0

120
Water dew-point (DC)
100~----~--~~------~----~----~------~--~

10

20

50

500

100
200
50C 100 y/x

1000

SP3 content of raw gas

Figure 6.13 Relationship between S03 content of gas and dew-point temperature. (Source:
Chemical Engineering Progress, August 1974 and April 1977). Example: SO, in raw gas,
200 ppm; gas temperature, 175 cC; moisture content, 11.6%; water dew-point, 50 'C; gas cooled
to 135 cC; residual SO" 20 ppm; S03 condensed, 180 ppm.

20
40
'"
60
E
~ 80
g100
c:

'w

.~ 200

o
400
600
800

0.5

1.0

1.5

Sodium in ash (%)

Figure 6.14 Effect of sodium in ash on emission for Australian coals, 0.4 to 0.54% sulphur.

175

CHEMICAL COMPOSITION AND REACTIVITY

13
0012

E
11
~

c10

.(3
0

Qi

>
c

Q)

Data points corrected for:


Temperature 130C
Moisture in gas 8% ulv
Inlet dust concentration 15 g/Nm3
Collector spacing 300 mm
Collection efficiency 99.5%

C,
E 6

>

"u

Q)

'I:

UJ

4
3
2

0.5

1.0
1.5
2.0
2.5
3.0
Total % sulphur in coal + % sodium in ash

3.5

4.0

Figure 6.15 Relationship between precipitator performance and sulphur in coal and sodium in
ash. Data points corrected for: temperature, 130 cC; moisture in gas, 8 % vIv; inlet dust
concentration, 15 g/Nm 3 ; collector spacing, 300 mm; collection efficiency, 99.5%.

emISSIOn was very dependent on the sodium present in the fly ash, as
illustrated in Figure 6.14. Lithium is also anticipated to react similarly, but
is only present in very small quantities, which are too small to have a
significant impact on performance. Potassium, although having similar
chemical properties to sodium, does not appear to react in the same manner.
Calcium and magnesium present in the ash tend to produce sulphates
which are not effective conductors, so interfere with resistivity reduction by
sulphur trioxide and thus must be considered as leading to increased
resistivity effects.
Performance measurements on fly ash precipitators have been correlated
with the coal and fly ash analyses and first-order precipitator sizing curves
have been developed. Figure 6.15 plots the migration velocity necessary for
a precipitator to give an efficiency of 99.5% [18J against the coal sulphur
and ash sodium content. Other curves incorporating silica, alumina, calcium
and magnesium, in addition to sulphur and sodium, have been developed
but do not appear to significantly affect precipitator sizing requirements
based on data from Figure 6.15.
Measurements of in-situ fly ash resistivity show good agreement with
those calculated from the Bickelhaupt relation and Figure 6.16 has been
plotted, showing how the resistivity varies predominantly with the sulphur
and sodium components. It is interesting to note that, at around a resistivity
of 10 10 Q-cm, the resistivity steeply increases for reducing values of sulphur
plus sodium, whereas for higher values the resistivity only reduces slowly.
This mirrors the performance curve as Figure 6.15; another interesting point

176

PHYSICAL AND CHEMICAL PROPERTIES OF PARTICLES

Gas temperature approx 130C


o Measured values (by H. Hall)

x Calculated values (Bickelhaupt)

108L-------~--------~--------~-------

% sulphur in coal + % sodium in ash

Figure 6.16 Relationship between fly ash resistivity and sulphur in coal and sodium in ash. Gas
temperature approx. 130 'c. 0, measured values (by H. Hall); x, calculated values (Bickelhaupt).

is that a resistivity of 2 x 10 10 Q-cm is the value which White [19], in 1963,


indicated as being an acceptable resistivity for good precipitation.
For precipitators operating at high temperature, the coal, ash and surface
chemistry is not so important, as volume conductivity is controlled more by
thermal molecular motion. Some unexpected performance problems were
however encountered on a US utility plant high temperature precipitator
which were found to be the result of sodium ion migration within the
collected dust layer itself (sodium depletion), the sodium concentrating on
the surface. This produced the classic reverse ionisation phenomena which
affected performance. This condition can be overcome by supplementing the
sodium content of the fuel artificially by adding soda ash or other sodium
products during combustion.
As an alternative to operating at high temperature to reduce particle
resistivity, investigations, particularly with NSW Australian low sulphur
coals, have shown that operating at a lower gas temperature, e.g. 100/110 c,
produces higher efficiencies by reducing the worst effects of back-ionisation.
For fairly inert dusts, e.g. from NSW, additional air heater surface can be
installed to reduce temperature without the risk of serious air heater
blockage. With other coals, a multitude of two fluid atomisers have been
installed in the ductwork upstream of the precipitator to effect the temperature reduction. In practice the problems of maintaining the atomisers such

CHEMICAL COMPOSITION AND REACTIVITY

177

that they don't result in duct blockage means that ideally a correctly sized
cooling tower should be supplied. Reference has already been made to the
application of spray driers ahead of precipitators; these effectively reduce
temperature and increase the moisture level, both aiding the reduction is
resistivity.
In the metallurgical field it is not unusual to find a high percentage of
submicron fume arriving at the precipitator; this fume arises because of
initial evaporation/volatilisation of metals and metalloids at the processing
temperatures, followed by recondensation at the lower back end temperatures. The presence of these fume particles can give rise to severe space
charge and corona suppression effects and the designer must be aware of
them, so the correct type of discharge electrode and matching HT equipment can be supplied to overcome these possible problems. In practice it has
been found that normally only the inlet field needs to have the high emission
electrodes, since if this field operates normally, the subsequent fields see a
very much reduced particulate loading and hence reduced space charge and
corona suppression effect.
The carryover of alkali chlorides giving rise to 'snow flake' type formations, particularly in the cement industry using wet or semi-wet feed
material, has a significant effect on precipitator performance, both in terms
of efficiency reduction and chimney appearance. The formation of snow
flakes is complex but is related to the quantity of soluble alkali material fed
to the kiln. As the alkali material is derived from volatilisation and
condensation, its particle size is very fine and consequently the highest
concentrations are found in the outlet field hoppers. If this can be extracted
separately and discarded, few precipitation problems are experienced; however, as in most plants the total precipitator catch is returned to the kiln,
the amount of alkali material in the feed builds up to a point where the snow
flake problems rapidly appear within the precipitator [20].
The term 'snow flake' was derived from scanning electron microscopy
where the material was found to have a crystalline structure resembling
snow. The material itself agglomerates into flocculant lumps of low bulk
density, so is not only subject to significant gas re-entrainment, but also
'lifts' within the field area, as a result of ionic wind or thermal diffusion, and
tends to collect across the collector top beams, which can lead to complete
shorting of the HT in extreme cases.
Deliquescent/hygroscopic materials, or those close to their melting point,
need special attention at the design stage, not that they cannot be satisfactorily precipitated as they are conductive and usually at equilibrium with
the gas phase moisture. If temperatures or process conditions alter, e.g.
during start-up and shut-down, then further moisture absorption can
produce a sticky dust, which may impede maintaining the internals 'dust
free' and so can affect operating conditions, or can give rise to hopper
dedusting difficulties. Although special operating procedures and plant

178

PHYSICAL AND CHEMICAL PROPERTIES OF PARTICLES

preheating can reduce the start-up and shut-down difficulties, it may be


advantageous to consider wet, rather than dry, precipitation.
In the case of high carbon dusts, controlled conditioning is deliberately
carried out to increase the cohesivity of the collected material to reduce
re-entrainment losses.
Corrosion, as a result of the particulate composition, is normally a
problem associated with wet or mist precipitators and its effect is fully
covered in chapters 13 and 14. High temperature corrosion, however, has
been experienced on older municipal incinerator precipitators operating at
300C where, as a result of chlorides being present in the feed, these when
precipitated can transform from the ferrous to more stable ferric form by
removing iron from the collector plates with their resultant thinning.
Nowadays, a change in incineration practice, to give a low back end
temperature, has largely resolved this particular problem, but the designer
should be aware of this reaction when considering materials of construction
for high temperature applications.
Various investigators have reported other elements in power plant fly ash
which affect the resistivity and hence precipitator performance, e.g. iron and
phosphorous. Iron is often an indication of the amount of pyrites in the coal
and hence combustible sulphur, while phosphorous has been shown to be
an effective conditioning agent in reducing the surface resistance of the fly
ash particles [21]. Their exact mechanism in reducing the ash resistivity,
however, needs further investigation.
It is likely that other elements present in the fly ash also influence
resistivity/performance, but the foregoing are the most pertinent.

References
1. Kauppinen E.!. et al. (1995) Fly ash formation on PC boilers firing South African and

2.
3.
4.
5.
6.
7.
8.

Colombian coals. EPRIjDOE International Conference on Managing Hazardous and


Particulate Air Pollutants. Toronto. Canada. Aug 15-17, in print.
Porle K. et al. (1995) Full-scale ESP performance after PC boilers firing South African and
Colombian coals. EPRI/DOE International Conference on Managing Hazardous and
Particulate Air Pollutants, Toronto, Canada, Aug 15-17, in print.
Lord Raleigh (1871) Philosophical Magazine, 41, 107.
Parker K.R. and Novogaratz D. (1991) Electrostatic control of air toxics. EPRI/EPA 9th
Particulate Control Symposium, Williamsburg, USA, Oct., Session EPRI TR 100471 2, Palo
Alto, CA, USA, 1992.
Lowe H.1. and Lucas D.H. (1953) The physics of electrostatic precipitation. British J.
Applied Physics, Supplement No.2.
Penney G.W. and Klinger E.H. (1962) Contact potentials and the adhesion of dust. Trans
A.I.E.E. 81 (I) p. 200-204.
Dalmon 1. and Tidy D. (1972) The cohesive properties of fly ash in electrostatic
precipitation. Atmospheric Enrironment 6, p. 81-92.
Ashton M.D., Farley R. and Valentin F.H.H. (1964) An improved apparatus for measuring
the tensile strength of powders. J. Scientific Instruments, 41, p. 763-5.

REFERENCES

179

9. Farley R. and Valentin F.H.H. (1968) Effect of particle size upon the strength of powders.
Powder Technology 1, p. 344-54.
10. IEEE Standard Criteria and Guidelines for the Laboratory Measurement and Reporting
of Fly Ash Resistivity. IEEE Std 548-1984.
11. Goard P.R.C. and Potter E.C. (1978) Operational resistivity measurements on freshly
generated fly ashes. CSIRO Symposium on Electrostatic Precipitation, Leura, pp. 3.1-8,
CSIRO, Sydney, Australia.
12. Nichols G. and Spencer H. (1975) Test methods and apparatus for conducting resistivity
measurements. Report prepared by the Southern Research Institute for the US. E.P.A.
SoRI, Bieringham, AL, USA.
13. Cohen L. and Dickenson R. (1963) The measurement of the resistivity of power station flue
dust. J. Scientific Instruments, 40, p. 72-5.
14. Durham M.D. et at. (1991) Identification of low resistivity reentrainment in ESPs operating
in dry scrubbing applications. 9th EPRI/EPA Particulate Control Symposium, Williamsburg, USA, Oct., Session 5A EPRI 100471 2, Palo Alto, CA, USA, 1992.
15. Porle K. et at. (1991) ESP operation following spray dryers with low resistivity particulates.
Proc. 9th EPRI/EPA Particulate Control Symposium, Williamsburg, USA, Oct., Session 5A
EPRI 100471 2, Palo Alto, CA, USA, 1992.
16. Bickelhaupt, R.E. (1979) A technique for predicting ash resistivity. EPA 600/7-79-204 US.
17. Parker K.R. (1990) The wet ESP and its role in modern pollution control. Proc. Xth
Australian and New Zealand Air Pollution Control Conference, Auckland, N.Z., Oct., pp.
23-30, Clean Air Soc., Auckland, N.Z.
18. Darby K. et at. (1991) The influence of sodium in fly ash on electrical resistivity and its
impact on precipitator performance. EPRI/EPA 9th Particulate Control Symposium,
Williamsburg, USA, Oct., Session 6A EPRI TR 100471 2, Palo Alto, CA, USA, 1992.
19. White H.J. (1963) Industrial Electrostatic Precipitation. Addison Wesley, USA.
20. Darby K. and Parker K.R. (1990) The use of electrostatic precipitators in the cement
manufacturing industries for the control of dust emissions. Proc. 4th International Conference on Electrostatic Precipitation, Beijing, China. Oct., 3, pp. 173-86, International
Academic Publishers, Beijing, 1991.
21. Darby K. and Whitehead C. (1974) The use of electrostatic precipitators in current power
station practice. Proc. Ins/. of Fuel Symposium on Changing Technology of Electrostatic
Precipitators, pp. 35-48, Inst. of Fuel, Adelaide, Australia.

Performance design considerations

c.
7.1

COTTINGHAM

Introduction

In the design and sizing of any electrostatic precipitator, both the customer
and vendor have certain responsibilities to ensure that the final plant will
meet the full expectations of both. This chapter will review the requirements
in detail in order to achieve the most competitive offer, with the minimum
of technical exposure to the vendor, wbile still meeting all the customer's
needs. The format to a large extent will be in the form of questions and
answers, which will explain some of the methodology used in the ultimate
sizing and configuration of the plant.
7.2

What are we trying to achieve?

The design and sizing of an electrostatic precIpItator to be ultimately


successful has to achieve the customer's requirements at the least cost to the
customer and vendor alike.
To achieve this the customer has to be absolutely clear in his requirements:
(a) understand what has to be achieved, usually in terms of emission, i.e. to
be aware of current and proposed future legislation;
(b) clearly define the design parameters, particularly at the inlet to the
precipitator or outlet of the plant to which the precipitator is to be
installed.
Although the above may be stating the obvious, there are many times when
the supplier is asked to commit time and effort to satisfy customers who
have little concept of what they really require.
The vendor has to assess the client's requirements and examine the
following in detail.
(a)
(b)
(c)
(d)
(e)

What process is being considered?


Are all the relevant process parameters defined?
What level of performance is required?
Are there any site constraints?
Does the client have any special technical requirements/preferences?

Only when all the above have been addressed can the vendor successfully
begin to size and offer a plant that is technically acceptable.

ASSESSMENT OF THE PROCESS

7.3

181

Assessment of the process

The designer of electrostatic precipitators (ESPs), can be confronted by a


bewildering array of processes and operating conditions (refer to chapters
12, 13 and 14). He has to be able to accurately assess the process conditions
and configure the ESP to satisfy the customer and ensure that the ESP has
minimum technical risk at the most competitive cost.
To illustrate the range of experience required, a list of a few of the
processes are given below.
Boilers
Pulverised coal
Heavy fuel oil
Fluid bed (sand and/or lime beds)
Stoker (grate or spreader)
Biomass (wood, straw, bagasse, etc.)
Emulsified tar
Steel
Main sinter strand
Sinter deduster
Furnaces (blast, BOF, etc.)
Electric arc
Gas recovery
Secondary ventilation
Pelletiser

CO recovery FCC
Black liquor recovery

Non-Ferrous
Converters (copper lead)
Furnaces (copper lead)

Cement
Kiln (wet, dry, semi-wet)
Finishing mill
Clinker cooler
Alkali bypass
Raw mill
The above represent only the more commonly addressed processes. As
can be seen, the designer has to be able to understand the wide and varied
range of applications that the ESP operates on. The designer has to be able
to assess the process parameters and determine how they relate to the ability
of the ESP to meet the performance requirements.
For the designer to be successful, it is essential that he has a consistent
method for analysing the process conditions. In trying to meet the client's
requirements, the designer may be confronted by various changes to the
design inlet parameters during the bidding process. If consistency is not
used, then the designer risks the chance of losing his way and hence his
credibility.
At all stages of the process analysis, the designer has to assess the impact
of the following primary parameters:
(a) gas temperature
(b) dust analysis
(c) gas analysis
(d) dust particle size

182
(e)
(f)
(g)
(h)

PERFORMANCE DESIGN CONSIDERATIONS

quantity of dust
dust resistivity
dust cohesivity
gas flow rate

For every process, and each one of the above, the designer should have
relationships that enable him to access the impact of the process on the
performance of the ESP.
In the early days of electrostatic precipitation, the designer had a library
of EMVs (Deutsch Effective Migration Velocities or Sizing Parameters) for
every process. Corrections and variations to the 'library standards' were a
matter of personal judgement and experience. This may have been acceptable 30 years ago, when both technical and commercial margins were
greater, but in today's environment more precise techniques are essential.
The experienced designer of today will have a background of reference
plants and test data that will enable him to guarantee the plant performance,
to much closer limits with much reduced technical margins. Coupled with
this, there have been much published data on the precipitation process in
the last 20 years, which can be judiciously used to expand/confirm his own
in-house data.

7.3.1

Typical assessment

We can best illustrate the techniques required by examining the following


worked example.
Let us assume we have an enquiry from a customer for a PC-fired boiler.
The design parameters are given as:
Gas vol.
Gas temp.
H 2 0 in gas
Ash in coal
Ash resistivity
Ash size

530Am 3 /s
130C
7.0% v/v
10% w/w
5.0 x 1011 Q-cm
50% < 20 11m

The designer will have his own database of tests that will enable a series of
relationships to be determined. Typically a relationship between effective
migration velocity (EMV) and the coal/ash characteristics (resistivity) is
determined from test data.
These tests will have been conducted under various operating conditions,
so the EMVs will have to be normalised to set levels of treatment time,
temperature, H 2 0, dust loading and collector plate spacing; otherwise any
relationship will be meaningless. The normalised levels will be determined
by the designer based on his experience.

183

ASSESSMENT OF THE PROCESS

1.2
1.1
>
::;:

1.0

;!;

(!J

0.9

0.8

z
<
J:

>
1=
<
...J
W
II:

0.7
0.6
0.5
0.4

% SULPHUR IN COAL + % SODIUM OXIDE IN ASH

Figure 7.1 Relative change in EMV vs. coal ash characteristics.

Figure 7.1 presents a typical relationship between EMV and some


coal/ash components [1]. For this the normalised levels are as given below:
Dust loading
Temperature

15G/Nm 3

H 20

8%v/v
50% < 20ltm
300mm

Particle size
Plate spacing

130C

Although the curve is plotted as the Deutsch EMV (w), the form would
be similar for the modified Deutsch (w k ), or an EMV based on any other
theory that the designer considers acceptable. If w is used, then the EMV
has to be corrected to a constant treatment time, but if the relationship is
based on Wk , then the EMV is automatically corrected for variations in
treatment time.
The EMV determined from the coal/ash characteristics is then corrected
to the customer's design parameters, by multiplying it by the relevant factors
related to temperature, H 2 0, particle size and dust loading. The designer
then has a design EMV that can be used to determine the ESP plate area
or treatment time.
A temperature relationship for a difficult fly ash is illustrated in Figure
7.2 [2]; this as explained in chapter 6 reflects the effect of temperature,
primarily on the electrical resistivity of the fly ash, the efficiency (EMV)
decreasing with increasing temperature to reach a minimum at around
160C, then rising as the resistance becomes more dependent on ionic rather
than surface conduction.

184

PERFORMANCE DESIGN CONSIDERATIONS


2.5.-------------------------,

>
~
w

2.0

Cl

z
<
::t:

1.5

>

i=
<
-'
w
a:

1.0

0.5 L-~----'-_~__'_~_ _L_~____"'__~__'__~__'_~_"__~----'


1 00
1 20
1 40
1 60
1 BJ
200
220
240
260
GAS TEMPERATURE DC

Figure 7.2 Relative change in EMV vs. gas temperature.

The main effect of moisture is to modify the electrical operating conditions; generally, the higher the moisture, the higher the operating voltage
and hence performance.
The effect of particle size is complex in electrostatic precipitation, since,
theoretically, the number of elementary charges that a particle receives is
proportional to the surface area and hence its diameter. Particles below
0.2 J1.m diameter receive their charge by diffusion processes as a result of
Brownian motion, while particles greater than 2 J1.m are charged by ionic
collision means. This gives the typical dip in the fractional efficiency curve,
as illustrated in Figure 7.3 [3] and in theory, chapters 3 and 8. This means
that there is a specific relationship between particle size and performance
which is typically based on the median or d 50 size of the particles and hence
specific surface.
Dust loading impacts on the precipitator size and configuration in the
following manner. First it affects the design efficiency assuming a constant
emission is required. Another aspect is the possible development of significant space charge or corona suppression, which requires special consideration in terms of discharge electrode design. Large dust catches, from high
dust loadings, demand proportionately sized hopper and dust evacuation
systems to prevent hopper overfilling, leading to mechanical damage to the
precipitator internals.
A similar methodology, using a data bank relating to the different
processes listed in section 7.3, would be used to establish the design
parameters/contact time, etc., for each different application. In these instan-

185

PLATE SPACING

1.3
1.2
>
:::;:
w

1 .1

c:J

c(

:I:

1.0

>

i=

c(
..J

0.9

a:

0.8
0.7

4 5

10'

4 5

100

4 5

10'

PARTICLE SIZE IN MICRONS

Figure 7.3 Relative change in EMV vs. particle size relationship.

ces, the variables would be related to functions other than the fuel and ash
analyses.

7.4

Plate spacing

Before moving on to consider the configuration of the ESP, we should


review the question of plate spacing.
Traditionally the ESP size has been evaluated using the EMV. The EMV
is a precipitation industry design tool that is variable and is influenced by
the process conditions given above and other factors. Before proceeding it
is worth expanding upon the background of the Deutsch equation and
EMV.
The basic equation of electrostatic precipitation was first developed by
Walther Deutsch in 1922 [4]. Deutsch used a simple wire and tube type ESP
to carry out his theoretical calculations, from which the collection efficiency
was derived as a function of tube dimensions and electric field strength:
exp - (2E"NL/Rv)
where E is the field strength, w is the particle velocity per unit field strength,
L is the tube length, R is the tube radius and v is the gas velocity.
As it was not possible to physically measure the factor 'w', practical
precipitation engineers replaced it by the empirically determined EMV. At

186

PERFORMANCE DESIGN CONSIDERATIONS

the same time, the tube dimensions and gas velocity were transposed into
collecting plate area (A) and gas volume (V). This changed the original
Deutsch relationship to the more well known:
exp - (EMV . A/V)

In this relationship, the influence of plate spacing and gas treatment time on
EMV is completely absent. Deutsch's original relationship contains both of
these factors by considering the tube radius, gas velocity and tube length.
The above relationship between EMV and A/V (specific collecting area),
became the accepted norm for many decades. The EMV was considered to
be independent of collector plate spacing and the trend in the 1960s and
1970s specifying SeA forced ESP suppliers (against more enlightened
suppliers' better judgement) to retain narrow spacing (220- 300 mm) in
order to be competitive.
There were some exceptions to this, e.g. on specialised applications where
severe corrosion demanded expensive fabrication materials; in these, the
plate spacing was successfully increased to reduce the cost of the installation.
In fact, in the early days of commercial precipitation, investigations were
successfully carried out using tube diameters of up to 48 inch (1.22 m), but
power supplies limited their practical usage.
It was not until the 1980s, before 'wide spacing' (400mm) was to be
considered acceptable on a large scale and, with the exception of some
unfortunate experiences in the United States, by the mid-90s, 400mm
collector spacing had been accepted throughout the world as a precipitator
'norm'.

7.5

Configuring the ESP

Once the required collecting area has been determined, the designer has to
decide how to configure the ESP. There is almost as much skill and
experience required to do this as is needed to determine the collecting
area/contact time for a specific duty.
The designer is confronted by a large range of decisions:
(a)
(b)
(c)
(d)
(e)
(0

(g)
(h)
(i)

How many series fields?


What size of collecting plate?
How many transformer rectifier sets?
How many electrical bus sections?
Any special insulator considerations?
Any special rapping considerations?
What type of inlet mouthpiece?
What aspect ratio is acceptable?
What gas velocity is acceptable?

CONFIGURING THE ESP

187

(j) Dust hopper considerations?


(k) Cohesivity/angle of repose of dust?
(1) Any corrosive dement in the process?
The choice of collector size is generally dependent on the quantity of gas
being treated but, as a general rule, the taller the collector plate the lower
the cost. This basic consideration has to be balanced against the number of
fields required, acceptable gas velocity range, minimum acceptable aspect
ratio and the effect of wind loads on the support structure design.
All ESP suppliers have a range of collector plate sizes, from a minimum,
of perhaps 1 m x 4 m, to a maximum size of say 5 m x 15 m. Over this entire
range the collector plate should: (a) be manufactured to the required
tolerances to ensure good electrical clearances; (b) be capable of being
rapped with sufficient energy to adequately remove the dust; (c) maintain
the electrical clearances within the life of the plant, i.e. remain mechanically
stable; (e) be manufactured and installed cost effectively.
Coupled with the collector design, the discharge electrode system must
also be capable of maintaining the electrical clearances in a cost-effective
way, whether the discharge electrodes are mounted in a mast, bedstead
frame, or are of the rigid 'unbreakable' type, as illustrated in Figure 7.4.
It will be appreciated from chapter 4 that the choice of discharge
electrode format to meet a particular process is decided from the particulate

Weighted wire
(shrouded)

Rigid frame
(bedspring)

Rigid frame
(strung mast)

Figure 7.4 Discharge electrode formats and mountings.

Rigid electrode
(Dura-Trade)

188

PERFORMANCE DESIGN CONSIDERATIONS

loading and the size analysis of the dust being presented at the inlet. As most
large precipitators comprise many kilometres of discharge electrode element,
the cost of special profiles can prove expensive. The designer must consider,
from the dust at the inlet, whether there is likely to be a significant space
charge or corona suppression effect which will demand the use of controlled
or high emission type electrodes. High emission electrodes usually have a
complex configuration, and are hence not only more costly than a simple
profile, but, if the corona suppression is not prevalent, then there will be a
very high power consumption. This may mean that the selected transformer
rectifier set is undersized to meet the required duty.
The ratio of the overall collector length to the collector height is called
the 'Aspect Ratio'. Each designer will have an acceptable minimum Aspect
Ratio depending upon the designer's experience, the design of the collector
plate, the nature of the dust, the gas velocity in the electrical field and the
quantity of dust being treated. For a given dust, the Aspect Ratio tends to
increase as the inlet dust loading increases, and/or the emission requirements decrease. This rule can be modified if the dust is of a type that does
not re-entrain easily.
The designer has to decide what is the minimum number of series fields
that are required to achieve the required performance. For example:
Required efficiency < 98.0%, two fields would be used.
Required efficiency ~ 99.0%, three fields would be considered.
Required efficiency> 99.7%, four or more fields would be the norm.
Choosing the number of fields is a compromise between minimum Aspect
Ratio and cost. Generally the greater the number of series fields, the higher
the overall cost, as each field is usually energised by its own transformer
rectifier set or sets and has its own hopper. One exception to this is when
precipitating a relatively easy dust in a small plant, where the cost of a
transformer rectifier set is a large proportion of the total cost; in this case,
a significant saving can be achieved by energising, for example, three
mechanically independent fields with two transformer rectifier sets, one on
the first field and one being shared by fields two and three. The plant readily
absorbs the power to achieve the performance, but the three independent
mechanical fields enable the rapping to be optimised.
The decision on the number of transformer rectifier (TR) sets is also an
important consideration when designing large ESPs [6]. The decision is
easy if the customer specifies the maximum plate area to be energised by
each TR, however, more often than not, the designer has to make his own
decision and he has the usual conflict between technical acceptability and
economics.
The amount of subsectioning within each field depends upon the design
of the ESP. As each subsection requires support insulators (generally two or
four), the fewer the number of sections, the lower the cost. There is always

189

CONFIGURING THE ESP

a mechanical limit to the size of each subsection, which may limit the total
number of subsections on a large power station plant, but the client may
also specify a minimum number, so that there is more security should a
subsection fail for any reason.
The gas velocity in the ESP treatment zone is an important design
parameter. Most customers consider a 'rule of thumb' value of about 1 mis,
as an acceptable precipitator gas velocity; however, many vendors have
experience of good performance with considerably higher gas velocities of
1.7 mls and higher for some applications. Generally, the higher the Aspect
Ratio andlor the greater the number of fields, the higher the gas velocity
that can be tolerated. Obviously there is a maximum velocity that should
never be exceeded, as rapping re-entrainment and possible gas scouring will
have an increasingly deleterious effect on the emission [7].
The designer often has a considerable problem in reconciling the economics of the configuration and the gas velocity. For a given design treatment
time and collector height, the only way the designer can alter the gas
velocity is to vary the ESP field length. This is illustrated by the following
table:

Collector size (m x m)
N urn ber of fields/length (m)
Number of gas passages
Aspect ratio
Gas velocity (mjs)
Treatment time (s)

Case 1

Case 2

Case 3

3 x 10
3/9
22
0.9
1.14
7.92

4 x 10
3/12
17
1.2
1.47
8.16

4.5 x 10
3/13.5
15
1.35
1.67
8.10

As can be seen, when the treatment time is virtually unaltered, the gas
velocity increase is a function of the length of the field. Changing the
collector height for the same treatment time will not alter the gas velocity.
There is always a possible conflict between acceptable Aspect Ratio, gas
velocity and economics. The conflict is particularly apparent when designing
very large ESPs and the designer is trying to utilise the largest collector plate
in his range. With an application that requires a low gas velocity and a
reasonable Aspect Ratio, the designer would be forced to reduce the height
of the collector and therefore increase his costs. The responsible designer,
however, will always put technical considerations before cost, so the plant
would normally have the reduced collector height configuration.
Many common processes, such as PC boilers, MSW incinerators and
cement kilns, present relatively minor process problems to the designer. The
configuration of the dust hoppers, inlet mouthpieces, type of insulators, etc.,
all come within a standard range of designs and experience. However, the

190

PERFORMANCE DESIGN CONSIDERATIONS

designer is also expected to produce designs for processes that may present
special difficulties outside the pure precipitation duty.
Some processes require special consideration to be made of the inlet
mouthpiece and gas distribution devices, to ensure freedom of dust deposition so that the gas distribution will remain unaffected. On non-ferrous
roasting processes, the dust can be very high in sulphates and is therefore
very sticky, so special provision is made to make sure the dust does not hang
up within the internals and hoppers. In this respect, the form of rapping in
terms of intensity needs consideration, particularly if the dust is very
cohesive requiring very intense rapping and therefore the type of suspension
and anvil may be of a special design to ensure maximum plant availability.
An area which needs careful consideration is that of high temperature
applications, e.g. those over 300C [8J, where differential thermal expansion
between casing and substructure may mean that the width of the casing is
somewhat limited, so that the foregoing optimum sizing considerations, e.g.
bus section size, number of TRs and other factors, have to be reassessed to
produce the lowest precipitator cost, while still minimising technical risk.
Other applications will demand special designs of insulators which would
also have an impact on the final configuration and cost.

7.6

Conclusions

At the end of the design process, which will have considered all the above,
the designer will have sized and configured the plant to meet the customer's
requirements at the lowest cost. This complex process is inevitably a
compromise but, as stated above, the last area to be compromised should
be the technical credibility of both the designer and the plant.
Although each major precipitator supplier will have his own data bank,
on which he will base the design and size of any new application, the
methodology and approach will be similar to that outlined. It is not the
intention of this chapter to contain sufficient information to enable the
non-specialist engineer to design a precipitator, but to explain how the size
was derived.
It is surprising that in spite of the many different precipitator designs that
have been used industrially, the size of any competitively sized precipitator,
for a given duty, is roughly similar, regardless of the supplier. While all
suppliers would consider their knowledge superior, in actual practice the
data must be comparable; otherwise there would be considerable differences
in the plant sizings.

REFERENCES

191

References
1. Darby, K., Cottingham, C. R. and Radai, J. (1991) The influence of sodium in fly ash on
electrical resistivity and its impact on precipitator performance. EPRI/EPA 9th Particulate
Control Symposium. Williamsburg, USA October 1991, Session 6A EPRI TR 100471 2, Palo
Alto, CA, USA.
2. Electricity Commission of New South Wales Research Note No. 59. Liddell Power Station
Investigation of Requirements of Electrostatic Precipitators, ECNSW, Sydney, Australia.
3. Parker, K.R. (1990) The wet ESP and its role in modern pollution control. Proc. 10th Clean
Air Society for Australia & Nell' Zealand Conference, Auckland, March, pp. 23-30, Clean
Air Soc., Auckland, NZ.
4. Deutsch, W. (1922) Bewegung und Ledung der Electricilatstrager in Zylinder Kondensator.
Ann. Phys., 68, 335-44.
5. Darby, K. and Novogoratz, D. (1990) Increased plate spacing in electrostatic precipitators.
EP RI/ EP A 8th Particulate Control Symposium, San Diego, CA, Session SA EPRI, Palo Alto,
CA, USA.
6. Ramsdell, R.G. (1973) Practical design parameters for hot and cold ESPs. American Power
Conference, Chicago, May.
7. Dalmon, J. and Lowe, H.J. (1961) Experimental investigations into the performance of ESPs
for PF power stations. Colloque International-La Physique des Forces Electrostatiques et
leurs Applications. Centre National de la Recherche Scientifique, Paris 1961.
8. Darby, K. and Parker, K.R. (1975) The electrostatic precipitator at high temperature.
National Society for Clean Air Annual Conference, Newcastle-on-Tyne, March, Clean Air
Soc., London, UK.

Electrical operation of precipitators


V.REYES

8.1

Introduction

The operation of any precipitator is closely related to its electrical energization, i.e. the way the power is delivered to the precipitator. Therefore, this
aspect will be covered in detail. The ancillary electrical equipment used in
the precipitator, such as rapping systems, insulator and hopper heaters and
purge air system, will not be discussed.
The electrical energization is the key factor for a satisfactory precipitator
operation. One of the pioneers and main contributor to this field [1] once
expressed it very clearly: 'a precipitator functions by the Grace of God and
electrical energization; if electrical energization is good, one's state of Grace
in other areas can perhaps be somewhat less than perfect'.
The technology and equipment used in the electrical operation of
precipitators have undergone a considerable improvement after the second
World War. This is mainly a result of considerable efforts in research and
development in Japan, USA and Europe. In the following, only the technology available at the present will be presented. This includes traditional DC,
intermittent and pulse energization. A fourth method is emerging, based on
the switch mode power supply technology. This is a very promising
technology but at the moment its use is limited to low current levels. For
this reason and for lack of space, this technology will not be covered.
Moreover, for the same reason, an historical review of the development of
the electrical equipment used in precipitators will not be included.
8.2

Precipitator performance and electrical energization

This dependence can be derived from the fundamental equation for precipitator efficiency [2]:
'1 = 1 - exp ( -

gill)

(8.1)

il

where A is the collection surface, Q is the gas flow rate and


is the
particle-migration velocity.
The migration velocity,
is a direct measure of the rate of collection of
the particles. From equation (8.1) the inverse proportionality between and
the collection surface A indicates that an improvement in means that A

il ,

il

il

PRECIPITATOR PERFORMANCE AND ELECTRICAL ENERGIZATION

193

may be correspondingly decreased, provided the collection efficiency remains at the same value. Therefore, ill is considered as a generalized
performance parameter and it is the natural link between electrical energization and collection efficiency.
From laboratory measurements on pilot precipitators, and from theoretical analysis, it has been found that ill is proportional to the product of the
mean and the peak precipitator voltage [2]. Collected data from different
plants have shown that ill is also proportional to the mean precipitator
current. Hence, ill may be expressed by
(S.2)

where Pc is the corona power and kl is a constant which depends on the


gas state, particle composition, precipitator geometry and size, etc. PcI A is
the corona power density. If ill is expressed in ft/s (1 ft/s = 30.5 cm/s)
and the power density in W/ft 2 (1 W/ft z = 10.75 W/m 2 ), as seen in Figure
S.l, the proportionality constant kl has been estimated to be 0.67 [2].
Pc takes into account the ill dependence on precipitator current and
voltage. As will be seen in section S.5, the precipitator current and voltage
delivered by commercial power supplies are not pure DC magnitudes, so a
simple calculation is not possible. By using the actual waveforms and
computer calculations, a useful expression for Pecan be determined [3]:
(S.3)

1.4 , - - - - - - - - _ - -_ _ _ _- - - - - - ,
(A)

1.2

~
3

'0

_____ ~(B)

~ 0.8
c

..8
f!

----

......

...........

...........,...................

0.6

0>

:E 0.4

(C)

0.2

oro--~0~.2--70.74--0~.6~-~0~.8~--1~-~1.~2-~1.4
Corona power density

[W/ft21

Figure 8.1 Migration velocity as a function of corona power based on empirical data. Reference
line redrawn from White [2].

194

ELECTRICAL OPERATION OF PRECIPITATORS

Equation (8.3) indicates that the collection efficiency depends directly on the
mean voltage, the peak voltage and the mean current of the precipitator.

8.2.1

Examples

To illustrate the dependence of the migration velocity, w, on corona power,


three practical examples with traditional DC energization have been chosen
and represented in Figure 8.2.
(A) A fly ash precipitator (one chamber, two bus sections) for a boiler
burning an English/Polish coal blend (S = 1%). In Figure 8.2, w is
plotted as a function of the corona power density, for this case, where
the dust resistivity is low. This curve shows a steady increasing w for
an increasing power density. The corona power density is limited by
sparking in both sections at around 60- 70% of rated current, but at
this level the curve becomes flat. This means that optimum precipitator
performance is obtained with approximately 14 W/m2 of collection
surface. The migration velocity used here is that modified by Petersen,
i.e. W B [4].
(B) A fly ash precipitator (two chambers, eight bus sections) for a boiler
burning high S (2.6%) American (Appalachian) coal. In this case, the
migration velocity w is lower than in the previous case, but the
corresponding curve in Figure 8.2 also shows an increasing w for an
increasing power density, which is typical for low resistivity conditions.
(C) A fly ash precipitator (two chambers, five bus sections) for a boiler
burning low S (0.26%) South African (Middelburg) coal, causing

24
22

~
.
.,

'5

~B

.-...-~-~

18
16

C'

14

12

Qi

>

10

c:
0

~Cl
:E

1_- -:" A

20

-I

- t .

-I-A:: British/Polish rIlediUmicoal


._

~---- -

___ _

C: South African low S coal

4
,

2
00

------r-

B: American high pcoal___

-----1-8

10

12

14

16

Corona power density [W/m'j

Figure 8.2 Migration velocity as a function of the corona power for three different coal blends.

CORONA SUPPRESSION AND SPACE CHARGE EFFECTS

195

resistivity problems. In this case, the optimal migration velocity w is


lower than in the previous case and it is clearly seen that above a power
density of 6 W1m 2 , the performance of the precipitator is almost
constant, with a slight decreasing tendency. This suggests that 6 W1m 2
is the limit for the useful corona power and the surplus power is wasted
in back-corona generation.

In the case of severe back-corona, occurring over most of the collection


surface in each bus section, the limit for the useful corona power is more
definite as the precipitator efficiency begins to fall, as suggested by Figure
8.14.
The curves in Figures 8.2 and 8.14 indicate the dependence of w on the
resistivity of the fly ash, which also determines the useful corona input
power. This must not be confused with the installed power, i.e. the rated
power that can be delivered by the high voltage power supplies. The useful
corona power can be defined as the power level giving the optimal w or
performance.
In Figure 8.1 the results of the three examples are also plotted. These
show a large divergence in relation to the reference line.
To summarize
The electrical energization is, in practice, one of the most important
factors in obtaining a high precipitator efficiency.
The particle-migration velocity w is the basic link between precipitator
efficiency and electrical energization.
The corona power can be expressed as a function of the mean current, the
mean voltage and the peak voltage.
The particle-migration velocity w may be related to different electrical
quantities, the corona power seeming to be the most appropriate.
The proportionality constant kl can be used as a reference, but the ESP
manufacturers base the sizing of their precipitators and the useful corona
power for a particular process from their own collected data. (Both factors
are closely related to the resistivity of the collected dust.)
8.3

Corona suppression and space charge effects

These effects are better understood by analysing the current-voltage relationship for a particular precipitator under different operating conditions.
The determination of the current-voltage (i-v) characteristics has been
thoroughly covered in previous publications [2,5, 8], so this matter will not
be reviewed in detail here.
The theoretical determination of the i-v characteristics for a duct
electrode (wire-plate) geometry is a very difficult task. This requires the
solutions of the quasi-static electrodynamic Maxwell equations, relating the

196

ELECTRICAL OPERATION OF PRECIPITATORS

electric field, the potential and the space charge with the resulting current
density field. The problem arises from two particularities of the electrical
conduction in gases:
the electric field that transports the charge also creates it, i.e. when the
precipitator voltage exceeds the corona onset level, the ion charges move
more rapidly and new ions are generated at a higher rate .
the moving ionic space charge is comparable with the electrostatic charge
on the surface of the electrodes, i.e. the ions distort the field which both
generates and transports them.
These characteristics of gas conduction produce a strong interaction
between electric field and space charge density, which makes it impossible
to calculate these two quantities independently of each other. Furthermore,
the differential equations governing the phenomena are highly non-linear.
8.3.1

Electrical characteristics with air load

The i-v characteristics have been determined by analytical methods [7J,


using an approximation in relation to the ion space charge effect. The
resulting equation is shown in Appendix 8.A. Here can be seen the influence
of the electrode geometry, temperature, pressure, ion mobility, etc., when the
precipitator is clean and no gases are passed through it (air load).
The i-v characteristics can also be determined by using numerical
analysis, normally by using computer calculations, which is more and more
extended, as explained in chapter 9. In one of proposed methods [9J the
current density is determined for different positions in the interelectrode
space for various values of applied voltage (see Figure 8.5). The average
current density is then determined for each voltage level, and these values
are plotted in the traditional way.
It is possible to find examples of good agreement between calculated and
measured i-v characteristics obtained from both methods, but due to
limited space, only one example wiII be given. The i-v characteristic shown
in Figure 8.3 corresponds to an industrial precipitator for a waste incineration plant having rigid electrodes and 400mm duct spacing, and indicates
a good agreement between the measured and the calculated i-v characteristic.
8.3.2

Characteristics with dust load

For the same precipitator, i-v characteristics have been obtained with the
precipitator in operation. The characteristics shown in Figure 8.4 correspond to the first field and the last field. These have been measured for 100%
boiler load, a dust load at the ESP inlet equal to 5 g/Nm 3 and a gas

197

CORONA SUPPRESSION AND SPACE CHARGE EFFECTS

-------

_______

0000

-----r--

-+

--

--

--

mjaSlJr~
calculated

0.1

'"c
Ql

"t:J

"E
~

air load

0.01

Ql

- - - -. - -- ---_.
- ---------~

.. _ - _ - - 1 - - -

--~

-~-----I

---

---~

----

0.001 20

40

30

50

80

70

60

Applied voltage [kV]

Figure 8.3 Average current density as a function of applied voltage in the case of air load and
clean precipitator.

temperature of 165 DC; again, good agreement between the measured and the
calculated characteristic for the first field has been obtained [9].
These i-v characteristics illustrate the effect of the particle space charge.
As the particle concentration falls along the precipitator, so does the particle
space charge density. The charged particles in the interelectrode space
contribute to the total space charge density, i.e.
(8.4)

E
:cr:

..

0.1

in

- ~---- -

Q)

"C

E
~

..

~., ~

-.:------~

.---t::::/-/"

0.01 -=-- - j - j --=--

-~--_____t______----t

~---,-

---.

---------

dust loaded ESP

~==--~=

e :~ ==r-=--l~~~~--==-- c~.
Q)

C1
I

c.... T - - - - - - - - - + - -

. ~ --.

0.00120.1:---+=---1---,;,---1----+-----7.:..0----180

Figure 8.4 i-v characteristics for first and last section with dust-laden gas and normal
operating conditions.

198

ELECTRICAL OPERATION OF PRECIPITATORS

where Pi is the ion space charge density and Pp the particle space charge
density expressed normally in C/m 3 .
Owing to the low drift velocity of the charged particles, in comparison
with the gas ions, the space charge effect of the particles is greater than that
of the ions, especially in the region of high particle concentration. In the
following it will be assumed that the particle space charge is negligible in
the last field. Then the corona onset potential v:, will have the value shown
in Figure 8.4.
The particle space charge in the first field partly shields the discharge
electrodes and the effect is to weaken the electric field near the discharge
electrode surface. In order to attain the critical corona onset field, it is
necessary to apply a higher voltage to the discharge electrodes. This is
equivalent to an apparent increase in the corona onset potential and the
result is that the i-v curve is displaced towards higher voltages with respect
to the i-v curve of the last field.
Assuming the particle charge density is constant in the first field, the
apparent corona onset potential
can be expressed [5] as follows:

V;

(8.5)
where Eo is the permittivity in the free space and s is one-half duct spacing
and Pp is expressed by:
(8.6)

where p is a dimensionless constant related to the relative dielectric constant


of the particles, E is the electric field and Sp is the particle surface area per
unit volume of gas (m 2 jm 3 ).
The effect of the particle space charge is a partial suppression of the
corona current. This is seen by comparing the two curves in Figure 8.4 for
the same applied voltage. It is seen, for the same voltage, that the current in
the first field is lower than that in the last field. Under extreme conditions,
with a very large particle concentration (or a large Sp) at the inlet, the
corona can be suppressed or quenched totally.
The corona suppression can also be observed, especially at the inlet field
of the precipitator. By using a computer model it has been possible to
determine the current density along the field in the gas direction for various
voltages [6,9]. This is illustrated in Figure 8.5 for the same precipitator. It
will be seen that the current is suppressed at the inlet of the field and then
starts increasing as the particle space charge decreases along the field
because of the precipitation of the particles. This allows the ion space charge
to increase and in turn the current density increases. When the ion density
increases, the particles receive a larger amount of charge and become

199

HIGH TENSION SECTIONALIZATION

200

54kV

150

--

<c

2;

~
CI)
c:

49kV
100

Q)

"'C

C
~
=:;

50-

()

36kV
0

Distance in gas direction 1m]


Figure 8.5 Average current density along the inlet field as a function of the distance for various
volta!;e levels.

precipitated faster. This interaction results in a sudden increase in the


calculated current density which can be interpreted as a threshold between
the low and high corona regions. For high dust concentration at the inlet,
or low applied voltages, the threshold moves towards the outlet of the field.
8.4

High tension sectionalization

One of the problems encountered by the designer of electrostatic precipitators, in order to achieve the required collection efficiency at minimum risk,
is whether to use a few large bus sections or more smaller bus sections, i.e.
to find the optimal degree of sectionalization.
From the previous section, 8.3, it becomes clear that a large electrostatic
precipitator cannot be designed as a single bus section, with all the discharge
electrodes energized by one common high voltage power supply, because of
the uneven current density along the precipitator resulting from the particle
space charge, which falls as the dust laden gas becomes cleaner and cleaner.
With large particle concentrations, the space charge effect causes partial
corona suppression. This effect is larger for the discharge electrodes near the
inlet and may be negligible for those near the outlet; therefore, it is
impossible to match inlet and outlet discharge electrodes electrically. If all
the electrodes are energized from one source, the outlet electrodes will
consume most of the current and the converse will happen with the inlet
electrodes.
Obviously, the discharge electrode arrangement must be sectioned in
order to obtain good precipitator performance. The maximum precipitator

200

ELECTRICAL OPERATION OF PRECIPITATORS

efficiency will be theoretically obtained by energizing each discharge electrode separately [2J, i.e. optimum energization.
There are other reasons for not using few large bus sections; probably the
most important is the reduction in the sparking voltage level, and in
consequence the operating voltage and corona power, resulting in a loss of
precipitator efficiency. Other reasons are that the construction details are
never perfect after the precipitator has been in operation for a long period
of time. These may be:

misalignment of the discharge electrodes


sharp points or corners in the collecting plates
vibration of the discharge electrodes
excessive dust build-up
electrode warping

Anyone of these factors can produce a localized reduction of the sparking


level.
Another reason for a poorer efficiency, when using very wide bus sections,
is bad gas distribution, or put in another words: it is more difficult to obtain
a good gas distribution when very wide and/or high sections are used.
Other authors have mentioned, as a problem, the mismatch between the
capacitance of a large bus section (C is proportional with the collection
area) and the reduced internal impedance when large TR sets are used. The
result may be very large current surges when sparking occurs, which is
considered as poor electrical stability. This, however, is no longer a problem,
as this can be solved as explained in section 8.5 and [10]. A difficulty that
still exists, in the case of localized sparks, is a potential reduction in the life
of the electrodes, because of spark erosion due to the larger energy
dissipated during the spark. The stored energy in the system immediately
before the spark is proportional to the capacitance of the bus section and
the square of the applied voltage (Q = iCV 2 ).
Finally, there are a number of aspects which also have to be mentioned.
One is the higher rapping losses where large sections are used. The other is
the case of back-corona, where smaller electrically optimized sections could
improve the precipitator performance [9]. This latter aspect is based on
theoretical analysis, but its benefits have to be demonstrated by full-scale
tests.
All these factors reflect a need for a larger degree of sectionalization.
Ramsdell (see [2J and [6J) has derived an exponential relationship between
the collection efficiency and the number of bus sections needed per
300000 m 3 /h of gas treated. This is based on performance data for more
than 30 fly ash precipitators. The relationship is plotted in Figure 8.6; the
curve shown is valid for the utility system analysed by Ramsdell, but for
other coal types and operating conditions it is displaced. In spite of the
curve being only roughly correct, it serves as a valuable guide in the design

TRADITIONAL DC ENERGIZATION
99

201

r------,-----,-~-____,

97
~

~ 95
~

-0
~

90 f------+-->'-----+---------l

~
~

o
(.)

70

50

4
2
6
Number of bus sections per 300.000 [m 3/h]

Figure 8.6 Necessary number of sections per 300 000 m 3 /h gas recommended in order to obtain
a required collection efficiency. Redrawn from [6].

of fly ash precipitators. Those precipitators which fail to meet the requirements expressed by Figure 8.6 may be expected to perform poorly or
marginally on the average [2].
There are many examples of improved performance when two bus
sections, previously connected in parallel and energized by one TR set, have
been separated and energized by its own TR set. The reason for not using
a larger degree of sectionalization is normally economic, as the cost of power
supplies and high voltage insulation is lower for large sections. This puts
economic pressures on the design engineer who tends to minimize the
number of sections. A more profound analysis, however, may indicate that
a smaller precipitator with more sections and TR sets could be more efficient
and cheaper. Finally, it should be mentioned that smaller and cheaper
power supplies could economically allow an increase in the degree of
sectionalization, so that more efforts in this area are necessary.
8.5 Traditional DC energization
The traditional DC energization used in the last two decades is obtained
with the power supply depicted in Figure 8.7. In order to control the corona
power, the line voltage is regulated by a thyristor controller (phase control
by a pair of antiparallel thyristors) before it is applied to the primary of the
high voltage transformer. By choosing a suitable turns ratio (n) the primary
voltage is raised to the desired secondary level and then rectified by a high
voltage silicon bridge rectifier. The rectified secondary voltage is applied
directly to a precipitator section without additional filtering.

202

ELECTRICAL OPERATION OF PRECIPITATORS

Phase control thyristors

High voltage oil tank


HV transformer

HV rectifier
Precipitator
bus section

ACline!

~~----------------~
Figure 8.7 Main circuit of a traditional power supply for electrostatic precipitators.

The high voltage rectifier is connected in such a way that the output
voltage has a negative polarity, i.e. negative corona is generated in the
precipitator.
Normally a current limiting inductor (Ls) is included in series with the
primary winding of the transformer in order to increase the short-circuit
impedance of the power supply to limit the current during sparking, arcing
or short-circuit in the ESP.
8.5.1

Basic principles

The output voltage of the power supply, applied to a precipitator bus


section, is controlled by varying the firing angle of the thyristors, i.e. by
delaying or advancing the firing instant in relation to the zero crossing
(cross-over) of the line voltage.
The principle is shown in Figure 8.8 for a 50 Hz power supply. This shows
the waveforms obtained when the thyristors are fired at a time equal to 3 ms
after the line voltage zero crossing, i.e. a firing angle cto = 180 x 3/10= 54 (Oel).
When a thyristor is fired, it changes from a state of high resistance to low
resistance and the primary current starts increasing and becomes as shown
in Figure 8.8a. Its magnitude and duration is determined by the current
limiting inductor (L.), the leakage inductance of the HV transformer and
the precipitator load.
The output current (io), the so-called precipitator current, has a waveform
corresponding to the rectified secondary current. The amplitude difference
between the primary and the secondary current is given by the turns ratio
(is = ip/n). The precipitator current and the resulting output voltage, the
so-called precipitator voltage, are as depicted in Figure 8.8b. These are
negative because of the negative corona, but in this chapter, for the sake of
simplicity, their absolute values will be used. These waveforms are typical
for a medium size power supply energizing a normal precipitator load.

203

TRADITIONAL DC ENERGIZATION

'E

.'

~ 400~--r...---r-~/v-~~---+---r---r--~--+-~

/~....+---+-+---+-+----l
2001----;"--+-/--+V-/--+---+'-:'-.\r

&

Or:--~-L+---r-~---+'---r~~--+---~~

~~

1.//
-4OO~--f---___I----+----+---+----+'..--r---+,,-""-.q/----l
-200

::::i

.......

~OOL-~--~--~--~~--~--~--~~--~

4 ms

8 ms

Time

12 ms

16 ms

20 ms

(a)

Precipitator voltage

2.5

.........................

""'"

.;..

II

c.

'il

!!! 0.5

a..

4ms

1OOk

V'

I X.
\ ...... II..;.. \
\
\
/

I y'"

'E
~ 1.5

- 0.5

Precipitator current

8ms

Time

12 ms

16 ms

-20k
20ms

(b)

Figure 8.8 Voltage and current waveforms with DC energization and a firing angle of 54
(3 ms).

The precipitator voltage has a considerable ripple, because of the inherent


capacitance of the precipitator section and the fact that no filtering is used.
The precipitator voltage is characterized by its:
peak value (Yapeak)
mean value (Yamean)
minimum value (Yamin)
The peak value depends on the charge delivered in one current pulse
(Qp), i.e. the area under the current pulse. Because the precipitator

204

ELECTRICAL OPERATION OF PRECIPITATORS

'E

~ 400r--~-~-~--7~-~-~-~

()

"&.

Q)

E -2001------j- ----L----r---c~-__t_-__t_--',~_;_7'I

(5

>

~ -400~-~-r_-~-~-r_-~-~-~~~~
::::;
-OOO~~--~--~------~--~~--~--~~

4 ms

B ms

12 ms

16 ms

20 ms

Time

(3)

2.5

Precipitator voltage
..........................

I
I

'E
~ 1.5

i3

".

.i

c.

.~ 0.5

c..

- 0.5

1\

Precipitator current
1OOk

I
I

I ....

............

I,'

1\

4ms

Q)

60kE
o
>

i5

1ii

20k

'5.
.~

a:

I
I

Bms

/1\

/
T. I

..... .......

,I

/1\1

1/ i

12ms

i
16 ms

-20k
20ms

Time
(b)

Figure 8.9 Voltage and current waveforms with DC energization and a firing angle of 108
(6ms).

capacitance is discharged in the time interval between current pulses the


voltage decreases. The voltage level can also be decreased by delaying the
firing angle tX o , resulting in a lower precipitator voltage as seen in Figure
8.9b. These waveforms illustrate the situation existing when the firing instant
occurs 6 ms after the zero crossing. Table 8.1 shows the most relevant
quantities and their values for both firing angles. These values were obtained
by computer-aided design evaluation.
It is clearly seen that both the precipitator voltage and the precipitator
current are reduced when a delayed firing is used. The extreme case

205

TRADITIONAL DC ENERGIZATION
Table 8.1 Current and voltages with two different firing angles
Firing angle

to(ms)

Primary current

i p,m,(A)

223

145

Precipitator current

io,m,(mA)
iopoak(mA)
iom,an(mA)

1400
2350
1030

920
1800
576

Precipitator voltage

VoP,ak(kV)

78
61
46

47
35
23

~moan(kV)
~min(kV)

corresponds to to = 10ms (ct o = 180) giving a stationary current and


voltage equal to O.
The earliest firing of the thyristor is influenced by the value of the
precipitator voltage at the firing instant which is equal to Vomin ' This is
greater than the corona onset voltage and is influenced by the precipitator
geometry and process conditions. With precipitator loads and a 50 Hz
power supply, the earliest firing is in the order of 2 to 3 ms, corresponding
to 36 to 54.
Another important quantity is the form factor (FF) of the precipitator
current, defined as
I

FF=~

(8.7)

IOmean

where
I arms

-1

iT i~(t)dt
0

(8.8)

and
(8.9)

The output current io(t) is periodic and Iio(t) I is its absolute value.
A calculation of the form factor of the precipitator current for to = 3 ms
gives a value of 1.4. This is a typical design value used in power supplies.
Since the precipitator current is not a pure sinusoidal wave, it is not easy
to calculate its mean and rms value. This can be done, however, by
computer simulation or by using approximations, as indicated in Appendix
8B.

206

8.5.2

ELECTRICAL OPERATION OF PRECIPITATORS

High voltage power supply ratings

The high voltage power supply for electrostatic precipitators is characterized


by the turns ratio (n) and the short-circuit impedance present in the main
circuit (X LsJ. It is predominantly inductive and is composed of the leakage
reactance of the transformer and the linear inductor L s ' as illustrated in
Figure 8.7. This inductor is added in order to increase X Lsc above 5 to 10%,
which is a typical value of the leakage reactance of the HV transformer. The
aim is to limit the current surges that might occur during sparking in the
precipitator and thus:
increase the lifetime of the precipitator internals
protect the electrical equipment, and
obtain a more stable electrical operation.
The value used by most of the manufacturers is in the order of 30-40%,
which reduces the current surges to 3.3 and 2.5 times the rated value,
respectively.
By knowing the short-circuit reactance, X Lsc' and the turns ratio, n, all
the important quantities of the power supply can be determined for different
loads. The ratings can be expressed in different ways, but in this chapter the
most usual practice in Europe will be used. The following quantities are
normally indicated on the rating plate:

Precipitator mean current (Ionom)


Precipitator peak voltage @no load (~nom)
Primary rms current (Ipnom)
Line voltage (VLnom ) and frequency (f)
Apparent input power (S)

rnA
kV pk
A
V and Hz
kVA

8.5.2.1 Precipitator mean current. The precipitator current has the waveform depicted in Figures 8.8b and 8.9b. Its mean value is given by
(8.10)
where io(t) is the precipitator current and T is the period of the line
frequency.
The rated precipitator mean current is the maximum mean current the
power supply is able to deliver to a load, without exceeding the design value
of the current form factor. Some manufacturers carry out tests on the power
supply with a RC-load, which simulates a precipitator load. Others use a
pure resistive load, which gives a lower current form factor. This means that,
at rated mean load current, the rms value of the secondary and the primary
currents have a lower value compared with the corresponding values of a
precipitator load.

TRADITIONAL DC ENERGIZATION

207

8.5.2.2 Primary rms current. The waveform of the primary current is


shown in Figures S.9a and S.9b. Its rms value is defined as
I Pnom

IT i;(t)dt

(S.11)

This can also be expressed by equation (S.12).


I Pnom = n FF Ionom

(S.12)

where n is the turns ratio, FF is the form factor used by the manufacturer
(typical value 1.35 to 1.4) and Ionom is the rated precipitator mean current.
8.5.2.3 Precipitator peak voltage at no-load. At no-load, the output
current of the power supply io = 0 and the primary current is equal to the
magnetizing current of the HV transformer. As this is negligible compared
with the rated primary current, the peak voltage at no-load is equal to

Onom

VI2nv,
L
Lnom

(S.13)

where VLnom is the rated rms value of the line voltage.


8.5.2.4 Apparent input power. This quantity is especially important in the
sizing of the electrical installation. The apparent input power is defined by
equation (S.14).

s=

I Pnom V,Lnom

(S.14)

i.e. the product of the line voltage and the primary current.
The active input power is given by
p = S cos ({Jl

(S.15)

where ({Jl is the phase angle between the line voltage and the fundamental
frequency component (50 Hz) of the primary current (cos ({Jl is normally
known as the power factor).
The active power cannot be expressed as a rated value, because it varies
with the precipitator load for the same rated precipitator mean current. The
active power is normally measured with a wattmeter or calculated by means
of computer simulation.
The power factor is normally better than O.S at rated current, assuming
the form factor is approximately 1.4.
8.5.2.5 Example. The waveforms depicted in Figure S.S are obtained with
a power supply having the following data:

Rated precipitator mean current:


Rated precipitator peak voltage:
Line voltage:

1000mA
90kV
400V rms

208

ELECTRICAL OPERATION OF PRECIPITATORS

Short-circuit reactance:
Design form factor:

35%
1.4

The precipitator capacitance used is 130 nF.


The turns ratio is
1 V

n =-

.Ji V

Omean

1 90000

.Ji

=--- =

Lnom

400

159

The precipitator rated rms current is

The rated primary rms current is


Ip,m,

nlo,m,

159 x 1.4 = 223A

The rated apparent power is


S=
8.5.3

Ipnom

VLnom = 223 x 400 = 89.2kW

Influence of the linear inductor

Because this aspect has not been illustrated sufficiently in the past, a brief
explanation will be given in relation to the importance of this component.
Advantages and disadvantages. The linear inductor's main function is to limit the current surges during sparking in the precipitator, but,
at the same time, it has a number of advantages related to the electrical
operation of the precipitator. To illustrate these advantages, Figure 8.10
shows the waveforms of the current and precipitator voltage when the linear
inductor is not used, i.e. the current is only limited by the leakage reactance
of the high voltage transformer. In Table 8.2 are shown the values obtained,
when the short-circuit reactance of the transformer is ~ 9%, for the same
rated mean current and voltage as obtained with a linear inductor. The
figures in parentheses correspond to the values obtained with a normal
short-circuit reactance of 35%.
Comparing these results with those obtained with a normal short-circuit
reactance, valid for the same precipitator mean current and precipitator
load, the following disadvantages can be pointed out:

8.5.3.1

The peak value of the precipitator current is higher and its duration is
shorter, giving a higher form factor (36% higher).
The higher form factor causes a higher primary current and apparent
input power for the same precipitator mean current.
The time occurrence of the precipitator voltage peak is closer to that of
the line voltage peak. Since sparking occurs around the peak of the

209

TRADITIONAL DC ENERGIZATION

.,,'

400

'

'1""""'Mi

,"-.'
,

'\,

/!

200

III

Primary current

Line voltage
..................

600

Ii

\"--J..
'

-600

1
4ms

1
I

B ms

'

'oj

r'.

-400

500

\i //:

-200

750

III

c:
::i

.......

12 ms

........

IF

250 'E
~
::::J

-250 it
-500

-750
20 ms

16 ms

Time
(a)

,~,~:'~,~~~~,t~~, y,?!~~,~~

Precipitator current

5,0 ~~=~::.::.::~-=:;:::=:=:::;::=--,---1'""----,---,100k

n'"

n)"'"

_ 1.0 '--_..Li----'_--'-_-L-_'-----'-_....L.._'------L_-"-20k

4~

~~

Figure 8.10 Voltage and current waveforms with DC energization and insufficient inductance
at a firing angle of 90 (5 ms).

precipitator voltage, the current surges will have a higher amplitude and
a longer duration. These characteristics are detrimental for stable operation of the precipitator, especially in relation to voltage recovery after
spark .
The precipitator current is obtained with a delayed firing angle, giving a
larger phase angle between the line voltage and the fundamental component of the primary current. This results in a lower power factor which is
not desirable by the power utilities.

210

ELECTRICAL OPERATION OF PRECIPITATORS

Table 8.2 Current and voltages with insufficient shortcircuit impedance (comparison with
values obtained with normal impedance level)
Firing angle

to(ms)

(3)

Primary current

ip,m,(A)

302

(223)

Precipitator current

i o,m,(mA)
iop<ak(mA)
iomoan(mA)

1900
4400
1000

(1400)
(2350)
(1030)

Precipitator voltage

~poak(kV)
~moan(kV)

90
61
38

(78)
(61)
(46)

Vom>n(kV)

As theoretical advantages, the following can be maintained:


The precipitator voltage waveform is more pulsating because the peak
value is higher. This could have a positive effect in the case of high
resistivity particles. (At the present this problem is solved by using
intermittent energization or pulse energization. See later sections.)
In the case of precipitator loads requiring a higher voltage, the power
supply is better suited to deliver its rated current because of the lower
voltage drop in the short-circuit reactance.
8.5.3.2 Physical implementation. The linear inductor may consist of an
air coil inductor or an iron core inductor with a suitable air gap giving a
linear characteristic. This is normally placed inside the high voltage tank
and its inductance cannot be changed. This is a good economic solution,
provided a good match exists between the size of the power supply and the
bus section.
Some manufacturers place the inductor inside the control cabinet, which
can then be provided with tappings. These allow one to change the
inductance value to overcome any mismatch between the power supply and
the bus section. This solution has the inconvenience of occupying space in
the control cabinet and causing acoustic noise.
In recent years, a variable inductor has been introduced, whose inductance value changes inversely with the value of the primary current inside a
certain range. An automatic control loop increases the inductance value
when the primary current decreases in order to keep the form factor
constant.
8.6

Intermittent energization

Intermittent energization (IE) is a recent method introduced in the early


1980s, with the purpose of saving energy and improving the collection

INTERMITTENT ENERGIZATION

211

efficiency with high resistivity dusts. This energization form is also known
under other trade names like energy control, semi-pulse, variopulse, etc.
IE operation emerged as a cheaper alternative to pulse energization
which was already developed and used commercially in the solution of high
resistivity dust problems.

8.6.1

Basic principles

Intermittent energization is obtained with the same electrical equipment


employed in traditional HV power supplies. The difference resides in the

(a)

~OOk

5.0

e 3.0 1--.;.....--\+--+-+--+---;...-+--+--+_---1 SOk S


5

I \

1.0

".

........ J

: \

I \

20k

- 1.0 L----I_---I._--I.._....I..._...!...._-"-_l...----I_---I._-l-20k

~~

~~

Time

(b)

Figure 8.11 Voltage and current waveforms with intermittent energization and a firing angle
of 54 (3 ms). Two out of three pulses suppressed (D = 3),

212

ELECTRICAL OPERATION OF PRECIPITATORS

automatic voltage control equipment, which allows the suppression of a


certain number of half-cycles of the primary current delivered to the
transformer by the AC line.
This suppression is obtained by not firing the phase control thyristors in
the respective half-cycles or alternatively, by using a firing angle 0(0 = 180.
The principle is illustrated by the waveforms in Figure 8.11, for a 50 Hz line.
This example shows six half-cycles of the line frequency and illustrates the
case where two out of three current pulses are suppressed.
Figure 8.11a shows the primary current in relation to the line voltage.
The corresponding waveforms of the precipitator voltage and current are
shown in Figure 8.11 b. The firing angle corresponds to 3 ms after the zero
crossing of the line voltage.

8.6.2

Comparison with traditional DC energization

The waveforms in Figure 8.11 show the following differences compared with
DC energization for the same firing angle in Figure 8.8:
The peak value of the precipitator voltage is higher.
The minimum (trough) value of the precipitator voltage is lower.
Due to the suppression of two current pulses, the mean and the rms values
of the precipitator current are reduced.
As a consequence, the following can be established:
The mean value of the precipitator voltage is lower.
The corona power delivered to a particular bus section of a precipitator
is lower.
The power consumption of the precipitator is reduced.
Table 8.3 Current and voltages with intermittent energization (comparison with values
obtained with DC energization)
Energization form

IE

(DC)

172

(223)

Primary current

I p,m,(A)

Precipitator current

Io,m,(mA)
IOP"k(mA)
Iom"n(mA)

1080
3100
476

(1400)
(2350)
(1030)

Precipitator voltage

~p,ak(kV)
~m,an(kV)

82
41
24

(78)
(61)
(46)

~m,"(kV)

INTERMITTENT ENERGIZATION

213

These results are summarized in Table 8.3 and discussed in the following
paragraphs:
8.6.2.1 Peak voltage. The peak voltage is higher because the area under
the current pulse is greater. As this area corresponds to the electric charge
Qp delivered to the precipitator section, the larger the value of Qp' the higher
the peak voltage because the precipitator load has an inherent capacitive
component.
8.6.2.2 Minimum voltage. Because of the longer time interval between
current pulses, i.e. without receiving electrical charge, the precipitator is
discharged towards the corona onset voltage. This causes a lower minimum
value.
8.6.2.3 Mean voltage. This is proportional to the area under the precipitator voltage in one energization duty cycle (in this example: three halfcycles of the line frequency). Because of the lower minimum value the mean
value also becomes lower.
8.6.2.4 Mean current and degree of intermittence. The mean value of the
precipitator current is reduced due to the suppression of a number of current
pulses. This suppression of current pulses is expressed by the so-called
'degree of intermittence' D.
D is defined as the number of half-cycles included in one energization
duty cycle divided by the number of current pulses in this time interval. In
the example shown in Figure 8.11 the energization cycle is three, so the
degree of intermittence D = 3. Another example: if the thyristors are fired
once and then kept blocked for the next ten half-cycles, then D = 11.
The degree of intermittence is also expressed by other names, e.g. the
so-called 'charge ratio' (). This is defined as the number of current pulses in
one energization duty cycle divided by the number of half-cycles included in
the energization cycle. For instance, the charge ratio in Figure 8.11 is:
() = 1:3

expressing that the IE operation, in this particular case, consists of one


current pulse out of three. The charge ratio is the reciprocal value of the
degree of intermittence.
Assuming the area under the current pulse is the same for IE and DC
energization, and if the mean current obtained with DC energization is I oe ,
the mean current obtained with intermittent energization lIE can be expressed by
I

_ IDe
IE-

(8.16)

214

ELECTRICAL OPERATION OF PRECIPITATORS

As shown in Figure 8.11 the area under the current pulses may sometimes
be higher with intermittent energization by a factor k. Then, the mean
current is
(8.17)
where k may typically vary between 1 and 1.5. The factor k in the example
shown in Table 8.3 is 1.39.
The mean current is then reduced by a factor kiD. Assuming that the
primary current pulse has a similar waveform in both cases, the rms value
is reduced by a factor equal to k divided by the square root of D.
There are two aspects where IE is inferior compared with DC energization:
the form factor of the primary current, and
the saturation of the transformer magnetic core
From Table 8.3, it is clearly seen that the relationship between the rms
value and the mean value of the precipitator current, i.e. the form factor, is
higher than the one obtained with DC energization. This results in a
corresponding higher form factor for the primary current, which is equivalent to a higher harmonic content.
Because of the pause interval introduced by IE, the value of the magnetic
induction variation (dB), which is possible to use without saturating the
core of the HV transformer, is lower than in the case with DC energization.
This problem can be avoided in different ways, e.g.

by
by
by
by

8.6.3

using
using
using
using

a larger transformer core


an auxiliary current pulse before the main one
a larger linear series reactor
a core with high remanence and low eddy losses

Collection efficiency

Intermittent energization (IE) reduces the corona power and in consequence


produces energy savings. The question arises, in which way does this
reduction in the corona power influence the collection efficiency of the
precipitator? The normal way to evaluate the collection efficiency obtained
with intermittent energization is to compare it with the efficiency obtained
with DC energization. A practical way is to compare the two migration
velocities and then to find the enhancement factor H defined as
(8.18)

215

INTERMITTENT ENERGIZA TION

400r---~-------r---'---'---'----'---'

~350

1--'-

c..

!z

300

C')

~ 250

-~---+---r-

Ol

.sc 200
o
.~

150

(J)

100

50

'e
iii
:J

-------r-

10

14

12

16

Corona power density (WI m2)

Figure 8.12 Stack emission as a function of the corona power with DC and IE for low
resistivity conditions.

where W IE is the migration velocity obtained with intermittent energization


and W De is the migration velocity obtained with DC energization.
The required collection surface is inversely proportional to W for a given
collection efficiency, so if H > 1, then H is the factor by which the collection
surface of a DC energized precipitator should be increased in order to
obtain the same collection efficiency provided by intermittent energization.

120

ii' 100
f-

Z
C')

E 80
0,

.sc

60

'iii
CIl

'e

40

(J)

iii
:J

20
0

4
6
8
Corona power density (W/m2)

10

12

Figure 8.13 Stack emission as a function of the corona power with DC and IE for medium
resistivity conditions.

216

ELECTRICAL OPERATION OF PRECIPITATORS

Intermittent energization results in a lower precipitator mean voltage and


a lower precipitator mean current. Then, for the case of low resistivity dusts
(see Figure 8.2), it can be assumed that this reduction in corona power will
result in a lower migration velocity and therefore a lower collection
efficiency. This is illustrated in Figure 8.12, where dust emission is plotted
as a function of the corona power density P cI A for DC and intermittent
energization (intermittence degree D = 3). The precipitator collected fly ash
from a boiler fired with an English/Polish coal blend causing no resistivity
problems. In this situation IE results in a lower performance and therefore
it is not advantageous for this application (H < 1).
In Figure 8.13, the same curves are plotted for a precipitator collecting
fly ash from a boiler fired with South African coal, which results in some
degree of back-corona. It is seen that both energization forms produce the
same minimum dust emission (H = 1), but, if a higher dust emission is
allowable (> 40 mg/m 3 NTP), then IE is better for the same power density.
In Figure 8.14 the dust emission is plotted as a function of the power
density for DC and intermittent energization (D = 7) for fly ash arising from
a boiler fired with Russian coal, which caused severe back-corona in the
precipitator. It is seen that IE is clearly better than DC energization (H > 1),
since it results in a lower dust emission at a lower power consumption.
These three practical examples show that:
Intermittent energization always results in a lower power consumption,
but it does not always give the highest collection efficiency .
The performance with intermittent energization is closely related to the
resistivity of the collected dust:

160r-------------__---------r--,-----~
~

140

c..
z 120

I-

'".

100

..sc::

80

.~

60

Ol

'E
Q)

t)

::l

Cl

40

- --- +

----~-----~-:r-I

20

o~~---+------~------+---~-+------~

10

12

14

16

18

20

Corona power density (W/m2)


Figure 8.14 Stack emission as a function of the corona power with DC and IE for high
resistivity conditions.

AUTOMATIC VOLTAGE CONTROL AND INSTRUMENTATION

217

in the case of high resistivity, IE is superior to DC energization.


in the case of medium resistivity IE is as good as DC energization and
sometimes better.
in the case of low resistivity, IE is inferior to DC energization, since it
results in a lower precipitator performance.
The above-mentioned statements are generally valid for a great number
of applications. The ESP manufacturers, however, need to consider other
factors when giving guarantees, regarding the enhancement factor H that
can be obtained by using IE. For instance, one difficult case is when the
precipitator has to collect high resistivity fly ash and at the same time the
un burnt carbon content of the dust is high.

8.7
8. 7.1

Automatic voltage control and instrumentation


Introduction

Other authors and contributors to this field [1,2J have in the past
emphasized the importance of optimizing electrical energization in order to
obtain the maximum collection efficiency. To accomplish this objective, two
important aspects have to be taken into account:
a good matching between the size of the TR set and the energized bus
section .
a good automatic voltage control (A VC) unit.
The first aspect has been covered in section 8.5. As explained, the key
parameter governing the corona power delivered by the TR set to a
particular bus section is the firing angle of the thyristors. This angle is
determined by the control unit for every half-cycle of the line frequency and
must have the correct value according to the existing operating conditions
within the precipitator.
Automatic control is closely related to the instrumentation used. Therefore, in the following, a brief review of the signals used in the AVC units will
be given.
8.7.2

Instrumentation

The signals used in the Ave units by different manufacturers are not always
the same. The Europeans have a long tradition of using the precipitator
current and voltage, the so-called 'secondary values', whereas the Americans
have preferentially used the 'primary values', but in recent years the
tendency has been to incorporate the secondary values in their AVC units.
Furthermore, the installation of opacity (or extinction) meters in the
stacks is more and more common, and in some countries they are compulsory, especially in connection with new plant. The signal delivered by these

218

ELECTRICAL OPERATION OF PRECIPITATORS

Phase control thyristors

High voltage oil tank

bus section
ACline

primary voltage
r--'--...L..,;.--.:...-.....:....-------, mA signal

Automatic voltage control unit

1--_ _ _--' kV signal

Figure 8.15 Signals used in the instrumentation and automatic control of a high voltage power
supply.

meters is used for continuous monitoring of the stack dust emission, but
sometimes it is also used by the control units. The purpose of the opacity
meter in conjunction with the AVes is:
optimization of the operation of the precipitator, and
energy savings under easy operating conditions.
The signals normally used by the control units are depicted in Figure 8.15.

8.7.2.1 Secondary values. These are the precipitator voltage and the
precipitator current. The voltage is measured by means of a voltage divider
and the current by means of a measuring resistor or current shunt.
In a modern control unit the following quantities are normally measured
and displayed:

precipitator mean current (lomean)


precipitator mean voltage (Yamean)
precipitator peak voltage (Vo peak)
precipitator minimum (trough) voltage (Yamin)

In the past, normally the mean values were used for the control tasks but,
in the last decade, the importance of measuring, for instance, the minimum
value has become more significant [10].
This value is vitally important in evaluating the operation of a precipitator collecting high resistivity dust. Other important tasks are the automatic
detection of back-corona and the automatic control of the degree of
intermittence [10,11].

AUTOMATIC VOLTAGE CONTROL AND INSTRUMENTATION

219

The peak value is important in determining the sparking level in the


precipitator and in optimizing the electrical operation, e.g. during voltage
recovery after a spark, and in computing the corona power in an approximated way (see equation 8.3).
In the past, these values of minimum and peak voltage were measured
with an oscilloscope using a voltage divider connected to the discharge
frame if one was not built into the TR set. This measurement was
cumbersome for less experienced plant personnel, but today it is a standard
feature in many modern AVC units using solid state circuitry (ICs).

8.7.2.2 Primary values. These signals are also shown in Figure 8.15.
Some control units use them in the automatic voltage control and/or
monitoring tasks. Nowadays, it is recognized that the use of the secondary
values is superior for the automatic control of the high voltage and in the
evaluation of the precipitator operation, e.g. spark detection, voltage recovery after spark and back-corona detection.
The primary values, however, can be used in important monitoring tasks,
e.g. in the determination of:

rms value of the primary current


rms value of the primary voltage
active power delivered to the TR set
apparent power delivered to the TR set

The primary voltage and current are measured by means of a potential


and current transformer, respectively. In this way these signals can readily
be connected to the AVC unit, as the transformers provide isolation and
adequate signal levels.
The active power delivered to the TR set has two components:
the corona power delivered to the bus section, and
the losses in the TR set (transformer iron and copper losses, silicon diode
conduction losses, etc.).
The active power delivered by the AC line has two main components:
the active power delivered to the TR set, and
the conclusion losses in the thyristors.
Because the last is negligible, compared with the corona power, the active
power measured by means of the primary values is approximately equal to
the active power delivered to the TR set.
The primary values are also important in monitoring tasks like:
state of the phase control thyristors
line overcurrent

220

ELECTRICAL OPERATION OF PRECIPITATORS

Figure 8.16 Simple energy management system (EMS) for an ESP.

line overvoltage
saturation of the HV transformer
too high a form factor, etc.

8.7.2.3 Opacity signal. The principle used is depicted in Figure 8.16. This
shows a three-field precipitator with two bus sections per field. Each section
is energized by a separate TR set, which in turn is controlled by an AVe
unit. The opacity (or extinction) meter is mounted in the stack and delivers
a 4-20mA signal to a converter, where the opacity signal can be filtered,
converted to a digital signal, etc., before it is connected to the individual
control units. This signal is a measure of the dust emission in the stack based
on a calibration performed by previous gravimetric measurements. The
measured dust emission is compared with a set point in each control unit,
resulting in a control action on the corona power in order to accomplish a
particular objective.
This is a simple and economic solution and must not be confused with
the more expensive approach, where the control units are connected to a
common communication bus, which is connected to a 'master' computer, or
to a plant computer, via a 'gateway' unit. This computerized central control
of precipitators is covered later in section 8.9.
8.7.3

Basic control principles

Under ideal operating conditions, the firing angle of the thyristors could be
controlled manually, but in practice this is impossible. Most of the processes

AUTOMATIC VOLTAGE CONTROL AND INSTRUMENTATION

Phase control thyristors

221

High voltage oil tank

ACline

~~------r-------------~
,____________ ~i~_1)9_~1)9_1!'___ .. __ . ______________ .... __ ~ ... ____ .,kV signal

: AVC unit

Figure 8.17 Principle of the closed loop automatic control of the precipitator current.

which use electrostatic precipitators are subject to both slow and fast
changes in the operating conditions. The gas flow, gas temperature, gas
humidity, fuel, raw material, etc., can change frequently. In order to keep the
collection efficiency as high as possible under difficult conditions, more
powerful and sophisticated control units are appearing all the time.
One basic architecture is illustrated by the block diagram in Figure 8.17.
Here it is supposed that the rnA signal is used as the feedback signal, i.e. the
precipitator mean current is the controlled parameter in a closed loop. In
other words, the firing angle of the thyristors is varied by a proportional-integral (PI) controller in such a way that the mean current follows
a reference signal (or a time varying setpoint) as closely as possible.
The firing pulses to the thyristors are delivered by an output stage,
providing an adequate signal level and isolation from the AC line. The kV
signal is also shown connected to the control unit, but is mainly used in
connection with spark detection and voltage recovery as explained in the
next section.
Nowadays, most of the control units are based on microprocessors and
peripheral circuits, which offer very powerful performance because of their
inherent memory and computing capabilities. The reference signal varies as
a function of time, according to a programmed control strategy.
The basic control principle is illustrated in Figure 8.18. In this example,
the mean current is increased linearly at a rate of rise R, until a spark occurs
or an upper limit is reached. R is normally expressed in %/min where 100%
corresponds to the rated current of the TR set.

222

ELECTRICAL OPERATION OF PRECIPITATORS

100
90

80
70

'*C

60

l3

50

Sparking level

Current upper limit

c:

CIS

Ql
~

30

20

10
00

10

15

20

25

30

35

40

Time (s)
Figure 8.18 Basic control strategy for the precipitator mean current.

The sparking level changes in the way shown. This is fairly constant at
the beginning, and then falls; it remains low during a short period and then
increases again. When a spark occurs, the current is automatically reduced
by a constant setback value S. In this example S has an absolute value
expressed as a percentage of the rated current (5%). Assuming a constant
sparking level, the spark rate SPR can be expressed as the reciprocal value
of the time interval between two sparks T..
From the zoomed area in Figure 8.18 it can be seen that the rate of rise
is determined by
R =

~(%/min)
T.

Then the spark rate can be expressed by


SPR

= ~ = ~(sparks/min)
T. S

(8.19)

Equation (8.19) indicates that a high spark rate can be obtained with a high
rate of rise R and a small setback S of the controlled variable. Conversely,
a low rate of rise and a large setback give a low spark rate. In the example
shown in Figure 8.18, the rate of rise R = 100%/min and the setback
S = 5%; then the spark rate, at a stable sparking level, will be 20 sparks/min.

AUTOMATIC VOLTAGE CONTROL AND INSTRUMENTATION

223

When the sparking level is decreasing, the time between sparks is shorter
and the spark rate is higher, but if the sparking level increases, the spark
rate becomes lower. When the controlled variable reaches the upper limit
the spark rate becomes zero.
The parameters Rand S or Sand SPR are normally found as settings in
all control units. S is used as an absolute or relative value. The way these
parameters are used and set by the specialists of the precitator suppliers is
different, each advocating having the best control strategy. Often this is
based on tradition, for the processes where the control units are normally
used, or in particular characteristics of their precipitator design.
Because of the variety of existing control strategies, e.g. current/spark
rate, voltage hill climbing, etc., only the general principles will be reviewed.
In assessing a particular control unit it is important that this has been
proven for difficult processes, e.g. those with fast varying operating conditions, like metallurgical plants, cement kilns, etc. It is also important that
the control unit has been developed by the precipitator manufacturer, i.e. by
people with experience in precipitator theory and operation.
With respect to the basic control strategy as shown in Figure 8.18, and
in order to maintain a high corona power level, during varying conditions:
the rate of rise R has to be high
the setback S has to be as small as possible
the spark rate SPR has to be high, but an upper limit must exist.
A limit for the spark rate exists, beyond which the collection efficiency
starts falling because of 'precipitation time' lost in voltage recovery. Moreover, too high a spark rate may be detrimental to the life of the internal
parts of the precipitator and the high voltage equipment.
8.7.4

Spark detection and voltage recovery

One of the important objectives in a modern control unit is to obtain a fast


recovery of the precipitator voltage after a spark; in this way it is possible
to maximize the voltage-time integral and maintain a high collection
efficiency. The fast voltage recovery is obtained if:
unnecessary turn-off time intervals (this turn-off time is also called
'deionization time', 'quench time', etc.) of the control thyristors are
avoided, and
the voltage is raised to the highest possible level within a few half-cycles of
the line frequency.
It is also important to perform this voltage recovery without a new spark
arising, i.e. 'multiple sparking' should be avoided. Voltage recovery is closely
related to spark detection, so this aspect will be briefly covered below.

224

ELECTRICAL OPERATION OF PRECIPITATORS

8.7.4.1 Spark detection. Most of the control units classify the sparks in
two types, according to their intensity:
a light spark (or spitting), where the precipitator instantaneous voltage
rises to a certain level after the spark within a very short period of time;
a severe spark (or arcing), where the precipitator instantaneous voltage
remains low after the spark for a certain period of time.
Figure 8.19 illustrates the two types of sparks and the voltage recovery
performed by a good modern control unit. Figure 8.19b shows that, even in
case of a severe spark, the voltage can be raised to a high level, without the
utilization of turn-off time and without the occurrence of multiple sparking.
The question whether or not to utilize a turn-off time is one of the less
understood problems in automatic voltage control techniques. There are
manufacturers of modern control units who recommend in their user
manuals to use a turn-off time in order to avoid the occurrence of arcs in
the precipitator. The alternative method to avoid such a problem will be
explained in the next section.
A fast voltage recovery is also closely related to the detection method
used. In the past, the primary values and the precipitator current have been
used, but the use of the instantaneous precipitator voltage has proven to be
superior, as discussed in reference [10].
8.7.4.2 Voltage recovery. The problem of recovering the preCIpItator
voltage within a few half-cycles of the line frequency without introducing
turn-off time does not have a simple solution. One of the difficulties is to
know how much the instantaneous voltage can be raised without the
occurrence of a new spark, i.e. the determination of the 'aimed level'. Then,
the next problem is to find the value of the firing angle which will provide
this aimed level.

I' I

~ ~j\,

-I ...rd

J J V V"
"

f'I

(1

, n
ESP
-r 1r..lV ~I r...J IH 1\ voltage

1\

IU

\I

"\I

Iv-'

I
(a)

J',

"

1'\

1\

"

1\ f\ 1\

1/

/ 1\ (

(b)

o
(I,

ESP

current

Time (10ms/div)

Figure 8.19 Classification of sparks according to their intensity. Light spark (a) and severe
spark (b).

225

AUTOMATIC VOLTAGE CONTROL AND INSTRUMENTATION

100

90

peak V Itage afer spa+ -

80

~CD

50

(;

40

lij

.c.
0
~

0..

B:

. -.~'~. -. ---:---1-+--- ; --- .- -

30
20

-~--+-----.-'- .-:~-~----- ~
j
;
~1
~o

10
0

--f--~
I

60

~
0
>

~--+

70

20

40

60

80

100

120

140

160

180

Firing angle [1

Figure 8.20 Control principle for a fast voltage recovery after spark.

This problem and its solution are illustrated by means of the curves in
Figure 8.20. Curve B shows a typical variation of the mean voltage as a
function of the firing angle during DC normal operation, while curve A
shows the attainable peak voltage in the first half-cycle after the spark.
Experience has shown that the aimed level can be represented by the curve
B, without causing multiple sparking, and at the same time giving an
acceptable precipitator voltage level. Sometimes a higher aimed level might
be used, but the probability of sparking in the recovery period is quite high.

8.7.4.3 Example. The problem will be illustrated in the following


example. Let us assume that the control unit is firing the thyristors at 80.
The aimed level for voltage recovery (52 k V) is shown by the dotted line.
After a spark, the preset setback gives an increased firing angle 0(1' this will
give a voltage level determined by the intersection of curve A and firing
angle 0(1 (70kV). It is clearly seen that this level is too high compared with
the aimed level and this will undoubtedly cause multiple sparking. The
correct firing angle is 0(0' determined by the intersection of the dotted line
representing the aimed level and the curve A.
At low voltage levels beyond the crossing of curves A and B, the problem
is reversed. If the closed loop control is not opened, then the firing angle
will be too high and results in too a low voltage level and a slow voltage
recovery. The recommended solution is:
to store the curve A in the memory of the control unit;
to open the control loop in case of spark and find the right firing angle
according to the aimed level and curve A;

226

ELECTRICAL OPERATION OF PRECIPITATORS


:
Ir.1 .... r
.
:
I
1- - ! I
j,J
I .
I i

A-

(a)

I !

.--~

: i

1
i

~.....l_

(b)

nme 20 ms/div
Figure 8.21 Reaction of a modern control unit to sparks at high (a) and low (b) current levels
(courtesy FLS miljo a/s) .

to close the control loop and perform the required setback;


to continue with the normal control strategy after the setback
formed.

IS

per-

To illustrate that a satisfactory voltage recovery can be obtained automatically, both at high and low current operation, and without using
turn-off times, the oscillograms shown in Figure 8.21 are included. They
speak for themselves and no further explanations need to be given.
Another important feature obtained with this method can be seen in
Figure 8.21:
(a) At higher current operation, the first pulse current used to raise the
voltage to the aimed level, and the immediate following ones, are lower
than the current pulses at normal operation.
(b) At low current operation, the current pulse used to raise the voltage is
higher than at normal operation.
It can be concluded that the electrical equipment is not subject to
overload, in connection with a spark or arc, if the above-mentioned method
is used. In this respect, it is necessary to remember that the condition for
obtaining this result is the use of a suitable high short-circuit reactance, as
mentioned in section 8.5.

8.7.5

Back-corona detection and corona power control

The occurrence of back-corona in one or more precipitator sections has


been normally determined by examining the corresponding i-v curve. In the
past, this curve was measured as a plot of the precipitator mean current vs.
mean voltage. The criterion used for determination of back-corona was the

227

AUTOMATIC VOLTAGE CONTROL AND INSTRUMENTATION

0.4
0.35

~
oS

::t"

mean

0.3
0.25

iii

CD

0.2

"tJ

E 0.15
~
:J

0.1

0.05
0

40

45

50

60
65
55
Precipitator voltage [kV)

70

75

80

Figure 8.22 i-v characteristics in the case of back-corona conditions.

slope of the i-v curve; if the slope was CIJ or negative, it indicated
back-corona. But in the early 1980s it was considered that this method was
not sensitive enough [10], and a better indication was obtained by using the
i-v curve, where the mean current is plotted against the minimum value of
the precipitator voltage, as illustrated in Figure 8.22.
This shows that the curve taken as a function of the minimum voltage
changes slope at a relatively low current level, while the curve based on the
mean voltage still has a positive slope. The curve based on peak voltage
shows a positive slope, irrespective of the presence of back-corona.
8.7.5.1 Automatic detection. Not all voltage control units include a
back-corona detector. The units which include back-corona detection invariably use different principles:
Slope of i-v curve. This corresponds to the method described above,
where the minimum voltage of the precipitator is used. This method,
however, has the disadvantage that the power levels have to be reduced in
order to find the inflexion point.
Voltage waveform at spark. This method is also based in the minimum
value of the precipitator voltage. By comparing its value before and after a
spark the occurrence of back-corona can be determined. Back-corona exists
if the minimum value after the spark is higher than the value before the
spark. This situation is illustrated by the oscillogram shown in Figure 8.23.

228

ELECTRICAL OPERATION OF PRECIPITATORS


ESP

voltage

Primary

current

TIme 50 ms/div

Figure 8.23 Precipitator voltage and primary current during spa rk in the case of back-corona.

By observing the oscillograms shown in Figure 8.19 and Figure 8.21, it


can be concluded that back-corona does not occur in these cases.
This method has the advantage that it does not need to reduce power
levels, because sparks are used. In the case of 'no sparking' conditions, a
blocking period, where the thyristors are not fired, is used instead.

8.7.5.2 Corona power control. The corona power has to be reduced in the
case of back-corona in order to extinguish it. As seen in the previous section
there are two means to accomplish this:
to change to intermittent energization and find the optimal degree of
intermittance, and
to reduce the electric charge (Qp) delivered to the precipitator, i.e. the area
under a current pulse, by delaying the firing angle of the thyristors.
This task has traditionally been performed by plant personnel, but over
the last decade the tendency is to do it automatically. One approach has
been to include this function into the Ave units; this is a cheap solution,
but it is difficult to obtain the optimal settings when only the typical
instrumentation signals, like those shown in Figure 8.15, are used by the
Ave unit.
The optimization of the degree of intermittence CD) can be performed in
combination with the automatic detection of back-corona. If during the
detection back-corona is found, then D is increased, but if no back-corona is
detected, D is reduced.
The optimization of the electric charge delivered by each current pulse,
Qp, is difficult. One method employed is to maximize the minimum value of

AUTOMATIC VOLTAGE CONTROL AND INSTRUMENTATION

~-----

229

Measurement interval --------:

Time

Figure 8.24 Determination of the figure of merit based on a comparison of the precipitator
voltage with a reference voltage.

the precipitator voltage while remaining at the optimal degree of intermittence.


A combined method for the optimization of D and Qp also exists [12].
Here D and Qp are varied, so a variety of combinations of these two
quantities is obtained. For each combination, a figure of merit is determined
and the combination giving the best figure of merit is selected for control.
Another method of determining the figure of merit is to compare the
instantaneous voltage waveform with a reference voltage during the time
interval where the corona discharge takes place, the reference voltage being
set to about the corona onset voltage. The method is depicted in Figure 8.24.
The figure of merit can be determined as the integral
I = f vp(t) . (vp(t) - V,ef )dt. Other ways of determining a figure of merit exist,
but these will not be covered here.
In order to improve the above-mentioned tasks, some units incorporate
the opacity signal in the optimization of the corona power, as shown in
Figure 8.16. But this is not 100% effective, as a particular unit does not
know if there has been a change in the settings of another control unit, or
a change in the operating conditions of the precipitator. Therefore, the
present tendency is to place this task in a supervisory computer, which has
all the relevant process and stack emission signals connected to it. With
these data, the computer can optimize the settings of the individual control
units accordingly. This 'supervisory computer control' of the precipitator is
covered briefly in section 8.9.

230
8.8

8.8.1

ELECTRICAL OPERATION OF PRECIPITATORS

Pulse energization

Introduction

The advent of commercial pulse energization systems in the early 1980s was
one of the major technical developments in the energization of electrostatic
precipitators. In fact, pulse energization was the first really new method
since Cottrell's development of the traditional power supply based on a high
voltage transformer/rectifier [ll
This method was developed to improve the collection of difficult high
resistivity particles, one of the main shortcomings of electrostatic precipitators. Pulse energization consists of short duration high voltage pulses
superimposed on a 'base voltage'. The pulse systems developed for a single
stage precipitator operate normally in one of the two following pulse widths:
a microsecond range or a 100 fls range. In the following only the latter type
will be covered.
The high voltage pulses are repeated at a certain frequency in the range
of 1 to 400 pulses/s (pps). A typical waveform of the applied voltage is
depicted in Figure 8.25 for a frequency of lOOpps, and for comparison
purposes the voltage waveform with traditional DC energization is included.
The differences are quite clear:
the narrow high voltage pulses have a high amplitude
the base voltage is kept close to the corona onset voltage

100r---~--------------------

__~------------r---,

80
, ....

-----

60

-------

"'-DC energi2lation

~ ~ :c~_/~'~~~~~~~~~~~~'~'~==~~~~~
20

10

12

14

16

18

20

Time [ms)

Figure 8.25 Typical voltage waveform obtained with pulse energization and comparison with
DC energization.

231

PULSE ENERGIZATION

the peak value of the precipitator voltage is equal to the base voltage plus
the pulse amplitude and its value is higher than for DC energization.
Historically, the first full-scale pulsing tests on precipitators were performed by White and Hall in the late 1940s [1,2]. The commercial
application of their pulse system was hampered mainly by the lack of a
reliable high voltage switch and by high power consumption. With the
advent of high frequency switching thyristors in the 1970s and refined
systems, in order to reduce power consumption, companies in USA, Japan
and Europe developed their own pulse systems and were able to demonstrate full-scale tests in the late 1970s [14].
Nowadays, 15 years later and with hundreds of commercially operating
pulse energization systems worldwide, there are a number of American,
Japanese and European companies supplying pulsers as standard products.
They are used in special applications where pulse energization is the best
economical/technical solution to a problem dust.
8.8.2

Electrical configuration

As depicted in Figure 8.25, a pulse system has to deliver a narrow high


voltage pulse superimposed on a base voltage, so a pulse generating circuit
and a base voltage power supply are required. Moreover, the pulse amplitude, the base voltage and the pulse repetition frequency have to be varied
according to a certain control strategy and this function is performed by a
special control unit.
A block diagram is shown in Figure 8.26, where the basic elements of the
system can be seen. In the construction of their systems, the various
manufacturers have used different approaches, mainly as a result of patent

ACline

I
Pulse
generating
network

Automatic
control unit

t--

Base voltage
power supply

Busbar section

Figure 8.26 Simplified block diagram of a pulse energization system.

232

ELECTRICAL OPERATION OF PRECIPITATORS

Figure 8.27 Main circuit of a pulse energization system with switching at low potential.

protection. But in principle, there are two main architectures; one based on
switching at low potential as seen in Figure 8.27 and one based on switching
at high potential [15] as seen in Figure 8.28. The first type normally uses
two high voltage tanks (one for the pulse generator and one for the base
voltage) and separate cabinets for the automatic control unit and power
control devices. The second type normally uses one control cabinet and one
high voltage tank resembling a traditional power supply. Both use an energy
recovery principle based on a series LC resonant circuit, where the precipitator, represented by its capacitance, is one of the components of the circuit.
The high voltage switch consists of a number of thyristors in series, each
having an antiparallel diode.

Figure 8.28 Main circuit of a pulse energization system with switching at high potential.

233

PULSE ENERGIZATION

8.8.2.1 Pulse system with pulse transformer. This is depicted in Figure


8.27. The base voltage is delivered by a separate power supply represented
by Voc. The pulse generating circuit includes a power supply Vps and the
series oscillating circuit consisting of the storage capacitor Cs , the coupling
capacitor Cc , the inductance Ls and the precipitator capacitance CF .
Before the generation of a pulse, Cs is charged to the voltage - Vps and
the precipitator is charged to - Voc. When the thyristor T is fired, oscillation is initiated and the current through the main circuit has the waveform
shown in Figure 8.29. During the positive half-cycle the current circulates
through the thyristor T and this is turned off around the zero-crossing when
the current falls below the holding level. Because of the energy stored in the
inductance Ls, the diode D is forced to conduct and the current circulates
in the opposite direction until it becomes zero. The current remains at zero
until the thyristor is fired again after a period of time, corresponding to the
pulse repetition frequency. The resultant precipitator voltage has the waveform shown in Figure 8.29.
The energy recovery, which is fundamental for the commercial utilization
of pulse energization, occurs during the negative half-cycle of the pulse
current. Here, the energy delivered to CF , and not used in the corona
generation, is returned to Cs and stored there during the time interval
between pulses and used in the generation of the next pulse.
The coupling capacitor Cc avoids a short-circuit of the power supply VDC
by the secondary winding of the pulse transformer. The pulse transformer
approach allows switching at a lower potential, but has the disadvantage of
a higher price, weight and volume. Assuming an ideal pulse transformer, no

i p(l)

............. vp+ v DC
vp (I)

OAr---~----T-------~------~~---r-4

L---~------------------------~~OV

Time

To

Figure 8.29 Idealized pulse voltage and current waveforms.

234

ELECTRICAL OPERA nON OF PRECIPITATORS

losses in the circuit and a very large storage capacitor, the pulse current can
be expressed by:
(8.20)
where Ip is the peak value of the pulse current and Wo is the angular
frequency of the oscillation. The pulse width T., can be expressed as a
function of woo
The pulse precipitator voltage is then:
vp(t) = - 1
CF

f ip(t)dt =

V (1
---.
2

- cos wot)

(8.21)

where Vp is the amplitude of the pulse voltage.


The quantities I p' Vp ' Wo and T., can be determined as follows:
(8.22)

(8.23)
(8.24)
where
V~s = nVps

(8.25)

L~=n2Ls

(8.26)
(8.27)

8.8.2.2 Example. For the typical values given below, the amplitude of the
pulse current, the amplitude of the pulse voltage and the pulse width are
calculated as follows:
Cc

Vps=4kV

500nF

n = 10

Ls = 251lH

CF

100nF

From equations (8.25), (8.26) and (8.27):


V~s =

10 x 4

~ =

10 2 x 25

Ceq

40 k V
X

10- 6

2.5mH

= 100 x 500/( 100 + 500) = 83.3 nF

PULSE ENERGIZA nON

235

From equations (8.22), (8.23) and (8.24):


J p = 40 x 10 3 /J(2.5 x 10- 3 /83.3 x 10- 9 ) = 231A
Vp

= 2 x 40 x 83.3/100 = 66.6 kV

I;, = 2nJ(2.5 x 10- 3 x 83.3 x 10- 9 ) = 9111S


8.8.2.3 Pulse system without pulse transformer. This is depicted in Figure
8.28. The base voltage is delivered by a power supply represented by Voc'
The pulse generating circuit includes a power supply delivering Vps and the
series oscillating circuit consisting of the storage capacitor Cs , the inductance Ls and the precipitator capacitance CF . In this configuration, the
semi-conductor switch is placed on the high voltage side and consists of
quite a large number of thyristors in series [16].
Before the generation of a pulse, Cs is charged to the voltage (Vps + Vod
and the precipitator is charged to - Voc. When the thyristors T are fired the
oscillation is initiated and the pulse current and the pulse voltage have the
waveforms as shown in Figure 8.29.
The expressions for determining the pulse current and the pulse voltage
are the same as before, i.e. equations (8.20) and (8.21). In this case, however,
the quantities J p , Vp ' Wo and I;, are expressed by:
(8.28)

(8.29)
(8.30)

where

8.8.3

Main features of pulse energization

During normal operation, the base voltage Voc is kept about the corona
onset level, and the pulse amplitude and the pulse frequency are varied
according to a certain strategy. The results of using pulse energization in a
precipitator can be better expressed by the i-v characteristics of the
respective bus section, as shown in Figure 8.30. Here, the current density is

236

ELECTRICAL OPERATION OF PRECIPITATORS

-~=---=-----+------~-------+----

-+----,-~

- -

- t - - - [-

tgQ-pps -

..c:

0.01

F==---- __

1--- -- -----

-~~~ - -- ~~=T~~-T------,----

i
0.001 -LO- - - 1....
0---20----+30---40.J.----S-0 - - - l
60

Pulse peak voltage [kV]


Figure 8.30 Typical i-v characteristics obtained with pulse energization (courtesy FLS miljo
a!s).

plotted as a function of the pulse amplitude for a constant base voltage (VDe>
and three different pulse repetition frequencies. The current density is
defined as the precipitator mean current divided by the collection area. The
precipitator mean current is also called the emission current (IF).
This typical family of curves shows the following interesting features of
pulse energization [17]:
the precipitator current can be varied by changing the pulse frequency in
spite of the precipitator peak voltage (VDe + Vp ) being kept constant;
the precipitator peak voltage is high due to the short duration of the
pulses;
the slope of the curves is rather flat (this is also the case with high
resistivity dust).
These features of pulse energization give the following advantages:
8.8.3.1 Current control capabilities. The i-v characteristic (Figure 8.30)
shows that the precipitator current (IE) can be controlled independently of
the precipitator voltage by varying the pulse repetition frequency. This
allows the current to be reduced to around the onset of back-corona,
without reducing the precipitator voltage, i.e. the precipitator can operate at
low current and high precipitator voltage. This means a more suitable
electrical energization for high resistivity dust compared with traditional
DC energization, where current control cannot be performed without
reducing precipitator voltage.

PULSE ENERGIZA nON

237

The dense ionic space charge produced by a pulse shields the discharge
electrode and reduces the electrical field strength at its surface. This causes
the suppression or limitation of the corona discharge during the rest of the
pulse period. The consequence is a fiat i-v characteristic.
8.8.3.2 Current distribution. With DC energization, the corona discharge
tends to be localized at discrete spots on the discharge electrode. With the
application of narrow pulses of high amplitude superimposed on a base
voltage around the corona onset voltage, the pulse peak voltage significantly
exceeds the corona onset level. This produces an intense corona discharge
during a pulse and a correspondingly dense ionic space charge. A discharge
electrode with very spotty corona under traditional DC energization can
literally be made to glow with pulse energization [18]. This results in a
better current distribution along the electrode, and this can be extended to
the whole pulse energized bus section. This improved current distribution
has been confirmed by measurements on a laboratory duct precipitator [17]
and also measurements on pilot precipitators [20].
These results show that with DC energization, the current density at the
beginning of the precipitator section is very low and increases along the
section in the direction of the gas (see section 8.3). With pulse energization,
the current density along the precipitator section is considerably more
uniform.
A good current distribution on the collecting plates is important in order
to avoid the initiation of back-corona due to localized spots of high current
density.
8.B.3.3 Electrical field strength in the interelectrode space. With DC
energization, free electrons are constantly generated producing an ionic
space charge density and a field strength that, in principle, does not vary
with time.
With pulse energization, where the base voltage is kept just below the
corona onset level, free electron and negative ions are generated only during
the pulse period. During this time, the ionic space charge crosses a part of
the interelectrode space and during the time interval between pulses, the
space charge moves towards the collecting electrode impulsed only by the
base voltage field. As a consequence, the space charge and the field strength
vary with time at each point of the interelectrode space.
Measurements on a pipe laboratory precipitator gave the results depicted
in Figure 8.31. The field strength in relation to the values obtained with DC
energization are plotted as a function of time. After a pulse, the electrical
field is determined by the base voltage and the moving space charge, its
strength increasing until the front reaches the collecting electrode. Thereafter the field strength decreases until all the ions have reached the collecting
electrode.

238

ELECTRICAL OPERATION OF PRECIPITATORS

4,------------------------------------------.

O~-------2---------4--------6~------~6~------710

Time [ms]
Figure 8.31 Electric field as a function of time after a high voltage pulse has been fired.

8.8.3.4 Particle charging. For particles of l/lm and larger, electric field
charging is the predominant mechanism, and the saturation charge is
determined by the maximum field strength created by the ionic space charge.
With DC energization, a particle at a certain position is surrounded by the
ionic space charge, and its saturation charge depends on the constant
electric field at that position. With pulse energization the particle charging
occurs when the space charge passes the particle and its saturation charge
is determined by the maximum field strength during the passage of the space
charge (see Figure 8.31). Because the maximum field strength with pulse
energization is much higher than with DC energization, this method
provides enhanced particle charging.
Measurements have shown that the best results are obtained with a high
pulse amplitude, because this causes a higher ionic space charge density.
8.8.4

Power consumption

A precipitator section can be represented by a capacitance (C F ) in parallel


with a current generator accounting for the electronic, ionic and dust space
charge current. Each pulse has to raise the voltage across the capacitance,
from the base voltage (Vod to the peak voltage (Voc + Vp ). This means that
a considerable amount of energy has to be used. Supposing that in the
circuit of Figure 8.27 the coupling capacitor Cc is much larger than CF , the
energy supplied by the pulse system only for charging CF is:
(8.31)

PULSE ENERGIZA nON

239

For typical values (Vp = 60 kV, CF = 100 nF), the energy required is
180J. If this has to be repeated 200 times/s, a large amount of power
(36 k W) has to be used. As the energy necessary for the corona discharge is
small compared with the energy needed to charge CF , the power consumption becomes excessive if the pulse system does not include means for energy
recovery. The system shown in Figures 8.27 and 8.28 includes this feature
in the series oscillating circuit. Here, when the voltage across CF is at its
maximum, the pulse current is zero. Then it reverses and in the negative
half-cycle the surplus energy is stored in the storage capacity Cs .
The power consumed by a precipitator section energized by a pulse
system with energy recovery can be expressed by [17]:

w., =

Pr = IE VDe

+ cI EVp

(8.32)

where the constant c has been found experimentally to be approximately


equal to 0.5.
8.7.4.1 Example. A pulse system is operating at 0.1 mA/m2, Vp = 60 kV,
Voc = 40kV and energizes a 3000m 2 bus section. Determine the power
consumption of the bus section. The precipitator mean current is:
IE = 0.1 x 3000 = 300mA = 0.3 A. Applying equation (8.32)
P

= 0.3 x 40 + 0.5 x 0.3 x 60 = 12 + 9 = 21 kW

This consumption corresponds to a power density of 7 W1m 2 , which is a


typical value for medium resistivity dusts. For high resistivity dusts, the
required power density is quite low, hence with a low pulse repetition
frequency (2 to 20pps) the precipitator mean current is correspondingly
reduced.
8.8.5

Collection efficiency

Similar to the case with intermittent energization, the improvement in the


precipitator performance is closely related to the resistivity of the collected
dust. This improvement is normally expressed by the enhancement factor
H = wp/wDe, where wp is the migration velocity obtained with pulse
energization and W De is the migration velocity obtained with DC energization.
The enhancement factor H obtained with very high resistivity dust,
reported by American [18], Japanese [20] and European [14] companies is
about 2. The enhancement factor H, as a function of the dust resistivity, can
be expressed by the curve shown in Figure 8.32. This comparison with DC
energization indicates that at low resistivity levels, both energization forms
produce the same result, and at high resistivity levels, pulse energization is
much better [21].
Another way to express these results is by saying that the precipitator
performance with pulse energization does not fall as much as with DC
energization for an increasing dust resistivity.

240

ELECTRICAL OPERATION OF PRECIPITATORS

j
1:
Q)
E

2l
c:
co

//

.c
c:
W

10 10

10

11

10

12

10 13

Dust resistivity [0 em)


Figure 8.32 Enhancement factor obtained with pulse energization in relation to DC energization as a function of the dust resistivity.

A comparison between pulse and intermittent energization at high


resistivity levels, expressed by the enhancement factor, gives a typical figure
of 1.2 to 1.5 favourable to pulse energization. The precise value used by the
precipitator manufacturer is assessed for each particular application, and is
based on experience for the particular process, process conditions, dust
composition, etc.

8.8.6 Applications
As previously mentioned pulse energization is used in the collection of very
high resistivity dust. In this case, the enhancement factor is high and can
compensate for the higher price of the pulse systems.
Typical applications are precipitators for:
Four-stage preheater cement kilns (without or with insufficient water
conditioning of the kiln gases).
Coal fired power and steam generating boilers.
Limestone, dolomite and magnesite kilns.
Sinter strands for iron ore agglomeration.
The application of pulse energization is not restricted to new precipitators. It can also be a very effective solution for improving the performance
of existing precipitators having resistivity problems. In this case, it is
important to bear in mind that the mechanical condition of the precipitator

SUPERVISORY COMPUTER CONTROL

241

has to be good. This means effective rapping of the discharge and collecting
electrodes, good gas distribution, good electrode alignment, etc. Otherwise,
the expected enhancement factor will not be obtained, because of low pulse
voltage and/or precipitator current limitations.
For example, when the precipitator is collecting high resistivity fly ash,
the current density is kept low by using a low pulse frequency in order to
eliminate back-corona. But where the carbon content of the dust, carried by
the flue gas, is high, the precipitation requires a high current density and
this conflicts with the need for low current density to eliminate back-corona;
hence the result is a higher stack emission.
B.B.7

Summary

Pulse energization is normally ideal for precipitators collecting high resistivity dusts. The improved precipitator performance in the collection of
medium and high resistivity dust, compared with traditional De energization and IE, is due to the combined effect of the following features:

better particle charging


higher collecting field strength
better current distribution
better current control capability

The enhancement factor obtained with pulse energization is mainly due


to the application of a high pulse amplitude. Therefore, the mechanical
condition of the precipitator has to be good. In the collection of fly ash, the
concentration of low resistivity particles, like unburnt coal, has to be low;
otherwise the expected performance will not be achieved.
8.9

Supervisory computer control

In the last decade, supervisory computer control for precIpitators has


become more common. A traditional solution has been a stand-alone
computer, implemented with minicomputers in the past, and now with
powerful pes. The use of pes has resulted in lower prices, which has meant
that these systems are becoming more affordable to customers.
With increasing plant automation and more advanced graphics operator
interfaces available in the control room, the integration of the precipitator
control in the plant computer system is an increasing demand. Most of the
plants built recently have a system similar to that depicted in Figure 8.33.
The workstation shown represents the main computer providing the major
services and where the application programs are run. Normally, a number
of operator view stations with display, keyboard and control facilities are
connected to the main computer via a common bus. This main computer

242

ELECTRICAL OPERATION OF PRECIPITATORS

ESP
supervisory
computer
Local
PLC

~
Plant computer system

Communication bus
Figure 8.33 Supervisory computer control of an ESP with stand-alone computer (courtesy FLS
miljo a/s).

communicates via another bus with the plant PLC (Programmable Logic
Controller) system, which in turn exchanges data with the local PLCs in
charge of the control of local areas or main equipment of the plant. In case
of communication problems, the local PLCs are designed to continue with
their alloted control tasks. This is the principle of the so-called decentralized
control system (DCS).
This principle is combined with application programs included in the
main computer providing the so-called Supervision, Control and Data
Acquisition system (SCAD A). The result is a system with powerful graphics
operator interface and control functions.
It is understandable, therefore, that for the plant management and
personnel interested in operating the whole plant from their operator view
stations, a stand-alone precipitator control computer in the control room is
not readily accepted.
A typical approach to this problem is the connection of the precipitator
control equipment to a common communication bus. This bus is then
connected to the plant computer via a so-called gateway unit, containing the
two communication drivers, necessary for communication and data exchange, between the plant computer and the precipitator bus. In this way,
the advanced control functions for the precipitator may reside in the plant
computer system, or if needed, in a precipitator computer connected to the
precipitator bus.

SUPERVISORY COMPUTER CONTROL

8.9.1

243

Stand-alone computer

A typical architecture is illustrated in Figure 8.33, which shows a six-bus


section precipitator, each controlled by an AVC unit. In the European
precipitators with rigid frame design, the AVC unit normally controls the
high voltage power supply plus the rapping systems of the respective bus
section. The six AVC units communicate with the precipitator supervisory
computer via a common communication bus. This supervisory computer
also receives data from the process and the stack opacity through a data
acquisition unit (normally a PLC). The ESP is operated from the control
room, through a local PLC, which takes care of the start/stop functions,
alarm indication, voltage and current display, etc., and is connected to the
plant PLC system via the system communication bus.
In this configuration all the advanced control functions are placed in the
supervisory computer. Among them, the following features can be found:
Menu based graphic operator interface, providing:
screen displays of the precipitator operating conditions;
screen displays of the operation of the HV power supplies and rapping
systems (voltages, currents, corona power, timers, etc.);
display and setting of parameters;
remote start/stop;
trend analysis, etc.
Optimization of the corona power (by means of D and Qp), according to
the existing operating conditions and dust resistivity.
Energy management system for costs savings whenever possible.
Optimization of the rapping sequences (and reduced rapping losses).
Automatic measurement and display of i-v characteristics.
Automatic precipitator start-up and shut-down routines.
Alarm handling and fault diagnosis.
One example of a commercial system for power plant preclpltators,
including all these features, is fully described elsewhere [22]. In this system
the optimization functions are performed by means of Fuzzy Logic, and the
aim of this so-called expert system is to act as a specialist being at the plant
24 hours a day.
A variant for this system is the direct communication between the
precipitator supervisory computer and the plant computer, shown by the
dotted line. This requires the existence of a communication driver in each
computer. This approach makes the data adquisition unit superfluous, as
the process data can be exchanged directly. This solution is attractive when
the communication driver, residing in each computer, has already been
developed. If this is not the case, the respective development can be a
time-consuming and expensive affair.

244
8.9.2

ELECTRICAL OPERATION OF PRECIPITATORS

Supervisory computer control via a gateway unit

A typical and simple solution to the above situation is the one depicted in
Figure 8.34. The common communication bus for the control units is
normally an accepted industrial standard. If this is not the case, this feature
has to be included in the AYCs or a suitable converter has to be used. When
this requirement is met, the gateway unit makes communication possible
between the standard bus and various PLC systems with the most common
trade marks. This communication problem is similar for other types of
equipment used in industrial plants, which has made it necessary to develop
the required communication drivers. This development has resulted in a
considerably lower price, because these gateway units can now be obtained
as stock components.
When the connection of the AYCs via a standard bus and a gateway has
been established, the supplier normally has other modules which can be
connected to the standard bus to exchange data with each other as shown
in Figure 8.35.
The standard bus runs through the whole plant and the different modules
can be placed physically at the point where they are required. As an
example, a PC acting as supervisory control of the precipitator can be used,
which includes the various functions mentioned in section 8.9.1. In the same
way, a remote terminal for operation of the AYCs can be connected to the
standard bus if neither local nor control room operation is required.
Furthermore, intelligent I/O units or small PLCs can be connected. Where
a more powerful PLC is required, this can be connected to the standard bus

Standard bus

Figure 8.34 Supervisory computer control from the plant computer system via a gateway unit
(courtesy FLS miljo ajs).

245

SUPERVISORY COMPUTER CONTROL

..................... _.. __ ...... __ ... -- ...~

Plant computer system

Figure 8.35 Supervisory computer control with dedicated ESP computer integrated in the
plant computer system by means of a gateway unit (courtesy FLS miljo a/s).

via a gateway unit. This architecture looks very attractive at the moment
especially because of its flexibility and modularity.
8.9.3

Advanced control functions

Among the features offered by a precipitator supervisory computer, one of


the most important is an advanced control strategy including functions like:
Optimization of the corona power, including
optimization of the degree of intermittance (D)
optimization of the charge delivered per current pulse (Qp)
optimization of the spark rate and current setback, etc.
Optimization of the rapping sequences, including
optimization of the off-time between rapping of the collecting plates
synchronization between rapping of adjacent sections
programmable power-off rapping sequences.
Energy management system.
A supervisory computer is able to fulfil these requirements in a satisfactory way because it is constantly receiving information about:

Process condition (gas temperature, flow, feed rate, 02' S02, etc.).
Stack opacity (dust emission).
Currents and voltages in the individual bus sections.
Status of TR sets, rappers, timers, parameters, etc.

246

ELECTRICAL OPERATION OF PRECIPITATORS

Based on these data the supervisory computer can optimize the relevant
settings in each automatic control unit (A VC) like:

Degree of intermittence.
Upper current limit (Qp).
Spark rate and current setback.
Rapping off-time, etc.

These settings are changed at regular intervals, or when required, and their
positive effect assessed by analysing the opacity signal, while simultaneously
checking that the process conditions have not changed, by means of a trend
analysis of their respective signals.
These relevant settings cannot be optimized in the same degree by the
Aves operating as stand-alone units, i.e. only relying on the electrical
feedback signals (rnA and kV).
For all these reasons, it is believed that the architecture depicted in
Figure 8.35 will become more and more accepted. The control room
personnel will operate the precipitator from their view stations in the
normal way via a gateway communication, but, at the same time, the
precipitator supervisory computer running in the background, will overtake
and perform more and more advanced control and monitoring functions,
like precipitator event and alarm indication and handling, fault diagnosis,
etc. These features, complemented with precipitator start-up and shut-down
automatic routines, will result in an intelligent and powerful supervisory
computer control. This will relieve the plant personnel from tedious work
routines and will provide cost advantages because of power savings, easier
and better maintainance, a higher plant availability and lower average dust
emISSIOn.

Appendix 8.A
J- V Relationship for a wide-plate geometry with air load

From the three Maxwell equations governing the electric field Poisson's
equation can be derived:
(8.A.l)

where p is the charge density (C/m 3 ) and Co is the permittivity of free space
(8.85'10 - 12 Fjm), which is valid for gases under normal precipitator conditions.
The solution of Poisson's equation for a wire-plate geometry shown in
Figure 8.A.1 is a formidable task.

247

APPENDIX 8.B
collecting plates

~o

2s

discharge electrode
collecting plates

Figure 8.A.I Electrode geometry.

The solution is simplified if it is assumed that the current is small and the
alteration of the potential by the ion space charge can be represented by the
same value found for the more simple wire-pipe geometry.
The average current density, as a function of the potential at the
discharge electrode (see [2,5]), can be expressed by:
.
Js

7rB o b
21 (dl
V(V - Y.,)
cs n ro)

(Aim)

(8.A.2)

where: b is the ion mobility (2.1 x 1O-4 m 2 /Vs for negative corona in air),
d is an equivalent cylindrical radius (d = 4s17r for sic ~ 0.6 (for other sic
values see the cited references)), and Y., is the corona onset potential
expressed by:
(8.A.3)

The corona onset field Ec has been found empirically [2,5] and for negative
corona in air it is expressed by:

Ec

15'(32.2 + 0.~5) (Vim)

(if ro is expressed in m).


The relative gas density 15'
1 atm. and 25C, i.e.
~,

IS

(8.AA)

conventionally expressed in relation to


298

u = (298

+ T)P a

(8.A.5)

Appendix 8.B
Approximated calculation of the mean and rms values of the ESP current

For the current waveform shown in Figure 8.B.l, the mean value, the
rms-value and the form factor can be expressed by the following approxi-

248

ELECTRICAL OPERATION OF PRECIPITATORS

i(t)

Time

T
Figure S.B.l Precipitator current waveform.

mated equations:
Imean

2Ipk(~)
n
T

I rms =

Ipkji;

FF =

~ fi

2V lr

Note: It is left to the ardent reader to check the results shown in Table 8.1
using the waveform from Figure 8.8b.

References
1. Hall, H.1. (1971) Trends in electrical energization of electrostatic precipitators. Proceedings

2.
3.
4.
5.
6.
7.
8.

of the Electrostatic Precipitator Symposium, February 1971, Birmingham, AL, USA, pp.
177-89, SoRI Publication, Birmingham, AL, USA.
White, H.1. (1963) Industrial Electrostatic Precipitation. Addison Wesley Publishing Company, Reading, MA, pp. 198-226.
Reyes, V. (1991) Determination of corona power in a precipitator section. Unpublished FLS
Miljo Company Report.
Petersen, H.H. (1990) A precipitator sizing formula. 4th International Conference on
Electrostatic Precipitation, Beijing, China. September, International Academic, Beijing,
1993, pp. 330-8.
Robinson, M. (1971) in Air Pollution Control. (W. Strauss ed.). Wiley-Interscience, New
York, pp. 241-52.
Oglesby, S. and Nichols, G. (1978) Electrostatic Precipitation. M. Decker, New York, pp.
39-54.
Cooperman, P. (1960) A theory of space charge limited currents with application to
electrostatic precipitation. Trans. AlEE, 79 1,47.
Cooperman, G. (1979) A new current-voltage relation for a duct precipitator valid for low
and high current densities. Trans. lAS 79, IEEE, pp. 146-7.

REFERENCES

249

9. Zamany, J. (1995) Numerical modelling of electrodynamic conditions influenced by particle


space charge and resistivity in ESPs of complex geometry for industrial applications.
Conference Electrostatics' 95, lOP, York, England, April, Poster Session. Inst. of Physics,
London, UK.
10. Reyes, V. (1987) Comparison between traditional and modern automatic controllers on full
scale precipitators. Proceedings of the EPRI/EPA Symposium, March, Nashville, USA,
Session 2A, EPRI, Palo Alto, CA, USA.
11. Reyes, V. (1990) Methods and apparatus for detecting back corona in an ESP with
ordinary or intermittent energization. US Patent 4,936,876. June 26.
12. Jacobsson, H. and Porle, K. (1994) Method of controlling the supply of conditioning agent
to an electrostatic precipitator. PCT Patent Application WO 94/20218, September 15.
13. Hall, H.J. (1990) History of pulse energization in electrostatic precipitation. J. Electrostatics, 25, 1-22.
14. Petersen, H.H. and Lausen, P. (1979) Precipitator energization utilizing and energy
conserving pulse generator. 2nd Symposium on the Transfer and Utilization of Particulate
Control Technology, Denver, USA, July, EPA Vol. II, pp. 352-68.
15. Kide, L. (1977) Electrostatic precipitator arrangements. US Patent 4,052,177, October 4.
16. Reyes, V. and Taarning, C. (1990) A semiconductor high voltage switch for pulse
generation. Proceedings of the Power Conversion Conference, Munich, Germany, June, pp.
348-61.
17. Lausen, P., Petersen, H.H. and Jorgensen, H.J. (1981) Theory and application of pulse
energization. 1st International Conference on Electrostatic Precipitation. Monterey, USA,
October, pp. 531-53, APCA Publishing Co., Pittsburgh, USA.
18. Feldman, P. and Aa, P. (1981) Operating results from the first commercial pulse energization system to enhance electrostatic precipitator performance. Proceedings of the American
Power Conference, Chicago, USA, April.
19. Petersen, H.H. and Lausen, P. (1981) Application of energy conserving pulse energization
for precipitators-practical and economic aspects. 3rd Symposium on the Transfer and
Utilization of Particulate Control Technology, Orlando, USA, March, pp. 291-302, Vol. 1,
EPA Publication.
20. Fujishima, H. and Tomimatsu, K. (1990) Applications of an electrostatic precipitator with
pulse energization system. 4th International Conference on Electrostatic Precipitation,
Beijing, China, September, pp. 419-30, International Academic Publishers, Beijing, 1991.
21. Schioeth, M. (1987) Five years' experience with pulse energized precipitators on power
plants burning a wide range of coal. Proceedings of the 3rd International Conference on
Electrostatic Precipitation, Abano, Padova University, Italy, October, pp. 197-207.
22. Reyes, V and Lausen, P. (1993) Utilization expert computer systems for control and
operation of electrostatic precipitators. 10th Particulate Control Symposium and 5th
International Conference on Electrostatic Precipitation, Washington, USA, April, Session
C4, EPRI TR 103048 Vol. 2, Palo Alto, CA, USA.

Precipitator sizing methods and models of


electrosta tic precipitators

c. PAULSON AND M.

REA

Editor's note

The method of sizing precipitators has traditionally been the preserve of the
precipitation using sizing factors (effective migration velocity), derived from
efficiency measurements using precipitators operating on similar process
plant. Originally the approach was based on the Deutsch equation, but as
higher efficiency/lower emissions were demanded to satisfy legislative levels,
sizing based on the Deutsch equation was found unsatisfactory and over the
past 20 years or so, suppliers have found it necessary to use a modified
equation, such as those derived by Matts-Ohnfeld of ABB Flakt or Petersen
of FLS Miljo.
While the use of this type of modified equation has resulted in significantly improved prediction levels, the sizing is still based on data derived from
units operating on similar processes and inlet conditions, etc. Most major
suppliers have, over the years, built up detailed data banks, collating
precipitator performance against specific process/inlet conditions/dust characteristics, and the use of these data banks has enabled precipitator
installations to be sized so as to minimise technical/contractual exposure
risks for the supplier.
In addition to determining performance characteristics from installed and
operating plant, many suppliers have operated pilot sized precipitators in
the field and laboratory, handling a few m 3 s - 1, not only to derive the
precipitation factor or EMV, but also to evaluate the effect of process
variables, e.g. temperature, gas velocity, contact time, moisture, etc. The use
of the pilot precipitator has, over the years, produced significant amounts
of data, but to be fully representative for plant sizing purposes, the inlet
conditions to the pilot should be identical to those the full-scale precipitator
will meet in practice.
The following chapter, after reviewing the modified equations of MattsOhnfeld and Petersen, describes the work carried out by CSIRO in
Australia in developing a further modified sizing equation for coal-fired
power station precipitators. This work, under the direction of Colin Paulson, was carried out in the laboratory using a small pulverised coal fired
combustor fitted with precipitators to produce a range of inlet conditions
similar to those met in practice. Although the modified equation gives good
results it is still based on measured data.

EDITOR'S NOTE

251

The rapid development of computers over the past two decades has led
a number of investigators to approach sizing from a theoretical numerical
standpoint. A number of programs have been developed, but to date,
although the investigators and various research organisations consider the
derived data to be reliable, they are not so readily accepted by the
precipitation industry for sizing purposes. Prof. Massimo Rea of Padova
University reviews some of the earlier programs and concludes with the
work being carried out by the University in conjunction with ENEL (italy's
Electricity Authority).
In addition to this reported work, other similar programs are being
developed by some of the major suppliers and research organisations, for
example the Electric Power Research Institute (EPRI) of the US. It is
suggested that, as this type of program is rapidly developing, the reader
should maintain contact with the various organisations to keep informed of
the latest status of programming.

9A

Precipitator sizing methods

c. PAULSON
9A.1
9A.l.l

Theoretical considerations
Basic dust-collection equation for gas in a duct

Consider a uniform duct along which passes a suspension of fine particulate


matter in a turbulent gas. It is assumed that the walls of the duct collect any
dust particles that arrive there, and it is further assumed that the turbulence,
whilst sufficient to prevent the appearance of a zone with severely reduced
particle concentration next to the walls, is not so vigorous that particles
already collected are continually being re-entrained back into the gas
stream.
Under these conditions it may be expected that the rate of arrival of
particles at a selected small area of wall will be directly proportional to the
volume concentration of particles in the immediate vicinity. Since those
particles impinging on the wall collect there, the volume concentration of
those passing by is correspondingly reduced. This loss of dust burden in the
gas is progressive along the duct and the rate of dust collection at successive
areas of wall correspondingly reduces.
At a selected cross-sectional plane distance / into the duct the above rate
process may be expressed by:
(9A.I)
where n 1 is the number of particles in unit volume at plane /, t is time and
k is a constant.
If the full length of the duct is L, then the total loss of particle
concentration is obtained by integration of equation (9A.l) from I = 0
(where the inlet particle concentration is n) to I = L (where the outlet
particle concentration is no). The result is
In(ndno)

kt

In(l - )

(9A.2)

where t is the average time a particle would take to traverse the duct, is
the fractional collection efficiency of particles, i.e. (no - nL)/n O and (1 - ) is
the fraction of particles not collected. This is termed the slip.
Given a constant throughput of dusty gas, t is directly proportional to L,
so that
In(1 - ) = constant x L

(9A.3)

THEORETICAL CONSIDERA nONS

253

Equation (9A.3) shows that, under these conditions, complete particle


collection (e = J) can only be attained in a duct of infinite length.
Now, if we assume that the duct is a cylinder of radius r, so that the
internal surface area, A = 2nrL, it follows that t, the treatment time, is
nr2 L/V, where V is the volumetric gas flow. It is also assumed that the dust
particles are small enough not to suffer significant slip relative to the
turbulent gas. From equation (9A.2)
In( 1 - e) = knr2 L/V = kAr/2 V = krxr/2

(9AA)

whererx = A/V, otherwise known as the specific collecting area (SeA) of the
duct.
From equation (9A.2) we note that k has the dimensions (time) -1, so that
kr/2 in equation (9AA) has the dimensions of a velocity. We can therefore
rewrite equation (9AA) as:
10g(1- e) = -ctw/2.303

(9A.5)

where w is the mean drift velocity of the particles to the cylinder wall and
the negative sign appears because k is negative from the outset in equation
(9A.l) because the dust burden is decreasing.
Taking a practical example, it is possible to achieve e = 0.5 using an
ordinary steel pipe about twenty-five times longer than its diameter with rx
set at 100 m 2 for every m 3 s -1 of dusty gas throughput. Equation (9A.5)
then shows that w is 0.007 m s -1. If it were desired to achieve e = 0.99 (a
99% collection efficiency) then the pipe would need to be lengthened to
almost 170 times the diameter. Ten times the original pipe length would
theoretically catch 99.9% of the incoming particles.
The impingement dust separator described above is impractical and
primitive, but is included here because it is a valuable introduction to better
devices. Thus, if a suitable transverse force is imposed on the particles
suspended in the gas stream, then clearly w is considerably increased and
the dust separator is much more efficient for the same size.
The following are examples of transverse forces which could be used in
gas cleaning devices.
(a) Magnetic forces. These are applicable only to particulates that respond
well to such forces and thus have restricted scope. Magnetic coatings on
non-magnetic particles are helpful.
(b) Centrifugal forces. These are applicable to the dusty gas stream as a
whole, and are the basis of the cyclone separator.
(c) Differential pressure forces. These are applicable when the wall of the
duct is permeable to gas but hardly at all to the particulate matter, as,
for example, in the vacuum cleaner and the industrial fabric filter.
(d) Electrical forces. These are applicable to charged particles in an applied
electric field and are the basis of the electrostatic precipitator.

254

9A.l.2

PRECIPITATOR SIZING METHODS

Electrostatic precipitation

9A.l.2.1 Basic principles.


three essential functions:

An electrostatic precipitator must provide

(a) the suspended particles must be given an electric charge;


(b) the particles must be subjected to an electric field to enable them to
migrate from the gas stream to a suitable collecting electrode;
(c) the collected material must be removed from the collecting electrode in
an efficient manner and deposited in a receptacle with the minimum
amount of loss.
These operations are usually achieved by applying a high DC voltage,
usually negative polarity, to a wire placed adjacent to an earthed plate and
passing the dirty gas between the wire and the plate. The high voltage forms
a corona around the wire causing the gas to ionise. The ions thus produced
attach themselves to the particles suspended in the gas and the particles
move in the electric field between the wire and the collector toward the
collector. Typical values for this collecting field are 2-4 kV cm - 1, and the
electric force acting on the particle can be many hundreds, or even
thousands, of times gravity.
The layers of collected particles are removed from the collecting electrode
by rapping the electrode which causes the material to fall into a hopper
below the collector. This rapping process is critical to precipitator performance because a balance must be achieved between keeping the electrode
clean and rapping too hard. Over-rapping causes particle re-entrainment
into the gas stream with a consequent loss of performance.

9A.1.2.2 The fundamental efficiency equation. Considering equation


(9A.5) in the context of the electrostatic precipitator the mean particle drift
velocity consists of two components: (i) the mechanical component (w m )
caused by unassisted impingement (0.007 ms - 1) shown above, and (ii) the
electrostatic component (we,) caused by the movement of the charged
particles in the electric field. So
10g(1 - ep ) = ex(w m
=

+ we ,)/2.303

log(l -

8m )

exw esl2.303

(9A.6)

where ep is the fractional efficiency of the precipitator and em is the


mechanical efficiency of the electrically dead precipitator. It should be noted
that both ep and em can be measured so that if ex is known then wes can be
estimated from equation (9A.6).
If em is small (or zero), which is very rare, then loge 1 - em) is zero and
equation (9A.6) becomes
(9A.7)

THEORETICAL CONSIDERATIONS

255

This is one expression of the well known Deutsch equation which was first
published in 1922 [1] and has been commonly used to estimate electrostatic
precipitator performance. Although the defect of omitting Em has been
recognised for many years this equation is usually quoted as
log(1 - E)

-!Xwe/2.303

(9A.8)

where E is the fractional efficiency of the precipitator and We is the effective


migration velocity.

9A.l.3

Improvement oj the Deutsch equation

The Deutsch equation is based on a number of assumptions. These include


spherical particles, even distribution of the dust in the gas, the gas velocity
and the electric current on the collecting plate, together with no particle
re-entrainment into the gas stream on rapping. These assumptions make the
equation unusable for practical purposes except for interpolating between
measured values.
It has therefore been necessary to develop improvements of this equation
to allow for the actual operating conditions prevailing in operating precipitators. A number of these improvements to the Deutsch equation are
discussed below.

9A.l.3.1 Matts-Ohnjeld equation. This improvement, which is widely


used to assess the performance of electrostatic precipitators, was developed
by A.B. Svenska FHiktfabriken [2]. In this method the Deutsch equation
(9A.8) is modified to
(9A.9)
where W k is the modified migration velocity.
This technique can only be used when the electrical conditions of the
precipitator are maximised and the temperature is constant. In theory, at
these conditions, W k will be constant for all values of !X (SeA). Therefore by
measuring the efficiency and SeA for a precipitator W k can be calculated and
then using this W k value the size of precipitator required can be calculated
for a different efficiency, or, of course, a new efficiency can be calculated if a
different seA is assumed.

9A.l.3.2 FL. Smidth equation. This equation has been used by F.L.
Smidth for sizing electrostatic precipitators since 1967 [3]. In this case the
Deutsch equation has been modified in the form
(9A.10)
where WB is the effective migration velocity in the first infinitesimal part of
the ESP and b is an empirically determined exponent found to be 0.22 in
most cases.

256

PRECIPITATOR SIZING METHODS

Like the Matts-Ohnfeld equation, because the exponent is empirically


determined, this equation gives a much better estimate of the precipitator
performance than the original Deutsch equation. This is despite the fact that
it requires a number of simplifying assumptions including disregarding
mechanical collection, re-entrainment, non-uniform gas distribution and
back-ionisation.

9A.l.3.3

Extended Deutsch equation (eS/ROJ


qB = (d 2 Ej4)(3k/(k

It is known [4] that

+ 2

where qB is the saturation charge, d is the particle diameter, Ec is the


charging field and k is the dielectric constant, and from electrostatic theory
Fl = qBEp

where F 1 is the force on the charged particle and Ep is the precipitating


electric field. So
where p = 3k/(k + 2).
From Stokes' law the viscous drag F 2 on a spherical particle is given by

F2

3nl]dw

where I] is the viscosity of the fluid and w is the particle velocity for steady
conditions F 1 = F 2. So
(9A.11)
so we see that
(a) as Ec and Ep are directly proportional to the applied voltage V, w is
raised to the square of the applied voltage. For example, a 25% increase
in voltage from say 32 kV to 40 kV gives an increase in w of over 56%;
(b) w is directly proportional to particle size, so that if the particle size can
be increased by 50%, w is increased by the same proportion;
(c) k and I] are constant at a fixed temperature, but they vary with
temperature. An increase in temperature from 120 DC to 330 DC causes
an increase in w of about 12% due to k and a decrease of 25% due to
1], so even over this large temperature variation the composite effect will
be small.
If (9A.ll) is substituted in (9A.8), the Deutsch equation becomes
10g(1 - e) = - C 2rxdpEcEp

and if Ec and Ep are proportional to V,


10g(1 - e) = -C 3 rxdpV 2

where C 3 is a constant.

THEORETICAL CONSIDERATIONS

257

Therefore, for a given feed material at constant temperature


(9A.12)
where C4 is a new constant. This equation shows that the collection
efficiency is directly proportional to the specific collecting area (:x), the
particle size td) and the square of the applied voltage (V2).
It has been determined that a precipitator which is electrically dead may
still collect particles at a relatively high efficiency. Previously it was believed
that the mechanical efficiency was low and mainly arose from gravitational
settlement. The advent of pilot plants and the modifications of the Deutsch
equation made it relevant to measure this efficiency. It has been found that,
on occasions, figures as high as 40-50% are evident and that this mechanical efficiency is nearly always significant. This mechanical effect, which is
mainly due to the turbulent motion of the gas causing particles to impinge
on dusty surfaces, is additional to the electrostatic effect.
To allow for this effect a term must be added to equation (9A.12). Thus
10g(1 - e)

10g(1 - eo)

+ C:xdV 2

(9A.l3)

where eo is the mechanical efficiency.


However, the charging of particles and electrostatic collection are not
possible until a corona is present so the term V 2 is not effective below ~,
the corona starting voltage. As the voltage is increased from 0 to Vg the
collection efficiency exceeds eo because some particles carry an electric
charge already and, as a precipitating field already exists, a small increase
in efficiency is found. Thus eo increases to eg at the corona starting voltage
~, and then above ~
10g(I - e)

log(l - eg )

+ C:xdV 2

(9A.l4)

This equation is called the extended Deutsch equation. The migration


velocity term in the original Deutsch equation (9A.8) has been replaced by
particle size and voltage components which can be measured. The collection
efficiency (e), the efficiency at Vg(c g) and the SCA (:x) can also be measured
and so, using the extended Deutsch equation, precipitator performance over
a range of operating conditions can be plotted with the log of the slip
[log(l - e)] as ordinate and oeV 2 as the abscissa. The ordinate may be
marked off as percentage slip or collection efficiency. This plot is called the
performance line [5-7]. The general shape of the performance line with
percentage slip as the ordinate is shown in Figure 9A.1.
The point A is the mechanical efficiency of the system and occurs when
V = O. Point B is the onset of corona and occurs at ~. The line AB is not
parallel to the abscissa because, as discussed above, some particles have a
small natural charge which allows some precipitation to take place. The line
BC is the effective performance line and if suitable values of:x and V are

258

PRECIPITATOR SIZING METHODS

100

.-__________

~--

__----__----_.o
50

10

90

la.5

95

>-

U
Z

...w
ii:

99
C

0.5

99.5
99.7

aV 2 x10' 4

Figure 9A.l Precipitator performance line.

selected then theoretically the line could be extended to very high efficiencies. This does not happen in practice because re-entrainment due to
rapping puts an upper limit on the efficiency that can be obtained and the
slope of the line decreases, as shown in Figure 9AJ, beyond C. Because the
extended Deutsch equation has the form y = a + bx, where b = Cd, the
slope of the performance line is proportional to the particle diameter at
constant temperature. A practical example of a performance line is given in
Figure 9A.2. In this graph each point represents a separate precipitator
efficiency test on a common dust.
9AJ4

Factors affecting electrostatic precipitation

9AJ4.1 Particle size. By substituting in the expression for w derived


from Stokes' law which is given in equation (9AJ 1) it can be shown that
w = O.095dEcEp/Y/ for an average non-conducting particle. Therefore, at
constant temperature and electrical conditions the migration velocity (w) is
proportional to the particle size (d).

259

THEORETICAL CONSIDERATIONS
100 . -______--------__-----------,0
50

50

.':

10

I:\.

90

:,

"

::::i

95

I ,

>
U
Z

(II

...wu::

'.'

99
0.5

99.5

0.3

99.7
0

u mV2xl0'4

Figure 9A.2 Precipitator performance.

The various equations derived up to now strictly apply only to particles


of the same size. However, in most cases, the dusts fed to precipitators
consist of particles with a range of different sizes with a distribution
characteristic of the material from which the particles originate. Assuming
that the particles collect independently according to equations (9A.13) and
(9A.14), then each size will collect at a different rate, the larger ones more
efficiently than the smaller ones [8]. Obviously, some single particle size is
required to represent the size distribution and it is usual to assume that this
is the mass median diameter (MMD), which is the diameter above and
below which 50% of the weight of the particles lie. Diagrams of the effects
of changing MMD on the efficiency of a precipitator [9] show that at an
seA of 60 m 2 m - 3 S - 1 a dust with an MMD of 5 Jim will exhibit a 4 % slip,
whilst at an MMD of 10 Jim the slip is reduced to 1.5% and at 25 Jim it is
0.6%.
This relationship is also demonstrated by the slope of the performance
line It has been shown that, for a range of dusts, the slope of the
performance line increases as the MMD increases [10]. This effect of particle

260

PRECIPITATOR SIZING METHODS

100 r-----------------------------,O
50

50

10

90

~ 5

95

a.
:;

tz

Ul

...w
ii:

99
99.5

0.2

L -_ _ _ _ _ _ _ _-L~_.J.~_ _L_~..L....~...J

99.8

Figure 9A.3 Effect of particle size on the slope of the performance line.

size on the slope of the performance line is shown in Figure 9A.3. The MMD
of the Dust A is 10 /lm whilst the MMD for Dust B is 6.5/lm.
The precipitator alters the particle size distribution of the dust, so that as
the material passes through the precipitator the dust becomes finer. Therefore, the latter sections of a precipitator collect material at a much lower
efficiency than the earlier parts.
9A.l.4.2 Temperature. In some industrial plants it is possible to select the
posltlOn of the electrostatic precipitator in the system. For instance in
pulverised-coal-fired power stations the positioning of the precipitator may
be either after the air heater, where the gas temperature is in the range
120-180C, or before the air heater, where the temperature is in the range
320-400 0c. Earlier it was claimed [11, 12J that precipitation proceeds less
erratically at the higher temperature Chot-side' precipitation) than at the
more common lower temperature (,cold-side' precipitation).
The argument for this was as follows. The 'cold-side' precipitation of fly
ash relies on surface electrical conduction for the successful passage of
current, because the glassy interior of the ash particles has a prohibitively

THEORETICAL CONSIDERATIONS

261

high resistivity at this temperature; so, if the surface conducts poorly, the
precipitator will suffer electrical congestion (such as back-corona) [4, 13]
with a consequent loss of efficiency. However, 'hot-side' precipitation should
not depend on surface conduction because at the higher temperatures the
electrical resistivity of the particle has been reduced to allow 'volume'
conduction through the bulk of the particle. Thus 'hot-side' precipitators
should not suffer from resistivity impediments and so should not exhibit the
variability of 'cold-side' precipitators in collecting different coals and ashes.
In fact, full-scale experience with 'hot-side' precipitators showed that they
were not, on the whole, more effective than 'cold-side' precipitators. When
this finding was coupled with the fact that the 'hot-side' precipitator had
operational difficulties due to the higher temperature, their use was discontinued except for special circumstances. A comparison of the performance of
precipitators operating at different temperatures has been made by Darby
and Whitehead [14]. They confirmed that in many cases the 'hot-side' unit
was no smaller than the 'cold-side' unit and at the same time the engineering
was more demanding.
The effect of temperature has also been discussed elsewhere [12,15-17],
but these publications do not discuss the effect of temperature on all the
parameters affecting electrostatic precipitation. To examine the effect of
these factors the extended Deutsch equation should be stated in full. So
expanding equation (9A.14)
10g(1 - e)

10g(1 - eg )

whilst below the corona starting voltage,


10g(1 - e)

+ C 2 PrxdV 2 11]

(9A.15)

10g(1 - eo)

+ C 2 PrxdVI1]

(9A.16)

These factors will be considered separately under gaseous, particulate and


electrical classifications.

(a) Gas-related temperature effects. Viscosity (1]) rises slowly as temperature increases giving greater drag forces on the particles and hence
reducing migration velocity. The increase in gas viscosity over the temperature range of, say, 120C to 330 C will decrease the slope of the performance line by about 25% which, if no other factors were affecting the
precipitator, could reduce the efficiency from 99% to 97%.
Temperature affects SCA (rx) if gas volume units are used but not if gas
mass units are used. For example, in a process generating a fixed mass rate
of gas and particles the SCA in units of m 2 1m 3 s - 1 (rxv) decreases with rising
temperature according to Charles' law. So, in a given precipitator, rxv at
120C is greater than that at 330 c by a factor of (330 + 273)1
(120 + 273) = 1.53, which is sufficient to reduce an efficiency of 99.5% to
95%. However, if the SCA is calculated in units of m 2 /kg s -1 (rx m ) there is

262

PRECIPITATOR SIZING METHODS

no change with temperature, but the gas velocity increases


installation, probably causing greater dust re-entrainment.

III

a fixed

(b) Particulate-related temperature effects. The mechanical efficiency


(eo) may be affected by the increase in velocity as temperature is increased
and also by a change in the surface characteristics of the particles, making
them either more or less sticky depending on their specific sources and
thermal history. Thus, it is difficult to estimate a priori the effect of
temperature on mechanical efficiency.
The dielectric constant (k) which is involved in the factor p (p = 3k/(k + 2)
increases with temperature. The effect on p of raising the temperature from
120 cC to 330C will be an increase in the slope of the performance line of
about 12%.
The particle size (d) will not be affected by temperature, but as discussed,
the particle stickiness may increase or decrease, thus affecting the clustering
of particles into agglomerates and this, in turn, will influence the effective
particle size and hence the slope of the performance line.
(c) Electrically related temperature effects. The fractional collection efficiency (e 5 ) at the corona starting voltage (~) is greater than the mechanical
efficiency (eo) because any natural charge possessed by some particles allows
them to collect electrostatically at voltages below ~. The difference between
eo and eg is small, however, and so the temperature effect on eg is likely to
follow that of eo.
The corona starting voltage (~) can be calculated from the equation
[18]:

(9A.l7)
where Eg is the electric field in k V cm - 1, a is the corona wire radius in cm
and J is the gas density relative to that at 25C.
As the temperature increases the gas density decreases and so Eg and
hence Vg are reduced. Furthermore, as the temperature increases the kinetic
energy of the gas molecules increases resulting in a higher current flow but
a reduced maximum sustainable voltage. This reduction in ~ causes the
performance line to move towards the ordinate, causing a reduction in the
value of the slip (or an increase in efficiency) at any given voltage. This is
offset if there is a fall in the maximum operating voltage (Vrna.) [12,19].
The effect of temperature on resistive impediments to precipitation is
difficult to assess because it is difficult to measure the resistivity of a dust in
the presence of an electric field in an operating precipitator. However,
considering the change from surface to volume conductivity discussed
earlier, the operational resistivity of highly insulating dust layers in precipitators is expected to rise to a maximum at some intermediate temperature
and then fall as 'hot-side' temperatures are approached. The sharpness of

THEORETICAL CONSIDERATIONS

263

this maximum will depend on the readiness of the dust surfaces to change
as they cool, and the temperature at which the maximum occurs will mainly
be determined by the dust composition.
These contrasting temperature effects make it very difficult to predict the
overall effect of increasing temperature on precipitator performance. For
instance pilot plant results for seven different bituminous coals [20] show
that there appears to be little advantage in precipitating the fly ash at 330C
rather than 120 0c. If current density is limited to the normal working
maximum at 120C of 2 x 10- 4 A m - 2 then the size of the precipitator
required for operation at 330C is larger in five of the seven cases. This
situation can only be reversed in favour of 'hot-side' precipitators if
considerably higher currents (up to 5 x 10 - 4 A m - 2) are accepted. Furthermore, information published on hot precipitators built in the USA [21]
shows that at 330 De precipitators designed to operate at 99% efficiency
have seA values up to 60 m 2 /m 3 s -1. This is equivalent to 100 m 2 /kgs- 1
which is the same size as a precipitator with an SeA of 90 m 2 /m 3 s -1 at
120C. This is not a small precipitator by international standards.

(d) Dust resistivity. If the electrical resistivity of the dust is high enough
there will be a large voltage gradient across the dust on the collecting
electrode causing electrical breakdown in the interstices between the particles. This initiates a 'back' -corona which will impede precipitation by
partly discharging particles during their approach to the affected area of the
collecting plate. Therefore, the electrical resistivity of the collected layer can
be critical to proper precipitation and so resistivity measurements have long
been popular in this method of particulate control.
In an operating precipitator the collected dust layer cannot impede the
precipitating action unless its resistivity (Pd) exceeds that of the gas
displaced during the collecting process (p g ).
For example, a typical gas at 125 e in a dust-free full-scale precipitator
will ordinarily conduct a current density of 2.5 x 10- 4 A m -1 on the
grounded collected plate with the applied field close to electrical breakdown
at 4 x 10 5 V m - 1 (0.4 k V mm -1). By definition the resistivity of the gas (Pg)
is the applied field across it per unit current density, so that under the
conditions cited the value of Pg is 4 X 10 5/2.5 X 10- 4 = 1.6 X 109 Q-m.
Thus, the dust layer must have a resistivity (Pd) exceeding 1.6 x 109 Q-m at
a field of 0.4 k V mm - 1 to interfere with its own collection in the precipitator
as described.
The importance of the volt/amp (or field/current density) characteristic
of a dust-free precipitator is that it enables Pg to be found, this being the
maximum dust resistivity that the precipitator can handle at a given
operating condition without the possibility of collection impediment.
The question then arises what impediment is incurred when Pd exceeds
Pg to a selected degree. An answer in principle to this question has been

264

PRECIPITATOR SIZING METHODS

Po/po'6
F-1 .25
100 , -__________________________
-.0

50

50

10

90

~
95

>
(.)
Z

(3

ii:
II.
w

99
A

0.5

0.2

L-~-'--_

_'__~__'_~_'_~___'__

345

99.5

___'____'_-:'

99.8

uV 2 x1 0"

Figure 9A.4 Effect of resistivity.

worked out in terms of the ratio r = Pd/P g , both resistivities being referred
to the same operating condition of the precipitator. It emerges [22] that the
factor F by which the original (dust-free) precipitating field in the precipitator is reduced following the collection of an excessively resistant dust
depends on the thickness of the layer according to:
F = 1 + (tld)(r - 1)

(9A.18)

where t is the dust layer thickness and d is the wire-to-plate distance and
assuming the original applied voltage can be maintained.
From (9A.l8) if tid is at its reasonable maximum of 0.05 and r is 6, then
F = 1.25, and an original precipitating field of 0.4 kV mm -1 is reduced to
0.32 k V mm - 1, just as if an applied voltage of 40 k V had fallen to 32 k V. In
a typical case, this means that a collecting efficiency of 99.5% is reduced to
96.6% (an increase in slip by a factor of six) as shown in Figure 9A.4. Note
that this assumes that no back-corona takes place.
At the instant when the dust layer is dislodged by rapping, the former full
collecting efficiency is almost restored, only to start falling immediately as
the dust layer accumulates again. It follows that high-resistivity dust

PRACTICAL CONSIDERATIONS

265

requires special optimisation of the rapping cycle to achieve the least


impediment to collection efficiency.
The electrical effect of a rising dust resistivity on the precipitator function
is that the applied voltage is increasingly transferred from the gas (where it
is effective for precipitation) to the dust layer (where it is not). This effect
reaches a limit when the voltage across the dust is high enough to cause
electrical breakdown of the dust, and the phenomenon of back-corona sets
Ill.

Back-corona adds considerably to the power demand of the precipitator


and causes instability. To avoid it the applied voltage must be reduced, and
the efficiency then falls at a faster rate with rising r values. Eventually the
voltage is so diminished that the normal corona cannot be sustained and
the precipitator ceases to function because particle charging has ceased.
The measurement in principle of the resistivity of a poorly conductive
powder layer is a comparatively simple matter, but to make that measurement relevant to the performance of a precipitator in which the powder has
been or will be collected is often extremely difficult. The difficulties lie in the
great sensitivity of high resistivity material to surface alteration caused by
adsorption or desorption of water vapour and ionogenic gases. The surface
may be subject to comparatively rapid and irreversible changes in air and
at elevated temperatures, and the resistivity will also be electric-field
dependent [23]. These features render most of the resistivity measurements
reported in the literature for such particles impracticable to interpret
usefully. As things stand the volt/amp behaviour of a precipitator both dusty
and dust-free is a good diagnostic guide to a resistivity problem in a
precipitator.
9A.2

9A.2.1

Practical considerations

Interpretation of test results

9A.2.1.1 Test results. The information required to apply either the W k


method, the W B method or the CSIRO extended Deutsch equation can be
obtained from tests carried out on either full-scale or pilot-scale plants. In
each case great care must be taken to replicate the conditions to be expected
on the plant being investigated. On many occasions this is easier on pilot
plants than on full-scale plant but one major problem with pilot plants is to
be completely sure that the dust being collected is exactly the same in
physical and chemical condition as that to be collected on the full-scale. It
must be remembered that the surface condition of the dust may have a very
large effect on the performance of the precipitator. Any dust removed from
a process may undergo considerable surface change when placed in a
different atmosphere or even just allowed to change temperature.

266

PRECIPITATOR SIZING METHODS

9 A.2.1.2 Application of equations. The precipitator's collection efficiency


or slip (percentage escaping from the precipitator) can be measured by
extracting isokinetic samples from the pipes entering and leaving the
precipitator. Commonly each sample passes through a preweighed filter
where the dust is separated from the gas, then through a cooler where the
gas is cooled to ambient temperature, part of the water in the gas is
condensed and the condensate collected. The saturated gas then passes
through a dry gas meter and is vented to atmosphere via a pump. The
weight of dust is determined by reweighing the thimble and the volume of
the gas is determined from the volume passing through the gas meter and
the volume of the condensate. Thus, the dust loadings into and out of the
precipitator are determined and hence the collection efficiency or slip can be
found.
The results have two major uses: (a) to supply information to precipitator
users and/or manufacturers on the precipitation characteristics of previously
uncollected dusts; (b) to investigate the electrostatic precipitation process.
To interpret these measurements the improved Deutsch equations described above are used. The interpretation of precipitator test results using
two of these methods is given below.
(a) wk method.

Equation (9A.9) states that


log(1 - s)

-(c('Wk )O.5/2.303

where for maximised electrical conditions s is the fractional collecting


efficiency; (X is the specific collecting area (m 2/m 3 s -1) and wk is the modified
migration velocity (m s - 1).
Full-scale electrostatic precipitators follow the wk relationship and, therefore, knowing the W k value for a dust, estimates can be made of the collection
efficiencies of other installations once the original collection efficiency is
known. However, there is always some uncertainty in this because there is
no guarantee that all installations will have exactly the same W k value even
when collecting the same dust.
With these limitations in mind, to demonstrate the use of W k a pilot-scale
investigation procedure was set up to determine efficiencies of collection,
and hence W k values, over a wide range of seAs for the collection of five
different fly ashes generated from a small pulverised coal combustor. The
testing was performed by varying the number of stages used in the
electrostatic precipitator at the combustion test facility and so varying the
specific collecting area.
The Wk values calculated from these tests were then used to make
predictions of the emissions from a given plant, assuming that 90% of the
fly ash reached the inlet of the electrostatic precipitator. Using this procedure results of the type shown in Figure 9A.5 may be obtained.

267

PRACTICAL CONSIDERATIONS

5000

1000

.,

.sa..

CI

::J

en 1 00

DUSTS

10
50

100

200

seA (m 2 /m3 s1 )
Figure 9A.5 Predictions of plant emissions using

Wk

values.

From the tests it was found that the W k values did vary from test to test
for the same coal. For instance, for one coal the results showed a variation
of Wk between 4 and 8 cm s - 1 whilst for another coal the results showed a
variation between 40 and 70 cm s - 1. This variation is sometimes due to tests
being carried out over a range of voltages according to the maximum
available at the time of the test. Therefore, it is important to carry out a
number of determinations of W k in order to obtain the most accurate
estimate of the value.
(b) Performance line method. Once corona voltage has been achieved
equation (9A.l4) applies.
log(l - e) = log(l - eg ) + C 1 cxdV2
where e is the fractional collection efficiency, eg is the fractional collection
efficiency at the corona starting voltage, cx is the SCA, d is the mass mean
particle size, V is the applied voltage and C1 is a constant. Considering
constant mean particle size
log(l - e) = log(l - eg ) + C 2 CXV 2

where C 2 is a different constant.

268

PRECIPITATOR SIZING METHODS


100

__________________________--,o
50

10

90

95

>
u
zw

i3

11.

iL

:i

IL.

en

99
0.5

99.5

0.1 '---__________________________---'" 99.9

seA (m2/m 3s1 )


Figure 9A.6 Effect of voltage change.

This final relationship can be used to plot a performance line [5,6,10].


The general shape of the performance line is shown in Figure 9A.l and is
explained above.
The performance line is useful for demonstrating the dependence of
precipitator efficiency on applied voltage and particle size. These effects are
shown graphically for ideal cases in Figures 9A.6 and 9A.7, in which
10g(1 - e) is shown as precipitator electrostatic collection efficiency. In
Figure 9A.6 the three lines correspond to three different applied voltages
covering the ratio of 2: 1. If a vertical section is taken at the extreme right
of the figure it is seen that a precipitator slip of 1% at constant Cl and applied
voltage V is reduced to 0.1 % by increasing the voltage by a factor )2 but
is increased to 10% upon decreasing the voltage by the same factor. It is
clear that the voltage is extremely important to maximise and that only one
or two kilovolts at 40 kV produce a significant increase in collection
efficiency.
In Figure 9A.7 the voltage variable is included in the abscissa, and the
three lines correspond to three different particle sizes covering the ratio 4: 1.
Once again taking the vertical section at the extreme right of the figure, it
is seen that at constantCl and V the finest dust (represented by the
uppermost line) corresponds to a slip of 10%, whereas the dust four times

269

PRACTICAL CONSIDER AnONS


100

__________________________--,o
50

10

90

95

>

zw
0

...iLw

II.

::::;
rn

99
0.5

99.5

0.1 '--__________________________---" 911.9

Figure 9 A. 7 Effect of particle size.

coarser (the lowest line) is precipitating with 0.1 % slip under the same
operating conditions. Bearing in mind that the average particle size of, say,
fly ashes from different coals can readily vary over a ratio of 2: 1, Figure
9A.7 demonstrates that a doubling of particle size alone is capable of
explaining a change in electrostatic slip from 10 to 1% for example.
The minimum number of test required to properly establish a performance line is six. The sequence of these tests is as follows:

Test

1
2

SCA
Value of most relevant
IX

3
4

IX

IY.

IY.

Value to give t: > 99%

IX

Voltage

Point
on Figure 9A.8

Vrn (maximum voltage

voltage)
Vrn
V. + (Vrn V. + 2(Vrn 0

B
C
D
E
F

V. (corona starting
V.)/3
V.)/3

270

PRECIPITATOR SIZING METHODS

100

r-----------------------------~O

50

II.

10

90

95

:J

>-

(J

III

u::
u.

w
A

0.5
0.3

99

' - -____________________________----l

99.5
99.7

Figure 9A.8 Tests for performance line.

The position of these points is shown in Figure 9A.8. This is the minimum
number of points required to establish a meaningful performance line, and
in practice ten and often more are required.
9A.2.1.3

Usefulness of equations

(a) Investigation of an unknown dust. When an unknown dust is investigated the first activity is to establish a performance line such as Figure
9A.2. This information can be used as a guide to the precipitator manufacturer provided there is a good relationship between this pilot plant and the
full-scale [24]. Quite often the performance of an unknown dust can be
compared with that of a well known one, and hence the precipitator
manufacturer can estimate from his experience the size and type of precipitator required for a new dust. When the performance lines for two dusts are
compared the variation in performance at constant conditions can easily be
seen (Figure 9A.3). For instance, from this figure we see that at a constant
operating condition of ~V2 = 4.6 X 104 , the slip for dust A is 0.6%, whilst
for dust B it is 1.9%. So that for these two dusts more than three times as
much dust would be emitted for dust B as for dust A. On the other hand if

271

PRACTICAL CONSIDERATIONS
100 ,--_ _ _ _ _ _ _ _ _ _ _ _ _ _--,0

tc..

50

50

10

90

95

120' C

:::;

UI

>
u
zw
13
ii:
u..
w

99
0.5

99.5

99.8
2

a vV2 x10'4

Figure 9A.9 Precipitator results for the same dust obtained at two different temperatures.

a slip of 0.6% was required for both dusts then the value of IXV 2 for the two
dusts would be 5.8 x 104 and 4.6 x 104 , respectively. Therefore, at the same
operating voltage the precipitator for dust B is more than 25% larger than
that for dust A.

(b) Investigation of the effect of temperature. The effect of temperature


on the performance of a precipitator has been investigated extensively using
the pilot plant and the performance line technique for interpreting the
results.
Figure 9A.9 shows some precipitator results for the same dust obtained
at two different temperatures plotted on the usuallX v V 2 basis. From this plot
it would be easy to interpret the results as showing a considerable improvement at 370 DC over those at 120C. For instance at IXv V 2 = 3 X 104 the slip
at 120 DC is 7% and at 370 DC is 0.8%. However it is not possible to compare
these results on this basis because as the temperature is increased the volume
of gas also increases and therefore IXv measured in m 2 m - 3 S -1 is temperature dependent.
When the temperature is varied the specific collecting area of the
precipitator must be based on the weight of gas being treated which is

272

PRECIPITATOR SIZING METHODS


1DD

5D

5D

1D

9D

95 >
()

120' C

ii:

u..
w

99

D.5

D.2

99.5

'---~_ _'--~'--_'--_'---'---'---'---'---'

99.8

Figure 9A.I0 Same results plotted independently of temperature.

independent of temperature. Thus, a new specific collecting area, which we


call !lC m , must be used with the units m 2 jkg s -1. Figure 9A.1O shows the same
results plotted independently of temperature and now it can be seen that, in
this case, the effect of temperature is not so great. At !lCm V 2 = 3 X 104 the
slip at 120C is still 7% but at 350 DC it is now 3.5% and if a slip of 0.6%
was required the !lCm V 2 value increases from 5.5 x 104 to 6.5 X 104 as the
temperature is increased from 120C to 350 c, an increase in precipitator
size of 18% at constant operating voltage.
The effect of temperature on the required precipitator size is, however,
not linear. There would be an increase in the required size and a subsequent
decrease in size as the temperature was changed from 120C through 180 C
and on to 370C. The severity of this effect could only be determined by
pilot-scale or full-scale testing.
(c) Investigation of the effect of dust composition. Electrostatic precipitation data at 120C for 29 Eastern Australian bituminous coals burned in
the CSIRO pilot-scale facility have been correlated with readily measured
characteristics of the coals and their laboratory ashes [25]. These data are
shown as crosses in Figure 9A.11.

273

PRACTICAL CONSIDERATIONS

130
120
110

'"

.sw

100
90

en

II:

80

I-

0(

l-

ii:

70

(3
W

II:
~

60
50
40
30

30

40

50

60

70

80

90

100

SI+AI+Fe ("!o)

Figure 9A.ll Electrostatic precipitation data correlated with readily measured characteristics
of Eastern Australian bituminous coals and their laboratory ashes.

It may be hypothesised that, if a fly ash is refractory, the particles will


firstly have less opportunity to enlarge by coalescence while molten in the
flame and secondly the collected ash is likely to be less electrically conducting. Thus silicon, through its refractory oxide quartz, and silicon and
aluminium through the formation of aluminosilicates may adversely affect
electrostatic precipitation. Furthermore, iron under oxidising conditions
may have the same effect.
Figure 9A.1l shows the correlation between IXm' the specific collecting
area, corrected to 15 % mineral matter in the coal, in m 2 kg - 1 S - 1 to achieve
an emission of 0.1 g Nm - 3 and the sum of the elements silicon, aluminium
and iron expressed on the ash content such that Si + Al + Fe + Ti +
Mn + Ca + Mg + Na + K + P + S = 100%. The figure shows that there
are two adjoining straight lines which make up the correlation. Since its
publication this correlation has been tested with other coals on the CSIRO
pilot plant and the results for the next 17 coals tested show excellent
agreement with the original correlation. These can be seen in Figure 9A.1l
as the results denoted by circles.

274

PRECIPITATOR SIZING METHODS

The conversion factor f to convert am (corresponding to 15% ash) into


the precipitator size at the actual ash percentage, A, may be obtained from

1.364 - 0.488 loglo[(100/A) - 1]

(9A.19)

Thus, it is now possible, knowing the basic coal and ash analyses, to obtain
a reasonable prediction of the size of precipitator required for a specific coal.
However, pilot-scale and/or full-scale precipitator tests may still be required
to confirm this size.

9A.3

Precipitator modelling

Mathematical modelling and practical testing are the two ways of simulating the performance of a full-scale electrostatic precipitator.

9A.3.1

Mathematical modelling

9A.3.1.1 Computer models. A number of computer models for electrostatic precipitators exist and typical examples can be found in the literature
[9,26-29].
The practical application of these models is difficult. In their 1975 paper,
Gooch and Francis [9] said: 'Calculation of overall collection efficiency of
polydispersed particulate in an electrostatic precipitator from theoretical
relationships gives results considerably higher than those obtained from
measurements on fulI-scale units for coal-fired power boilers.' Corrections
to the idealised or theoretical collection efficiency to estimate the effects of
non-uniform gas flow, re-entrainment and gas by-passing the electrified
sections reduce the overalI values of calculated efficiency to the range of
values obtained from field measurements. These calculations suggest that
the theoretical model may be used as a basis for quantifying performance
under field conditions if the major non-idealities were to be quantified.
9A.3.I.2

Regression equation models. Several regression equation models


have been published. Two examples are outlined below.
Frisch and Coy [11] proposed an equation which had the form
1] =

1-

e[-k(P,/A)"(A/Q)'(v)'(x)d]

where 1] is the fractional efficiency, PjA is the power density (Wm-2), A/Q
is SCA (m 2/m 3 s - 1), V is the average treatment velocity (m s - 1), x is the
mass median particle size (/lm), k is the regression constant and a, b, c, and
d are regression coefficients. This equation was combined with: corona
characteristics, charging mechanism, resistivity, fuel specifications, temperature, etc.

PRECIPIT ATOR MODELLING

275

Tassicker [30] proposed the equation


w = ao +

a In(g) + azlogp + a P + a c~o) + as C~oy


1

a6

+ d + a7 z + a s log 1oz
where w is the effective migration velocity (m s -1), A/Q is the SCA
(m z/m 3 s - 1), P is the resistivity (n m - 1), T is temperature ( 0c), d is mass
median particle size (m), z is ash content (%), an are the regression
coefficients.
It can be seen from these two equations that these mathematical models
are complicated and require a considerable amount of information which on
many occasions will not be readily available. It should furthermore be noted
that these two equations have completely different forms, with the first one
being a product whilst the second is a sum.
9A.3.2

Practical testing

Many organisations have pilot-scale rigs and laboratory-scale equipment


designed to assess the electrostatic precipitation characteristics of dusts. In
some cases, such as fly ash generated in the combustion of pulverised coal,
this dust must first be produced in the laboratory.
9A.3.2.1 Laboratory scale equipment. Coal ash may be produced in a
small laboratory furnace, designed for small amounts of coal which consist
of

a gas mixture of air, oxygen and propane


a pulverised coal feeder
a burner for the gas/coal mixture
an electrically heated vertical tube furnace
a heated ash collector.

The ash is generated by dropping the coal particles through a propane flame
and collecting the resulting ash. In this manner an ash similar to that
produced in a power station is generated [31-33].
The characteristics of the dust are then measured. Key measurements
include the resistivity of the dust, the effect of the dust on corona characteristics in a laboratory-scale electrostatic precipitator, determination of the
dielectric constant of the dust which affects its ability to be charged and so
collected, measurement of its particle size distribution, and other parameters
which affect re-entrainment.
These are then used to assess the precipitability of the dust [30,31,
34-36]. With or without these results it is still important, whenever possible,

276

PRECIPITATOR SIZING METHODS

to carry out pilot plant tests to obtain operating information on the


performance of the dust in an electrostatic precipitator before building a
full-scale plant.

9A.3.2.2 Pilot-scale equipment. A pilot-scale electrostatic precIpitator


may be either a transportable piece of equipment which is taken to a
full-scale plant and then fed with dirty gas by a side-stream of the main gas
flow, or may be a fixed installation in a laboratory attached to a dust feeding
system.
For some processes the dust must be generated as part of the electrostatic
precipitator assessment process. For instance, fly ash from the combustion
of pulverised coal can only be properly tested if freshly generated fly ash in
its original flue gas is considered. In this case a test furnace would be used.
The time/temperature history of this test furnace must be carefully designed
to be similar to the time/temperature history through power-station boilers.
The flue gases exit from the test furnace into the heat exchanger followed
by a splitter. The purpose of this splitter is to allow varying quantities of
flue gas to be passed to the test electrostatic precipitator so that the specific
collecting area can be altered without changing the number of stages on-line.
Electrostatic precipitators with a tubular configuration are commonly
used for these tests. These have number of advantages, which are itemised
below:
no sneakage
minimum edge effects
minimum corona wind problems
minimum temperature control problems
less mechanical dropout with upward facing entries to each zone.
The main difficulty which has been foreseen with either a wire cylinder or
wire plate configuration has been mechanical drop-out, which is common in
pilot-scale installations. It is felt that this could be minimised with a tubular
configuration by directing the gas carrying the fly ash upwards at the
bottom to each stage by fitting turning vanes. Commonly the installation
comprises a number of vertical tubular stages to allow for variations in the
specific collecting area. The operating temperature of the precipitator should
be variable up to about 400C allowing simulation of cold-side or hot-side
precipitators.
Each precipitator stage should be energised by a single transformer
rectifier set. Intermittent energisation is a low cost control strategy which
can be easily fitted to the electrostatic precipitator and is extremely
beneficial in minimising the difficulties associated with the collection of high
resistivity fly ashes. Using this energisation one or two half-waves are
imposed from the transformer rectifier set for the test electrostatic precipi-

277

PRECIPITATOR MODELLING

tator followed by an adjustable period of no energisation. During this period


the voltage at the emitting electrode terminal decays to a minimum value
which is controlled to the maximum level that can be sustained before
currents run away due to back ionisation.
The major factors taken into consideration in the design of pilot-scale
electrostatic precipitators are:

SCA/velocity relationship
rapping
gas distribution
electric field
particle size and distribution.

All these factors cannot be taken into account at the same time, but the two
most important - the electric field and the SCA/velocity relationship - must
be given priority. A description of a typical pilot-scale electrostatic precipitator and its mode of operation has been given by Darby [37]. This paper
also details the corrrections which may need to be applied to the pilot-scale
results to translate them into full-scale design data.

100

PILOT-SCALE
FULL-SCALE

10

90

>
u

~ 5

95

II.

::::;

zw

(3

u::

u.
w

II)

x xX

Ix

4< I

" , l

..

\x x

99

0.5

_ 99.5

0.2

99.8

aE 2 x10- 2

Figure 9A.12 Good relationship between pilot-scale and full-scale plants_

278

PRECIPITATOR SIZING METHODS

If these pilot-scale test results are to be applied to full-scale plant it is


important that there is a good relationship between the pilot and full-scale
plants. An example of such a relationship is shown in Figure 9A.12 from
which it can be seen that, in this case, the relationship between the two
plants is very good.

References
1. Deutsch, W. (1922) Bewegung and Ladung Der Elekrizitiitstriiger 1m Zylingerkondensator.
Ann. Phys., 68, 335.
2. Matts, S. and Ohnfeldt, P.O. (1963) Efficient gas cleaning with SF electrostatic precipitation. SF Rev. 1963-1964, 6,7. 105-22.
3. Petersen, H.H. (1990) A precipitator sizing formula. 4th International Conference on
Electrostatic Precipitation, Beijing, China, September, International Academic Publishers,
Beijing, 1992, pp. 330-8.
4. White, H.1. (1963) Industrial Electrostatic Precipitation. Addison-Wesley Publishing Co.,
Reading, MA, p. 297.
5. Paulson, C.A.J. and Potter, E.C. (1974) Reduction of particulate emissions to air by
improved assessment of electrostatic precipitators. 2nd National Chemical Engineering
Conference, Queensland, Australia, July, pp. 404-11, Inst. Chern. Engs, Australia.
6. Paulson, C.A.l., Potter, E.C. and Kahane, R. (1974) New ideas on precipitation technology
from the CSIRO combustion rig. Institute of Fuel Symposium on the Changing Technology
of Electrostatic Precipitation, Adelaide, Australia, November, pp. 26, Inst. Fuel, Adelaide,
Australia.
7. Potter, E.c. and Paulson, C.A.l. (1974) Improvement of electrostatic precipitator performance by carrier-gas additives and its graphical assessment using an extended Deutsch
equation. Chern. Ind., pp. 532-33.
8. White, H.J. (1975) Role of electrostatic precipitators in particulate control- a retrospective
and prospective view. J. Air Pol/ut. Control Assoc., 25(2), 102.
9. Gooch, 1. and Francis, N. (1975) A theoretical-based mathematical model for calculation
of electrostatic precipitator performance. Symposium on Electrostatic Precipitation for the
Control of Fine Particles, Pensacola Beach, FL, USA. EPA-650/2-75-016, September.
10. Paulson, C.A.l., Kahane, R. and Potter, E.C. (1976) Electrostatic precipitation of flyash
from a range of Australian coals. Institute of Fuel Conference on Energy Management,
Sydney, Australia, November, Inst. of Fuel, pp. 20.1-12, Sydney, Australia.
11. Frisch, N.W. and Coy, D.W. (1974) Sizing electrostatic precipitators for high temperature
collection of flyash. Institute of Fuel (Australian Membership) Symposium on the Changing
Technology of Electrostatic Precipitation. Adelaide, Australia, November, 28 pp., Inst. of
Fuel, Adelaide, Australia.
12. Walker, A.B. (1974) Hot-side precipitators. Symposium on Electrostatic Precipitators for the
Control of Fine Particles, Pensacola Beach (Florida). EPA-65012-75-0J6, September.
13. Schmidt, W.A. (1949) Electrical precipitation and mechanical dust collection. Ind. Eng.
Chern., 41, 2428.
14. Darby, K. and Whitehead, C. (1974) The use of electrostatic precipitators in current power
station practice. Institute of Fuel (Australian Membership) Symposium on the Changing
Technology of Electrostatic Precipitation, Adelaide, Australia, November, 35 pp., Inst. of
Fuel, Adelaide, Australia.
15. Matts, S. (1975) 'Cold-side' precipitators. J. Air Pol/ut. Control Assoc., 25. 146.
16. Hall, H.1. (1975) Design and application of high voltage power supplies in electrostatic
precipitation. J. Air Pol/ut. Control Assoc., 25, 132.
17. Tassicker, 0.1. (1975) Some aspects of electrostatic precipitator research in Australia. J. Air
Pol/ut. Control Assoc., 25, 122.
18. Cooperman, P. (1960) A theory for space-charge limited currents with application to
electrical precipitation. Trans. lEE, Pt. 1, 79, 47-50.
19. Shale, c.c. (1967) New concept of electron detachment for air in negative corona at high
temperature. US Bureau of Mines Information Circular 8353.

REFERENCES

279

20. Paulson, CAJ., Potter, E.C. and Kahane, R. (1978) The influence of temperature on
electrostatic precipitation performance. CSI RO Conference on Electrostatic Precipitation,
Leura, Australia, August, pp. 12.1-15, CSIRO, Sydney, Australia.
21. Kiff, J.W. (1976) Performance data for Western Precipitation's hot precipitators. Western
Precipitation Seminar on High Resistivity Flyash Collection, Sydney, Australia, March,
35pp. Joy Manufacturing Co., Sydney, Australia.
22. Potter, E.c. (1988) Principles of practical gas cleaning. 3rd CSIRO Conference on Gas
Cleaning. Medlow Bath, NSW, Australia, August, pp. 2.1-17, CSIRO, Sydney, Australia.
23. Goard, P.R.C. and Potter, E.c. (1978) Operational resistivity measurements on freshly
generated flyashes. CSIRO Symposium on Electrostatic Precipitation, Leura, Australia,
August, pp. 3.1-8, CSIRO, Sydney, Australia.
24. Potter, E.C. and Paulson, CAl. (1975) What size precipitator? A new basis for designing
plant to specified dust-extraction performance. Proceedings International Clean Air Conference, Rotorua, New Zealand, February, pp. 159-75, Clean Air Soc., Sydney, Australia.
25. Paulson, CAJ., Potter, E.C. and Vale, J.W. (1986). Correlation of some readily-measured
parameters of coal and flyash with electrostatic precipitator performance. The World Clean
Air Congress, Sydney, Australia, August, pp. 420-7, Clean Air Soc., Vol. 3, Sydney,
Australia.
26. Theodore, L. and Pardini, J. (1971) Application of modelling and simulation techniques to
the design of electrostatic precipitation. 64th Annual Meeting of the Air Pollution Control
Association, Atlantic City, NJ, USA, June, APCA, Pittsburgh, USA.
27. Theodore, L. and Eastmont, T. (1972) Simulation of an electrostatic precipitator- effect of
velocity, particle size, particle mass flow rate and electrostatic force distribution on
collection efficiency. 65th Annual Meeting of the Air Pollution Control Association, Miami
Beach, FL, USA, June, APCA, Pittsburgh, USA.
28. Theodore, L., Reynolds, I. and Navarette, R. (1973) Results of a new technique for
calculating collection efficiencies of electrostatic precipitators. 66th Annual Meeting of the
Air Pollution Control Association, Chicago, IL, USA, June, APCA, Pittsburgh, USA.
29. Reynolds, J., Mercando, A. and Theodore, L. (1976) The effect of voltage in two-stage
electrostatic precipitator efficiency: comparison between model and experiment. 69th
Annual Meeting of the Air Pollution Control Association, Portland, OR, USA, June, APCA,
Pittsburgh, USA.
30. Tassicker, 0.1. (1974) Performance of cold-side and hot-side electrostatic precipitators
treating high resistivity flyash. Institute of Fuel (Australian Membership) Symposium on the
Changing Technology of Electrostatic Precipitation, Adelaide, Australia, November, p. 19,
Inst. of Fuel, Adelaide, Australia.
31. Sullivan, K.M. (1975) A comparative study of laboratory fiyash and power station fiyash.
Australian Coal Industry Research Laboratories Ltd., P.R. 75-10, ACIRL, Sydney,
Australia.
32. Sullivan, K.M. (1975) A comparative study of laboratory fiyash and power stationfiyash. Part
2. Australian Coal Industry Research Laboratories Ltd., P.R. 76-12, ACIRL, Sydney,
Australia.
33. Baker, J.W., Sullivan, K.M. and Tassicker, 0.1. (1977) Assessment of a laboratory
technique for predicting the precipitability of flyash derived from a coal bore core.
Proceedings of the Fourth International Clean Air Congress, Tokyo, Japan, May, Clean Air
Soc., Tokyo, Japan.
34. Baker, l.W. and Sullivan, K.M. (1976) The examination of the electrostatic precipitability
of coal. Institute of Fuel (Australian Membership) Conference on Energy Management,
Sydney, Australia, November, pp. 19.1-15, Inst. of Fuel, Adelaide, Australia.
35. Tassicker, 0.1. and Sullivan, K.M. (1973) Estimation of precipitator performance for
collection of flyash by examination of low sulphur bore cores. 66th Annual Meeting of the
Air Pollution Control Association, Chicago, IL, USA, June, APCA, Pittsburgh, USA.
36. Baker, J.W. and Sullivan, K.M. (1978) Developments in electrical testing of fly ash in
relation to electrostatic precipitation. Proceedings of the Sixth International Clean Air
Conference, Brisbane, Australia, May, Ann Arbor Science Publishers, Michigan, USA,
pp. 223-38.
37. Darby, K. (1981) The use of pilot testing in field and laboratory. 1st International
Conference on Electrostatic Precipitation. Monterey, CA, USA, October, APCA, Pittsburgh, USA.

9B

Models of electrostatic precipitators


M.REA

9B.l

Basic concept

The performance of an ESP, like that of any other industrial process,


generally depends on some input or process variables and on the design and
operating parameters (Figure 9B.1).
In the case of an ESP the main input or process variables are:
the flow rate, the temperature, the pressure, the chemical composition of
the gas;
the size distribution, the chemical composition of the solid particles which
determine several physical parameters like the resistivity, the cohesivity
and adhesivity, etc.
The main design parameters are:

the
the
the
the
the
the

cross-section which determines the gas velocity,


duct width and height,
applied voltage and its function in time (its control),
shape of electrodes (both the emitting and the collecting electrodes),
type and frequency of rapping,
design of the hopper and the uniformity of the air flow.

Process
variables

MODEL

Design
parameters

Figure 9B.l General representation of a model.

Performance
variables

BASIC CONCEPT

281

Finally the main performance variables are:


the collection efficiency
the derivative of the collection efficiency in respect of the input or process
variables, commonly called 'flexibility'.
In this respect any model simply represents the set of equations which
relates the performance variables to the input variables and the design
parameters. An appropriate model is needed to define the design parameters
in order to attain the required performance with given input variables. It is
quite difficult to define the equations relating the output variables to the
input variables and the design parameters, as even an exhaustive identification of the input variables represents a difficult task. It is obvious that
simplified models can be profitably used which take into account efficiency
as the output variable and consider only those input variables and design
parameters whose variability mostly influences this output variable. The
difficulty of solving the set of equtions is today reduced by the availability
of powerful numerical methods and computers; it is worth mentioning that
as the computing power in solving any set of equations has increased,
scientists are prone to take into consideration more input variables and are
less stimulated to identify those input variables and those physical phenomena which have most influence on the efficiency (output variable). Several
input variables are multidimensional, i.e. they are distributed in space and
are frequently variable in time; this brings a great increase in complexity
when determining and solving the set of equations relating to any model.
Sometimes it can be a reasonable approximation to neglect the distribution
in space and/or the distribution in time; in these cases unidimensional or
two-dimensional models are proposed.
The phenomena playing a role in electrostatic precipitation may becharacterised by time constants dissimilar even by several orders of magnitude; in these cases it is advisable to initially solve the equations of the
fastest developing phenomena and then place the results as constant
parameters in the equations which describe the slow varying phenomena.
For instance, the equations describing the corona discharge are solved in
order to determine the time-averaged value of the space charge and of the
electric field distributions; these values are taken into account in the
equations describing the charging of the solid particles and their drift/
migration to the collecting electrodes.

9B.1.1

The Deutsch equation

The oldest and simplest model still widely used is the Deutsch equation
(Figure 9B.2). Its proposer, Mr. Walther Deutsch (1885-1957) was a great
German scientist and technician [1]. He proposed that the particles are
driven to the collecting plates by the electrical field with a constant velocity

282

MODELS OF ELECTROSTATIC PRECIPITATORS

Process
variables

MODEL

r---

Migration
velocity

~
Gas flow Ir-rate
.

>

Deutsch
equation

~ Collection
t-y
efficiency

Specific
collection
area

Figure 98.2 The model based on the Deutsch equation.

(migration velocity). Of course this velocity will depend on several other


process variables, but it can be experimentally determined or evaluated by
comparison with other similar installations. In these hypotheses the collection efficiency can be evaluated using the following 'Deutsch' equation:
1J

( W.S)

( W.S)

= 1 - exp( - w SCA) = 1 - exp - v. A = 1 - exp - G

where 1J is efficiency, W is migration velocity (velocity of the solid particle


toward the collecting plate), v is mean gas velocity, S is the surface of the
collecting plates, A is overall cross-section of the electrofilter, G is gas flow
rate = V A and SCA is specific collection area = S/G.
The Deutsch equation may be used in two different ways:
once the efficiency has been measured, the migration velocity (which is
then called effective) can be computed;
once the effective migration velocity has been estimated, for instance
because of the similarity with another precipitator, the collecting surface
(S) can be evaluated in order to achieve the desired efficiency with the
given gas flow rate (G).

BASIC CONCEPT

283

It is interesting to mention how the equation has been derived because it


points out the influence of some process variables and it explains some
suggested later modifications by other workers [2]. Let us assume that:

the particles are completely charged;


because of gas turbulence, the particles are uniformly distributed at any
cross-section;
the gas velocity does not influence the particle velocity toward the plates;
the particle always moves at its electrical terminal velocity;
there is no interaction between particles;
the collision of ions and neutral gas molecules has no effect;
at a short distance from the collecting plate the gas flow is laminar and
the particles are driven to the plate with velocity, w.
With reference to Figure 9B.3, let Hand Wbe the height and width of the
gas duct respectively, L be the length in the direction of the gas flows, and
v the velocity of the gas stream. After a time increment t, during which the
gas stream has moved a distance L = vt, all the particles N present in the
boundary layer, at a small distance d from the plate, will be driven to the
collection plate. If w is the drift velocity of the particles to the plate, it
follows that d = W t and N = H W L. The incremental equation for
particle removal will be:
,1.N

w,1.L
vW

Figure 9B.3 Single duct of an electrostatic precipitator.

284

MODELS OF ELECTROSTATIC PRECIPITATORS

which, after integration, gives:


N

( W.S)
vA

No exp - - -

where No represents the particle concentration at the inlet and N the


particle concentration after a length L of gas path. The efficiency, expressed
as the ratio of the collected particles to the inlet particles, in the above after
rearrangement, becomes the Deutsch equation.

9B.l.2

Charging of particles and the modified Deutsch equation

Several improvements of the Deutsch equation were proposed in order to


take into account other process variables which were recognized to play an
important role while still taking advantage of the simplicity and efficiency
of the Deutsch equation.
The most important modification of the Deutsch equation is that
suggested by Matts and Ohnfeldt [3J; its purpose is to take into consideration the influence of the particle size distribution on the collection
efficiency. In order to understand the reasons for this it must be remembered
that the corona discharge developed by industrial precipitators produces
negative ions which impact with the solid particles flowing between the
collectors and transfer to these some of their electric charge. Collision of the
negative ions with the solid particles is generally due to random motion, but
the electric field distorted by the solid particles may either favour or hinder
the collisions depending on the charge on the particle and on its size.
In addition to the motion of the particle towards the collecting plate,
there is the resultant effect of several forces (see Figure 9B.3) acting on the
particles, some of which depend on the size of the particle, itself.
Fe=q

electrical force

F~ = 6nalJw

viscous drag force

Fg=rng

gravitational force

dw
F., = rndt

acceleration force

where is the applied electric field, q is the charge on the particle, IJ is the
viscosity of medium, a is the radius of the spherical particle, w is the velocity
of the particle, rn is the mass of the particle and g is the acceleration due to
gravity.
If the variable q(a), evaluated following the model of charging and the
resultant of the previous listed forces acting on the particle, is solved, the

THE MODERN APPROACH TO COMPUTER MODELLING

285

migration velocity of the particle w becomes


w =

E
k T
--In
6nry e

--p-

[0 +

1["

a v N~ . e 2 .
kT

t)]

where Ep is the electric field close to the plate, k is the Boltzmann constant,
T is the temperature, e is the charge on the electron, v is the gas velocity
and N~ is the negative ion density.
In particular this implies that migration velocity is a function of particle
size a and that the Deutsch equation should be evaluated for each class of
particle size. If a log-normal distribution of particle size is assumed, the
Deutsch equation becomes:
ry = 1 - exp( -Wk SCA)k

where k represents a factor taking into account several process variables.


Another important modification to the Deutsch equaton is that proposed
by Cooperman [4] which takes into consideration a diffusional force acting
on the particles due to the different concentration along the precipitator,
which reduces the effective migration velocity. The suggested equation takes
the following form:
ry = (1 -

_13_)
- exp[ -w(1 1-:x

:x). X SCA]

where :x is the coefficient of variable erosion, i.e. re-entrainment due to


rapping, is the coefficient of constant erosion, i.e. re-entrainment due to
gas scouring and X is the ratio between dust concentration close to the plate
and average value.

13

9B.2
9B.2.]

The modern approach to computer modelling


Early models

The development of computers during the 1970s gave rise to the development of computer-based models for electrostatic precipitation. Several
scientists of the Southern Research Institute in the US have been involved
in this task guided by Oglesby and Nichols [2]. The basic approach was to
subdivide the precipitator into small incremental lengths L and to subdivide
the particles into a number of particle size increments a so that the Deutsch
equation becomes:
ryi.j= 1-exp(-wi.j SCA)

where w i j is the migration velocity of the jth particle size and SCA j is the
specific collection area for the jth incremental length. The collection

286

MODELS OF ELECTROSTATIC PRECIPITATORS

efficiency i for the ith particle size becomes:


'1'

= Lj '1i,j' Ni,j

Ni,j

where Ni,j are the number of particles of the ith particle size per cubic meter
of gas entering the jth increment of length.
The overall collection efficiency becomes:

where Pi is the percentage by mass of the jth particle size. In fact this model
takes into account the space distribution of flow and the size distribution of
dust.
In these models a number of important phenomena like rapping reentrainement, flow pattern were represented by correction factors. Some
other phenomena had to be neglected such as:
the influence of the dust on the space charge density and electric field
strength;
the influence of the dust layer thickness and resistivity on the electric field
including the back-corona;
the formation of the dust layer, i.e. the probability of the charged particles
being collected by the dust layer or causing the re-entrainement of a part
of the dust layer;
the erosion of the dust layer produced by the gas flow;
the re-entrainement of dust produced during rapping of the collecting
plates.
The approach which considers the basic physical equations can be
applied to the elementary particles (solid and gaseous) present in a small gas
volume, or on a small surface of the collecting plate and numerically, then
integrates this equation all over the electrostatic precipitator, is becoming
very popular with a number of investigators.
The main problem with the models produced by this approach is the
dimensions of the cells into which the precipitator has to be divided; in order
to respect the physics of many important phenomena, the mesh should be
so fine that even large computers would be unable to perform the computations. The technique being adopted is to use a coarse mesh for the
precipitator and a finer mesh for each cell.
9B.2.2

Model by Canadas et al. [5 j

Figure 9B.4 represents the flow chart of a recently proposed computer


model [5]. Referring to the original paper for details the fundamental
variables considered as substantially independent are: the applied voltage,

THE MODERN APPROACH TO COMPUTER MODELLING

287

Input ESP inlet conditions


(mass loading. particle size distribution
ash resistivity. gas flow)

Input section conditions


(geometry. applied voltage waveform)

Parameters at inlet of length increment


(relative radius of active and ionising voltage)

Calculate relative particle charges


Calculate current intensity

Calculate migration velocity of particles

Calculate dust layer growth


Calculate voltage drop across dust layer

Correct relative particle


charges and voltage drop

Calculate outlet dust concentration


and particle size distribution
Repeat for each increment
Repeat for each time interval
Calculate overall performance

Figure 9B.4 Flow chart of the model proposed by Caiiadas et al. [5].

the ionising voltage, the radius of the active zone, the current intensity
arriving at the plates, the voltage induced by the particle space charge, the
relative particle charge of each particle size, the migration velocity of each
particle size, the dust layer thickness, the particle layer resistivity and the
voltage drop across the deposited layer.

288

MODELS OF ELECTROSTATIC PRECIPITATORS

The basic equations considered by the model are as follows.

1. the electric field is computed as a function of the applied voltage, the

2.

3.

4.

5.

6.

distance between the collecting plates and discharge electrodes, the


electrode system confuguration factor and the shape factor of the electrostatic field;
the relative particle charge is computed as a function of the ion mobility,
the distance between the collecting plates and discharge electrodes, the
voltage waveform, the electrode system configuration factor, the charging
coefficient, the ionising voltage, the effective velocity of the gas molecules
and the particle diameter;
the current/voltage characteristic is computed as a function of the ion
mobility, distance between the collecting plates and discharge electrodes,
current attenuation factor, net available voltage, voltage induced by
particle space charge, collecting capacity of plates, electrode system
configuration factor;
the migration velocity is computed as a function of the permittivity of the
gas, the particle charging coefficient, the peak voltage value, the maximum value of the field shape factor, the mean voltage value, the shape
factor of the electrostatic field in the deposit layer, the distance between
plates and discharge electrodes, the electrode system configuration factor,
the relative particle charge and the particle diameter;
the dust layer thickness is computed as a function of the applied voltage,
the particle layer resistivity, the current density in the deposit layer, the
specific gravity of the dust layer, the distance between the collecting
plates and discharge electrodes, the collecting capacity of the plates, the
electrode system configuration factor, the gas viscosity, the particle
concentration, the shape factor of the electrostatic field in the deposit
layer, the maximum particle charge attainable by ion bombardement, the
relative particle charge and the particle diameter;
the voltage induced by particle space charge is computed as a function of
the density of particle space charge, the distance between the collecting
plates and discharge electrodes, the permittivity of the gas.

H is evident that a number of empirical coefficients are used, such as the


shape factor of the electrostatic field, the electrode system configuration
factor and the current attenuation coefficient. This means that the model is
suitable for predicting, for a given precipitator, the variation of the collection efficiency from the actual value due to some variation of the operating
conditions or of the structural dimensions.
9B.2.3

Modelling at Padova university [6J

An even more detailed model which claims to b.e able to predict the
collection efficiency by simply referring to the design characteristics of the

THE MODERN APPROACH TO COMPUTER MODELLING

289

------------------------------------------------------------------------------,,,

: Sect. 2
r--------------------------------------

:L ____________
Sect. 3
_

,,

: Sect. 4

------------------------------------.-----------------------------------------Figure 98.5 Model developed by Prof. Gallimberti, University of Pad ova [6].

ESP and to planned operating characteristics is being developed at the


University of Padova under a research contract from ENEL [6].
The flow chart represented in Figure 9B.5 shows that the model is
organised into the following sections and subsections:

1. gas and particle motion: it evaluates the actual fluid-dynamic conditions


of the gas motion and the velocity field which retards the dust particles;
2. electric field and discharge processes, subdivided into the modules:
2.1 electric field which solves the Poisson equation to determine the
electric field;

290

MODELS OF ELECTROSTATIC PRECIPITATORS

2.2 back-corona which takes into consideration the emission of positive


charges from the collecting plates;
2.3 glow-corona which simulates the continuous flow of positive and
negative charges in DC operation;
2.4 streamer corona which simulates the production of charge due to
impulse streamer discharge;
2.5 breakdown which simulates the critical conditions resulting in electrical breakdown.
3. particle charging and migration, subdivided into the modules:
3.1 particle charging which models the progressive charging of the solid
particles;
3.2 particle migration and current field which takes into consideration
the forces applied to the particles, the charge exchanges and the
turbulent diffusion;
3.3 space charge distribution evaluates the charge distribution needed for
the updating of the electric field;
4. particle collection, subdivided into the modules:
4.1 particle collection simulates the laminar flow close to the collecting
plates;
4.2 rapping and re-entrainement simulates the sudden injection of particles from the dust layer due to rapping or the continuous one due
to scouring;
4.3 process efficiency, determines the overall collection efficiency of the
precipitator.
This model, recently completed, is presently under test. Up to now it has
been able to predict with high accuracy the voltage/current characteristics,
when the precipitator is loaded with different monodisperse particle concentrations and differently energised: conventional AC rectified voltage or DC
voltage with short voltage pulses superimposed. It has also successfully
verified the prediction of the back-corona onset as a function of the applied
voltage and the dust layer resistance and the prediction of the breakdown
voltage at a given voltage and dust layer thickness.
Any model cannot avoid a number of numerical constants which combine the effects of several minor physical process; the main feature of this
latter model is that it requires very few 'adjusting parameters', namely the
following:
the roughness of the emitting wire, which also takes into account the
possible dust layer formation of the wire;
when the corona discharge produces streamers (when voltage pulses are
superimposed) the number of streamers per unit length of emitting wire;
the turbulent diffusion which depends on the gas composition and
temperature;
the re-entrainement due to the erosion of the dust layer;

REFERENCES

291

the dust re-entrainement due to the rapping; this parameter has to be


computed in advance following a model of the dust layer.
The input quantities required by the model may be grouped into the
following items:
the data characterising the geometry;
the data characterising the gas (composition and velocity distribution at
the inlet cross-section) and the particulate (composition and size distribution at the inlet cross-section);
the data characterising the applied voltage.
References
1. Mayer-Schwinning, G. (1987) Opening Lecture at the Third International Conference on
Electrostatic Precipitation, Abano, University of Padova, Italy.
2. Oglesby, S. and Nichols, G. (1978) Electrostatic Precipitation, Marcel Dekker, New York.
3. Matts, S. and Ohnfeldt, P.O. (1963) Efficient gas cleaning with SF electrostatic precipitators,

A.B. Svenska Fliiktfabriken Report, 6(7), 105-22.


4. Cooperman, P. (1969) A new theory of precipitator efficiency. Paper no. 69-4, Fourth
APCA Meeting, APCA, Pittsburgh, USA.
5. Caiiadas, L., Navarrete, B., Salvador, L. and Rodriguez-Aragones A. (1993) PRELEC: a
mathematical model of electrostatic precipitation, 10th EPRI Particulate Control Symposium
and the Fifth International Conference on Electrostatic Precipitation, Washington, USA, pp.
21.1-15, EPRI TR 2, 103048, Palo Alto, CA, USA.
6. Bellagamba, B., Lami, E., Mattachini, F., Gallimberti, I., Turri, R., Gazzani, A., Tromboni,
U., Pasinetti, A. and Sala, R. (1993) A mathematical model for simulation of large scale
electrostatic precipitators" 10th EP RI Particulate Control Symposium and the Fifth I nternationa I Conference on Electrostatic Precipitation, Washington, USA, pp. 25.1-14, EPR! TR 2,
103048, Palo Alto, CA, USA.

10

Sampling and analysis for particles and heavy


metals in gas streams
G.B. NICHOLS and E.B. DISMUKES

10.1

Sampling and analysis

The determination of the performance of air pollution control devices, as


well as the emission rates to the atmosphere, requires the sampling and
analysis of the materials carried by the gas stream exiting from the process
plant. There are a number of specific standards that govern the measurement methods established by the appropriate government agencies responsible for the control of emissions, as well as standards developed and
specified to be followed to determine the performance level of control
devices, either for establishing contractual guarantees or for information
required for other reasons. The specific standards that apply depend upon
the particular country where the information is to be acquired. For example,
in the United States, the sampling and analysis methods are described in the
Code of the Federal Regulations (40 CFR), while in Europe the DIN
standards generally apply. It is not the intent of this chapter to delineate all
the specific codes that apply, but rather to provide a discussion that applies
to sampling and analysis in general that provides insight into the methodology. Regardless of the purpose for the measurement and the specific
standard that applies, it is necessary to obtain a representative sample of the
materials contained in the gas stream. The standards that apply in various
locales are given in references 1-4.
There are a number of different measurements needed to completely
characterize the emissions from a particular source. The first is a simple
measure of the mass of the condensed liquid or solid particles emitted. These
consist of particles and droplets suspended in and carried by the gas stream.
The stream velocities in ductwork are usually sufficiently high that particles,
other than the large ones, do not settle out significantly, but there is always
a tendency to develop some stratification in the ductwork. This stratification
is related to the differential gravitational settling velocities of the large
heavier particles, compared with the lower settling velocities of the smaller,
less massive ones. Bends, obstructions, internal supports, leaks and fans also
contribute to the formation of non-uniform distributions of particles in the
cross-section of the ductwork. Thus, it becomes necessary to establish a
protocol that samples the entire cross-section of ductwork in a statistically
representative manner.

SAMPLING AND ANALYSIS

293

Items of interest in a gas stream emanating from any process may include
solids, liquids and vapors. A total characterization of the emission requires
that these items be sampled representatively. The appropriate method to use
for sampling either particles or droplets requires sampling the entire duct
cross-section with enough sampling points to be statistically acceptable and
to extract these samples isokinetically. This means that the gas velocity
inside the sampling nozzle be identical with that just prior to entering the
nozzle. Individual sampling standards will identify the minimum number of
sampling points to adequately sample a given gas stream. In general, the
greater the distance from turns or other obstructions, the smaller the
number of sampling points; there is however, a minimum acceptable
distance, upstream and down, from a flow upsetting obstacle for sampling.
This should be addressed in the applicable sampling standard.
The necessity for isokinetic sampling is associated with the behavior of
the constituents in the gas stream. If the item to be sampled is a gas, logic
would suggest that isokinetic sampling is not required, except for the case
where in-leakage or other factors may lead to a non-uniform distribution of
the gases. The mass and inertia force of the gas molecules of interest are the
same as the other gas molecules in the stream and will be sampled at the
same rate as the carrier gas. However, both solid and liquid materials
suspended in the gas stream will have significantly greater gravitational and
inertia forces acting on them than the gas molecules. Figure 10.1 illustrates
the three conditions of sampling discussed. Figure 10.la represents
isokinetic sampling, i.e. the velocity in the sampling nozzle matches that in
the ductwork just prior to sampling. Next, consider the case where the gas
stream in the sampling nozzle has a velocity greater than the velocity in the
duct as illustrated in Figure 10.1 b. At the inlet to the nozzle, the gas flow
lines will converge into the nozzle. The inertia force acting on the particles,
which increases with size and mass, will cause the larger particles to flow
through the gas in the converging flow into the nozzle and some portion of
the larger particles will be missed, leading to an underrepresentation of the
large particles. This will lead to a collected mass sample less than that in the
main gas stream and a bias in the particle size distribution of the sample
collected.
Conversely, when the velocity in the sampling nozzle is less than that in
the main gas stream, the sample collected will also be unrepresentative of
that in the main flow stream. Figure lO.lc illustrates this undersampled case.
In this example, the gas stream from the main duct will be diverted around
the nozzle. The inertia force on the particles will now cause the larger
particles to flow through the diverging gas stream and into the nozzle. This
sampling error leads to oversampling the particles, with a sample collected
that has more mass per unit volume than the main gas stream. Typically,
isokinetic sampling within 5-10% of the local velocity is adequate to collect
an acceptable sample.

294

GAS STREAM SAMPLING/ANALYSIS: PARTICLES, HEAVY METALS

(a) Isokinetie sampling

---*
-----:... "
-.--- ........."

...

...---

~--t~(b) Oversampling

===

~=======.
(e) Under sampling

Figure 10.1 Illustrations of isokinetic, over and undersampling.

The test methods used throughout the world for determining the mass
loadings and/or emissions are similar in that the sampling trains used
consist of some type of probe equipped with a streamlined nozzle, a filter, a
gas flow measuring device and a pump or other means for pulling the gas
stream through the sampling system. The probes typically are equipped with
thermocouples and pitot tubes to measure the gas temperature and velocity.
The sampling system is referred to as in-situ if the filter is located in the gas
stream proper and extractive if the filter is external to the flow stream.
Bubblers and liquid filter traps may also be added for specialized testing
needs. An example of an extractive sampling system is given in Figure 10.2
[5]. If the filter holder was mounted just behind the sampling nozzle the
example would represent an in-situ probe.
Special care must be exercised if the gas stream contains condensable
materials and the sampling method selected is extractive. Some methods
specify the temperature of operation for the external filter depending upon
the purpose for the measurement. If the purpose is to evaluate the performance of a particle control device, the filter should operate at or slightly above
the process gas stream temperature. If the purpose is associated with the

295

SAMPLING AND ANALYSIS

NOTE
Impinger train optional:
May be placed by an
equivalent condenser
Heated
area

Filter holder

Check

Probe

I '~II~
Impingers

Ice bath

Thermometers

Vacuum

I(jI gauge

Vacuum
line

Manometer

Dry test meter

Air-tight pump

Figure 10.2 EPA Method 5 particulate matter sampling train.

determination of compliance with regulations, the applicable standard must


be consulted. If the sampling system is operated at a temperature below the
process stream temperature, the condensables may be collected on the filter
medium and be reported as particulates. In both extractive and in-situ cases,
the particulate material that adheres to the inside surfaces of the probe
nozzle and probe liner upstream from the filter must be retrieved and added
to the filter catch for a representative sample to be collected.
There are several other types of mass measurement systems that are
available for specialized purposes. Some of these include Beta beam absorption measurements through a previously collected sample, light scattering
transmissometers and absorbers, and multiple light frequency instruments.
These instruments are typically used for monitoring performance rather
than for detailed performance or mass measurements. As such they are only
mentioned for completeness. Their operation will not be described. References 6-10 identify some instruments of these types.
A second characteristic of the particles or droplets suspended in the gas
stream is the particle size distribution of the particles as they exist in the gas.

296

GAS STREAM SAMPLING/ANALYSIS: PARTICLES, HEAVY METALS

Inlet
Jet stage
(7 total)

Collection
plate (7 total)

O-ring

Filter holder

Jet stage

Figure 10.3 University of Washington Mark III inertial impactor.

Both laboratory and in-situ methods are used to determine the particle size
distribution, but the laboratory methods will not be discussed in this
chapter.
Inertial impactors and multiple cyclones are useful for determining the
particle size distributions of particles suspended in a gas stream. Both
systems operate under the principle of inertial separation. As mentioned
above, isokinetic sampling is required to obtain a representative sample. The
flow path of the particle-laden gas stream is caused to make abrupt turns
such that the inertial forces acting on the particles cause them to move
across the gas flow stream lines to deposit the particles on an impaction
surface. Inertial impactors have a number of stages designed with increasing
stream velocities in each section. These streams flow onto a flat plate where
the gas stream is forced to make a right angle turn with the appropriate size
of particle being retained on the impaction surface and the smaller particles
carried on by the stream. In the next stage, where the velocity is higher,
smaller particles will now impinge on the impaction plate. Finally, a
back-up filter serves to collect the remaining particles to complete the
process. Figures 10.3 and lOA illustrate an inertial impactor and a multiple
cyclone sampling system, respectively [5].

SAMPLING AND ANALYSIS

297

Cyclone 1

Outlet
Figure 10.4 EPA/Southern Research Institute five-stage series cyclone.

The cyclone system operates in a similar manner. Each stage of the


cyclone system is designed to operate at increasing velocities to collect
reducing smaller particle fractions. The larger particles are collected in the
inlet cyclone and the smaller ones downstream. Back-up filters are also used
with cyclones when appropriate. Cyclone systems are typically used in high
mass loading situations or where large particle samples are desired for
analytical or other purposes.
There are conflicting requirements for conducting particle size distribution measurements in a duct system with a non-uniform gas velocity
distribution. First, there is the requirement for isokinetic sampling to collect
a representative sample of the particles. This leads to the requirement for
adjusting the sampling velocity in the sampling probe to match the local gas
velocity at the sampling point. There is also the requirement to be able to
determine the particle size distribution of the collected sample. Since the
cutpoint for each stage on either an inertial impactor or series cyclone is
related to the gas velocity through the stage, it is necessary to maintain a
constant velocity through the system while sampling. This precludes adjust-

298

GAS STREAM SAMPLING/ANALYSIS: PARTICLES, HEAVY METALS

ing the flow at each point to maintain isokinetic velocities at each sampling
point. In practice, the inertial impactor or cyclone is operated at the average
gas velocity for the cross-section of ductwork being measured, thereby
keeping the cutpoints for each stage constant for the measurement.
Currently, there is an interest in determining the respirable fraction of the
particles suspended in a gas. In some instances, particles smaller than 10 fim
diameter (PM-lO) are considered to be appropriate, while in others, those
smaller than 2.5 fim (PM 2.5) are selected. The actual decision about which
is appropriate to regulate is still pending. The measurement may be required
to be made either in the ambient air or in a confined gas stream. Either
inertial impactors or cyclones could be appropriate for this determination.
In addition to the need for sampling for mass emissions and particle size
distribution, there is a current interest in sampling for specific substances
contained in the exiting gas stream. These items are referred to as air toxics,
volatile organic substances and heavy metals. The heavy metals subset of
the items of interest require sampling and analysis techniques significantly
different from other aspects of sampling. These techniques will be discussed
in the following section.
10.2

10.2.1

Heavy metals

General considerations

At temperatures pertinent to particulate control, heavy metals occur in flue


gas both as particulate matter and in the vapor state. There is, of course, a
significant range of temperatures where particulate-control devices operate.
The range is from temperatures as high as 371C (700 OF), for hot-side
ESPs, to temperatures as low as 52 DC (125 OF), which are reached in wet
units. This range of temperatures permits some metals to exist primarily as
vapors and others primarily as particulate matter, depending upon circumstances.
Even at the highest temperatures of concern, certain metals are insufficiently volatile to occur other than as solids. Barium is certainly one such
example; even at the lowest temperatures of concern, on the other hand, one
specific metal- mercury - is volatile enough in the elemental state to exist
wholly in the vapor state. The factors concerning mercury that determine
the primary state of actual occurrence include both thermodynamics and
kinetics. The prevailing evidence is that, regardless of the interplay of these
two factors, mercury occurs in the flue gas mainly in the vapor state. The
volatilities of three of the more volatile metals, their oxides and the
dichloride of one (mercury) are depicted in Figure 10.5. The sources of these
data are References 11-15.
The analysis of heavy metals in flue gas must be based upon some of the
same fundamental sampling strategies that are critical for particulate matter.

299

HEAVY METALS
+2

+2

-2

-2

S-4

.9-4

N
0 -6

"0 -6

>

>

C>

C>

.Q

.Q

-8

-8

-10

-10

2.0

2.4

2.8

3.2

3.6

2.0

1000/T, K-1

2.4

2.8

3.2

3.6

1000/T, K- 1

+2r---~------------------,

o
Hg

-2

-8
HgO

-10

200 150 100


I

-12

2.0

2.4

50C
I

2.8
3.2
1000 IT, K-1

3.6

Figure 10.5 Volatilities of metals and compounds with relatively high vapor pressures.
Reprinted from Fuel Processing Technology, 39, E.B. Dismukes, Trace element control in
electrostatic precipitators and fabric filters, 403-16, (1994) with kind permission of Elsevier
Science-NL, Sara Burgerhartstraat 25, 1055 KV Amsterdam, The Netherlands

In general, sampling for metals occurring either as particulates or vapors


must be done by traversing the entire duct in question, to ensure that
non-uniform concentration distributions are taken into account. There may
be the temptation with vapors to compromise this principle, based on the
argument that vapor concentrations are far less likely to be stratified than

300

GAS STREAM SAMPLING/ANALYSIS: PARTICLES, HEAVY METALS

particulate concentrations. Still, the sampling strategy must allow for the
possibility that temperature gradients will change vapors to particulates, or
vice versa, and must allow for the possibility that air in-leakage will create
concentration gradients even in vapors.
Testing analytical data for material balance should be regarded as an
important, if not essential, aspect of analyzing heavy metals in flue gas. The
data currently obtained for heavy metals are often in doubt as to accuracy,
and the testing of data for material balance is a useful technique for
determining which metals have been determined successfully and which
have not. The material balance exercise may embrace an entire plant,
beginning with the fuel and extending to all waste streams. On other
occasions, just incoming and outgoing streams at a control device may
provide an adequate basis for material balance considerations [16].
10.2.2

Sampling methods for multiple types of heavy metals

The state of the art in the United States for multi-element sampling of heavy
metals is incorporated in the so-called Method 29 of the US Environmental
Protection Ageny. This method has not yet achieved status as an officially
sanctioned method; that is, it has not yet been formally included in the Code
of Federal Regulations [1]. It should ultimately appear in Title 40, Part 60,
but it will not appear earlier than the Fall of 1996. Nevertheless, the method
has long been unofficially known as Method 29, and has been used for
several years for sampling programs under official US Governmental
auspices. It remains in use today as the primary method for sampling flue
gas for multiple metals.
Method 29 is fundamentally just a modification of Method 5 for
measurement of total particulate concentrations (Method 5 has appeared in
40CFR 60 for more than two decades). Method 5 extracts flue gas
isokinetically from a gas duct through a heated probe into externally
mounted devices that, first, collect particulate matter on a heated filter and,
then, collect any condensable vapors in ice-chilled impingers.
40 CFR 60 actually describes the basic Method 5 and a total of eight
variations designated as Methods 5A through 5H. The basic method
stipulates that the filter be maintained at 121C (250 OF); two of the
variations permit higher temperatures in an effort to mininize the presence
of sulfate salts or sulfuric acid as a component of the filter catch. Method
29 retains the use of the 121C (250 OF) temperature; thus, in principle, it
distinguishes operationally between particulate and vaporous metals on the
basis of an arbitrary filtration temperature.
Water is the only collection medium employed in Method 5 and its
variations. Other media containing acids and oxidizing agents, on the other
hand, are used in Method 29. There are essentially two different sampling
media in the impingers of Method 29. The first impinger traversed by the

301

HEA VY METALS

Glass filter holder


in oven

~uartZ-Ii~ed
I/

Quartz
nozzle
Thermocouple

probe

Temperature
Check valve
/ / Flexible PTFE umbilical Temperature

~~~~~~~~f

or glass connection

"'

Type-S
pitot tube
Duct
wall

/
Pitot
manometer

Temperature
By-pass
valve
Orifice
'

Orifice
manometer

Vacuum
gauge

....9

Dry gas Pump


meter

Figure 10.6 EPA Method 29 multiple metals sampling train.

sample gas stream contains a mixture of hydrogen peroxide and nitric acid;
the second contains a mixture of potassium permanganate and sulfuric acid.
The sample collected in the peroxide-nitric acid mixture is analyzed for all
metals of concern; the permanganate-sulfuric acid mixture is analyzed only
for mercury. See Figure 10.6 for a sketch of the Method 29 sampling train
[17].
One of the alternative methods for sampling of heavy metals is the
so-called REST method of John Cooper [18]. In comparison with Method
29, the REST method is much simpler to use and provides samples that are
far more easily analyzed. In essence, it employs a Teflon or quartz particulate filter followed by charcoal impregnated filters to retain any metals that
occur in the vapor state rather than in particulate matter. The method
produces analytical results with non-destructive instrumental analysis of the
filters; it avoids the complex sample digestion and dissolution scheme
required for samples from Method 29.

302

10.2.3

GAS STREAM SAMPLING/ANALYSIS: PARTICLES, HEAVY METALS

Sampling methods for mercury alone

Method 29 can, by choice, be used for sampling mercury when this is the
only metal of interest. Some time after Method 29 had been in use for total
mercury, there came the belief that the two liquid media in the impingers
can provide data on different vaporous species of mercury. The peroxide
impingers seem to collect principally the vapors of oxidized mercury
(believed to be mainly HgCI 2 ), whereas the permanganate will definitely
collect the elemental form of the vapor that remains in the gas stream after
the oxidized vapor has been removed. In summary, Method 29 offers the
prospect of determining particulate mercury (oxidized mercury in chemical
forms such as the oxide), vaporous mercury in the oxidized state (likely
HgCI 2 ), and vaporous mercury that is in the free elemental state.
The so-called Bloom method [19J was the first widely used method for
sampling mercury for the specific purpose of speciating the vapors. This
method depends upon the use of solid rather than liquid collecting media.
A cartridge containing soda lime is believed to be selective for collecting
oxidized vapor; a back-up cartridge packed with iodated carbon collects the
elemental vapor that remains (as well as any oxidized vapor that is not
collected by the soda lime).
A more recently developed method for mercury, described by Keith
Curtis of Ontario Hydro [20J, may be regarded either as a modification of
Method 29 or a modification of Method lOlA [1]. As a modification of
Method 29, the change is the use of an aqueous solution of KCl in place of
the peroxide~nitric acid solution. The rationale for aqueous KCl is that
HgCl 2 is soluble in water and stabilized in that medium as the HgCl~ 2
complex ion with excess chloride ions. As a modification of Method lOlA,
the change is the introduction of the KCl solution as a sampling medium
ahead of the permanganate. In either event, this method provides samples
of particulate mercury on a filter, oxidized mercury (HgCI 2 ) in the KCl
solution, and elemental mercury in the permanganate solution.

10.2.4

Metal analysis in the laboratory

Method 29 has been used to determine some 15 to 20 trace metals.


Antimony, arsenic, barium, beryllium, boron, cadmium, chromium, cobalt,
copper, lead, manganese, mercury, molybdenum, nickel, selenium, and
vanadium comprise the list of analyses of interest in recent investigations of
utility plants. Method 29 recommends that most of the metals of interest be
determined by inductively coupled plasma atomic emission spectroscopy
(ICAPES), which provides the advantages of simultaneous measurement of
numerous target metals and, as a rule, acceptable levels of sensitivity. In
certain instances, however, because of limited sensitivity on the part of

REFERENCES

303

ICAPES, other methods are selected instead. For mercury, the choice is
cold-vapor atomic fluorescence spectroscopy. For arsenic and selenium,
atomic absorption spectroscopy with a graphite furnace is preferred.
10.2.5

Prospects for real-time monitoring

Costs of metals analysis in terms of money and time make it highly desirable
to produce instruments capable of real-time, continuous emission monitoring. The particular concerns about mercury have led to emphasis on
continuous emission monitors for this metal. ADA Technologies, Inc., is an
American firm that claims to have CEM for mercury [21] that is soon to
be commercially available. The analytical principle for mercury calls for
conversion on all compounds to the elemental vapor and measurement of
the vapor by ultraviolet absorption.

References
1. Title 40 Code of Federal Regulations Part 60, Reference Methods US Government Printing
Office, Washington, revised annually and dated July 1 each year.
2. BSI Standard ISO 9096.
3. VDI Standard 2066.
4. Jl5 Z 8803-1970, Japan.
5. McDonald, J.R. and Dean, A. (1980) A manual for the use of electrostatic precipitators to
collect fly ash particles. EPA publication EPA-8-600/8-80-025, May.
6. Opacity Meter, Monitor Labs Inc. 74 Inverness Dr. East, Englewood, CO 80112, USA.
7. Forward Scattering, Insitec, 2110 Omega Rd. Suite D, San Ramon, CA 94583, USA.
8. Back Scatter, Environmental Systems Corp., 200 Tech Center Dr. Knoxville, TN 37912,
USA.
9. Beta Gauge, Graseby-Anderson, 4801 Fulton Ind. Blvd., Atlanta, GA 30336, USA.
10. Triboelectric, Auburn International, Inc., 8 Electronics Ave., P.O. Box 2008, Danvers, MA
01923, USA.
11. Smith, J.D. (1973) Arsenic, antimony and bismuth, In: J.C. Bailar, Jr et al. (Eds.)
Comprehensive Inorganic Chemistry, Vol. 2, Pergamon Press, Oxford, pp. 547-683.
12. Behrens, R.G. and Rosenbkat, G.M. (1972) Vapor pressure and thermodynamics of
octahedral arsenic trioxide (arsenolite). J. Chem. 1hermodyn., 4: 175.
13. Neumann, K. and Lichtenberger, E. (1939) Molecular-weight determination and vapor
pressure of selenium. Z. Phys. Chem., A184: 89.
14. Pupp, C. et al. (1974) Equilibrium vapor concentrations of some polycyclic aromatic
hydrocarbons, As 4 0 6 and Se0 2 , and the collection efficiencies of these air pollutants.
Atmos. Environ., 8: 915.
15. Chase, M.W., Jr et al. (1985) JANAF Thermochemical Tables, 3rd edition J. Phys. Chern.
Ref. Data 14: Supplement No. 1.
16. Vann Bush, P. et a!. (1995) Sampling and analytical challenges for air toxics assessments.
EPRI/DOE International Conference on Managing Hazardous and Particulate Air Pollutants, Toronto, 15-17 August, in print.
17. Methods Manual for Compliance with the BIF Regulations, EPA/530-SW-91-0JO, December 1990.
18. Cooper, J.A. (1994) Recent advances in sampling and analysis of coal-fired power plant
emissions for air toxic compounds. Fuel Processing Techno!., 39, 251.

304

GAS STREAM SAMPLING/ANALYSIS: PARTICLES, HEAVY METALS

19. Bloom, N. (1991) Mercury speciation in flue gases: overcoming the analytical difficulties.
Conference on Managing Hazardous Air Pollutants-State of the Art, Washington, DC, 4-6
November, EPRI, Palo Alto, CA, USA, pp. 148-60.
20. Curtis, K. (1994) Ontario Hydro Technologies, private communication, October.
21. Schlager, R.J. et al. (1995) Continuous monitors for measuring emissions of mercury and
particulate matter. EPRI/DOE International Conference on Managing Hazardous and
Particulate Air Pollutants, Toronto, 15-17 August, in print.

11

The commissioning of electrostatic precipitators


D.A. STYLER and J.e. WESTBURY

11.1

Introduction

As the marketplace has become more and more competitive over the last
several decades, so has the need to design an electrostatic precipitator that
will meet its design duty with a margin in which both supplier and purchaser
are confident.
As a prerequisite to this need, the designer must be able to size the
precipitator with confidence. To do this he needs to be happy in the
knowledge that his base data are formulated on results obtained from plant
that is in extremely good condition, both mechanically and electrically.
The starting point in ensuring that this state exists, and has a better
chance of remaining so, is the initial commissioning of the equipment. A
dictionary defines the word 'commissioning' as getting something 'ready for
active service, to assign to perform a task or function'.
A plant that has been commissioned properly and is seen to be performing well in its early life induces a sense of well-being in both the vendor and
his client. In the case of the former, it strengthens and enhances the validity
of his technical database. For the latter, it promotes the idea that the
precipitator is well worth maintaining properly by using and keeping to a
managed maintenance programme.
Taken to an extreme, commissioning can be considered as a function
which is not only necessary but should commence from the point at which
the first piece of steel or concrete is laid down on the site. If this part has
not been manufactured or installed correctly, then the remainder of the
construction and commissioning exercises can only be a compromise, to a
lesser or major degree.
Even though there are a substantial number of manufacturers producing
their own type of electrofilter, all have the same basic elements contained
within them. These fundamentals would be discharge electrodes, collecting
electrodes, electrode rapping equipment, high voltage insulators, gas distribution equipment and a casing to house them all in. It is because these
principal parts are common to all manufacturers that the fundamental
procedure for commissioning will remain more or less the same for each.
When considering the procedure undertaken from initial commissioning
to hand-over as a fully operational plant to the client, the process can be
split into three major parts. Each of these parts, at various stages, will be
running either in isolation to, or in parallel with, each other.

306

THE COMMISSIONING OF ELECTROSTATIC PRECIPITATORS

These three parts can be identified as follows:


(a) mechanical commissioning
(b) electrical commissioning
(c) process commissioning.
The timing of these functions within the commissioning programme and
whether or not they overlap in any way may well vary, not only from
contract to contract but also between two ostensibly identical contracts, as
a result of prevailing site conditions or client-imposed criteria.

11.2 Mechanical commissioning


The action of mechanical commissioning can be resolved into four phases:
construction, post construction, cold and then hot commissioning. These
phases can also be split further into a number of component parts. As with
the commissioning programme, they too can be carried out as discrete
actions or be involved in combinations of actions from different components.
As has previously been mentioned, it is important to ensure that the
construction of the precipitator is monitored in as much that the criteria
that affect the performance of the unit are checked and approved as the
programme progresses. In order to follow the construction phase activities
through, Figures 11.1 and 11.2 show a sectionalised, large, steel cased
electrofilter and an exploded diagram of the main features of the casing,
respectively. They both identify and name the various component parts for
easy reference.

11.2.1

Construction stage

11.2.1.1 Substructure and casing. Once all the civil engineering work
which is directly associated with the filter structure has been passed as
acceptable, the first stage of this construction procedure can begin. Table
11.1 indicates the typical tolerances which must be achieved for this civil
engineering stage, and a guide to the tools and instruments to obtain them.
The first stage of construction will be the placing of the understructure onto
the foundations. As the support columns are positioned, their verticality and
diagonal relationships must be verified. The diagonal measurements are
taken to ensure that the casing will not end up as a parallelogram. The
degree of accuracy required for verticality is that which can be obtained by
the use of a theodolite. Provided that the substructure fabrication was to
tight tolerances and any shimming, required to obtain the correct height, has
been inserted, the pads at the uppermost extremities of the columns will now

- - -- ----1

Sliding bearings

Casing side cladding _ _ _ _ _ __


Side access dOOfS
Main plaliOlm
Casing side lagQlng
Stiffened casing plate

Main access stairway

Interlocked top houSing dOOf

Wealherproof top houSIng


Mouthpiece wil.h access d
lOCk-<lfl earth SWitch - - - - Rapping motOi switch boxes
Rectilier set and inSlUlator cover
Top housing access platfOlm

If ,,.

"

'"

"'""'"""

Figure 11.1 Typical sectionalised view of large steel cased precipitator.

"

- Collector bottom gUides


Discharge electrode bottom frame
Dousl hoppers
Understructure

" - Gas flow

Collector rapping frame


Discharge top frame
with tumble hamme1 rapping
Collector rapping rods
Mouthpiece With access dOO<
ColiectOl support member
Catch space COllector
,'- DiSCharge electrode
tnlet gas distributors
tnlet !Jos distrobutoon sphUers
Internal caSing b raCing

Collector and
dlSCharge rapping droves
_
_ --- - - _
- la\jged top plate
--idtl!lltll~:;;:;;~~----.: Discharge suspension
,~
<::::
Access dOO<

308

THE COMMISSIONING OF ELECTROSTATIC PRECIPITATORS

, - - - - - Top housing assembly


Mesh screen
Portal frame

Purlins-------o~......::--

Doorway _ _

.....

Sheeting --,..,.,~
rails
Outlet--,......
mouthpiece

Collector rapping base channel


Lead through insulator base
Top plate assembly
Insulator support
Top plate access door
Top plate support
steelwork
Side column
-A':l1T"<:,--- Top centre girder
Side plate
lA"Ir9It--Top end girder
~m""l-tt--Af't"""""--tl--Collector support angle

Bottom side---rt.Irc-....
girder
Bottom-----=----:;~~

centre girder
Hopper-----'

Inlet flare
Stub column
Bottom end girder
Access door
' - - - - - - Hopper beam
Figure 11.2 Exploded view of typical precipitator casing.

permit the casing proper to be built otT horizontal datum points. Referring to Table 11.1 will indicate the requirements and how best to achieve
them.
Placed on top of the understructure will be the sliding bearings which will
accommodate any movement of the casing relative to it. One support
column, which is located as close to the centre of the footprint of the filter
as possible, is used as a fixed point. The casing is bolted to it at this point
and any movement caused by thermal expansion will be away from it. By
allowing the casing to slide over the understructure by equal amounts in all
directions away from it, potential structural failure is avoided. The diagrams
shown in Figure 11.3 highlight the above points.

309

MECHANICAL COMMISSIONING

Table 11.1 Typical tolerances necessary in civil engineering construction and mechanical stages
of building ESP and instruments to obtain them
Item

Measuring device

Foundations

Tape measure, Theodolite,


Spirit level
Tape measure, Theodolite,
Spirit level
Tape measue, Spirit level,
Water level
Tape measure, Spirit level,
Water level
Theodolite, Plumb line
Tape measure, Spirit level,
Water level
Tape measure, Spirit level,
Water level
Tape measure, Spirit level,
Water level
Tape measure, Plumb line
Tape measure
Tape measure, Plumb line
Tape measure
Tape measure, Plumb line,
Spirit level
Tape measure, Plumb line,
Spirit level
String line, Spirit level,
Tape measure
String line, Spirit level,
Tape measure
String line, Spirit level,
Tape measure
Tape measure
Tape measure

Columns-Verticality
-Diagonals
Ring beam
Bottom girders
Casing walls
Top girders
Roof plate
Collector support channels
Collector-spacing
Collector squareness
Discharge electrodes
C/E to D/E spacing
D/E suspension supports
D/E suspension/carrying frame
Rapping drive chains
Rapping drive bolted coupling
Rapping drive flexible coupling
Hoppers
Flares

Tolerance
5mm
2mm
5mm
1 mm on overall length
1 mm on overall length
lmm
1 mm on overall length
1 mm on overall length
5mm
2mm
6mm
2mm
lOmm
3mm
5 mm on all dimensions
lmm
lmm
lmm
3mm
3mm

There are, however, some units which can be built using a 'monolithic'
design. This is a system whereby all expansion is taken in the substructure
columns. The use of sliding bearings is, therefore, not needed for this
application. It must be remembered, nevertheless, that the same setting-out
procedure for the substructure must be undertaken.
The ring beam, shown in Figure 11.3, which consists of girders positioned
longitudinally and laterally, creates a rigid framework onto which the dust
hoppers and superstructure of the precipitator are attached. Once again it
is essential to ensure that the diagonal measurements of these beams are
verified (see Table 11.1.) If this operation is not carried out, then difficulties
may be experienced during later stages of the commissioning in terms of bad
alignment of the internal components. In extreme cases this situation may
require major surgery to correct the problem at a time in the commissioning

310

THE COMMISSIONING OF ELECTROSTATIC PRECIPITATORS


Typical ring beam

~irl-_'--

Restraining guides parallel


to direction of expansion

~ _. ,--1--J---...J.....L.-, ;- _.
I
I

I
I

I
I

j-

--

'
I

"---":
..

______ 1

..

Sliding bearing surface

I ... ~----- .... -

.r-:=----------------------;:=.
:

:
:
:

,
,

~ ...- - -

I:
I:
I:
I:I:
:
,,

1:::___ - - - - - - - - - - - - - - - - - - - : : -

III
Casing expansion

Figure 11.3 Bottom ring beam and sliding bearing details.

MECHANICAL COMMISSIONING

311

Figure 11.4 Relationship between ring beams and hoppers.

programme when access is difficult and time is at a premium. Figure 11.4


illustrates the relationship between the ring beam and the hoppers.
While access is available to all points of the casing and hoppers on their
internal surfaces, the commissioning engineer must take the opportunity to
inspect and confirm the integrity of all weld runs which are regarded as seal
welds. These are the welds which stop the ingress of ambient air or the
egress of process gas once the system is put into service.
The same degree of care should be taken when erecting the side walls and
adding the top girders, for the same reasons as have been indicated in
previous paragraphs. Failure to implement checks at this stage can have
similar repercussions, particularly where designs of the internals involve the
use of gravity to achieve and maintain good electrical clearances between
the discharge and collecting electrodes. Table 11.1 indicates the degree of
care required.
11.2.1.2 Internal components. Many designs of electrofilter use the top
girders to suspend the collecting electrodes. Figure 11.5 depicts the way in
which they can be lowered into the casing and Figure 11.6 their location on
the top girder (see Table 11.1).
Small departures from the manufacturer's tolerances can be absorbed and
are those which, under any other condition, could cause a reduction in

312

THE COMMISSIONING OF ELECTROSTATIC PRECIPITATORS

Figure ll.S Lowering of finished collector plates into casing.

electrical clearances. This is because most types of 'plate' electrode are


located at the top level and have adjustable guides at the lower level.
However, the discharge electrode system often does not have this luxury. It
may require gravity to maintain its alignment and verticality between the
collector plates. The hackneyed industrial phrase 'One out. All out' may well
apply to this situation, should the control criteria not be undertaken.
It is important when positioning the high tension discharge electrode
support system, located at the roof plate level, to once again ensure that
alignment will easily be achieved. It is imperative that the commissioning
engineer has completed his inspection of the finished plant and granted
provisional acceptance of its construction before there is any move to weld
into position or make any permanent fixture of this emission electrode
support equipment. Table 11.1 indicates the means to achieve this. An
arrangement showing the post support insulators of a rigid mast type of
discharge electrode system can be seen in Figure 11.7.

MECHANICAL COMMISSIONING

313

Motor~Jollector rapping assembly

~"

Roof deck

Upper topgirder
Drop rod-

__6

====
Collector'--++plate

f.'==

Figure 11.6 Location of collector plate on top girder.

Where the type of precipitator requires it, the rapping equipment for the
gas distribution system and both discharge and collecting electrodes must
be checked over. Again, depending on the manufacturer, these systems can
be located entirely within the casing, with only the drives passing through
it, or be a combination of these two, whereby certain parts can be found
within and others attached to either the casing walls or roof plate. Figures
11.1, 11.6, 11.7, 11.8, 11.9 and 11.10 give varying views of typical rapping
devices.
It is essential that all rapping drive shafts have not only been connected
in their correct sequence, according to the construction drawings, but that
they are aligned across the full width of the precipitator. This is usually done
by string line (see Table 11.1). Where flexible couplings or some form of dog

314

THE COMMISSIONING OF ELECTROSTATIC PRECIPITATORS

Cam tip
rod --t~~i:i~
iPost
nsulasupport-~~~l:jB~cC~~f~~![j~
tor
Drop

...

~~~~I-

I III

Anvil

Figure 11.7 Discharge electrode assembly.

tooth drive are used to connect the individual sections of the shaft,
alignment is still required but is not such a problem. If, however, the system
relies on the relative position of the drive shaft to the rapping hammer, then
alignment becomes far more important. Specifically, this refers to the type
whereby a cam attached to a rotating shaft lifts and drops the rapper. The

315

MECHANICAL COMMISSIONING

Discharge electrode rapping gear


_~.

Drop rod and


s ealed tube
Temporary
"'/
suspension
rods

"

Suspens
spring

r-.

Ir+

_Q~

~
I

ionr-.~
I~

Gas flow

II

a; ,

:+t
I

Anvil

"'~

't.

<

.r

4>

JIl.II
r"'"

Discharge elect rod e


assemblies
Collector
assembly

Figure 11.8 Discharge and collector electrode rapping gear.

316

THE COMMISSIONING OF ELECTROSTATIC PRECIPITATORS

Figure 11.9 Collector rapping assembly.

relative position of cam to rapper is critical because the height of the lift
and, therefore, the potential energy imparted into the rapper is directly
linked to this geometric relationship.

11.2.1.3 Hoppers. By the very nature of their shape, hoppers and, most
notably, the pyramid type, can be the cause of blockages to the flow of dust

~-1

__________________

~L =~

Figure 11.10 Assembled precipitator internals showing clearance space for inspection and
maintenance.

MECHANICAL COMMISSIONING

317

whilst they are being dedusted. The most common reason is damp dust in
the bottom of these hoppers.
At their extremities, hoppers have a large surface area but a very low
contained volume. Any hot dust retained in the hopper bottoms would not
have the heat capacity to maintain its temperature, or indeed, the temperature of the hopper platework, at a level on or above the water dew-point
of the process gas. This fact would be of particular concern when the process
involved has a high moisture content, or it cycles through the dew-point
frequently. Commissioning engineers would recognise the problem of cycling as a common feature on initial start-up of new process plant!
To overcome the potential for hopper blockages, heating devices are
usually installed, covering their outer surfaces for several metres up from
their dust discharge point. Further, to maintain the highest thermal efficiency possible, the heaters, whether they are of the electrical or steam coil
type, and the remainder of the hopper surface area, will be lagged. It is,
therefore, a prerequisite in the commissioning programme to be able to
power up the heaters to verify that they are functioning prior to them being
lagged over. It is a disheartening sight to see new cladding being ripped off
to remedy a fault that could have been rectified before its installation!
11.2.1.4 Flares and gas distribution equipment. Both the inlet and outlet
flares, sometimes referred to as nozzles, contain the means of achieving good
gas distribution within the electrical fields and, as such, require initial,
detailed checks (see Table 11.1). This equipment may also have rapping gear
attached to it, as a precaution against dust build-up and the resulting
maldistribution of gas. The gas distribution system can take on several
designs. Probably the most common is the holed screen type, where several
screens, in series, are installed in the inlet flare. Another type involves the
use of splitters and gas turning vanes. Figure 11.11 illustrates several views
of these types of gas distribution equipment. Whichever form has been used
and with respect to the rapping gear, the same comments apply as for those
made for the collector and discharge electrode rapping, described earlier in
this chapter.
11.2.1.5 Lagging, cladding and heat sinks. Although not directly concerned with the initial start-up of the precipitators on processes where
lagging and cladding of the casing and hoppers is required, it should be
inspected for integrity and efficacy. Ideally this would be whilst access is still
available, before scaffolding is removed.
Where the material of construction of the casing is steel, it would not be
immediately apparent that a problem, which resulted from badly installed
lagging, existed. The engineer who is inspecting the lagging and cladding
should bear in mind that there will be a time lag of some months before any
evidence will show itself. Particular attention should, therefore, be paid to
areas where there are likely to be large heat sinks through the casing. The

318

THE COMMISSIONING OF ELECTROSTATIC PRECIPITATORS


Gas flow
Typical nozzle assemblies
Splitters

Gas flow ....

Figure 11.11 Precipitator inlet gas distribution devices.

predominant points are where items of equipment are positioned on the


outside of, and in direct contact with, the casing, or the extreme corners of
the electrofilter.
Whilst it is a common design feature to install some form of heat barrier
between any parts that would contact the casing and the casing itself, it is
not unheard of to find that this vital piece of the precipitator has been
missed out.

11.2.1.6 Plant access. The final checks that have to be made in this phase
of the cc)mmissioning are in respect of the access to the filter. These items
would be all the staircases, cat ladders, landings, the treads, grating or
plating used to walk on and the associated hand railing. Typical location of
the access can be seen in Figure 11.1. With increasing legislation occurring
on the topic of Health and Safety at Work it cannot be over-emphasized
that all the above-mentioned components have the integrity of all of their
fixings checked whilst site construction staff and alternative means of access,
if required, are still available.
11.2.1.7 Casing pressure test. There have been occasions when end-users
have requested a pressure test of the filter casing. In cases where the process
involved generates toxic gas which passes through the precipitator, it is not
an unreasonable one. However, where toxic gases are not involved, there is
not really any need to perform this test. Furthermore, it is most unlikely that
the manufacturer of the electrofilter would have either designed it to be
totally gas tight or even budgeted to cater for an eventuality such as this.
Indeed, for these types of application, it is common to expect a small
percentage change in volume from inlet flare to outlet flare, as a result of

MECHANICAL COMMISSIONING

319

insulator purging or slight egress or ingress of gas or ambient air! With this
in view, a test of this nature is usually considered as a gross waste of time,
money and effort for both the client and the manufacturer.
Should it be unavoidable, however, for whatever reason, it is best carried
out whilst the casing is empty but otherwise finished. It will entail the
complete sealing of all orifii, including flares, access doors, all points where
any piece of equipment passes through the casing, the dust extraction ports
at the hopper bottoms and any poke rod holes in them. As access may be
required to parts of the casing to which, under normal conditions, it would
not be possible to get to, the construction programme and possibly the
method, would have to be rearranged to cater for this test. An example of
this would be the collector plates. Many designs are capable of being
constructed outside of the casing and then lifted and lowered into their final
working position as a finished article. Where a pressure test has to be part
of the construction programme, the casing roof plate would have had to
have been put into place and totally sealed. The construction of the collector
plates, therefore, would have to take place within the casing, so that the gas
tight seal, created between the roof plate and the rest of the casing, is not
broken.
Having reached the completion of the 'construction' phase of the commissioning, the period which links the cold and hot commissioning operations starts. This is best described as the 'final inspection' stage.
11.2.2 Post construction stage
11.2.2.1 Final inspections. This is the point in the programme at which
all of the component parts of the precipitator are confirmed to be in their
correct positions and that the mechanical fixings and welding required to
maintain this state are completed to the specification laid down in the
drawings and in the contract documentation.
In an ideal world, no further inspections should be required as the
attention given to the filter should have been such that the whole plant had
been confirmed as complete during the staged inspections whilst being
constructed! Unfortunately, there is no such thing as an ideal world! The
reason for this sad state of affairs is simple. It is nothing more than homo
sapiens!
With the best will in the world, it is a practical impossibility for any
individual to inspect a piece of equipment of such complexity and not miss
something that mayor may not have some direct bearing on the ultimate
performance of the end product. However, on a more serious note, there is
another, almost uncontrollable, element which could affect its condition.
This is the fact that it is impossible to isolate the component parts from
either human or mechanical contact after their completion within the

320

THE COMMISSIONING OF ELECTROSTATIC PRECIPITATORS

construction schedule, if only because other sections still have to be attached


to them.
In the final inspection programme, it will be necessary to recheck many
of the actions already carried out in the construction phase. Although it is
a repeat action, at least the commissioning engineer now has far more
control over the plant as a whole. He will usually be able to specify where
other members of the team can have access to the finished unit, such that
there will not be any activities taking place that would be detrimental to the
overall integrity of the precipitator.
The areas of specific concern for this portion of the programme are the
discharge electrode and collecting electrode systems, their respective rapping
gear, where fitted, sprays and washdown equipment found on wet plant,
high voltage insulators and the gas distribution equipment, situated in both
the inlet and outlet flares and their rapping equipment, if fitted.

11.2.2.2 Discharge and collector electrodes. In the first instance, the


discharge electrodes must be checked to ensure that they have not been
subjected to any physical damage. Broken or slack emitting wires or strips
or distortion of the framework in which the emitters are installed should all
be attended to, be they of rigid mast, bedstead frame or weighted wire type.
Any broken wires or strips should be removed. Provided that there are not
too many pieces in a zone, several metres in the many thousands that are
installed in a precipitator will not cause a shortfall in the efficiency by not
permitting the full potential current input to be realised.
Frames or rigid masts found to be creating a reduction in the electrical
clearances between them and the collector plates, or any other part which
is at earth potential, must be realigned. Where the verticality of the
discharge electrode is achieved by the use of gravity, i.e. a wire and weight
arrangement, any misalignment can be more difficult to remedy. The
questions that would immediately come to mind in this situation would be:
(a) Are the collectors plumb, or worse?
(b) Are they and/or the casing out of plumb?
The first would be relatively easy to correct but the second highlights the
necessity to ensure that the casing has been constructed, fully inspected and
passed before any of the internals are added to it. This procedure has been
described previously in this chapter.
It is an important fact to remember that in the process of making any
corrections, it is not inconceivable that the alignment in areas already
checked and approved as satisfactory could be moved out of tolerance.
Even though the collector electrode system is, by the very nature of its
physical construction, more robust than the discharge electrode system, it,

MECHANICAL COMMISSIONING

321

too, can be vulnerable. The more commonly found problems can be caused
by such things as scaffolding equipment being moved or taken in or out of
the casing. Often lengths of up to 3 m are being swung around and if they
happen to connect with plates which are of a thinner section-in the order
of 1.2 to 1.5 mm-it is not difficult to imagine that, at best, distortion could
occur and, at worst, the plate could be punctured. The danger with this
latter point is that it could produce jagged edges. Should they not be found,
or are just left because it is thought that this is of no real consequence,
reduced power input could result or, even worse, premature electrical
breakdown, once the process is on-line, rendering the bus section inefficient.
Thin or thick plates can be susceptible to distortion at their mid-height,
particularly on the very tall collectors. It is not difficult to understand that,
at this point, a small induced load in a direction across the gas flow can
create a large deflection and possibly permanent set because of the high
leverage rates obtainable. Should any activity have been carried out at this
level, following their installation, the problem of reduced electrical clearances becomes a distinct possibility!
11.2.2.3 Discharge and collector electrode rapping gear. With respect to
the rapping equipment, the manual checks which are described below are
essential actions.
Before any form of rotational checks are undertaken, tests should be
made to ensure that all hammers are free to swing on their pivot pins or,
where the rappers are in rod form, should be able to drop freely under the
influence of gravity. If there is any evidence of frictional resistance to their
action, its cause must be sought and remedied. Failure to do this will result
in inefficient rapping of the electrodes, collector or discharge, and will allow
unacceptably large amounts of dust to remain on them which, in turn, will
impact on performance.
Upon completion of this stage, hand barring of the rapping gear can be
done by mechanically turning the rapping drive shafts as close to the motor
drive as possible. This usually entails disconnecting any drive chains or
direct couplings to the motor and then, by use of a large wrench, turning
the shaft slowly so that the action of all rappers can be observed through
several revolutions.
Other types of rapping, which involve the lifting and dropping of either
a complete discharge electrode frame or a complete collector sheet, or even
both, may require the availability of power before this check can be made.
Whilst it is possible to carry out this action manually, it is very arduous
because of the masses of steel involved.
Yet another type is the magnetically induced lift and drop rapping. This
operates in a manner similar to the mechanical drop rod but uses a solenoid
device and employs one per rapping point. For obvious reasons, rotational

322

THE COMMISSIONING OF ELECTROSTATIC PRECIPITATORS

checks will not be needed for this application. However, power will be
required before the full operation can be assessed.
11.2.2.4 Irrigation and flushing sprays. Where sprays are used as the
means to remove the dust from the internals, it is advisable to flush the
pipework out prior to fitting the spray nozzles. All new sections of pipe are
subject to the accumulation of weld spatter, bits of paper, matchsticks,
etcetera, somewhere in their length. As some of these are rather bulky,
compared with the spray nozzle orifice, it is vitally important to ensure that
the pipes are clean and debris free. Several methods are available which will
rid the system of these unwanted articles. One would be to remove the
blanked-off end of every pipe and then turn on the spray pumps. Another
would be to backflush the pipes fitted within the scope of the contract back
to the point where supply commenced, by means of a separately piped
source. Venting of the waste water using this method should always be at
the lowest point available in the pipe network and through an orifice as
large as possible. This because the water put in the pipes to clean them will
be at a relatively low pressure, compared with that generated by the spray
pumps and will not, therefore, have the same cleaning or moving power.
Once clear uncontaminated water is seen to be flowing from the pipework, the joints, end caps, etcetera, can be refitted and the spray nozzles
installed. Before finally boxing up the plant, the sprays should be inspected
with the pumps on, so that the spray operation can be verified against its
design characteristics, such as volumetric throughput, spray shape and
whether the pump will generate the pressure required at the spray level.
11.2.2.5 High voltage insulators. Some of the more commonly used
materials for insulator manufacture, such as silica or porcelain, are very
susceptible to breakage or cracking if given the lightest of taps in the wrong
place. It must also be remembered that porcelain has an ability to absorb
water. Therefore, if any of the insulators that are manufactured from
porcelain are found to have any form of defect, action must be taken to
either create an electrically stable seal or make a complete replacement
where the exposure areas are too large. The main point of concern is where
penetration of the salt glaze has occurred and the problem spots are open
to the atmosphere or process gas.
As many insulators could be exposed to human contact after their
installation, a close inspection of their integrity is essential. Many construction engineers will ensure that as soon as the insulators have been positioned, they will be given protection by creating some sort of box around
them. These boxes would not then be removed until the commencement of
the final inspection phase.

MECHANICAL COMMISSIONING

323

In some respects, the gas distribution equipment, which is located in the


inlet and, sometimes, the outlet flares, can be subjected to the same problems
as the collector plates during the construction programme. The examples
given previously with respect to the collector plates are equally valid here.
As good gas distribution has a vital part to play in ensuring that the
efficiency of the electrofilter is achieved, any damage that is noted must be
rectified.
11.2.3

Cold commissioning

At this junction, there is a need for both the mechanical and electrical
commissioning programmes to come together. From a mechanical commissioning point of view, the immediate requirement is that the rotation of the
moving parts, such as the rapping gear and insulator purge air fans, can be
checked. From now on, the two programmes will be running in parallel and
interfacing with each other on an ever increasing degree of necessity.
With the provision of both low and high tension power to the electrofilter, the basic function of the unit can be verified.

11.2.3.1 Rotation and function tests. Once rotational checks have been
completed on all drive motors, the mechanical operation of the various
moving components can be observed. In all cases, the driven parts must be
watched through several cycles to ensure that their performance is as seen
during manual tests made earlier in the commissioning programme. If there
is a drive where the motor would be running for long periods when the filter
is on-line, a proving period of 24 h is advisable. This is particularly
important for all rapping gear when both the driving and driven items can
be closely observed.
11.2.3.2 Gas distribution tests. Some manufacturers of precipitators will
model test the installation prior to its construction in order to establish the
positions of any gas distribution devices that may be required. Others will
wait until this stage of the commissioning to do tests within the casing itself.
Whichever method is adopted, it is important to know that the full
cross-section of the filter is being subjected to an even distribution of process
gas. A matrix of points across the casing area and at several positions within
its working length can be assessed with the use of an anemometer, once the
fans have been run-up. Personnel can be boxed into the unit (provided that
the fans are only moving ambient air, of course) to ensure that this situation
does exist. People involved in these tests are not under any risks by being
enclosed in the casing. There may be an initial sensation in the ears when
the fans are first switched on which is similar to that of being in an aircraft.

324

THE COMMISSIONING OF ELECTROSTATIC PRECIPITATORS

There will be a slight change in pressure within the ear which, provided the
individuals concerned do not have any health difficulties with their ears,
should not present any problems. If the tests are carried out at a time when
the ambient temperature is low, warm clothing should be worn. The
converse will obviously apply should the tests be in hot weather. A
telephone or radio link must be available so that the people inside the casing
can not only contact each other but also speak with the safetymen and even
the control room.
Whilst in the casing for the distribution tests, it is also an opportune time
to verify that there is no sneakage of gas into the hoppers or, where it is a
design feature, into the dead space above the field. This can be done with
the use of smoke pellets.
11.2.3.3 Casing integrity. Before the process is started, the integrity of the
casing and access points can be assured by running-up the fans, induced or
forced draught, so that a complete check of all locations can be made where
either process gas can egress or ambient air ingress, from or to the
precipitator, respectively. This is best done with the fans producing a gas
pressure which, ideally, would be near to the operational level.
The importance of this action cannot be underestimated. Where an
electrofilter is being used on a process which involves high gas moisture
levels in conjunction with temperatures above water dew-point, corrosive
gas or even toxic gas, it does not require a great deal of imagination to
realise that process gas must be kept in and ambient air out of the casing.
In the instances of high moisture and temperature, severe corrosion can take
place in an amazingly short period of time if ambient air is allowed in. The
ingress of dry and relatively cold air will create moisture as it meets the
hotter, damper gases. The water which condenses out will usually deposit
itself on some local steelwork, casing, door frames, collector sheets, etcetera.
The resulting problems could be structural failure of the casing or high
voltage electrical instability if the ingress occurs after the process has gone
on-line.
Three of the simplest methods of performing this task are to either listen
at each location, move one's hand around the joints or gently blow smoke
at the seal. In the first method, it is common to hear a whistling noise as air
passes from one side of the seal to the other if its integrity is not 100%. The
second method relies on the fact that the hand would detect a cooling
sensation where air is striking it, blowing onto the palm if it were egressing
or between the fingers if it were ingressing through a bad seal. The third will
be fairly obvious. If the casing is under suction, the smoke will disappear
into the offending joint or get blown away very quickly from a pressurised
casing.

MECHANICAL COMMISSIONING

325

11.2.3.4 Pre hot commissioning inspection. At this point, it is advisable for


the commissioning engineer to make one final inspection of the whole plant.
It must be made clear to all personnel involved in the construction and
commissioning of the precipitator that the commissioning engineer and the
person nominated as safety officer are to be the only people who are to enter
any space which, when the process is on-line, is not accessible. These areas
would typically include the casing proper and, where applicable, penthouses
or high voltage insulator compartments. Should the process require a
pressure test of the casing, the final inspection should be made prior to the
test. This will be because once the seals of access doors have been made gas
tight they should not be broken until there is a real need for access into the
casmg.
11.2.3.5 Final casing integrity test. No explanation is really needed to
obviate the inherent dangers associated with the lethal combination of a
leaking, pressurized casing and toxic process gas. However, as an example,
if the casing of a filter cleaning blast furnace gas is not gas tight, highly toxic
carbon monoxide could collect in a pocket where personnel have access.
Very severe headaches at best, or death at the worst, could result! With this
in mind, a pressure test of the casing is required. It will often revolve around
the criterion that a pressure, approximately 1.5 times that of the process
pressure, should be maintained for a period of several hours with only a
relatively small percentage loss. It must be remembered that a pressure test,
conducted in conditions where the ambient temperature is fluctuating or the
sun is shining onto the casing, could easily give a very false idea as to the
real state of its ability to contain the gases. The variations can be quite
dramatic, particularly when there is a large casing surface area to gas
volume ratio and the surfaces are unlagged. Indeed, the differences can be
so great as to invalidate any test which is meant to verify the integrity of
the casing. The periods which are the best for conducting such tests are
either on a cloudy day or during the night, after the time when the ambient
temperature has stabilised.
All potential leakage points should be identified and assessed by the
commissioning engineer prior to the test, then whilst it is in progress these
and other areas can be monitored. A common method of monitoring is by
the use of soap and water spread liberally around the joint or suspect area.
If any leaks are occurring, bubbles will very quickly form and be seen.
Remedial action will then have to be taken to rectify the problem until a
complete seal is achieved.
Once back outside the casing and having accounted for all personnel, the
engineer can 'box' the precipitator up by completely securing all access
points and doors through the interlock system.

326

THE COMMISSIONING OF ELECTROSTATIC PRECIPITATORS

11.2.4 Hot commissioning

Hot commissioning of the precipitator is the final stage of putting the plant
on line. As the procedure is intimately bound up with Process Commissioning, its description will be found under the heading, later in this chapter.

11.3 Electrical commissioning

11.3.1

An overview

At first sight the process of electrical commissioning occurs at the end of a


long line of operations resulting in the building of a precipitator. It could be
said that the electrical commissioning engineer only switches on the finished
product that so many people have had a hand in producing.
This concept of electrical commissioning is a long way from the reality.
The act of electrical commissioning relies on a process of groundwork and
long hours which lay down the foundations of an installation that is right
first time. It requires a logical approach to the commissioning process and
final checking and optimising of the finished product.
The commissioning process relies on the following functions:
(a) Familiarity-it is important that the engineer is fully familiar with all
the component parts of the control systems that operate the precipitator, and understands how these systems function.
(b) Monitoring-the engineer must monitor the installation of the cabling,
control systems, motors, rectifiers and all the electrical equipment
required to operate the precipitator.
(c) Experience-the engineer must also have a working knowledge of the
factors affecting the performance of the precipitator or 'know a man'
that does.
11.3.2

Managers and commissioning

Generally managers do not commission physically, but they facilitate the


process and as such can help or hinder the commissioning. At this stage, it
is worth pointing out a number of management prerequisites that are
needed in order to ensure a smooth electrical commissioning phase.
1. Once electrical commissioning engineers have been assigned to a contract

they must feel that they have responsibility for all the commissioning
activities. They should know their areas of responsibility and know
whom to contact in the event of problems or when requiring advice.
2. They should be given the time and resources to prepare for the electrical
commissioning process well in advance and be made available for the
entire commissioning period. Swopping and changing engineers leads to
inconsistency of approach, often with communication failures between

ELECTRICAL COMMISSIONING

327

the various people and companies providing services, leading to a general


reduction in motivation and ownership.
3. They must be well briefed and provided with all the relevant information,
or at least all that is available at the time.
11.3.3

Familiarity revisited

The first stage of the electrical commissioning is undertaken back in the


office, well before the job is erected. During this period the engineer must
become absolutely familiar with the electrical schematic and layout drawings detailing the circuitry and its operating logic. Should the drawings not
be completed at this phase of the contract, it is a prerequisite that the
engineer makes contact and holds discussions with the electrical contract
engineer. This will enable as full an understanding of the 'system' as is
possible to be achieved. Only in this way can the equipment installation
procedure be monitored effectively, because this is what the majority of the
commissioning engineers' function comprises.
It is during this stage that quality rears its head. Nearly all customers,
worldwide, require a quality plan of some sort to be developed. (In the UK
this is essential if the company is BSI approved.) A quality plan defines the
activities to be carried out, the type and extent of checks to be made, and
the controlling documents or check sheets to be used to record the outcome
of the checks. Quality plans allow each process relating to the contract to
be monitored against measurable goals. The electrical commissioning quality plan should be written by the commissioning engineer responsible for the
job and approved by the departmental head. In this way ownership of, and
familiarity with, the commissioning process is instilled into the commissioning engineer at an early stage in the process.
In general the quality plan can be split into two areas broadly defined as
follows:
1. Pre-energisation checks where each component of the system is checked

off-line (electrically isolated) to ensure that what is supplied conforms


with the material specification, that it has been installed correctly and
that it has not been damaged during installation.
2. Live testing where equipment is powered up and checked to see that it
functions in accordance with its operating specification and logic.
These areas will be covered in more detail later in this section.
11.3.4

The programme- who writes the programme and when?

The process of erecting a precipitator requires the programming of a large


number of activities, in a logical order, to achieve a cost-effective and
efficient build. A significant part of the electrical commissioning activities
comes toward the end of the contract and as such can be subject to a

328

THE COMMISSIONING OF ELECTROSTATIC PRECIPITATORS

number of factors which can lead to programme delays. This can be critical,
in that most of the final commissioning checks involve live testing of
equipment against specific dates for plant start-up, and any delays to
programme can incur financial penalties.
In order for a relevant commissioning programme to be developed, the
commissioning engineer must be involved in, or be familiar with, the
following:
(a) Have a comprehensive understanding of the requirements of the contract, specifically relating to the electrical control and layout of the
systems and components involved.
(b) A knowledge of the delivery and completion dates relating to all of the
major electrical components and erection activities that will effect the
commissioning processes.
(c) Be involved in the final inspection of major electrical components, such
as control panels and rectifier sets. If the commissioning engineers have
been thorough in their preparation this will have been written into the
quality plan previously discussed. This process increases job ownership
and responsibility. There is no one better to check a component than
the person who will ultimately be responsible for making it work.
There are no excuses if the component does not work when it gets to site
and this can really focus the commissioning engineer's mind.
The first stages of the programme development are provided by others
and usually provide for an overall erection programme showing the start
and finish dates for all the major activities and their relationship, but little
detail. The commissioning engineer must be involved in the detailed
development of the commissioning programme. This commissioning programme must fit around the precipitator build and mechanical commissioning functions and a great deal of discussion is necessary with all those
involved in order to achieve a credible programme. At this stage communication between departments is critical.
The commissioning engineer must have a thorough understanding of the
programme layout and requirements in order to identify the areas which are
likely to cause delays. Late deliveries, construction delays and bad weather
can all add up and mean that the programme falls behind schedule. All
activities then tend to slip back and the final commissioning gets sandwiched between this date and the immovable plant start-up data. If this
situation is left, commissioning activities tend to be concentrated into a mad
rush of 24 h working and frayed tempers.
Commissioning engineers must study carefully the progress of all facets
of the erection and evaluate the effects of these delays on their commissioning phases. They can then adjust their programme accordingly to make up
for the delays.

ELECTRICAL COMMISSIONING

11.3.5

329

Monitoring

11.3.5.1 Phase 1. Getting to know you. At this time all parties should be
aware of the names of site engineers, through which all verbal and written
communication and instructions should be channelled so that any misunderstandings can be avoided.
The first phase of the electrical commissioning engineer's monitoring
process starts with a walk around the site with the electrical contractors site
engineer responsible for the cable and equipment installation. The walk
around usually familiarises everybody with the site layout and builds a
working relationship between the commissioning engineer, the contractor's
engineers and the client. It should be based on the programme, the
specification and the quality assurance plan. From this there should start to
develop the methods by which new cables and equipment should be
installed on new installations, or alternatively, how existing cables and
equipment should be removed and new cables and equipment installed on
retrofit situations, within the timescales available.
The commissioning engineer should bear in mind that this should include
the introduction of a safe working practice by the electrical contractor, with
safety audits being carried out for each activity within the programme.
Safety is everyone's responsibility.
As the installation proceeds the commissioning engineer's activities
generally concern checking the following:
(a) Monitoring the progress of the electrical contractors in relation to the
agreed programme. Identifying shortfalls and delays and developing
actions, in association with the electrical contractors, in order to make
up for lost time. Ensuring that the contractor has enough suitably
qualified staff on the site to achieve the programme dates. Contractors
sometimes take on too much work and can be tempted to remove
personnel from site in order to fulfil other commitments. It is essential
that clear agreed targets are developed and that the engineer is firm in
the commitment to achieve then.
(b) Ensuring that the work is carried out in accordance with the contract
conditions, particularly with respect to quality and safety. If the programme falls behind, the temptation to cut corners increases and if the
electrical commissioning engineer is not careful, both quality and safety
can suffer. During this period various problems will occur relating to the
provisions within the contract. In many cases the original contract will
not cover all aspects of a strip out and installation and so the dreaded
'claim of extras' will arise. This is where a clear head and a thorough
knowledge of the specification is essential. All the groundwork carried
out in the office should now payoff.

330

THE COMMISSIONING OF ELECTROSTATIC PRECIPITATORS

Commissioning engineers should keep in mind that it is not their


responsibility to authorise extras to contract. The engineers should
submit a report to the contracts engineer detailing the nature of the claim
for extras and their comments on the validity or otherwise of the claim.
Bear in mind that arguments on-site can affect the progress of the work
and could cause delays. It is usual that these extras claims be acknowledged without prejudice and an agreement be obtained with the electrical
contractor whereby he submits any claims for extras at the end of the job.
(c) Progress meetings with the client in order to brief him on current
situation with respect to the electrical work. Again the electrical commissioning engineer will be part of a team fulfilling the contract and
cooperation will be important in achieving the targets and satisfying the
client.
During this period the commissioning engineer's presence on site should
generally be 1 or 2 days per week. This will allow them to keep track of the
progress. Status reporting back to head office is essential during this period.
There are many other functions occurring at this time and a clear overview
of the current build status by the contract engineer is essential for the
smooth running of the overall contract. These reports should be of the
exception type, i.e. only identifying deviations either from the programme,
which may have knock-on effects, or problems which have occurred. The
maxim should be 'be brief and concise'.
11.3.5.2 Inspections. Usually during this period manufactured equipment
is becoming ready for shipment to site. This is when inspections occur and
are usually at suppliers' premises. The most important equipment requiring
inspection by the electrical commissioning engineer consists of the rectifier
sets and control panels. These are critical to the commissioning programme
and a thorough inspection is advisable and should be mandatory on the
quality assurance plan. This inspection is carried out by the commissioning
engineer and representatives from the suppliers and sometimes the ultimate
client.
11.3.5.3 Inspecting the rectifier sets and control panels. It is usual for the
supplier of the equipment to carry out the inspection checks which are
witnessed by the commissioning engineer. These checks are usually based on
100% inspection, which means that every panel and rectifier is tested. The
panels and rectifiers are usually inspected as matched pairs and clearly
marked up as such for transportation to site. The procedure for these
inspections should be agreed prior to them being carried out. Electrical
commissioning engineers should familiarise themselves with the inspection
procedures and ensure that remedial actions in the event of faults are clearly
defined. At this time delays could jeopardise the build programme and there

ELECTRICAL COMMISSIONING

331

are too many contractors waiting to claim lost time because of late
deliveries.
The inspections should generally include the pre-energisation checks set
out as follows.
(a) Control panels
1. Carry out a dimensional check of the cubicle to ensure it is to drawing,

2.
3.
4.
5.
6.
7.

remembering that this panel will eventually have to fit in a prepared


opening on site. Check also that it complies with the specification in
respect of dustproofing. A large number of control panels have to be sited
in operational areas where the level of airborne dust is high and the
necessity for sealed control panels therefore becomes important.
Check that the safety interlocks are fitted and operate as intended and
that door alignment and seals are acceptable.
Check the painted surfaces for compliance with the specification and
finish quality, checking for damage to the panels or scratches in the
paintwork incurred during the build.
Carry out a cable and component check to ensure that the specified cable
sizes and components have been used.
Check the layout of the components to ensure that they have been
positioned in accordance with the backplate layout drawing. This should
include any provision for segregation of different power supplies.
Check that the cables have been terminated correctly and are securely
fixed. Loose connections can at best burn out components and at worst
cause a fire.
Check that all cables are identified in accordance with the electrical
layout drawings using specified ferrules and markings. The client will
usually provide a cable number inventory and terminations and the
equipment should be checked to ensure the numbering conforms with
that specified.
(b) Rectifier sets

1. Check that the rectifier set dimensions and paint finish comply with
drawings. If cooling fins are fitted, check that they conform to the

required radiated area requirements.


2. Check that the LV terminals are secure and undamaged.
3. Check that the HV insulators are undamaged and fitted correctly.
4. Check for oil leaks at joints. It is usual to pressure test the rectifier tank
overnight.
5. Ensure that a sample of rectifier oil has been checked for dielectric
strength and moisture levels. Excess moisture can lead to premature
arcing within the rectifier tank and can cause failure. A rectifier set cannot
easily be repaired on site, because replacement of the internals requires a

332

THE COMMISSIONING OF ELECTROSTATIC PRECIPITATORS

clean environment; this can be an expensive option, requiring a sheeted


area set aside for the work.
6. Check that the rating plates are as specified.
Live testing. This section is not intended to provide a blow-byblow checklist for the live testing of the rectifier and control panels. It is
intended, however, to identify the major components of the testing process.
These live tests must be carried out in accordance with a written test
procedure agreed with the supplier beforehand. They must be performed in
an area set aside for this work, which is safety interlocked to restrict access
during the tests.
The tests should be carried out after the equipment has been thoroughly
inspected in line with the visual and dimensional checks given above. A
record of these test results should be made on the relevant quality assurance
check sheets as the tests proceed.
The engineer responsible for the testing must ensure that all connections
to the equipment are adequately protected by earthed screening or isolated
in safety interlocked areas and all accessible equipment is fully earthed
before the equipment is energised. It should be remembered that very high
voltages are present (typically 100 k V peak voltages).
The control panel contains a combination of load carrying and electronic
circuitry. In order that high voltage insulation tests can be carried out, all
wires connected to electronic equipment such as the automatic voltage
control unit and the thyristor control cards must be disconnected to prevent
damage to the electronic circuitry. These wires should be securely isolated
and 'made-safe'. Only after these precautions are complete should the cables
be meggered and flash tests carried out.
Applied primary voltages should be checked and recorded together with
output voltages from the transformers used within the control panel. The
range and operation of the kV and Ma meters should be checked by
injecting a 4- 20 rnA signal. The range and operation of the ammeter should
be checked by injecting 0-1 amp signal.
The panel usually provides for a number of local and remote control
functions. All these should be simulated in order to check that they function
as required. All fuse and link connections should be checked by removing
the appropriate fuse or link and ensuring that the correct components are
de-energised. Operational checks on the panel and transformer/rectifier
equipment can then proceed, during which the stop/start, trips and automatic voltage control functions are checked to ensure correct operation.
Once all these tests are complete and the relevant test sheets have been
signed off, the panels and rectifier sets are ready to dispatch to site. It is
usually a good idea to check the packaging to ensure that it is adequate for
the journey. Damaged equipment can cause delays on site which could be
expensive. The on-site storage of the control panels and rectifiers should be

11.3.5.4

ELECTRICAL COMMISSIONING

333

considered. They will require a sheltered area, under cover, where they will
not be damaged by passing traffic.
11.3.5.5 Site checks as the installation proceeds. During the precipitator
build stages, checks on the installation of the electrical equipment and the
cabling are necessary. These checks ensure that the equipment has been
correctly installed and covers the following items.
TR sets and control panels. After these items have been installed and
wired, a visual inspection is required in order to check for any damage
during transit or installation. These are intended as a final re-inspection,
designed primarily to establish that all items are installed correctly and are
to specification. They cover functional checks of the panel interlocks and
door locks, together with an inspection of the cabinet and paintwork for
scratches or dents to establish whether any rectification is required. Checking for rust at this stage is also recommended as this can be an indicator of
inferior paint finish which can be identified to the supplier for rectification.
Site electrical checks should cover the following:

Ensure that all the switchgear, contactors, relays, thyristors and automatic
voltage control equipment within the panel are installed to drawing and
in line with the specification.
Check that the meters are correctly installed and have the correct ranges
and units.
Check that the correct indication lamps are installed.
Ensure that the equipment within the panel is correctly terminated, checking
particularly for any loose connections which could cause overheating
during operation.
Check the type and version of control software installed to ensure it
complies with the specification.
Visually inspect the data recorded on the transformer panels, to ensure it is
as specified.
Check the level of the transformer oil. If this is low then a leak is most likely
the cause. Visually inspect the rectifier for evidence of oil leaks and rectify
if possible. Leaks from flanges and gaskets can usually be stopped by
pinching up the retaining bolts. Further checks will be required to ensure
that the leaks have stopped; if they have not, then a new seal gasket may
be required. Remember that leaks usually get worse as the unit comes
on-line and the rectifier temperature increases. Changing a gasket during
the commissioning stage is preferable to having the unit off-line to do so
later.
Check that all the rectifier transformer temperature trips are set and
function as required. Finally check that the rating label is fitted as
specified.

334

THE COMMISSIONING OF ELECTROSTATIC PRECIPITATORS

11.3.5.6 Rapping motors. Once the rapping motors are finally installed,
the oil levels within the gearbox should be checked and topped-up as
necessary. If the oil level is very low, check for leaks or damage and report
if found. Check all stop and start buttons, door switches and shear pin
switches for correct operation using a multimeter; replace any sticking or
damaged switches as necessary.
As cables are pulled in, usually many metres, mistakes can be made and
the cables can be terminated incorrectly. The commissioning engineer
should ensure that all cables are numbered and ferrules fitted as they are
'belled' (for electronic equipment) or megger tested (for power cabling) by
the electrical contractor. It is critical that at this stage all cores are identified
and recorded as this will help in the event of problems later. The rapping
motor resistance to earth should be checked and recorded together with the
resistance between the motor windings. Finally check that the data on the
motor rating plates is as specified and that the plates themselves are securely
fitted.
The rapping motors are normally controlled from a motor control centre
(MCC). If new, this panel should be checked to ensure it is placed as
required by the contract and that the equipment within the panel complies
with the specification.
During this stage it is possible that modifications to existing MCCs have
had to be made to accommodate new equipment. Every effort is usually
made to ensure that the site drawings are up to data, but undocumented
changes can take place over the years and the commissioning engineer
should check thoroughly that the existing MCC wiring is to drawing. If not,
the drawings should be modified to reflect the 'as built' status. It is
important that drawings are kept up to date as the job proceeds. It becomes
a mammoth task to detail, or even remember, all the wiring changes carried
out throughout the electrical installation and commissioning stage. A simple
red line diagram will suffice, which can then be used to update the final 'as
built' drawings supplied to the client at the end of the contract.
11.3.5.7 Insulator and hopper heating. With most plants, it is usual to
provide electrical heating to ensure that the lead through insulators and
dust hoppers are pre-heated before the plant goes on-line. This prevents (1)
thermal shock and condensation forming on the insulators which lead to
arcing and subsequent cracking, and (2) blockages in the hoppers due to
damp dust forming as a result of condensation.
The heaters are controlled by monitor panels and the engineer should
check that these are sited as specified. Again, the panel equipment should
be checked to ensure that it conforms with that specified and all the cables
are correctly terminated and numbered in line with the wiring layout
drawing. Check to confirm that all the cables have been meggered or bell

ELECTRICAL COMMISSIONING

335

tested and that records are available for the quality assurance manual. The
resistance of the heater elements and fuse ratings should be checked and
recorded. Finally the heater thermostats should be set and recorded.
11.3.5.8 Lifting hoists. In most cases where the rectifiers are sited at the
top of the precipitator, lifting hoists are provided to remove the transformers
for maintenance. The cables to these hoists have to be checked to ensure
they are correctly installed and have been meggered out. The electrical
controls and limit switch operation must also be checked and recorded
using a multimeter.
11.3.5.9 Earthing. The final checks relate to the main earthing system.
This usually consists of heavy copper strip/tape clipped directly to the
structure of the precipitator and linking all the current carrying components
to earth. The joints in the copper tapes are brazed or soldered together and
every joint should be ducter tested to ensure continuity to earth is maintained. The final connections are made to the plant earth system, therefore
tying everything down to the same potential. After testing, each joint should
be permanently marked, and a record kept of the tape routing on a general
arrangement of the plant. Ducter test results should be recorded on a check
sheet.
A second earthing system consists of the portable earths, which are used
when the precipitator plant is off-line and access to the internals is required
for maintenance. These are usually earthing balls connected directly to the
electrode frames and sockets which are attached to poles and copper
earthing tapes which are in turn connected to the main earth system
described above. The engineer should check that the balls are fitted in such
a way to ensure that the electrical clearance between the electrode and
collector plate is maintained as specified. The portable earth assembly
should be checked to ensure that it fits together correctly and that the earth
pole location points for storage of the equipment during plant operation are
sited correctly. These portable earths are a lifeline for personnel entering the
precipitator and should be checked thoroughly to ensure that they function
correctly.
11.3.5.10 Safety checks before energising. Having checked that the plant
is erected correctly and to specification, a number of safety checks have to
be carried out before the unit can be energised. A final inspection of the
internal chambers, hoppers and rapping rooms should be carried out in
order to check that no debris remains left over for the erection phase. Once
the inspection is complete, the interlocking system should be checked. Every
interlock should be checked to ensure the system gives safe access to the
precipitator internals and that none of them can be defeated.

336

THE COMMISSIONING OF ELECTROSTATIC PRECIPITATORS

All earths should be in place and all switchgear should be isolated and
locked off to prevent any attempt at switching on. The system should then
be closed off in order to check that the interlock system effectively bars
personnel from every access to the unit at this point (this includes all access
doors, top houses, panels, switchgear cubicles) and that it cannot be defeated
in any way, that all keys are correctly sited and that no duplicates are
available. If there are problems at this stage, under no circumstances should
one proceed to energise the unit.
This is the time that all the check sheets, filled in so conscientiously
during the build, are checked, the quality assurance manual is inspected and
any checks not completed are carried out.
11. 3. 5.11 Live testing. Live testing is perhaps the most satisfying and
potentially the most dangerous part of the commissioning process, when all
the equipment, that has taken so long to erect, is put into operation.
The rapping drive chains are disconnected and the rapping motors
powered up. A visual check is made to ensure that the motors are rotating
in the correct direction. Again this is important as the system will jam up
and break the shear pin if the rotation is not correct; should the rotation be
incorrect, it is a simple matter to switch off the power to the rapping motors
and change two of the phases over. After the rotation checks are complete,
the local and remote stop/start switches, shear pins and rapping room door
switches are checked to ensure they are operational.
After this, the drive chains can be refitted and the rapping motors
energised over a 24 h period to bed-in the rapping shaft bearings to prove
the system. On completion of the 24 h trial, and before the rapping system
is switched off, the motors are checked to ensure that they are drawing the
correct power consumptions. High loads could indicate resistance in the
drive system caused by misalignment of the rapping drive shaft; this must
be investigated and corrected. If the problem is left unresolved, there is a
likelihood of motor failure within a short period of time from initial
operation. It is not easy to gain access to the internals or rapping room once
the precipitator is on-line, because it is a high voltage area and hence is
covered by the Plant Safe Access Permit system. Access times are usually
after midnight in the small hours and are not pleasant, so fixing the problem
at this stage is far more convenient.
Once the rapping system is fully functional, all the motor control circuits,
lamps, push buttons, timers can be tested to ensure they are operational. It
is becoming more common that the rapping control is effected by a PLC
system that monitors and modifies the speed of the rapping shaft rotation
and hence the frequency of the rapping blows. These systems should be
checked to ensure that they function as specified and deliver the correct
rapping speeds.
In order that the panels can be live tested, a lamp load must be applied
across the transformer terminals in order to simulate the transformer to

ELECTRICAL COMMISSIONING

337

check the operation of the panel control circuits, the automatic voltage
control and thyristor circuits. (The actual transformer primary must be
disconnected for this test.)
The process allows the panels to be checked, without applying high
voltages via the transformer rectifier, and permits testing of the basic control
functions of the AVC units. At this time, all the overloads can be set to the
specified levels and the values recorded together with calibration and
operational checks of the panel kV and rnA meters. Once the panel has
tested out satisfactorily, the transformer load can be reconnected and load
and short circuit tests carried out, checking that the panels and TR sets
energise the correct fields. This is critical in that if the high voltage cables
are fed to the wrong field, then the interlock systems are defeated and there
is a serious danger of personnel electrocuting themselves by gaining access
to energised fields unknowingly. The panels should then be electrically
isolated until still air tests.
Prior to the precipitator being put on-line, i.e. process gases passed
through it, the insulator and hopper heating must be energised. The heaters
should be switched on and the monitor panel circuits checked to ensure they
operate as expected. It is worthwhile checking that the insulators are getting
hot and that the thermostats are operational. During this time the heater
currents should be taken and checked against specifications to ensure that
no excessive loads are being drawn. Again if there is a discrepancy it must
be followed through and resolved.
The lift hoist should be checked to ensure that the davit moves freely and
that the controls operate correctly. It is most important to check that the
limit switches function and control correctly and that the rail stops are in
place in the event of an overshoot. Finally, the hoist should be checked to
ensure that it gives the correct length of lift and that it has a Load Test
Certificate for the specified duty required.
11.3.5.12 Precipitator still air testing. The final checking to be done is the
still air testing. This is coordinated with the mechanical commissioning
functions. Only when the unit has been boxed up, in cooperation with the
mechanical commissioning engineer, can this be carried out in order to
prove that both the electrical functions and the mechanical alignment of the
internals of the precipitator are satisfactory.
With the main power to the control panels on, switch the sets to local
and the AVC to manual, then energise the panel to provide power to the
transformer. Using the manual raise selector, increase the power to the AVC
unit and observe the primary and secondary voltage and current readings
on the panel meters. Readings at a number of steps, up to set maximum
rating or flashover condition, should be taken and recorded. These secondary readings should be plotted on a linear graph to check the relationship
between the voltage and the current and the results filed.

338

THE COMMISSIONING OF ELECTROSTATIC PRECIPITATORS

Generally as the voltage rises toward flashover the rate of ionisation, and
hence current flow, should increase exponentially until full transformer load
is reached or flashover occurs. During this period care should be taken to
raise the power slowly and steadily in order not to cause premature
flashover within the casing. If the panels cannot be energised or trip on
start-up, it is probable that there is an internal fault causing an electrical
short. Hard arcing can usually be located by listening to the sound produced
by the flashover. In any event there is no alternative but to re-enter the
precipitator and find the fault, sometimes a long frustrating job but essential
for the unit to be brought on line.
Once all the fields have been energised and voltage current characteristic
readings have been taken satisfactorily the unit is ready for service.
The still air for each field should be almost identical, provided the same
type of discharge electrode, size of bus section and energisation method are
constant. Any deviation in the curves, assuming that the panel meter
readings have been checked as satisfactory, can only be associated with the
mechanical alignment of the internals or a peripheral reduced clearance.
When the precipitator is in operation, the VII characteristics will change
as a result of temperature, gas composition and, more important, dust
loading. The dust loading will give rise to either corona suppression or,
more likely, space charge effects. The effects of these are discussed in detail
in earlier chapters.
Typical VII characteristics are illustrated, along with the still air results
in Figure 11.12, for a three-field precipitator handling a fairly easy duty. The

'y
Ii,'
.I
I '
l

Process, field 3 Process,


;-field2

....i

.......

....... / '

............ ,./-

.....;;.:.~.::.

......::::::,............

......

/'

/'

......../

,,'

/'

....

/'

, ,

,,\'
Process, field 1

;;

,..,"

kV

Figure 11.12 Typical voltage/current characteristics for still air and process conditions.

PROCESS COMMISSIONING

339

effect of space charge is to reduce the corona current flow mainly on the
inlet field with a corresponding increase in voltage because the space charge
effect reduces as the gas becomes cleaner, the outlet field corona current
increases and the voltage reduces.
Taking and recording these characteristics, as soon as the process reaches
some form of normal operation, provides a useful check on any future
precipitator performance. Since any change could indicate unacceptable
levels of dust deposition on the internals, a mechanical or electrical fault of
some kind or a change in process conditions, for example a fuel change on
a boiler plant. As such the curves provide the precipitation engineer with a
simple comparison tool by which to judge operation.
11.4

Process commissioning

This part of the commissioning programme is probably the most interesting


of all. It is the point at which, to paraphrase a well-known expression, 'The
proof of the pudding will, if everything has gone alright, be in the heating.'
The culmination of the combined efforts of designers, draughtsmen, project
and contract engineers, construction staff and commissioning engineers
should now be ready for the 'acid test'. Will it do the job it was built to do?
Whilst being interesting, this can also be a very frustrating part as well.
Up until this period, the commissioning engineer has had to maintain a level
of progress against which the only outside influence has been a construction
programme. With the start of the process commissioning stage, the precipitator no longer remains centre stage, as it were. It becomes only one piece
of equipment in the many which go to make up the complete process line.
At this stage, its importance in that line, and thus the importance of place
that the commissioning engineer has in the overall process commissioning
picture, is measured by how much influence it has in the continuity of the
operation of the process. The less importance that the filter has in the
system, the higher the degree of frustration. This is because, should there be
a need to make any alterations or rectifications, they may be found
impossible to do because the process cannot be 'halted' for some period of
time. The frustration is born from the loss of control previously held but
now not being able to 'pull rank', as it were, and instigate the actions
required immediately. Perhaps a gauge of its importance can be assessed by
considering what happens if, for some reason, the filter fails. What is the
knock-on effect? Does it completely stop everything? Can parts or all of the
process carryon until such time as there is a window in the programme
which will allow access to the filter?
For obvious reasons, if the unit goes on-line with no problems, it can be
very satisfying. Only when it does not, do the varying degrees of frustration
start to show. Is it the filter that has caused the problem and can it be

340

THE COMMISSIONING OF ELECTROSTATIC PRECIPITATORS

rectified quickly? Has some other part of the process caused it and does it
have to wait until such time as others say it can be rectified? Typical
situations in the everyday story of commissioning folk!
To limit the potential for external sources being the cause of filter
problems, it is essential that when bringing the plant on-line, no short cuts
are taken to expedite or shorten the initial start-up period. If there is an
agreed method for this procedure, it is of paramount importance that this
is adhered to. Not all processes can be commissioned in exactly the same
way. This section is aimed at helping to ensure that it is not the precipitator
that is the cause of any procedural hitch in the process commissioning
programme.
11.4.1

Hot commissioning

As with the previous sections, the operations will be taken in the order that
they would normally occur when putting the plant on-line.
11.4.1.1 Initial activities. Depending on the process concerned, some
preparation may be required in the electrofilter prior to process gas being
admitted to the chamber.
Where rapping is used as the means of removing dust from the gas
distribution equipment and both discharge and collector electrodes, it is
advisable to switch them on as soon as power is available to the plant.
This applies to all heaters, when fitted, irrespective of their location on
the electrofilter. It is necessary to allow a period of several hours to elapse
from this action to that of allowing process gas into the chamber. This will
ensure that the parts of the plant that are being heated are up to their
operating temperature and as such will give the required protection for
which they were installed. At the same time that heaters are energised on
other components of the filter, any purge air fans and, where fitted, their
respective heaters, should also be powered up.
11.4.1.2 Precipitator purging. In cases where the gas to be cleaned is
potentially explosive, the casing must first be purged of ambient air. There
are several ways of accomplishing this. One is to use dirty process gas whilst
the other is to fill the chamber with an inert gas, such as nitrogen.
The purging process can be a time-consuming activity but unless it is
carried out correctly, observing all the necessary safety precautions, the filter
is a potentially explosive device. It is commonplace to ensure that all
potential pockets within the gas pass should be fitted with vent pipes. This
will allow air to be replaced by the purge gas. In designing the purge pipes,
however, the differences in densities between air and the purge gas has to be
known. It would be ridiculous to have vent pipes pointing upwards from the

PROCESS COMMISSIONING

341

highest extremities of the pockets being purged if the purge gas is less dense
than the air it is supposed to replace. It would merely vent straight to
atmosphere leaving the air still in place.
An extra short length of pipework, which will include an isolation valve,
is spurred off each vent line. The object of this is that it is possible to sample
the gases being driven off and thus establish the level of oxygen still
remaining in the chamber. It will be necessary, of course, to be able to close
off the line venting to atmosphere by the use of a further valve so that the
gas to be sampled is redirected to the sampling point. Once the predetermined safe level of oxygen has been achieved on several consecutive readings
from all test points, then, and only then, can it be considered safe to
continue with the commissioning programme.
It is worth repeating that there must be an oxygen level which is recorded
as lower than the maximum allowed in the chamber and that it has been
found at each and every sample point from at least two consecutive tests. It
must also be borne in mind that one test will be a complete circuit of all test
points, so at least two circuits must be made in order to verify the oxygen
content. A typical purge pipe arrangement to be found on a coke oven gas
detarrer, tube type precipitator can be seen in Figure 11.13. Points at which
the tests referred to above can be carried out should be adjacent to the
valves labelled A, B, D, E, G, Hand J.
As a precursor to the admittance of process gas to the precipitator, it will
be necessary to ensure that the dust extraction equipment is operational.
Even though it is unlikely that there will be anywhere near full-load
conditions prevailing, there will be dust being carried forward in the gas
stream. If the filter is attached to a process which has a high moisture
content in its gas, it is imperative that the dust is removed 'as quickly as
possible. This will avoid the possibility of it sticking to the collector plates,
hopper walls or even blocking the hopper discharge point.

11.4.1.3 Warm-up and condensation. Where there is any form of heat


input into the process which is involved in producing the dust, there will
inevitably be moisture as well. Typical examples are cement making processes which could produce gas with as much as 30% moisture in it, on a
volume per volume basis. Municipal waste incinerators can have similar
levels whilst power station electrofilters, controlling emissions from boilers
fired with pulverised coal, can have between 8 and 15% moisture.
If it is considered that in some processes the volumes of gas involved,
albeit at full production rates, can be as high as four million actual cubic
metres per hour, it can be appreciated that this could mean a tremendous
amount of water being bound up in the gas stream. It can and often does
run into many tonnes per hour. Should the moist gas condense out on any
of the internal surfaces of the filter, it will form a very good, damp surface
on which to allow dust to adhere.

342

THE COMMISSIONING OF ELECTROSTATIC PRECIPITATORS

Vent pipes taken


above roof level

w
Steam
heating
chests

H
;-

Gas"}:"L.-:
outlet ;(

_I~'I

Gas
'inlet

Tube section

Test and
pressure
gauge points

Steam supply

~ 1" B.S.P.----t
2" B.S.P.

Tar outlet

----------

I.

:Ki~

D :~lT:5--' Heated seal pot to be


i ii y
filled with tar or
L"_J
water before purging

Figure 11.13 Detarrer purge pipe arrangement.

From this, it will very quickly be realised that as soon as gas is passed
into the precipitator chamber, it will be impinging on surfaces that will
almost certainly be at a temperature below the gas dew-point. With some
processes, water, in abundance, can then be seen running in rivers from
places like hopper outlets or expansion joints until such time as the
dew-point temperature has been exceeded by all components in the casing.
To alleviate the potential for water to build up in the hoppers or in dust
conveyor casings, inspection plates could be left open at the lowest point

PROCESS COMMISSIONING

343

available. When moisture has stopped coming from them, they can be boxed
back up. This will also allow them to become warm and reach a temperature
above the dew-point.
It is not uncommon for a time in excess of 4-6 h to be required to reach
a point where the internals and casing have obtained a temperature which
would not cause condensation when hot, moist gas comes into contact with
them. If the process involves the necessarily slow build-up of temperature,
perhaps because brickwork or refractory has to be dried out in a kiln of
some sort, then this period of 4-6 hours could well stretch to several days.

11.4.1.4 Preheat. With current environmental legislation, many processes


will not have the luxury of being able to bypass the filter for the initial
start-up period. Indeed, they may well have to energise the precipitator at
the same time that gas is passed through it, should dust be generated during
this curing period. If this is the case, then it would be advisable, if at all
possible, to have a system which can give some sort of preheat to the casing
and internals, prior to the admission of process gas. This could take the form
of a closed circuit heating system which is continually passing the gas
content of the precipitator over some heating source. Obviously, this would
only be sensible if the inlet and outlet flares are blanked off by good
dampers. The contained gas would then be drawn from the outlet flare,
ducted over the heating bank and returned to the casing through the inlet
flare. By maintaining the internals at some temperature as close as possible
to the dew-point of the process gas, the prospect of dust build-up is
minimised at worst and eliminated at best. The dampers would remain
closed until the process gas started to flow. Even though it would not have
reached the dew-point at that particular point in time, this would be
tempered by the fact that the internals were already hot enough to cope with
the lower temperature.
1l.4.1.5 Gas bypass. In the event that bypassing is allowable, it can only
be possible on two counts. Firstly, and definitely most importantly, the local
pollution control authorities will have to have sanctioned the action. This
would mean that the maximum emissions would be at a concentration
which did not infringe any preset legislative values. Secondly, some suitable
ductwork has to have been built into the system. If this latter route is not
possible, another alternative would be an emergency vent stack which can
be sited just before the inlet to the filter. As will be discussed at a later point
in this chapter, this type of stack has a more important role to play in the
operation of the electrofilter.
To aid not only the commissioning engineer but also the plant operative,
after hand-over has taken place, it is useful to position thermocouples in the
ductwork immediately prior to the inlet flare and further ones in the outlet
flare. The benefits to be had from these instruments are that they give a good

344

THE COMMISSIONING OF ELECTROSTATIC PRECIPITATORS

indication as to the temperature of the internals of the precipitor. By noting


the temperature differential across it, a close approximation to the real
temperature can be made. From this information a better judgement can
result in assessing when to start energising the rectifiers and thus minimise
any potential problems which would occur from powering up when the
internals were still wet.
11.4.1.6 Electrofilter energisation. Having reached a temperature whereby the plant internals will be dry, the high tension can be switched on. It is
common practice to phase in the transformer rectifier sets such that the inlet
field will be energised first, followed by the second and then subsequent sets.
With modern P LC controls and given the fact that most new precipitators are now computer managed, it is a relatively easy task to link the
thermocouples into the control loops and use them to automatically bring
in the sets at predetermined temperatures. Indeed, it is also possible to hold
the power levels down until such time as the temperature would allow an
increase and the dust burden being presented requires it.

11.4.1. 7 Thermal expansion and thermal stresses. As the temperature of


the process gas increases, there will be physical changes occurring to and
within the casing and ductwork alike. The two most important of these
changes are expansion and stress relieving.
With respect to the expansion aspect, there are a number of things that
will occur. Considering ductwork first, there is the relationship of either inlet
or outlet ducts to their respective precipitator flares. Whilst it is normal
design practice to include expansion joints immediately before the inlet and
after the outlet flares, provided any movement of the ducts relative to the
flares does not impose any stresses onto them there will not be any
problems. However, should due consideration of this relative movement not
have been given, serious damage can occur. A typical example could be the
approach angle of the inlet duct. In particular, if it comes in at some angle
which is oblique to the axis of the filter, tremendous side loads could be
exerted on the inlet flare. Because of their relatively low mass, when
compared with that of the precipitator, ducts will achieve the temperature
of the gas much more quickly. Therefore, relative movement will also occur
as a faster rate. Strategically placed expansion joints should eliminate this
problem. Nevertheless, as the heating-up process continues, regular checks
should be made to ensure that the joints are fulfilling their duty. For
instance, if a sliding joint is installed, is it sliding freely? If one that is flexible
is used, has enough space been allowed for between the two solid parts, so
that either it does not get pinched and holed or the two sections that it is
separating do not touch?
There are some designs of rapping which involve the location of drives
outside of the casing. It is feasible that there could be considerable

PROCESS COMMISSIONING

345

differential expansion between the casing and any driven shaft. This would
be of particular importance in situations in which there is a big difference
in temperature between the process gas and that of the ambient existing in
the locale of the drives. On applications where the gas temperature is
between 300C and 400 C, it is not inconceivable that the differential could
be as high as 200C. This would mean that at stable temperatures, a solid
one-piece rapping drive shaft could be around 20 mm 'shorter' than the
casing where the casing is 10 m wide. If this design were installed, it would
be extremely detrimental to the rapping efficiency of the lift and drop rod
style. The cam lifter for the rod would be so far off-centre that the lift, and
thus the potential energy available for the rap, would be greatly reduced.
The example highlights the necessity to split the shaft into smaller pieces
and have them linked together via a coupling which can absorb the
expansion without moving the shaft. Provided that the individual sections
have a fixed point located half-way along their respective lengths, the effects
of differential expansion can be minimised.
Collector plates can also be vulnerable if they have not been installed
correctly. Because of their thin section, they can reach the temperature of
the gas extremely quickly and thus their expansion can be quite large
relative to their surrounding, more heavily sectioned, components. Clearances which will allow for these differentials should be verified during the
earlier inspection stages.
One further case for consideration is the way in which the precipitator
casing is mounted on its sliding bearings. As has been previously mentioned,
allowance has to be made for the relative differences in expansion between
the casing and its understructure. Although it is most unusual to experience
any problems with this part of the filter, it should be included as part of the
checking procedure required during the warming-up period. Unfortunately,
these bearings are sited in the most inaccessible areas and, to exacerbate this
even further, the individual bearings will normally have some sort of
protective skirt around them, making them even more difficult to see.
However, in the unlikely event of there being any problems, the results
would become obvious. Skirts could become torn at their welds or the
casing appearing to be sitting skewed on its understructure are several
potential examples. If this does occur, causes must be found immediately.
The aspect of stress relieving can be addressed from several fronts. Firstly,
maintaining the theme of high temperatures from the previous few paragraphs, there will be the potential for the thermal relief of welded joints or
stresses built into plates as a result of fabrication. Unfortunately, the results
will be quite difficult to find. However, the fortunate aspects are that damage
very rarely occurs from this cause and, if it does, it is not usually of a
magnitude that will be deleterious to the efficiency of the filter. Should
anything happen that could affect the operation of the electrofilter, it would
appear as plate buckling and/or weld tearing. The former will usually be

346

THE COMMISSIONING OF ELECTROSTATIC PRECIPITATORS

contained and effectively limited by the casing stiffeners on a well-designed


unit. The latter case will be the more serious result. If it does occur, the most
effective way to trace it is to listen to the casing! Not as silly as it may at
first sound but, as has previously been described, egress or ingress of gas
from or into a casing can cause an audible sound which is readily traceable.
It may not be easy to get at but at least its location can be traced! It is worth
repeating the warnings, given earlier in the chapter, of the potential major
problems which could arise as a result of a leakage casing.
The other type of stress relieving stems from the vibrations induced by
the rapping gear. As with thermal stress relieving, the potential for it to
cause any problems which are serious enough to affect the performance are
quite remote. Settlement of mechanical fixings is the major concern here.
Fraze which has not been removed before fitting bolts and their washers and
nuts, will loosen as it gets compressed. Bolts that have not been tightened
or torqued up correctly during the construction phase undo. Often the first
that is known about these events is when the components of the fixings start
appearing from the hoppers. If this sort of thing does occur, it should be
investigated at the earliest possible time- before any, much larger, parts of
the filter start 'relocating' to the hoppers!

11.4.1.8 Equipment optimisation. Once full load has been achieved on the
process, optimisation of such equipment as rapping and automatic voltage
controllers can commence. The former will be in conjunction with the
electrical engineer whilst the latter will form part of the electrical commisSIOnIng sequence.
11.4.1.9 Rapping. The optimisation of rapping timings is usually a function which is carried out over a fairly prolonged period of several days, at
best, but it could run into weeks. The object of this exercise is to ensure that
stack emissions are kept to the best achievable level, whilst maintaining the
cleanest possible internals without creating severe 'rapping spikes'. Rapping
spikes are the result of a surge of re-entrained dust, rapped from collectors,
going forward into the stack and being detected on stack dust monitors.
They will show up on the trace as a severe peak on what would, it is hoped,
be an otherwise consistently low recorded emission level. Should a stack
monitor not be fitted, the results of the spikes can also be recognised by
regular, short, sharp periods of discoloration in the plume emitted from the
stack.
For processes that have equipment in their train where thermal stabilisation is not required, the procedure for bringing the filter on-line is a
distinctly less protracted affair. Precipitators that are cleaning saturated
gases, for instance, can be put straight onto maximum duty, provided that
the essential precursors have been observed. Typical of these would be that
all heaters had been on for sufficient time to allow the hardware that they

PROCESS COMMISSIONING

347

are protecting to achieve a temperature that is at least equal to the water


dew-point of the process gas.

11.4.1.10 Wash-down sprays. On plant that relies on water or a 'liquor'


wash-down to clean the internals, a procedure which is analogous to
rapping optimisation is required. The difference here will be that it is the
frequency of flushing that has to be assessed. Often this will be an
intermittent event which can be carried out at some convenient point during
the operation of the process. Where this is not possible, however, the
experienced engineer will soon start to recognise the telltale signs of
efficiency reduction and from that be able to judge when to go into the
cleaning cycle. As the depth of dust on the internal components starts to
increase, so a gradual decline in the secondary current, or increased
flashover, will be witnessed.
It is quite common for the vendor to suggest a frequency of wash-down
which will be greater than is actually needed as a starting point. Over a
period of time this frequency can be regularly assessed and adjusted to suit
the prevailing conditions. As with rapping optimisation, a regime which is
best suited to the process conditions can take a long time to establish.
11.4.1.11 Emergency dampers and stack. Earlier in the chapter, it was
mentioned that an emergency stack has a very important role to play. To
expand on this statement, imagine a scenario where there is a sudden
emergency which requires that the electro filter be physically isolated from
the process and that the system is at, or close to, being at full production
rates. The inertia of the process would normally be such that the only way
to alleviate the situation would be to vent the gases to atmosphere. This
would be done through a dampered stack at the inlet to the filter and sited
either in the ductwork immediately prior to the flare or taken out of the top
of the flare itself. A typical reason would be a case where the flue gas
temperature exceeded the design maximum for the precipitator and therefore it would need diverting to eliminate the potential for mechanical
damage. It these pieces of equipment are part of the system, the emergency
damper will usually be activated from a thermocouple and controller
situated upstream from the emergency stack. The commissioning of this
portion of the plant would be a parallel action of both mechanical and
electrical disciplines.
11.4.1.12 Instruction in operation and maintenance. There is an increasing
awareness on the part of the end-user to ensure that all personnel that are
likely to have any form of contact with the filter are fully conversant with
the techniques of running and maintaining it. To this end, it is becoming
more common for the commissioning engineer to be asked to give some sort
of 'teach in' to the operatives. For obvious reasons it is best left until the

348

THE COMMISSIONING OF ELECTROSTATIC PRECIPITATORS

end of the commissioning period. Only then can most of the quirks of the
filter be known and thus passed on as pertinent information, along with all
other relevant operating details. The opportunity should also be taken to
highlight maintenance procedures and routines.
11.4.2

Back to the real world!

Once the process has reached and held the design load for several days, all
its associated plant has reached thermal stability and there have not been
any problems with the electrofilter, the commissioning engineer can consider
reverting back into a normal human being and all that that entails: sleeping
in a bed, not dozing in a chair in some control room in a remote part of the
site; eating breakfast at breakfast time, not three o'clock in the afternoon;
recalling how good a pint of beer tastes, when drunk out of a glass in a pub,
not out of a tin at four in the morning in a hotel bedroom!

12

Dry type precipitator applications


K. PORLE AND K.R. PARKER

12.1

Introduction

Dry precipitator applications far outnumber the use of wet or mist precipitators. One of the advantages of dry type separation is that the particulates,
removed from the process gases, are recovered in a more easily handled
state. The precipitator itself, as the gas temperature is normally above
dew-point, can be constructed from ordinary carbon steel, and the higher
gas temperature also means that chimney discharges are adequately dispersed without resorting to expensive reheating strategies.
The largest single application is undoubtedly in the removal of fly ash
arising from the combustion of carbonaceous fuels in the power industry.
The usage is worldwide and precipitators fitted to a 2000 MW coal-fired
station would collect in excess of 1 million tonnes of fly ash per annum from
a 20% ash coal.
As indicated in chapter 1, the dry precipitator can be designed to satisfy
the most stringent legislative regulations, in fact on numerous processes,
emissions of less than 10 mg/Nm 3 have been measured, for example see
Figure 1.3, chapter 1.
While the applications using dry precipitation are too numerous to
address in detail, this chapter will review the main types and examine the
various factors which determine the size, form and operation of the
precipitator in order to meet the design performance levels.

12.2

Power generation industry

Without electricity, modern-day man would find life very different, particularly in the industrialised nations where one expects power to be readily
available at the flick of a switch. In spite of the gravest fears of the
environmentalists that the world is irrevocably polluting itself, power
produced from the combustion of carbonaceous fuels far exceeds all other
generation sources.
The continual demand for power and the need for countries to be
self-sufficient has led to all forms of fuel being used for large-scale electricity
generation, from low rank peat through to high rank anthracites, plus oil
(hydrocarbon) based derivatives. This wide range of fuel types has meant

350

DRY TYPE PRECIPITATOR APPLICATIONS

that special designs of combustor and boiler have been developed to


satisfactorily fire a specific fuel. Each fuel produces different waste gas
conditions, in terms of temperature, chemical composition (analysis) and
dust loadings, which will affect the size, design and possibly the operation
of any precipitator.
Although some combustor designs can satisfactorily fire more than one
basic fuel type, generally each design is specific to a single type or source.
Because of the differing gas conditions, the size and design of precipitator
will vary to optimise the performance for any specific fuel type.
Some countries which are rich in fuel reserves have very viable and
successful export markets for both oil and coal. The internationally traded
coals are mainly from Australia, South Africa, Colombia, Indonesia, Canada, the United States and Poland. There are other exporting nations; China,
for example, with its vast coal reserves, will become an important player in
a few years' time; however, its own expanding usage could mean China will
become a nett importer in the early part of the next century.
The coals which are internationally traded tend to be typical steamraising coals, having a high calorific value, lowish ash and usually a low
sulphur content. These fuels can readily replace/supplement coals traditionally fired in many countries on existing boiler plant.
The greatest precipitator experience has been obtained from units firing
steam-raising class coals. In the past two decades, however, different
countries have used indigenous lower quality fuels to satisfy increasing
power demands, for example brown coals in Australia and Eastern Europe
and subbituminous coals in western USA and Canada. Other geographic
areas having significant quantities of local coal have also developed plant to
be self-sufficient. These different fuels have enabled the precipitator suppliers
to extend their databank knowledge significantly and the ultimate user
should have every confidence that a precipitator installation, provided by
one of the major suppliers having similar relevant experience, should
function entirely satisfactorily in terms of performance and availability.
In addition to the foregoing coals, the power industry also uses hydrocarbon-derived fuels, such as fuel oil, orimulsion and refinery residues; all
produce waste gas conditions which will have an impact on the size and
design of the precipitator. These changes will also be reviewed.
12.2. J

Bituminous coals

These coals represent the greatest fuel usage in pulverised coal (PC) power
generation and are typified by a reasonable Hardgrove Grinding Index,
45-64, a high calorific value and contain a high volatile content 18-34%,
which assists and stabilises combustion.
Dependent on the country of origin and form of mining, the fuel, as fired,
can contain as much as 45% ash, e.g. India, but many mines use benification

POWER GENERATION INDUSTRY

351

methods, such as coal washing, to improve the quality of the coal, particularly where large quantity shipments are involved.
A 2000 MW station would consume some 5 to 6 million tonnes of coal
annually depending on the quality, usually denoted by its calorific value.
Not only does coal washing reduce the amount of ash, but, dependent on
the benification method adopted, can also remove iron pyrites and hence
gives a lower sulphur content, an important feature when considering
limiting the acid gas emissions on environmental grounds.
12.2.1.1 Medium sulphur bituminous coals. It will be appreciated from
chapter 6, section 6, that the performance, or required size of any precipitator, is affected by the quantity of sulphur trioxide present in the flue gas,
which acts as a self-conditioning agent for lowering the electrical resistivity
of the fly ash particles.
For coals having a sulphur content of around 1.5%, which is typical for
UK coals, the precipitator is of fairly standard design (see chapter 4) and its
size or design migration velocity can be readily and confidently predicted
from the typical curve presented in chapter 6, Figure 6.15, after making due
allowance for required efficiency, temperature, dust loading and collector
spacing effects.
12.2.1.2 High sulphur bituminous coals. For coals having sulphur contents in excess of say 2.5%, such as those from eastern US deep mine
sources, the quantity of sulphur trioxide produced during combustion is
more than sufficient to self-condition the fly ash particulates. In fact it is the
excess sulphur trioxide which can give rise to difficulties and can influence
the design of the precipitator.
A high sulphur content will, dependent on the excess air, combustion
conditions and temperature within the system, produce potentially high acid
dew-points, which can result in the acid condensation should the temperature fall below dew-point temperature in the precipitator. If this occurs then
acid particles will be precipitated along with the fly ash and could produce
very cohesive deposits. These will require improved and effective rapping to
ensure that the build-up will not interfere with the precipitator electrical
operating conditions.
Other areas requiring special consideration are those associated with
potential acid deposition, where difficulties have to be resolved in order to
improve precipitator availability; e.g. (a) hopper evacuation: as gas bypassing is controlled in a well-designed precipitator, the only heat arriving is by
radiation plus some conduction; therefore, to avoid dedusting difficulties the
hoppers should have steeper valley angles and the externally supplied heat
should be increased to ease the situation; (b) to minimise insulator tracking
problems, it is important that the insulator heating and purging system is
fully effective; (c) casing corrosion: the thermal insulation of the casing needs

352

DRY TYPE PRECIPITATOR APPLICATIONS

special attention during installation; not only must all heat bridges be
designed out, but areas around access doors and all corners, e.g. roof and
hopper joints, must be adequately lagged and weatherproofed to eliminate
cold spots, which would result in severe local corrosion.
Another problem, which can affect the chimney emission is that of acid
condensation within the outlet ductwork or in the chimney itself. To
minimise this it is important to maintain the gas temperature above acid
dew-point by eliminating all areas of cold air ingress by ensuring that all
seals are effectively maintained. Little can be done to avoid an acid plume
as the gases leave the chimney, except to ensure adequate dispersion to
reduce the visual impact.

12.2.1.3. Low sulphur coals. As the quantity of sulphur available in the


coal reduces, the amount of sulphur trioxide for self-conditioning of the fly
ash particles is lowered to a point where precipitation becomes difficult and
the size of the unit to meet a specific duty will increase, as per the typical
design curve of Figure 6.15, of chapter 6.
Generally, except for a more difficult fly ash to precipitate, the design
features of the precipitator are fairly standard, except that the casing is much
larger, as is the number of internal components. The main phenomena
arising is the development of reverse or back-ionisation on the collectors,
because of the higher resistivity of the fly ash, as described in chapter 6,
section 6.5.
If, as a result of fuel changing to a lower sulphur coal on environmental
grounds, an existing precipitator is found to be too small to perform
satisfactorily, or there is a change in legislation demanding a higher
efficiency, then it is possible to improve performance by the addition of small
quantities of sulphur trioxide (see chapter 15) as a conditioning agent to
reduce the fly ash resistivity.
For some Australian and other low sulphur fuels, which have an ash
exceeding 90% silica plus alumina content, but low sodium etc., in order to
improve fly ash coheson, while reducing its resistivity, dual conditioning
using sulphur trioxide plus ammonia may be necessary, or in some instances
only ammonia is injected, such as to optimise performance by eliminating/reducing re-entrainment of deposited material.
A theory arose amongst some suppliers and users, who appreciated that
the problem of reverse ionisation was the result of high resistivity ash
deposition on the collectors, that, if these could be kept free of dust the
problem would disappear. This lead to the development of high 'g' rapping
specifications, which although assisting in some instances, did not fully
resolve the problem, since with the general low cohesivity of the fly ash, the
collectors were already relatively clean and all the higher rapping did was
to increase the amount of potential re-entrainment and to result in higher
maintenance demands on the rapping systems.

353

POWER GENERATION INDUSTRY

99.9
99.8

.................

~99.5
>0
c::
<D

'0

99

.........

98

-........ _...... -CoalS

95
90
75

100

125

150
175
Gas temperature (OC)

200

225

250

Figure 12.1 Effect of gas temperature on ESP efficiency.

Investigations in Australia, and elsewhere, into changing the temperature


operating regime, have confirmed that when operating at a lower precipitator inlet temperature, the resistivity can be lowered by a factor (see Figure
6.16). The effect of operating at a temperature of 100C, because of a
significant reduction in reverse ionisation, decreases the emission by almost
a factor as indicated in Figure 12.1.
Other methods, for improving the performance of a precipitator firing
low sulphur coal are related to changing the method by which the precipitator is energised, such as the use of intermittent energisation or some form
of pulse charging [1]. These systems, which reduce the reverse ionisation by
controlling the amount of 'charge' arriving at the collector in a given time
interval, are fully described in chapter 8 dealing with the electrics.

12.2.2

Anthracite coals

This coal has a low volatile content, e.g. 6-15%, and is usually very hard
and, as such, is difficult to successfully grind and fire. These properties mean
that combustion is more difficult and high carbon carryover with the fly ash,
of up to 8 %, is the norm. This increase in carbon, where a normal boiler
would have a carryover of less then 2%, can result in severe carbon
re-entrainment or 'slip' which impacts on the overall emission from the
precipitator. In extreme cases particulate emissions have been found to
comprise up to 90% carbon.
To minimise the effect of this carbon re-entrainment or rejection, the
designer/supplier of the precipitator should consider the following measures:

354

DRY TYPE PRECIPITATOR APPLICATIONS

(a) configure the precipitator to have a lower gas velocity than 1.5 mis, (b)
ensure the HT rectifier equipment and discharge electrode system can
provide sufficient power to 'clamp' the particles to the collectors once they
have been deposited, (c) possibly skew the gas distribution to collect as
much carbon material as possible in the inlet field (see chapter 5) and (d) if
high carbon carryover is experienced in operation, apply ammonia conditioning to increase the cohesivity of the particles (see chapter 6, section 6.4).
While the above apply to the precipitator design and operation, improving the coal grinding efficiency, by the addition of dynamic classifiers to the
mills, will improve combustion and hence reduce the carbon carryover and
could possibly eliminate the necessity of the above precipitator modifications.
This latter approach is an important consideration, not only for anthracite coals, but with legislation demanding reduction in NOx emission, some
low NOx burners, employing air staging and lower flame combustion
temperatures, result in higher carbon carryover. So if difficulties are experienced on existing precipitators, after retrofitting with low NOx burners, then
the fitting of dynamic classifiers to reduce the quantity of large coal particles
has been found to significantly reduce the carbon carryover, which not only
reduces fuel costs but, more important to the user, is the improved
precipitator performance.

12.2.3

Subbituminous coals

These coals are typically strip-mined adjacent to the power plant and are
normally of low sulphur content, but contain high moisture levels, 20- 30%;
those from the Western US can also have significant quantities of lime in
the ash. The gas volume arising from the combustion of these fuels is larger
than from bituminous coals, as a result of the higher moisture level, of some
12%, and increased excess air used for combustion.
In spite of the higher moisture content, which should have aided
precipitation, at one stage the ash arising from the combustion of these coals
was believed, by some US manufacturers, to be non-precipitable, or at least
not with an economically sized precipitator, as a result of the severe reverse
ionisation and re-entrainment difficulties exhibited by these fuels. Pilot
precipitator investigations by European suppliers, however, produced design
data from trial burns of these fuels which exploded this myth, and there
are now many major power plants fitted with electrostatic precipitators
very successfully operating in the western and central parts of the United
States and other areas of the world firing these and similar subbituminous
coals.
Possibly except for the larger size of precipitator demanded by the larger
gas volume, fuel and ash analyses, the precipitators themselves do not
contain any special design features. To accommodate the larger precipitator,

POWER GENERATION INDUSTRY

355

its layout, particularly on retrofitted plant, is difficult, especially where


limited boiler centres prevent a conventional 'in line' arrangement. With the
larger units, the configuration, to avoid expensive double decking, take the
form of a 'chevron', where the precipitators are angled to the boiler centre
line, or 'back to back' arrangements, where the flow is at 90 to the boiler
centre line. Difficulties with ensuring uniform dust distribution arose with
the 'chevron' configuration, so many suppliers favour the latter 'back to
back' approach.
As a cost alternative to the larger cold-side precipitators, there was a
trend in the mid-1970s to install the precipitators upstream of the air heater
and operate at a temperature of above 300 C to overcome the effects of the
reverse ionisation, etc. Engineering difficulties and performance problems of
sodium depletion with the hot-side units [2], together with the reduced
thermal efficiency of the units and higher power consumption, has resulted
in a number of original hot-side units being recently converted to conventional cold-side installations.

12.2.4 Brown coals


These are much 'younger' coals being relatively soft and can contain, as
mined, in excess of 60% water; they have a low ash content but can be high
in soluble salts, typically sodium sulphate. The major reserves of brown coal
are in Australia and Europe, where large areas are strip-mined to supply
local power plants.
The fuel, because of its softness, cannot be ground in conventional coal
mills and is only shredded, either in pin mills or more commonly in a fan
having breaker bars fitted to the internal walls of the casing. Earlier boiler
plant of the chain grate type used predried brown coal briquettes, but more
recent plants are basically of the PC design, employing hot primary air to
predry the fuel in the shredders before feeding the burners.
The waste gases from a brown coal or peat-fired station, with the large
quantity of water intrinsically bound with the fuel, have moisture levels
exceeding 20% and tend to have gas temperatures of around 200C;
consequently, the gas volume approaches double that of a bituminous
coal-fired unit.
Although the ash content is fairly low, the problems of comminution
usually means that combustion, in spite of a high volatile content, is far from
complete and results in a high carryover of carbon in the form of char. This
char, being devolatilised coal particles, has a low mass but large surface area
so is subject to potential re-entrainment.
To counter this, the design gas velocity should be limited to say less than
1.4 mls and the minimum aspect ratio, i.e. field length to collector height,
should be around unity. In addition (a) the TR equipment must be
generously rated, not only because of the higher voltages resulting from the

356

DRY TYPE PRECIPITATOR APPLICATIONS

high gas moisture levels, but also to provide additional clamping forces for
the char particles, (b) the rapping regime is also critical so that rapping
spikes are minimised. Rapping spikes, because of the size and colour of the
char, are very visible and are often referred to as 'Indian Smoke Signals'
where re-entrainment is prevalent. With the char being readily re-entrainable, some installations employ complete flow isolation during the rapping
period, using so-called 'off-flow' rapping (see chapter 4, section 4.7) to
combat this.
Although the ash content is fairly low, the silica to alumina ratio is
different to that from most other coals, and, where the alumina is high, e.g.
40%, as opposed to a more normal 20%, a large proportion of the ash
particles are expanded to produce a very voided structure, often referred to
as 'popcorn', which, again because of the low mass/large area, is readily
re-entrainable, so the control of rapping and gas velocity is of paramount
importance for the successful operation of these units; otherwise precipitation, per se, is very easy.

12.2.5

Lignites

This class of fuel falls midway between brown coal and bituminous forms;
again it is a young coal and is typically strip-mined adjacent to the power
plant. It can be found fairly universally, Germany, Greece, North America
and Australia having possibly the largest of the known reserves.
Lignites being somewhat older than brown coals are 'harder' and can be
reasonably ground in conventional milling systems and with their high
volatile content are not difficult to fire. The fuel as mined, has a moisture
up to 50% and sometimes a high ash, often containing a lime content of
30%, with occasionally a significant sodium chloride level.
Generally the gas volume from a lignite-fired unit, because of the higher
back end temperature and moisture level, is larger than from a bituminous
coal unit. Dependent on the lime and salt content of the ash, the precipitability can vary from moderate to difficult. The lime reacts with any free S03
produced and the salt tends to volatilise and pass through the precipitator
as submicron condensed fume, or, more likely can be deposited on the
discharge electrodes to give symptoms of corona suppression, i.e. reduced
charging corona current flow.
Both these situations detract from the potential collecting efficiency based
on coal and ash analysis predictions. By making due allowance for the high
calcium and salt by fitting controlled corona or high emission electrodes,
precipitator performance and reliability is completely satisfactory.

12.2.6

Oil-based fuels

12.2.6.1 Fuel oils. A number of fuel oils are fired in the power industry,
ranging from light oil containing 0.5% sulphur, through to heavy (3600

POWER GENERATION INDUSTRY

357

second) or higher oils, containing a minimum of 3% sulphur. The ash content


is very low, typically < 0.1 %, so particulate pollution from the ash rarely
arises. The need to fit precipitators on heavy fuel oil installations results from
the carryover of carbon and the production of S03 during combustion.
With light fuel oils the additional treatment at the refinery means that the
final product is lower in sulphur and has a much lower viscosity. This results
in improved atomisation and combustion, such that S03 and carbon
carryover are significantly reduced and with only the true ash loading, further
treatment is often not necessary, in many countries, to meet legislation.
In order to fire heavy fuel oil successfully, its viscosity must be reduced
by heating the oil to a temperature of 170C, usually by steam, so that it
can be pumped and sprayed through the burner system. While good
atomisation can be obtained with most spraying systems, there are always
some large droplets produced which fail to burn out completely, some giving
rise to a coke-like structure, others forming 'cenospheres', which are really
gas bubbles in an ash surround. The formation of these cenospheres is not
completely understood, but since they are hollow, having a very small mass
but a large surface area, like the coke particles, are subject to re-entrainment
from the precipitator collecting plates.
The high sulphur content of the heavy fuel oils can, dependent on the
excess air in the combustion zone, produce high acid dew-points, so the air
heater exit temperatures around 160C tend to be the norm; this increases
both the volume and viscosity of the gases to be treated by the precipitator.
With the low dust loadings normally arising, 500mg/Nm 3 , but containing up to 80% carbon, the design precipitator gas velocity needs to be
moderate for high efficiency collection, with an aspect ratio of around unity.
If the gas approaches dew-point temperature, the ash becomes very cohesive, so in spite of potential re-entrainment difficulties, the rapping needs to
be effective, albeit at a fairly low impact frequency.
This is particularly true for installations which employ either magnesia
or ammonia to control the free S03 levels in the discharge gases. The
injection rate of ammonia is critical to prevent the formation of ammonium
bisulphate/bisulphite, which will either fuse within the precipitator giving
adhesive ash or pass through the unit as a vapour to condense later, giving
rise to an unacceptable whitish/blue plume.
With correct control of the injection rate, e.g. a nominal stoichiometry of
around 2 (the actual value being temperature dependent), this mode of
operation is desirous, since not only does it effectively lower the acid
dew-point temperature, but it also reduces the potential quantity of acid
which can be adhering to the ash and carbon particles, making rapping and
hopper dedusting easier. To cater for possible hopper dedusting difficulties,
the hoppers must be adequately heated and thermally insulated and all
areas of exposed steel work attached to the hoppers must be covered.
To avoid insulator tracking problems, with the potential high acid
dew-point and carbon, it is imperative that the design of the insulator/

358

DRY TYPE PRECIPITATOR APPLICATIONS

support system has an adequate and effective heating and controlled


purging to ensure maximum plant availability.
Potential spontaneous combustion of the fine carbon particles on exposure to oxygen can be minimised by eliminating all points of air ingress,
particularly into the hoppers where large quantities of hot carbon dust are
collected. When the dust is removed from the hoppers, ideally it should be
stored in closed containers or fed back to the combustor under nitrogen. In
bringing the unit off-line as much dust as possible should be removed from
the internals and hoppers to reduce any fire risk, which could result in
serious mechanical damage to the precipitator internals.

12.2.6.2 Orimulsion. Orimulsion is a new fuel and is a 70: 30 mixture of


Orinoco Basin bitumen in water. The particle size of the bitumen is around
15 f1m in the emulsion and as such creates few combustion difficulties in
terms of 'burn out'. The ash content although higher than from fuel oils is
still low at around 250 mgjNm 3 but contains high vanadium, nickel and
magnesium components.
With the high sulphur content, around 2.7% as fired, the quantity of S03
produced is very dependent on combustion zone oxygen levels, i.e. carbon
monoxide in the waste gases. The amount of S03 in the gases produces two
distinct ash forms, a light buff and a green colour, the green colour being
associated with vanadyl sulphate, which has a high specific gravity. The high
S03 levels give potentially high dew-points, so effective heating and purging
of the insulator systems is essential to avoid tracking problems.
The fine droplet size produces a fine ash, some 70% being submicron [3]
and, as such, high emission or controlled corona electrodes are required to
avoid corona suppression effects detracting from performance. For the
capture of the fine fume, the gas velocity, although not too critical, should
be no higher than 1.5 m/s. Normally the precipitator would have a minimum
of three separately energised series precipitation fields for an emission lower
than 35 mgjNm 3 .
While the fine fume can create space/corona suppression problems, the
operating gas velocity, temperature, etc., means that the TRs have to be
adequately rated, particularly at the outlet end of the precipitator, where
minimal space charge, as a result of the 'clean gas' conditions, can result in
high discharge currents being consumed.
Although E.C. legislation is presently demanding an emission of 50 mg/
Nm 3 dry at 6% 02' the fine particle sizing requires an actual emission of
around 25 mgjNm 3 to reduce possible environmental complaints regarding
high chimney emissions.
Hopper dedusting can, if air is allowed to ingress, cause difficulties,
particularly when the oxygen levels at the combustor are such as to produce
deliquescent vanadyl sulphate. This can rapidly absorb atmospheric moisture to produce a gelatinous material which is difficult to handle. The dust

THE CEMENT INDUSTRY

359

itself, being high in vanadium, is classed as a 'hazardous material' and


requires 'safe disposal', either to special lined lagoons after conditioning or,
better, to a chemical plant for the recovery of the vanadium.
12.2.6.3 Other-refinery-byproduct derivatives. This class of fuel is becoming of interest to power producers and refineries as power generating plants
offer a safe means of disposal for these products, following legislation
preventing their destruction by simple incinceration at the refinery.
The products fall roughly into two main forms, petroleum coke, which is
solid at atmospheric temperature, and 'other residues', which like pitch, are
really very high viscosity fluids, and although appearing to be solid, slowly
flow depending on the temperature conditions.
The petroleum coke can be readily ground, although its hardness can
affect mill maintenance requirements, and fed into the combustion zone as
a pulverised fuel. The 'cokes' normally contain very little ash, have a low
volatile content but a high sulphur content, up to 9% being found. Although
as a pulverised fuel, combustion is acceptable, the high sulphur content can
lead to excessively high acid dew-point temperatures and it is normal to find
petroleum coke used to supplement other fuels having a lower sulphur.
Some units have been developed to fire only petroleum coke and have
operated successfully with a high back end temperature; however, with the
high sulphur, desulphurisation of the flue gas is mandatory for this form of
plant.
The use of 'other refinery waste residues' has been considered as an
alternative fuel for PC units, but the general consistency makes it more
suitable for fluid bed combustors, which will be described later.

12.3

The cement industry

Dry precipitators are used in the cement processing industry on a large


variety of different applications. Their use dates back to 1912, when the first
application was to reduce the emissions from the wet process kilns at the
US Riverside Cement Works (see chapter 2). In the search for overall
improved thermal efficiency, the wet process and the later semi-wetjsemi-dry
systems have been superseded by large throughput completely dry process
plants.
12.3.1

Wet process manufacture

In the wet process, the feed material is mixed, ground and metered to the
kiln as a slurry containing about 40% water. The slurry is dried as the
material works its way down the kiln and the waste gas exiting the smoke

360

DRY TYPE PRECIPITATOR APPLICATIONS

box has a moisture content of around 30%, a temperature of 200C and a


dust loading of some 10 g/Nm 3.
Precipitation of this ash is relatively easy, since at 200C and high
moisture level the resistivity of the ash is around 10 10 Q-cm. The major
process problem which can arise is the result of refeeding all precipitated
hopper material back to the kiln, particularly if the raw material contains
fairly high chlorides, which volatilise in the kiln and come forward as
submicron particles and form 'snow flakes' as described in chapter 6, section
6.7. To overcome this difficulty a continual bleed-off of the outlet hopper
dust catch is required [3].
12.3.2

Semi-wet processing

In the case of semi-wet/semi-dry processing, the slurry after mixing and


grinding is partially dewatered by some form of pressure filter to give a cake
having a moisture level of around 20%. The cake is shredded and fed on to
a two-stage moving grate; the first section dries the cake and in the second,
a precalcining stage, the temperature is raised from about 300C to 800 0c.
The precalcined material then passes through a short rotating kiln for
ultimate calcination to produce the clinker.
The gases leave the drying stage at a temperature between 100 and 160C
and a moisture content of around 12%. The dust loading, because the bed
acts as a barrier filter, is relatively low, so very high collection efficiency is
not required to meet legislation.
At the lower gas temperatures, precipitation is relatively easy, but below
11 0 C, severe corrosion of ordinary carbon steel components can rapidly
develop, leading to reduced plant availability. To counter this situation,
internals fabricated from aluminium/magnesium alloy have been found to
give a satisfactory life.
At slightly higher temperatures, stainless steel is an alternative material,
but pitting corrosion with high chloride raw materials may necessitate the
use of high grade 317L stainless.
If the kiln feed is reduced or the cake has a low moisture level, then with
temperatures approaching 160C, since precipitator performance is dependent on the relative humidity of the gas (resistivity effect), it may be necessary
to use water conditioning to lower the temperature and increase the
moisture level.
12.3.3

Dry process production plant

For modern high output plants, dry processing of the materials is now the
preferred method of cement production (Figure 12.2). In this, the raw
material feed is mixed, dried and ground in a mill using the process gas as
both the carrier and drying media. The ground material is then passed

..

Figure 12.2 Rotary kiln with supension preheater dry process production.

Bulk cement

Conditioning
tower

362

DRY TYPE PRECIPITATOR APPLICATIONS

through a series of cyclones, in the form of a preheater/precalcining tower,


prior to entering a short rotary kiln for final calcination.
Two conditions arise necessitating treatment of the gases by electrostatic
precipitators; the first is for the gases leaving the raw meal mill, carrying up
to 1000 g/Nm 3 of material, i.e. the full raw material supply to the plant, and
the second, the gases leaving the preheater cyclone tower installation, when
the mill is out of service for maintenance, etc.
The gases leaving the raw mill, which originate from the cyclone tower,
are at a temperature of around 90C with a relatively high moisture level.
On many installations, where a spindle type of mill is fully gas swept, dust
concentrations can approach 1000 g/Nm 3. To reduce the possible effects of
severe space charge with this dust loading, some form of mechanical
precollector can be fitted upstream of the precipitator, usually in the
mouthpiece, to reduce the dust concentration to more manageable levels.
Except for the mechanical precollector, the precipitator is of fairly standard
construction, designed for an efficiency of some 99.95% plus.
For the kiln only gas condition, which occurs when the mill is out of
service for routine maintenance, the gases are at a temperature of around
350C, a moisture level of 8% and a dust loading of some 80 g/Nm 3 . The
initial plants treated these gases in a 'hot' precipitator and although a large
number performed satisfactorily, the transient period of bringing the mill
off-line led to resistivity problems, with the dust remaining on the collectors
giving transient high emissions. Once the dust on the internals had been
conditioned by the changing process gas the precipitator usually worked
well.
To eliminate this transient situation, when emissions could well exceed
statutory levels, most modern plant now have cooling/conditioning towers
to reduce the gas temperature to 150C, which at the same time reduces the
dust resistivity to overcome the transient changes.
On some installations, the application of pulse charging techniques (see
chapter 8, section 8.8) has been successfully applied to improve this transient
situation [4]; however, the costs of these units are high and there is,
unfortunately, no guarantee that pulse energisation will universally resolve
the problem.
The design and operation of the conditioning tower is critical to plant
operation and availability, since it is imperative that the water injected for
cooling is fully evaporated to eliminate a wet tower bottom and the
carryover of droplets into the field area. If this occurs, then the bottom of
the tower could build up, ultimately effecting gas throughput and water
carryover to the field would result in a cemented build up which the rapping
would not be able to remove.
The types of spray most suited for this conditioning are either two fluid
atomizers or high pressure spill back nozzles. Even though these will
produce the small droplets required for rapid evaporation, the gas
flow through the tower must be uniform, and to the same standard as

THE CEMENT INDUSTRY

363

for a precipitator (see chapter 5) to avoid wall wetting and ineffective


conditioning.
The precipitator is often fitted with controlled high emission electrodes,
in at least the first and possibly the second fields, to minimise the risk of
possible corona suppression and space charge effects, arising from the
volatilisation of alkali-type materials present in the feed or the high dust
loadings. The TR equipment is typically fitted with intermittent energisation
facilities in order to optimise both performance and power consumption.
With the high dust capture rates the design and sizing of both the hopper
and dedusting systems is critical and must be carefully evaluated to avoid
the hoppers overfilling to such an extent that the internals are damaged.
Hopper sealing against air inleakage is important, since with the high
suctions experienced under normal mill operation of up to 1000 mm w.g.,
any ingress would re-entrain significant quantities of hopper dust, and in the
outlet hopper region would result in excessively high chimney emissions.
12.3.4 Alkali bypass plant
To comply with cement product specifications, it is necessary on some
installations to remove/reduce the level of alkali materials in the final
product. This is achieved by drawing off some of the gases at the kiln smoke
box; at this point the gas is around 800 e and the alkali components are
largely in a gaseous phase. After leaving the smoke box the gases are cooled
by peripheral air injection, to around 300 e, thereby condensing the alkali
compounds.
To optimise the performance of the alkali bypass precipitator it is
necessary to further cool and condition the gases, by means of a cooling
tower, to a temperature of 150C. This conditioning/cooling tower, although usually smaller than the main kiln tower, because of the lower gas
volume being handled by the bleed-off system, must nevertheless be carefully
designed along the same lines as for the main process plant.
The alkali bypass precipitator is of a standard design, except that to
avoid corona suppression from the condensed alkali material, the discharge
electrodes are preferably of the high emission type. Generally the same
electrodes are fitted to all fields. Dust extraction from the hoppers can be
more difficult and extra care must be exercised in the hopper design; also
the dust extracted from the hoppers must be handled separately for safe
disposal, possibly to a fertiliser manufacturer where the alkali components
are potassium based.
12.3.5

Clinker cooler precipitators

In the measures taken to improve the thermal efficiency of cement production, heat from the calcined clinker, which was originally wasted as the
clinker was cooled in planetary air coolers situated around the kiln, is now

364

DRY TYPE PRECIPITATOR APPLICATIONS

recovered. Cooling is achieved by feeding the hot clinker onto a moving


grate, where the clinker, at a temperature of some 1000 DC, is cooled by
undergrate air supplied by an FD fan system, as indicated in Figure 12.2.
Part of the heated air is used as primary combustion air, while any excess
is vented through precipitators to atmosphere. Although this air is hot, the
amount of heat is variable and many operations cannot justify further
thermal recovery plant.
The hot clinker fed on to the grate is very variable in terms of quantity
and depends on the kiln discharge rate; if a kiln 'slip' should occur, then the
mass of clinker can change by more than a factor very rapidly.
During this situation, the amount of heat extracted by the air is very high
and tempertures can change from a normal 120C to over 400C in a short
time interval, during which, because of the high gas velocities arising
through the high temperatures, large increases in dust carryover can occur,
the loading changing from some 5 to over 25 g/Nm 3 .
Originally, it was thought that the clinker dust entrained in the cooler
gases would be very difficult to precipitate; not only has the dust been
exposed to temperatures in excess of 1000C during calcination which
would have destroyed all self-conditioning aspects, but the only moisture
available for reconditioning is that contained in the atmospheric cooling air.
This would mean that the maximum moisture level would be unlikely to
exceed 2% and generally a very much lower figure would prevail. Therefore,
from Figure 6.10, chapter 6, one would expect a dust resistivity of some
1013 Q-cm and above, which should give a difficult dust, i.e. reverse ionisation conditions, within the precipitator. Pilot precipitator investigations
showed, however, that in spite of the high resistivity, the pilot operated
satisfactorily without serious back-ionisation manifesting itself. This was
attributed to the dust being fairly coarse, having a high specific gravity and
a low cohesivity, and as such, was readily removed from the collectors
preventing the full impact of back-corona.
Although the pilot work indicated that the dust was precipitable, the
initial full-scale installations were fitted with water conditioning sprays in
the cooler outlet ducting, but the short contact times and dust build-up
problems quickly led to the water injection being discontinued. Unfortunately, these units did show some increase in emission during kiln slipages,
because of the rapid increase in gas volume and temperature impacting on
the electrical operation, a factor which the spraying system should have
resolved. Later plants designed to cater for this high volume/temperature
condition, however, have worked well and are probably the most costeffective option to handle the rapidly changing conditions.
12.3.6

Cement mill precipitators

In order to have a saleable cement product, the clinker after calcining and
cooling has to be ground to form cement as most people know it. Grinding

GENERAL STEAM-RAISING PLANT

365

is either achieved in a traditional ball mill or, more recently, in vertical


spindle mills of some type. The gas volume from both is relatively small and
consequently the precipitator is usually of a standard modular size rather
than a custom-designed larger unit.
In the case of the ball mill, although most are air swept to entrain the
ground material, the gases first pass through cyclones which reduce the
effective dust loading by up to 90% before reaching the downstream
arrestment plant. Most spindle mills are not fitted with cyclones and the
whole of the product arrives at the downstream collection plant.
Like the raw meal installations, to reduce the dust loading into the first
precipitation stage, many cement mill precipitators are fitted with simple
mechanical impact collectors in the precipitator mouthpiece. Again because
of the potential corona suppression, high emission discharge electrodes are
often fitted throughout the unit.
To minimise the amount of heat developed within the mill, water sprays
are normally fitted to limit the temperature. This water effectively conditions
the mill dust and many precipitators operate with a temperature of 90C,
but with a dew-point of around 60 0c. These conditions are ideal for
precipitation, so few problems exist, unless the water cooling system
becomes ineffective for some reason. The main difficulty is associated with
failure to remove dust from the hopper, which because of the small size
of precipitator does not have much storage capacity; therefore, overfilling
can lead to mechanical and electrical problems which impact on performance.

12.4

General steam-raising plant

Although some of the following applications produce steam for either


process heating or electricity generation, the size of the combustion plant is
much smaller than those used on power stations, many units being less than
100 MW thermal output.
These combustors are fired by a wide range of fuels having very different
characteristics and heating values. To ensure these burn satisfactorily,
special designs of burner and associated firing zones have been developed
to optimise combustion and, hence, thermal efficiency.
The types of fuel include colliery and oil refinery wastes, bark, straw and
other biomass products and conventional municipal wastes. To successfully
fire these fuels, although the chain grate stoker can be used, reciprocating/
cyclic grates and fluidised systems, such as bubbling, deep and recirculating
bed designs, offer improved and more efficient combustion for some fuels.
Some of these combustor designs have offered various challenges to the
precipitator supplier and designer, but the performance of these precipitators, when correctly sized and designed, can satisfy all emission regulations.

366

12.4.1

DRY TYPE PRECIPITATOR APPLICATIONS

Moving grate combustors (chain grate, reciprocating and cyclic beds)

The combustion zone, on a moving bed grate system, is established about


two-thirds of the way along its length. Raw fuel, usually in large pieces, is
fed on to the moving bed and is slowly moved toward the combustion zone.
Combustion is achieved by control of both undergrate and overgrate air, the
overgrate air acting as secondary air to burn otT carbon monoxide and any
carbon particles. The majority of the ash from the fuel remains on the grate
to fall otT into an ash pit at the remote end of the bed.
The waste gases from the grate tend to have a fairly high oxygen and a
temperature of around 180C with a moisture level mainly depending on
the moisture and hydrogen content of the fuel. Ash carryover, because of the
low velocity in the combustion zone, is typically 2 g/Nm 3 with a fairly
narrow size range. For an efficient combustion control system, the ratio of
undergrate to overgrate air is critical; too little overgrate air can result in
poor combustion and the carryover of carbon particles, which can impact
on the performance of the precipitator because of potential re-entrainment
problems.
Generally with optimised and efficient combustion, few precipitator
difficulties arise handling fly ash from a coal-fired moving grate type
installation. Not only is the dust loading low, but the particle size tends to
be larger than from a PC type combustor. Also, with the ditTerent combustion conditions, the quantity of free S03 arriving at the precipitator is
proportionately higher, so self-conditioning of the particles to reduce their
resistivity is rarely a problem.

12.4.2

Fluidised bed units

These fall into two main designs having ditTerent operating regimes, either
at atmospheric or high pressure. The latter is a special case and will be
covered in chapter 16, dealing with new and future designs and applications.
In the case of atmospheric designs, some retain the material, typically
sand, fuel and limestone, close to the bed itself as in the bubbling bed, or,
as in the case of the high turbulence deep bed approach, in the freeboard
area. While others, e.g. the recirculating bed, actually ensure that the
material is carried out of the freeboard area completely, passing into a
cyclone, the underfeed from which recirculates the material, after dedusting,
back to the bed.
An advantage of the fluidised bed over other combustors is that it can
accept a wide range of feed stock types and sizings and with the highly
turbulent bed situation, lime or limestone added to the bed material acts as
an etTective desulphuriser. While this can influence the size of the precipitator, which is usually decided by the calcium salt carryover, this, in its own
right, limits the resistivity etTects of only partly conditioned fly ash.

367

BIOMASS-FIRED STEAM-RAISING PLANTS

Primary superheater

Cyclone
Economiser

Fumace
To air heater
precipitator etc.
Coal feed

Refeed system
hopper lock

Figure 12.3 Principle of a circulating fluid bed combustor.

In the case of the bubbling bed, the main combustion occurs within or
just above the bed, so little dust is carried out of the freeboard area whereas,
with the deep bed approach, the major combustion takes place within the
freeboard area and, because the velocities are higher, results in an increased
loading but still below a typical PC operation. As combustion, particularly
in the bubbling bed, is largely sub stoichiometric the volatiles tend to react
and combine with the fly ash and bed material, so are retained and not
carried forward with the waste gases.
With the circulating bed (Figure 12.3) only the cyclone 'slip' passes
forward to the downstream arrestment plant and while the particles are
cyclone slip, they are invariable coarser than the d so of a PC installation, so
present little problem to effective precipitation, even when limestone is
added for desulphurisation.

12.5 Biomass-fired steam-raising plants


In attempting to reduce dependency on fossil fuels, there are many installations now using biomass or 'renewable' fuels for combined heat and power

368

DRY TYPE PRECIPITATOR APPLICATIONS

(CHP) steam-raising duties. The type of combustors are similar to those


described in the preceding section but, depending on the properties of the
biomass fuel, some prior to entering the combustion zone require preparation, e.g. drying or comminution to give a greater surface area.

12.5.1

Wood chip combustion

As one would expect, wood chips are an ideal source of fuel but can present
the precipitator with several problems; the most onerous is the volatilisation
of various organic materials which can coat the normal particles with an
insulating layer promoting back-ionisation. This condition, although very
obvious from the precipitator electrics, is difficult to predict since it depends
on combustion conditions within the plant; if it should arise, the precipitation engineer must cooperate with the supplier of the combustion equipment
to modify the combustion conditions.
Generally the particle size of the particles from a wood chip fired
installation are very fine and have a low bulk density. In addition to possible
corona suppression, hopper sizing and dedusting must be carefully assessed
during the design stage to eliminate potential difficulties.
The carryover of still smouldering or burning particles can also create a
fire risk, but again this is dependent on the overall combustion conditions.
The high moisture content of the gases associated with wood chip combustion, if one can avoid the problem of volatiles, is ideal for precipitation and
many successful precipitators are in operation worldwide.

12.5.2

Chicken litter etc.

With the trend to raise all types of poultry in battery conditions, considerable quantities of waste material have to be disposed of. The large amount
of straw and litter arising from the batteries has been recently used as a
biomass fuel and there are a number of CHP plants designed to burn this
type of material. The resultant fine ash, like that from wood chip installations, need the hopper and dedusting systems to be carefully designed;
otherwise the precipitator is of a fairly standard configuration.
Again because of the high moisture content of the furnace gases, in spite
of the fine particle sizing, requiring high emission electrodes, the precipitability of the ash presents few actual precipitation difficulties. The only real
area of concern, particularly with chicken litter, is the high concentration of
alkali salts, such as potassium chlorides and phosphates, which can result in
'snow flakes', which tend to settle out in the roof area and may short-circuit
the electrics.

12.5.3

Municipal wastes

Most countries favour the idea of incineration to dispose of wastes arising


from normal living, rather than landfill which can create its own problems.

IRON AND STEEL WORKS

369

The idea of combustion is not new and during the 1960s and early 1970s a
large number of continuous grates were built to incinerate municipal waste.
Many of these did not have energy recovery but the present economic trend
is to use waste heat boilers for steam-raising as CHP.
The early units cooled the hot gases, often with temperatures of 1000 C,
by cooling/conditioning towers to achieve an inlet temperature to the
precipitator of around 300C. At this temperature no serious precipitator
difficulties occurred when combustion was reasonable. The only problem
which arose was in the carryover of unburnt flakes of char from incompletely combusted paper. This was overcome by the addition of simple char
collectors, either in front, or in the outlet flare of the precipitator.
An unexpected difficulty did arise from operating at 300C, in that if the
chloride content of the waste was high, then the deposition of ferrous
chloride, at this temperature, is fused and reacted with the carbon steel to
form the higher ferric chloride by attacking the steel; temperature control
and operation became critical to avoid this situation.
An operating condition which manifested itself was that, on some
installations which only fired waste for 5 days, there was a tendency for the
last operating shift to just switch off the fans and electrics when combustion
was still taking place on the grate. This led to a situation where the
downstream equipment was left over the weekend full of gases below
dew-point (nominally 70 o q, which resulted in general corrosion of the
plant. By introducing a proper purging system and leaving the electrics on,
this situation with proper attention to thermal lagging and weatherproofing
was overcome.
All recent and modified plants now include thermal recovery by passing
the gases through a waste heat boiler. This serves to reduce the temperature
to below 200 C, and in spite of the lower moisture content of the gases,
precipitation is still excellent and few difficulties are experienced [5].
Recent legislation governing the emission of acid gases and heavy metals
has meant that the municipal incinerator, because of the potential high
chloride and heavy metal content of the waste, is subject to severe emission
regulations. Most plants are now fitted with some form of scrubber system,
either ahead of, or behind, the precipitator. For those units employing either
upstream spray driers or dry scrubbers, precipitation is still at an acceptable
level, although because of the increased absorbent loading, the size of unit
has been increased. Again the inlet conditions to the precipitator are still
acceptable and no performance difficulties have been encountered.
12.6 Iron and steel works

12.6.1

Sinter plants

To improve the overall production efficiency of the blast furnace, the charge,
which originally comprised loose basic raw material, coke and flux, is now

370

DRY TYPE PRECIPITATOR APPLICATIONS

partially reduced by a sintering process before feeding the furnace. In this


process, the raw materials are premixed, wetted and pelletised before being
fed on to a moving grate in a controlled manner. The bed moves forward
to pass beneath a combustion zone, which is usually fired by a mixture of
coke oven and blast furnace gas. This combustion zone serves to dry and
ignite the bed which, with the high temperatures, results in some reduction
of the iron ore by the coke, etc., to produce a basic sponge iron sinter
product. The sinter then proceeds through a controlled cooling section of
the grate to strengthen the product, prior to its discharge from the grate into
a breaker and screening system. The hot screened and sized sinter then
passes to air-type coolers before going to stock. The undersized particles are
returned for reprocessing on the sinter plant.
Assuming the raw product emissions have been controlled, the pollution
arising from the sinter plant itself can be considered as arising from two
main areas. The first is that from the hot grate bed, where combustion gases
are drawn through the grate under high suction, e.g. 1000 mm w.g. or more.
The second is that arising from the sinter breaker, screens, coolers and
general handling operations. In addition to the high suction conditions, the
main strand gases are hot and comprise the combustion gases plus air, while
that from the second source is predominantly atmospheric entraining air.
12.6.1.1 Main strand gases. Dust arrestment, in the case of the main
strand gases, it now almost exclusively achieved with electrostatic precipitators to reach the emission levels now demanded by legislation. Originally
only cyclones were installed, but even with the low dust loadings of 2 g(Nm 3 ,
cyclone 'slip' not only resulted in severe fan wear, producing out of balance
conditions on the high duty fans, but also the high density of the particulates
up to 4000 kg/m 3 caused serious local fall-out difficulties.
The electrostatic precipitator used for this duty is of a fairly standard
three-field configuration, except that the casing must be designed to cater
for the high fan suction, e.g. 1000 mm w.g. plus possible temperatures up to
300C. The high 10 fan suctions also impose restrictions on the hopper
sealing device, since if this leaks, inleakage air can re-entrain large quantities
of hopper dust which will raise the average emission levels.
The dust loading at the precipitator is typically low and 'fine', since with
most sinter plant designs there is always a large 'wind box' area beneath the
grate where any heavier and coarser particles inertially fall out. The quantity
of dust, because the bed acts as a barrier filter, is small comprising mainly
raw materials together with volatiles driven ofT during combustion. Some of
these volatiles will condense into submicron-sized alkali fume before the
precipitator, possibly resulting in corona suppression conditions which will
impact on precipitator performance.
Although the gas temperatures can vary widely during 'upset' conditions,
the general operation of the plant is fairly stable. Moisture levels, with the
water from the mix and from combustion, gives a concentration of around

IRON AND STEEL WORKS

371

6% v/v in the waste gases, which in addition, comprise some 6% carbon


dioxide, around 9% oxygen and a nitrogen balance. Provided the gas
moisture is maintained at 6%, few precipitator performance problems are
met under normal operation.
During short-term strand stoppages, the main ID fans should be damper
isolated to maintain the relative humidity within the precipitator. If air is
allowed to be drawn through the system it serves to 'dry' the material on
the collector plates and electrical re-entrainment can result. This is unacceptable, since not only does this result in a short-term emission situation,
as the 'dust sheds', but when the unit returns to service any dust remaining
on the plates has to be reconditioned with the strand gases for the
precipitator to reach optimum performance.
The size of precipitator, or design constant (EMV), is governed by the
amount of lime added to the mix; this not only depends on the ore
composition but whether the furnace is to operate with a basic or acid slag
in the bosch (hearth). The terminology used is the sinter basicity, as the
number increases above unity, i.e. to give a basic flux, the amount of lime
added increases and, as some comes off the bed as CaO, the precipitator
design constant must be reduced. Another aspect of adding lime is that,
during processing, the lime tends to drive off alkali materials from the mix
as sodium and potassium based salts, which increases the quantity of
submicron fume passing through to the precipitator.
A further operating problem on installations which have attempted to use
petroleum or finely ground coke, is that of carbon being pulled off the bed
and passing through to the precipitator This can result in selective carbon
re-entrainment, which affects the emission, or in the worst scenario, a carbon
fire in the hoppers which will mechanically damage the intervals.
12.6.1.2 Sinter-plant dedusting. Although the entraining gas in this application is predominantly air, it can contain a dust loading of some 20 g/Nm 3.
This dust can contain very sharp pieces of sinter, which could cause
destruction of any bag filter media during normal operation or in cleaning;
this possibility makes the electrostatic precipitator suitable for this appliction.
Originally, since the process gas is air with only ambient humidity, but
heated by the various processing stages, it was considered that sinter
dedusting operating conditions would make precipitation uneconomic because of the inherent high resistivity of the particulates. Pilot test investigations, however, showed that the dust could be readily precipitated. This was
attributed to the high specific gravity and poor cohesivity of the dust, which
when precipitated easily shed from the collectors when rapped thereby
preventing reverse ionisation or back-corona conditions being established.
Following these findings, many successful dedusting precipitator installations have been supplied and are the preferred method of air pollution
control.

372

DRY TYPE PRECIPITATOR APPLICATIONS

The electrostatic precipitator is of conventional design with usually three


fields. Unlike the main strand operating conditions, suction at the precipitator is lower, which reduces the amount of casing stiffening required and
simplifies hopper sealing requirements in terms of inleakage, although
effective sealing is still mandatory to eliminate hopper dust re-entrainment.
Since the gas is basically warm ambient air, the precipitator casing is
generally left without thermal insulation, which significantly reduces installation costs.
If the installed precipitator should sufTer from any performance shortfall,
usually because of extending the extraction system to include basic raw
material dedusting operations, e.g. lime into the system, then precipitability
can be improved by increasing the relative humidity slightly. The minimum
moisture level for good precipitation should be around 1.5% v/v, i.e. a dew
point of 13 C, so the amount of water to be evaporated to achieve this is
only some 21/s even for large installations.
12.6.2

Pelletising plants

A few iron and steelworks employ a mixture of sinter product plus pelletised
raw material as the blast furnace feed to further improve production output.
The quantity of pellets used depends on the permeability of the furnace
shaft; too low a permeability will limit gas throughput and, hence, iron
production.
Without describing the actual pelletising plant itself, the basic principle
is that the blast furnace raw feed material is mixed and formed into a 'green'
pellet. This green pellet is fed into the pelletiser for drying, hardening and
finally cooling. Similar to the sinter plant, the drying and hardening stage is
carried out under combustion conditions and the final cooling by atmospheric air, so separate electrostatic precipitators are used and similar
operational comments apply as for the sinter plant precipitators.
12.6.3

Steel making

The older form of decarburising the pig iron was to use open hearth furnaces
having air as the oxygen source; later, however, oxygen enrichment of the
air offered improved efficiency. The next development was to replace the
open hearth furnace by a rotating vessel which was oxygen blown, e.g. the
Kaldo process, where again electrostatic precipitators could effectively
handle the emissions.
Most modern steel making employs direct injection of oxygen into the
hot pig iron contained in a pivoting ladle, i.e. BOS converter in major
steelworks, or into an electric arc furnace in the case of mini steel plants,
where scrap is melted by electrical means to provide the hot metal.
12.6.3.1 BOS converter steelmaking. Carburisation is achieved by injecting oxygen into hot metal to burn ofT the carbon contained in the pig iron

IRON AND STEEL WORKS

373

as carbon monoxide. In the earlier plants, the carbon monoxide was burnt
off in the mouth of the converter giving rise to large volumes of very hot gas
which had to be treated/cooled before being cleaned.
Although heat was recovered by both a water-cooled hood, extracting the
waste gases over the converter mouth, and waste heat boilers, the gas was
finally cooled to a temperature of between 150 and 250 DC, by means of a
cooling/conditioning tower ahead of a dry precipitator, which with the
well-conditioned dust operated highly successfully.
Later process developments reduced the total gas volume to be treated
by using controlled combustion approaches over the mouth of the converter, where only part of the combustible gas is burnt off to form a 'dead burnt'
gas zone around the main central gas core. This gas can contain up to 70%
carbon monoxide and is a valuable fuel source when cleaned. The length of
time during which this CO-rich gas exists is about 10 min, or some 50% of
the total processing time for a typical converter operation.
Since the gas composition rapidly changes during the processing, from
air through to very rich CO gas, reducing CO content and finally back to
air, control of the gas flow control valves, either delivering gas to the gas
holder, if BOS gas recovery is practised, or to the flare stack is critical.
Cleaning of the main converter off gases can be achieved by electrostatic
precipitators, but to cater for the potential explosive conditions that can
arise, the precipitator casing is designed as a cylindrical pressure vessel
having a special dust evacuation system. To further minimise risks, in
addition to strict control of any bypass valves, etc., the HT supply to the
precipitator is closely related to the process oxygen injection timings to
reduce the risk of sparking which might initiate an explosion.
12.6.3.2 Secondary pollution control. In addition to the above major
pollution source, significant secondary pollution can arise during the following converter operations:

(a)
(b)
(c)
(d)

initial charging ..,f scrap into the converter


charging of hot metal into the converter
sampling and alloy additions to the converter
pouring of the finished steel from the converter.

These operations require the convertor to be tilted from beneath the main
extraction hood and the need not to impede access, for the cranes feeding
the scrap or hot metal, severely limits the positioning and location of
extraction hoods. Although local hoods are fitted, in spite of these operating
under high suctions, much of the fume escapes into the building to create
not only an unacceptable working environment, but unless captured gives
rise to external pollution.
The capture of these secondary fumes is normally achieved by converting
the root area immediately above the converter into a hood; the gases,
usually after being mixed with the converter local hood gases, can then be

374

DRY TYPE PRECIPITATOR APPLICATIONS

Figure 12.4 Secondary fume collection and cleaning - BOS plant.

passed through an electrostatic precipitator to remove the fine iron oxide


fume. As the dust concentration is usually less that 1 gjNm 3 , the electrostatic
precipitator is more economic than a bag filter, since it only has to achieve
a mid-90% efficiency, and with the large extraction volumes, the fan power
cost of a bag filter is extremely high. In the case of precipitators the fan
power costs are minimised by mounting the precipitator at high level to use
the thermal lift of the gases themselves, since the pressure loss through the
precipitator is only a few mm w.g. as indicated in Figure 12.4 [6]. The local
extraction hoods are handled by their own smaller fans, which assist the
flow through the total plant.
12.6.3.3 Electric arc steelmaking. For mini steel plants, the hot metal is
usually derived by melting scrap in an electric arc furnace; the furnace has
carbon-type electrodes supplying a three-phase current or high frequency
DC to the scrap causing it to melt. The main furnace gas extraction is
accomplished by means of a 'fourth' or 'second' hole in the furnace roof; the
gases from this point traditionally pass through a combustion area, where
any volatiles or unburnt carbon monoxide are completely combusted.
This combustion process gives rise to very high temperatures, and the
original gas cleaning plant, which treated only these furnace off gases,
employed precipitators following cooling/conditioning towers to reduce the
gas temperatures down to a suitable condition for effective precipitation.
Although this satisfied the main gas condition, it did nothing to prevent
the secondary emissions which arise when the furnace roof is off for
charging, alloying and pouring. To collect these fumes, a large hood is built
above the crane rails, used for charging the furnace, which is operated under
suction to evacuate the hood as it fills with secondary fume. As this
containment volume is very much larger than the 'fourth or 'second' hole

NON-FERROUS INDUSTRIES

375

gas flow, depending on whether the furnace is AC or HFDC operation, the


latest approach is to mix and treat these two gas streams together. By
thorough mixing of the gas streams, some plants dispense with the combustion stage and rely on the dilution effect of the hood gases to minimise any
explosion risk from volatiles or carbon monoxide, i.e. the final gas concentrations are too lean to ignite. As an alternative to the ESP, many plants
now use bag filters for this combined duty.

12.6.4 Operations involving the casting of hot metal


The hot metal route following any necessary alloying, which is carried out
at a small ladle station, is either to produce a cast slab or billet, or a
continuous casting machine.
For mini steel plants, although most of the decarburisation is accomplished in the electric furnace, if special treatment is necessary, this is usually
done in small ladles using molten steel from the arc furnace. In this
situation, if a combined furnace hood and main gas extraction system is
employed, it would be possible to duct the converter ladle station off gases
into the electrostatic precipitator. As the ladle off gases are unlikely to have
controlled combustion hooding, this approach would give a 'dead burnt' gas
containing little or no carbon monoxide. A similar situation arises on major
steelworks, but the volumes emanating from the ladle station normally
justify their own gas cleaning plant. In this case either electrostatic precipitators or bag filters can be used, but with the higher gas temperatures and
possible difficult fume, the former is the preferred system.
Where the steel is cast into slabs, prior to further operations, e.g. rolling,
the slab undergoes a scarfing or scalping operation, where the surface is
removed by passing the slab beneath an oxy-hydrogen burner. The temperature of the burner is sufficient to locally melt the surface and volatilise the
metal; the resultant fume comprises a large quantity of iron oxide fume
together with iron particles plus other tramp elements from the surface.
The machine in which the scarfing occurs is fully enclosed and placed
under suction to contain the fume. This fume was originally treated, after
quench scrubbing to eliminate the hot iron particles, before final cleaning in
a wet precipitator. Later plants employ a dry precipitator which, although
somewhat larger in size, offers the advantage that the discharge gases are
hot and buoyant and so eliminates the cold steam plume from the wet
precipitator.
12.7 Non-ferrous industries
There are many dry precipitator applications in the non-ferrous field; these
differ from those in the iron and steel industry as a result of a generally much

376

DRY TYPE PRECIPITATOR APPLICATIONS

lower metal concentration in the ores. The ores usually require some form
of pretreatment before the metals can be extracted; this can involve the use
of water benification where the heavier metal-rich components settle as
deposited material, and many undergo some form of chemical treatment.
The following outlines some of the processes involved in the 'winning' of
copper and nickel metal, which can be considered as being typical of other
non-ferrous processing.
12.7.1

Copper and nickel recovery

There are numerous applications from the drying of the concentrates,


through primary extraction and finally refining. While the drying and
refining equipment is fairly standard, the methods of smelting differ from
plant to plant.
One of the major differences is that the copper and nickel ores are mainly
used as an oxide or frequently a sulphur bearing concentrate. While this is
of assistance in the roasting operation, in that sulphur is exothermic when
converted to sulphur dioxide, the waste gas will contain high concentrations
of sulphur dioxide, e.g. 11-15%, plus high levels of free S03 resulting in high
acid dew-point temperatures. These can well exceed 200 DC, depending on
the SOz, gas moisture and excess air available. These high acid dew-points
mean that, to avoid acid deposition within the precipitator leading to
adhesive dusts and corrosion, the operating temperatures are usually held
in excess of 300C.
Except for the additional design duties for the casing, the possible fitting
of controlled corona-type electrodes to eliminate corona suppression, and
good rapping, the precipitator itself comprises fairly standard components
when used for roasting applications.
A typical process route is (a) to produce the concentrate, (b) dry the
product, (c) roasting or calcining to develop a matte and (d) converting the
matte to a raw product.
12.7.1.1 Concentrate driers. Because of the high sulphur content of the
concentrate, drier operation is critical, in terms not only of how the heat is
applied for drying, but also the temperature to which the concentrate is
exposed and the oxygen concentration of the waste gases if fires and other
problems are to be avoided.
The waste gases would typically be at a temperature of around 120C, a
moisture of 20% and a dust loading up to several hundred gjNm 3 . The main
difficulty with this application can lie with the hopper dedusting operation;
not only is the loading high, but the dust also has a high angle of repose.
The avoidance of air ingress, which would result in a fire, must also be of
concern to both the designer and user.
12.7.1.2 Roasters. In the roasting operation, the dried concentrate is
allowed to fire under controlled conditions, such that the heat of reaction

NON-FERROUS INDUSTRIES

377

produces a calcined product ready for smelting. Most roasters, because of


the high exit temperatures, have waste heat boilers to reduce the temperature to 350 0c. The gases then pass through conventional 'high temperature'
precipitators to remove the dust, which can be up to 40 g/Nm 3 , depending
on the process, before it passes on to a sulphuric acid production unit.
12.7.1.3 Roasting/smelting. After drying the concentrate, the next step is
to produce a calcine, either by roasting in a multi hearth furnace or fluid bed
system. The calcined product is then fed into a fossil fed reverberatory
furnace, similar to the open hearth furnace used in steel making, where a
matte containing some 50% copper is produced. To improve thermal
efficiency and the dependence on fossil fuels some plants use an electric
furnace, similar to the electric arc furnace used in steel making.
Whether the calcine smelting, to produce the matte, is carried out in a
reverberatory or electric furnace, the waste gas moisture and SOz levels at
10% and 1%, respectively, are much lower than from the roaster. In this
case, the waste heat boiler can be more efficient, as the exit temperature can
be below 300 C, rather than 350 C before the precipitator as in the case of
roasting.
For processing using the Outokumpo or Inco 'Flash Smelter', which
basically differs only in shape, the dried concentrate and flux are injected
into the furnace with the combustion air/oxygen, which then reacts with the
concentrate. The heavier solids drop into the slag/matte in the base of the
furnace while the waste gases pass over the surface to leave the furnace at
high level. The gases leaving the flash smelter have a much lower volume,
because of the more controlled combustion conditions, and contain between
50 and 150 g/Nm 3 of dust. This is carried in the gas, which has a relatively
low moisture and some 10-15% SOz' The gases are again cooled by means
of waste heat boilers to 350 C before passing through a precipitator ahead
of an acid plant.
12.7.1.4 Converters. The matte is converted to the raw copper, 95% or
higher, usually in a Pierce Smith Converter. This is basically a tilting furnace
through which air or oxygen-enriched air plus flux is blown into the molten
matte.
In converting to the basic refined metal, the molten matte undergoes
various stages of refining by the oxygen. The flux, usually added with the
air, is used to protect the copper and to collect the oxidised iron and other
materials released from the matte. After a certain time the slag is run off by
tilting the converter, and further refining occurs with possible intermediate
slag run-off opertions. The 'blowing' is continued until the required purity
metal has been achieved. The converter is then normally tilted backwards
to run off the 'blister' copper.
Because of the various processing stages and vessel rotational requirements, the thermal properties of the waste gas are variable and while waste

378

DRY TYPE PRECIPITATOR APPLICATIONS

heat boilers are sometimes used, other plants employ cooling/conditioning


towers for this duty. Again the temperature of precipitator operation is
around 350C with the gas containing some 10 g/Nm 3 of dust. For
the waste heat boiler scenario, the moisture is around ambient level,
as compared with 15% when a cooling tower is employed, but in either
case effective precipitation is achieved. With a sulphur dioxide concentration of up to 12%, the cleaned gas normally passes to a sulphuric acid
plant.
12.7.1.5 Arsenic recovery. On plants which have ores containing high
levels of arsenic, after cleaning at 350C, to remove the heavy metal
components, the gases are further cooled to condense the vapour-phase
arsenic trioxide by a cooling/conditioning tower to around 105C before the
gas enters another dry precipitator, or a bag filter, to recover a fairly pure
arsenic trioxide.
The design and operation of this section of the plant is critical because
of the toxicity of the arsenic. Special procedures are necessary if the unit
is to be internally examined or maintained, so absolute reliability is
essential.
12.7.1.6 Further refining. If further refining is necessary, this is normally
carried out in smaller units, which can be directly or indirectly fired, electric
arc or induction-type furnaces. These can be individually hooded and the
extracted gases passed through a dry precipitator or bag filter of conventional design, operating at a temperature of around 100C.

12.8

Aluminium smelting

Aluminium is normally extracted from a melt of fused sodium aluminium


fluoride by electrolysis using a high current low voltage DC supply. The
anodes are composed of carbon/pitch and the cathode, carbon blocks lining
the furnace or cell. Pollution arises during the electrolysis, which takes place
at a reasonably high temperature, and the waste gases collected from each
cell contain fairly high loadings of alumina together with hydrogen fluoride.
These waste gases are initially treated by injecting an absorbent into the gas
stream, ahead of a dry precipitator, which not only collects the alumina, but
also the absorbed HF as a sodium aluminium fluoride which is returned to
the cell for reprocessing.
While this method of treatment is only on the edge of the true nonferrous applications, it nevertheless can be widely used for other acid gas
removal systems, where an absorbent is injected upstream of the precipitator, which collects the offending gas as a dry product.

PAPER AND PULP INDUSTRY

12.9

379

Paper and pulp industry

Processing of raw timber into paper is a complex operation, involving a


large number of applications to which the dry precipitator is eminently
suitable. Again, without presenting all the facets of the complete process,
most of the applications are related to pollution arising from combustion
processes which provide steam employed in the manufacturing of the final
product.

12.9.1

Bark firing

Bark from the timber used for further processing, is fired, after shredding as
a 'hog' fuel, on either moving grate or fluidised bed combustors. Similar to
wood chip combustors, there is a risk of burning or smouldering particles
being carried forward and it is not unusual to find a cyclone upstream of
the precipitator to collect this coarse material, which from a moving grate
can comprise up to 70% carbon. With the fluidised bed system the improved
combustion produces much less carbon so cyclones are not absolutely
necessary.
The gas temperature is typically 200 DC, has a moisture level of some 17%
and a dust loading of 3 g/Nm 3 from the moving grate, and 10-15 g/Nm 3
from the fluid bed units. Except for the fire risk and possible re-entrainment
of carbon particles, few precipitation difficulties are encountered.

12.9.2

Cellulose pulp production

During the manufacture of pulp, the chemicals used for digesting the wood
are expensive and need to be recovered on economic grounds. The chemicals, caustic soda and sodium sulphite, are used in a strong solution, referred
to as cooking liquor, which, after digestion, turns black from the organics,
mainly lignin, contained in the timber.
To recover these chemicals from the used liquor, it is first concentrated
by evaporation to produce a water content of 20-40%. This is then sprayed
into an oil-fired recovery boiler, where the resultant smelted product, mainly
sodium carbonate and sulphite, collects in the bottom of the furnace and is
tapped off to be ultimately re-used after treatment as the cooking liquor.
Some of the sodium from the smelt combines with the fuel gas phase sulphur
products to form sodium sulphate, or carbonate if there is insufficient
sulphur present. Due to impurities, the waste gas will contain carbon
particles, sodium chloride and potassium salts, which are collected and
recycled back to the dissolving tanks.
There are two main precipitator locations, either upstream of the economiser where the temperature is some 320C, or downstream at a temperature of 160 dc. In both cases, the dust loading of largely volatilised salts is

380

DRY TYPE PRECIPITATOR APPLICATIONS

around 15 g/Nm 3 , which means that the precipitator must be fitted with
high corona emission electrodes. Although the precipitator internals are
rapped with conventional impact systems, two hopper arrangements can be
found: in the US the trend is to use a 'wet' bottom, where the dust falls into
a bath of molten black liquor, or in Europe where wide dry scraper chain
conveyors are favoured.
A further process modification can also be found, where the hot dirty flue
cases are brought into contact with the black liquor in a direct contact
evaporator (DCE) ahead of the precipitator, where the evaporator takes the
form of a rotating tube bundle passing through the black liquor to provide
an adhesive impact surface to the dust carried in the hot gases; another
approach is where the black liquor is sprayed into a 'wet' cyclone, which
acts as both an impact collector and spray drier. Either method produces a
gas temperature of around 180 cC and a final dust loading of 10 gjNm 3
ahead of the precipitator. Other differences to the normal approach are a
higher moisture content because of evaporation and a much lower plant
odour.
Good thermal insulation and sometimes preheating of the precipitator
casing, if of steel construction, is necessary to minimise corrosion and sticky
dust difficulties. To avoid these difficulties, particularly corrOSIOn, some
casings are constructed from lined reinforced concrete.

12.9.3

Lime sludge burning

In the black liquor recovery system, the sodium-based products from the
smelt are dissolved to form a 'green' liquor, to which quick lime is added to
produce the 'white' liquor used in the cooking process. The resultant
calcium carbonate deposited from the 'green' liquor stage is extracted as a
lime sludge or lime mud. This mud is then used as a feed material for a
conventional rotating kiln for conversion back to CaO. The impurities
driven off during calcination can contain a high proportion of sodium salts
which are initially volatilised and then condense in the cooler feed end of
the kiln. The waste gases to be treated by the precipitator are typical of a
wet process cement kiln, i.e. a temperture of some 180 to 200C, a moisture
content of 33% and a dust loading of 13 gjNm 3. Except for the problems
which may arise because of the condensed submicron sodium salts, precipittion is not difficult.

12.9.4

Magnesium sulphate burning

As an alternative to the black liquor route, some processes employ a


magnesium or 'red' liquor approach to digestion. The preparation and firing
of this used liquor is similar to that used for the black liquor, except that
during combustion the furnace is operated to maintain a dry bottom, i.e. all

REFERENCES

381

products are carried forward with the waste gases. The fly ash comprises
mainly magnesium oxide and some sulphate, carried in a gas stream having
30% moisture and, to ensure satisfactory precipitation, a controlled gas
temperature of 160C. The dust collected in the hoppers is reacted with
warm water to form magnesium hydroxide, which is then used to scrub the
S02 from the gases, the resultant bisulphate being returned to the cooking
liquor stage.
12.10

Conclusions

An attempt has been made in this chapter to indicate some of the


applications to which dry electrostatic precipitators have been applied and
to give some of the operational characteristics which can impact on the
efficiency of the installed plant.
The listing above by no means covers all the applications to be found,
but it is hoped that by examining those most widely used, a general
impression will be gained of the sort of factors which need to be considered
in the sizing, design and operation of any dry precipitator application.
References
1. Porle, K. and Karlsson, R. (1995) Long-term experience with pulsed energisation of ESPs
at a Danish power station, EPRI/DOE International Conference on Managing Hazardous
and Particulate Air Pollutants, May 15-17th, Toronto, Canada, in print.
2. Bicklehaupt, R.E. (1980) An Interpretation of the deteriorative performance of hot-side
precipators, JACPA, 30, 812.
3. Parker, K.R. and Cottingham, C.R. (1993) An electrostatic precipitator designed specially
to collect orimulsion fly ash. Proc. 10th Particulate Control Symposium and 5th International Conference on Electrostatic Precipitation, April, Washington, USA, pp. 7.1-16, EPRI
TR 103048 2, Palo Alto, CA, 1993.
4. Darby, K. and Parker, K.R. (1990) The use of electrostatic precipitators in the cement
manufacturing industry for the control of dust emissions. Proc. 4th International Conference
on Electrostatic Precipitation, September, Beijing, China, 3, pp. 173-86, International
Academic Publishers, Beijing, 1991.
5. Parker, K.R. and Russell-Jones, A. (1985) Control of Refuse Incinerator Emissions to the
Atmosphere. Conference on Energy Recovery from Incineraton, 6th Feb., pp. AI-18. Inst.
Mech. Engs, London, U.K.
6. Darby, K. and Parker, K.R. (1977) Problems of secondary ventilation in the iron and steel
industry. Proc. 4th International Clean Air Conference, May, Tokyo, Japan, Clean Air Soc.
of Japan, Tokyo, Japan.

13

The wet electrostatic precipitator:


design and applications
K.R. PARKER

13.1

Introduction

Although the wet electrostatic precipitator or WESP, operates on the same


basic principles as the dry type of unit, the major difference is that the
charged particulates, on arriving at the collector, are removed by a flushing
liquid rather than by mechanical rapping. Otherwise they both function in
the same manner. The wet collector interface makes the wet precipitator
ideal for the collection of particulates that are either sticky or are carried in
a gas stream which is close to, or at, saturation temperature. Under these
circumstances the dry form of precipitator would rapidly build-up with
deposited material, which could not be removed by mechanical means, and
hence its performance would deteriorate.
The history of the wet precipitator, as will be appreciated from chapter
2, dates back to the mid-191Os, when difficulties were experienced in the dry
collection of various materials, either because of electrical resistivity problems, or the material to be collected was in a gaseous phase which
condensed on cooling in the atmosphere to produce an undesirable plume.
Possibly the first form of wet precipitator was with a wetted collector,
where water was allowed to run down the collector surface as a film, upon
which the dust was deposited and removed with the water. This approach
was developed to reduce the drop in precipitator performance when handling dusts having high electrical resistivity.
On some of the US metallurgical smelters, it was found that after
removing the metallic solid-phase materials at relatively high temperatures,
there was still an air pollution problem resulting from condensed volatile
compounds, such as arsenic and cadmium. Whilst it was appreciated that
the dry precipitator could collect these materials in a condensed form, the
practical problems of maintaining a specific gas stream temperature, by
means of water sprays, or air cooling (which further impacts on the size of
the precipitator), led to the use of wet precipitators, usually following a
quench tower or upstream scrubber used to cool and saturate the incoming
gases.

DESIGN CONSIDERATIONS

383

13.2 Design considerations

Whilst the principle of operation is basically identical to the dry unit, there
are significant differences in the practice of wet electrostatic precipitation
which impacts on the engineering design of the plant. The following
discusses the dominant aspects of these differences.

13.2.1

Dust deposition and removal

There are basically three approaches used for removing dust deposits in wet
precipitation. The first, as indicated, is a film of water flowing over the
surface of the collector. The second, which may be used in conjunction with
the film flow approach, is simply to periodically sluice the field with water
from large capacity sprays situated above the electrode system, and finally
a fully irrigated plant, where a multiplicity of small medium pressure sprays,
operating continuously, results in the total electrode system being wetted.

13.2.1.1 Collector film flow. The modification of the early vertical tube
type units to film flow, rather then mechanical rapping, was fairly simple. As
the tubes were suspended from a header plate, forming weirs at the top was
not too difficult. These weirs are essential to ensure that each of the tubes
is fully wetted. A typical flow rate for a 150 mm diameter tube would be in
the order of 9ljmin.
To avoid the electric field pulling liquor draining from the bottom of the
tubes to the discharge electrodes, the bottom of the tubes were belled
outwards and cut at an angle to lead the liquor away from the discharge
electrode. Had this not been done, then flashover or short-circuiting would
have resulted. To further minimise this effect, the electrode emission characteristics were locally reduced by having the electrode wires ferruled at both
ends to increase their diameter at the point where they pass into the
collector tube. These points are illustrated in Figure 13.la.
With the plate form of precipitator, the simple weir approach proves
impractical, since not only is one attempting to uniformly cover the full
collector length but also equally cover both sides. This is particularly so for
large collector plates, which are used to minimise costs for large gas flow
treatment. In practice, almost regardless of the design used, as one is relying
on surface tension to give reasonably uniform flow, collectors having
dimensions greater than 3 m in length and 6 m in height can give difficulties
in maintaining complete wetting. By significantly increasing the amount of
liquor flowing over the plates it is, however, possible to use plates having
heights greater than 6 m; some plants in fact have plates up to 12 m.
To minimise the difficulty in maintaining complete film flow along a
typical thin collector, many designs use a thicker, or double-sided collector,
lined up with the water distribution arrangement along the top. A typical

384

THE WET ELECTROSTATIC PRECIPITATOR

Feed

Discharge
frame
Weir
Water
level
Tube
plate
Collector
tube

Ferruled
wire

I-------i/

'----------r

~----_IJ _. Liquor

-~Belledend

IL,

collector
Wire
tensioning
weight

Thick FRP
collector

END

(a) Film flow tube unit

collection
gutter

SIDE ELEVATION

(b) Film flow plate unit

Spray pipes
, . - _............._---.,/

Collector
plate

D.E. cross
connector _________ ~~:;:::::::::~::::::::::I

SIDE ELEVATION

END
(c) Spray irrigated unit

Figure 13.1 Different water washing approaches. (a) Film flow tube unit. (b) Film flow plate
unit. (c) Spray-irrigated unit.

DESIGN CONSIDERATIONS

385

wet collector would have some form of hollow top beam which would have
numerous orifice inserts in the beam, as indicated in Figure 13.1 b. These
inserts are adjustable in height to cater for differential water pressure along
the collector length together with fabrication and erection tolerances, the
height and hence flow rate being set once the plant is ready for operation.
The nozzle orifii are normally 2 to 3 mm diameter to accept recycled dirty
water usage and operate with a head of around 5 mm w.g.; consequently,
flow control along the collector length is difficult, not only to set up, but to
maintain in practice.
A typical flow rate, for a 2.5 m long collector, would be some 40 ljmin;
again, with this largish water flow, drainage from the collector bottom
presents difficulties if it is not to be pulled from the collector surface by the
electric field. Many designs incorporate a sloping collector bottom edge
which, with surface tension effects, leads the water to a safe position, i.e.
away from the influence of the electric field, before discharging it into the
hopper or sump.

13.2.1.2 Wash-down plant. In an attempt to reduce the problems in


maintaining a continuous film of water over both faces of an entire collector,
the full wash-down approach was developed. A typical application would be
for a gas above saturation temperature creating dry spots which would
result in operational problems.
In this design, large-capacity low-pressure sprays are mounted above the
collector and discharge system, each spray having a covering capability and
throughput of some 200 ljminjm2 at a distance of 1500 mm below the spray.
Having such throughout and operating at low pressure provides sufficient
large coarse droplets to completely cover the system to flush off deposits
from both collector and discharge electrodes. With this large quantity of
liquid passing into the field it is sometimes necessary to electrically isolate
the section being washed down and also, in some instances, to reduce power
on the immediate downstream field. This is because liquor splashing and
carryover can lead to electrical instability. Using the latest control technology, however, means that the power to the TRs can be automatically
reduced so as to eliminate the need to electrically isolate the field being
washed down.
Although the need to lead the liquor to a position outside the influence
of the field is not as critical with this approach, as the field is not effectively
energised during the wash-down period, there is a finite time to be
considered before full optimum energisation can be re-established. There are
many plants which have been installed using this form of wash-down, but
as there is always a drop in efficiency during and immediately following a
wash-down, if the unit has to meet a specific maximum emission, then this
loss of performance must be allowed for in the plant sizing.

386

THE WET ELECTROSTATIC PRECIPITATOR

While this loss of efficiency during the wash-down period has a cost
disadvantage in terms of sizing, the approach does mean that the discharge
electrodes and collectors are simultaneously washed, and therefore the
performance following a wash-down is optimised.
With film flow designs, unless off-line wash-down is also incorporated in
the approach, there are no means of removing deposits from the discharge
electrodes except during a plant outage when manual washing is possible.
Deposits on the discharge electrode system can be readily identified by a
gradual deterioration in performance, reducing corona current and, ultimately, increased flashover. In practice it is normal to find both forms of
wash-down system being provided to cater for this event.
13.2.1.3 Full-spray irrigation. To eliminate the off-line and reduced performance scenario associated with the wash-down approach, it was found
that by injecting atomised water droplets into an electric field it was possible
to maintain energisation without running into a flashover situation. Investigations showed that only moderate atomisation was necessary and involute type sprays, having orifii of 2 mm diameter operating at 0.5 bar g
pressure, were more than adequate. These sprays are mounted above the
collectors, typically with the sprays on the duct centre line, such that both
sides are wetted. Some spray droplets also reach the discharge electrodes
thus helping to keep them clean.
Since the droplets are quickly precipitated onto the collectors, it is
necessary to have a multiplicity of sprays along each collector to ensure full
liquid coverage (see Figure 13.1 c). A typical coverage rate is in the order of
4ljmin/m of collector.
Again, because of water draining continuously from both collectors and
discharge electrodes it is necessary to control this flow, such that breakdown, resulting from continuous filaments of liquid being pulled by the
electric field off either electrode system, is minimised. Typically, some form
of 'shedder' is included in the design for this purpose (see Figure 13.1c).
13.3

Discharge electrodes

The form of discharge element found on wet precipitators, because of the


relative ease by which corona normally arises, tends to be sturdier than that
used in the dry precipitator design. The actual profile of discharge electrodes
can be round, square or twisted square. In the case of potential corona
suppression applications and those likely to suffer from high space charge
effects, star or other high emission type electrodes are used (see chapter 3).
The collector height and width limitation generally means that the design
of the discharge system is smaller and simpler, whether it be a weighted wire,
mast or a bedstead frame type of mounting. However, like the collector, the

CASING/HOPPER DESIGN

387

design should allow for safe water shedding to avoid electrical breakdown
problems.
In many designs, with the limitation on collector size, the top frame
supporting the discharge system is of fairly light construction. With full
wash-down, this offers less masking, so there is a more effective coverage.
This avoids 'dry spot' formation and, hence, efficiency and reliability are
improved.
13.4 HT insulators
The wet saturated gas conditions found on many wet precipitator applications exclude the use of the dirty gas insulator design used on some dry
installations. With 'wet conditions', the HT insulators supporting the
discharge electrode frame are normally mounted outside the gas stream in
heated chests. This ensures the insulator temperature is maintained well
above the water dew-point temperature of the gases. Many designs also
incorporate a gas purging system to prevent process gas contacting the
insulator itself.
A typical design would employ either steam or electric heating coils
surrounding the lower end of the insulator in a sealed, thermally insulated
compartment, mounted either above the discharge suspension system or
outboard on the side of the casing, as indicated in Figure 13.2.
The insulator material is typically of high density porcelain or alumina.
This would be used as either lead through bushings or 'flower pots' for
inboard mounting, or as vertical support insulators for outboard designs.
With the outboard design, these usually have hot gas purging to keep the
process gases from entering the chest and depositing material in base or on
the insulator surfaces.
As many wet precipitators are used in potentially explosive gas conditions, such as cleaning of blast furnace gases, the insulators for these
applications are of the lead through type and are normally mounted inboard
and above the discharge system and do not incorporate purging. In addition
to being mounted vertically and above the main gas flow, to avoid
deposition problems on the insulator most use an extended bus ring below
the chest to extend the hot area. The use of a lead through type of design
simplifies the insulator gas sealing requirements which are essential for
explosive/toxic gas applications.
13.5 Casinglhopper design
With the exception of the lower bottom section, the casing design is similar
to that for a dry unit. The major difference lies in the hopper area, which

(a) 'Flower pot' insulator


general application

Discharge
electrode
suspension

Bus ring

through bushing

Bus ring

general application

(c) Outboard post insulator

.., t'.

Heated
air purge

general application. (b) Lead


toxic gas/pressure application. (c) Outboard post insulator - general application.

(b) Lead through bushins


toxic gas/pressure application

Discharge
electrode
suspension

Bus ring

Roof plate

Insulated
heater
chest

Gas seals
(100%)

Clamping
ring

Insulated box

Casing

Figure 13.2 Typical discharge electrode system support insulator arrangements. (a) 'Flower pot' insulator

Roof
plate

.\

Purge in
top plate

Air purge and


control

389

CASING/HOPPER DESIGN

usually takes the form of a shallow trough incorporating a single end


discharge point. A shallow trough, as opposed to a pyramid hopper, means
that the amount and weight of liquid that can be retained is limited, so dead
loads are minimised.
Evacuation of the liquor can be effected by either overflow from a
hydraulic leg (water seal arrangement) or by pumping between high and low
level detectors in a sealed sump, which forms part of the trough. Whilst it
is desirable to operate with low particulate to liquid ratios, if the particulate
is of high density and likely to settle out in the trough, additional liquor is
injected into the base to minimise/prevent settlement. This usually takes the
form of irrigation of the trough surfaces which are most likely to suffer from
sludge build-up.
With the lower temperature of operation most wet precipitators tend to
have no thermal insulation on the casing, except for the most arduous
ambient environments. In this case, the insulation would be to prevent
freezing of the flushing liquor system or to minimise external casing
corrOSIOn.
In the case of pressure type applications, the casing, in addition to the
dead loads of the internals, can be subjected to high process gas pressures.

Access door
Wash down spray pipes
Top frame --7-----4-J...
Spray irrigation
pipes

Inlet
~

Distributors

Bottom frame
Figure 13.3 Typical electrofilter for cleaning high pressure gases from blast furnace (up to
2 atm.).

390

THE WET ELECTROSTATIC PRECIPITATOR

Two typical exmples of this pressure induced stress would be high pressure
blast furnace gas cleaning and oxygen blown steel making processes; both
can produce potentially explosive gas mixtures. For this type of application
the casing takes the form of a horizontal cylindrical pressure vessel which,
whilst designed to withstand the maximum required pressures at the lowest
casing cost, the packing of the internals is not as economic as an ovoid
(Figure 13.3) or flat-sided vessel. There is, however, a cost advantage in that
the space between the outer walls of the precipitator and cylinder shell can
accommodate the quench or scrubber system.

13.6

Water treatment

As considerable quantities of liquor, usually water, are involved on any wet


precipitator installation, regardless of design, it is essential to recycle the
liquor at rates of several thousand literes per min, to minimise operating
costs. By using sprays having large orifii, the liquor only needs to be
relatively clean; concentrations of up to 500 ppm of total suspended solids
(TSS) can be accepted. Lower concentrations are obviously advantageous
in reducing possible blockages within the system, but the use of clean water
or further cleaning has a significant cost disadvantage.
Typically the liquor extracted from the sump is initially cleaned in a 'hot
well'. Here the denser, coarser particles are allowed to settle and are then
removed by scraper chain from the bottom of the well. From there, the
water on large installations, such as blast furnace gas cleaning duties, will
pass to clarifiers where the overflow concentration is reduced to below
100mg/m 3 (100 ppm). Clarifiers are circular ponds of some 20m in diameter. The underflow from the clarifier is usually treated in hydroclones, the
sludge passing to a belt or vacuum filter for further treatment, while the
hydroclone overflow, depdending on its cleanliness, is re-introduced back
into the water circuit. A typical water treatment circuit is illustrated in
Figure 13.4.
Blockages within pipework are generally the result of suspended solids.
If, however, the dissolved solids/pH of the liquor results in the depositon of
calcium salts, a more difficult situation will have arisen as the deposition can
occur anywhere in the system. The total dissolved solids in suspension
(TDS) can be controlled by reacting the circulating liquor with lime or
limestone in a separate vessel to precipitate calcium salts. Once these salts
have been brought out of solution they can then be removed.
The type, size and, hence, cost of the water treatment system is very site
specific. For the complex system indicated, however, the cost is a very
significant proportion of the complete installation cost, at times being more
than 50% of the total.

391

MATERIALS OF CONSTRUCTION

To

From
precip

precip

--.,
,

!.----~- .... ------

Hydroclones :

,,

Clarifier

- -- --~ --- --'- ---- -,

c:i'

Skip

Cooling
tower

Sludge filter
vacuum or belt

Figure 13.4 Typical water treatment plant - schematic.

13.7

Materials of construction

The majority of wet electrostatic precipitators operate at, or close to, the gas
saturation temperature, often with the gas containing relatively high oxygen
levels. This means that the choice of construction material is of paramount
importance if the plant is to have a reasonable life without suffering from
accelerated corrosion.
In applications where gases are deficient in oxygen, such as blast furnace
gas, A36 carbon steel can give many years satisfactory life. However the
liquor circuit may need to be treated with a biocide to eliminate any sulphur
reducing bacteria. These have been known to attack carbon steel, resulting
in expensive component replacement in a very short time frame.
For casings, since corrosion is normally from the inside only, it is
common practice for larger plants to be of a lined/protected carbon steel
construction. This protection has over the years taken many forms;
examples are FRP, glass flake, chlorobutyl rubber, silcone rubber and 'wall
papered' stainless steel or Hastelloy coatings. For wall papering, the base
metal casing has battens of the higher grade material welded to it. On top
of these battens there is a complete overlay of the higher grade material,
typically 1.5 mm thick, which is welded to itself using the alloy as filler to
reduce possible weld corrosion.
With FRP and rubber type coatings it is essential to ensure that the
preparation of the base material is good and the adhesion sound. Any

392

THE WET ELECTROSTATIC PRECIPITATOR

material imperfections, porosity, cracks or penetration through fabrication


or site damage, will let liquor through to the base material and allow
corrosion to proceed unabated. The 'wall papering' approach was developed
to overcome these difficulties, particularly with material porosity and poor
adhesion. If, however, the weld metal is attacked and liquor penetration
occurs then the casing will be damaged, so correct weld preparation and
welding techniques are essential to minimise these risks.
For precipitators where very high concentrations of, say, hydrogen
chloride or fluoride are to be found in the carrier gases, reinforced concrete
casings lined with acid resistant tiles and adhesives have been used to
protect against corrosion. The advent of acid gas legislation has, however,
eliminated the need for these special approaches, since it is now usual to
remove these gases upstream of the precipitator in a separate system.
Because the weight of the internals in a large precipitator can be
significant, the choice of material is critical to the procurement and
fabrication cost. If one takes ordinary carbon steel as unit price, then the
316 series of stainless steel would increase the cost by a factor of 6. The
higher molybdenum 317 series would have a multiplier of around 10, whilst
that of the high nickel alloys, e.g. Hastelloys, would be 15 or more,
depending on grade. Even higher values would result if pure metals were
considered, e.g. titanium. Consequently, the overall cost of a wet precipitator
can vary by an order of magnitude, dependent on the duty, the materials
used, and the corrosion resistance required.
The selection of materials for the internals of a wet precipitator is a
complex process. In order that pitting and crevice corrosion are avoided,
consideration has to be given to the oxygen and chloride concentrations in
the process gas and wash-down liquors. With respect to the liquor, it must
be borne in mind that the pH is controlled by the process technology of the
on-site water treatment plant.
In the context of corrosion resistance the following table shows a
selection of materials and the range of chloride concentrations under which
they can successfully operate. The final choice, however, may well depend
on the supplier's previous experience for similar applications or data from
corrosion test spools. This is always provided there has been sufficient
'up-front' time to establish the data.
Chloride concentration
Below 14000 ppm
Up to 14000 ppm
Up to 24000 ppm
30000 ppm and above

Material selection
316 Series stainless, e.g. 316L
317 Series stainless, e.g. 317LMN
Duplex stainless, e.g. Alloy 255
High Nickel Alloys, e.g. C22jC276

ELECTRICAL ENERGISATION

393

The procurement and fabrication costs increase, as previously indicated,


the more alloyed the material. Unlike the mist type precipitator, which is
dealt with in the following chapter, the wet precipitator does not handle
particularly corrosive fluids, such as strong acids, and therefore alloys can
be used to reduce the overall costs.
One area to be aware of in the use of alloys is in selecting the best method
of fabrication. Ideally, any welding should be carried out using filler of
identical composition. If composite fixings are used, they should be electrically insulated to avoid the formation of electrolytic cells and, of course, the
fittings must be corrosion resistant. Although sacrificial, very heavy carbon
steel structural beams are sometimes used in lieu of lighter but more
expensive alloys. The down-time for replacement of the sacrificial components must be carefully assessed in any life-time costing analysis prior to
making a decision on materials.
The collectors, on some irrigated plate designs, take the form of a
sandwich layer of FRP approximately 75 mm thick, so that a uniform flow
of flushing liquor can be obtained from the water distribution system,
without the need of a double skin metallic plate format.
On intermittent wash-down plant, carbon fibre coated conductive FRP
has been successful and, on fully irrigated plant, glass veiled FRP plates
have been used. PVC, polypropylene and similar materials have also been
applied; these, however, although showing excellent corrosion resistance,
can suffer from thermal distortion and flashover-induced fires. Consequently
they do not have a wide usage.

13.8

Electrical energisation

The system of energisation is similar to that employed on the dry type of


plant. However, as the fields tend to be smaller, the TR set capacity tends
to be lower because the total plate area energised is reduced. The 'wet
operating conditions', however, produce a sIghtly higher operating voltage
and corona current than would be found on dry plants. For precipitators
handling potentially explosive gases, the supply to the field is often via a HT
cable, but for other wet precipitators, the feed can be either HT cable or bus
duct connection.
With the intermittent full wash-down form of plant the HT and washdown valves are typically electrically interlocked, such that during washdown the TR is switched off or operates at a very reduced power. Power is
only slowly returned to optimum after a programmed time delay, any
instability arising in the adjacent fields being catered for by the TR
automatic voltage control systems.

394

THE WET ELECTROSTATIC PRECIPITATOR

In the case of potentially explosive gases, the gas stream is monitored by


an oxygen analyser which, if a certain value is exceeded, automatically
switches otT the TR power supplies and earths down the fields. This action
prevents any residual charge on the electrode system, or electrical spark,
initiating an explosion.

13.9 Typical applications of wet precipitators


In the precipitation industry, the distinction between the wet and mist
precipitator application is sometimes blurred. Often the process gases are at
the saturation temperature and water droplet deposition will occur in both.
For the purpose of this chapter, it is assumed that wet precipitators are
installed to handle gases containing solid phase or adhesive liquid particulates which require a liquid film to remove them. The mist precipitator,
described in the following chapter, basically handles condensed liquid phase
particulates which are self-draining from the collectors. Nothing, however,
is so simple! Many mist precipitator application gases contain some solid
phase material, so it is usual for the mist unit to be fitted with some form
of intermittent wash-down system to periodically remove any agglomerated
material.
13.9.1

Applications in the iron and steel making fields

13.9.1.1 Blast furnace gas cleaning. One of the earliest large-scale applications of wet precipitators was in the cleaning of blast furnace gases.
However, with the advent of high pressure blast furnace practice, the gases
can be satisfactorily cleaned by downstream high energy scrubbers; therefore
the need for wet precipitators has disappeared. Although the precipitation
industry responded with designs capable of handling pressures up to several
bar, the economics of the scrubber for high pressure installations wins out
against the precipitator, as the latter needs electrical power and obviously
has a higher capital cost and space requirement.
Both ovoid and cylindrical casings can be found in the field and both
have pumped liquor evacuation systems because of the pressure situation.
To avoid frequently handling the large quantity of liquid associated with
intermittent wash-down, the high pressure design either incorporates fullspray irrigation or film flow collector systems, backed by intermittent
wash-down for very infrequent cleaning.
In the past, with low pressure applications, i.e. < 1000 mm w.g., the
casings were flat sided and intermittent wash-down was invariably used for
cleaning. The liquor from the wash-down was received in an open sump,
which acted as a hydraulic leg seal.

TYPICAL APPLICATIONS OF WET PRECIPITATORS

395

With gas flow rates often in excess of 100 Nm 3 / S, dust loadings of


1 gjNm 3 and emission requirements of < 10 mgjNm 3 , it is normal to find an
arrangement of several parallel multifield units serving a single furnace.
Each field is normally supplied by its own transformer rectifier set for
operational and performance implications.
The commissioning of any potentially explosive gas plant is complex and
must be carried out by competent engineers. A preset programme of purging
must be implicitly followed to ensure that all areas of the precipitator and
associated ductwork are free of oxygen, i.e. < 0.5% vlv, before electrically
energising the unit. A similar situation occurs in bringing the unit off-line,
when the reverse is practised. In this case, as man access usually follows, the
unit must be completely purged of all process gas. With this in mind, the
designer must ensure that there is no potential for trapping dangerous gas
in pockets within the casing.
The need to arrange for insulators and access doors to be adequately
sealed is of prime importance and the provision of explosion relief doors on
the lower pressure applications must be considered. The risk on modern
high pressure applications is significantly reduced and explosion relief doors
are not normally supplied.
13.9.1.2

Cleaning of gases from oxygen steel making

(A) Ladle processes. Mention has already been made of the use of wet
precipitators for oxygen blown ladle steel making, employing controlled or
suppressed combustion. In this case the CO- and H 2 -rich gases are drawn
off the central section of the ladle hood and are treated in a cylindrical
shelled wet precipitator usually having high energy wet scrubbers in the
space between the collector side walls and the vessel shell. In addition to the
'special design', the control of the total extraction system is critical to the
safe operation of such an installation. A typical BOF blowing cycle would
be 3 min for scrap and hot metal charging of the ladle, 2 min reheat, 10 min
oxygen lancing, 2 min for sampling/alloying, possibly a further short period
of oxygen lancing and then 2 min pouring of the finished steel. During this
cycle only some 8/9 min produces gas which is cleaned by the wet
precipitator before passing to the gas holders. At other times the gas, after
scrubbing, passes through flare stacks to the atmosphere; hence, the timing
and sequence of valve operation is critical to the safe operation of the
system.
(B) Secondary BOS gas cleaning. The gases arising during suppressed
combustion are only partially cleaned by the high energy scrubbers during
normal operation and contain up to 100 mgjNm 3 of particulates. This gas
is stored in gas holders and before further use is normally SUbjected to
further treatment to achieve a cleanliness of better than 10 mgjNm 3 .

396

THE WET ELECTROSTATIC PRECIPITATOR

This cleaning is carried out by wet precipitators located downstream of


the gas holder. As the pressure is less than 1000 mm w.g. the precipitator
casing is normally flat sided. Even though the lower pressure simplifies
sealing of all casing penetrations, the same basic design principles apply
because of the potential risk of explosive gas concentrations due to possible
air ingress. To minimise this, the units are fitted with oxygen monitoring
systems which, in the event of detecting oxygen levels in excess of say, 1%
v/v, will switch off the HT and earth down the fields to reduce any
possibility of flashover triggering off an explosion.
Again because of the explosive nature of the gases, a strict purging and
operating procedure has to be followed to ensure the 'safe operation' of such
plant.
These units can either be of the intermittent wash-down or irrigated type.
The usual inlet dust concentration is < 100 mg/Nm 3 , so the frequency of
wash-down is very much reduced compared with the blast furnace gas
cleaning plant, where inlet loadings of 1000 mgjNm 3 are commonly encountered.
(C) Electric arc furnace gases. A number of wet precipitator installations
can be found on electric arc furnace plants. These tend to treat both the
primary gases leaving the furnace together with the secondary emission
gases contained by hoods above the furnace. As the secondary gases form
the largest gas load, any CO or Hz generated from the smelting and not
burnt off is heavily diluted; therefore, there is little risk of explosive gas
mixtures being generated. This simplifies the design and construction of the
precipitator in spite of the oxygen lancing situation.
13.9.1.3 Scarfing and scalping operations. To remove scale and other
impurities from the surface of steel billets prior to rolling and further
production operations, they are normally subjected to scarfing or scalping.
In this process an oxy-hydrogen burner is used to remove the top surface of
the billet. With the high water content of the combustion gases and the low
temperatures created by high levels of excess air, the traditional system was
to use wet precipitators. The modern approach, however, to avoid the water
poIIution aspect, is to use dry precipitators for this application.
With the low suctions used to capture the fume, the wet precipitator is
of flat-sided construction and it is normal for these units to have intermittent wash-down. The process is such that the scarfing usuaIIy occupies a
time of no more than 20 min and then several minutes elapse while changing
or turning the billet. The wash-down usuaIIy takes place during this down
period, so that the precipitator performance is always optimised for the
commencement of the fresh biIIet face.
Depending on the quality and chloride content of the wash-down liquor,
corrosion risks in the case of these units in spite of the high oxygen levels

TYPICAL APPLICATIONS OF WET PRECIPITATORS

397

can be handled either with low grade stainless, or by using heavy gauge
sacrificial replaceable components.
13.9.2

Applications in the chemical industries

There are numerous applications of wet precipitators to be found in the


chemical industries. These units generally follow scrubbers for the polishing
and removal of toxics before discharge to atmosphere. Some are used,
however, for super cleaning of process gases, e.g. chlorine from electrolytic
cells or ethylene from oil refinery operations.
The following indicates some of the typical applications. Whilst it is not
intended to be a full listing, it serves to illustrate the more unusual
applications.
13.9.2.1 Lead refinery operations. Because of the high toxicity and fine
particle sizing of condensed lead fume, strict limits are laid down with regard
to discharge levels and, in many operations, the wet precipitator is used to
remove and polish the exit gases from scrubbers.
The size of these units tends to be fairly small and they are positioned on
the downstream side of the scrubber fan. Consequently they have flat sides
and, as polishers, have either one or two fields. In order to maintain
optimum efficiency these units are normally spray irrigated to avoid
down-time periods for cleaning.
Generally the gases and liquors can have high acidic components, such
as chlorides, so the internals are normally constructed from high grade
stainless, whilst the casing, if carbon steel, is coated with a rubber lining.
As the gases being treated are largely combustion gases plus air, there are
no special requirements, except the normal safety precautions necessary for
the operation of all precipitator installations.
13.9.2.2 Cleaning of chlorine gases from electrolytic cells. Although this
was a special application, the construction details are of interest. The unit
was of a vertical tube type, where the collector tubes were fabricated from
hard rubber, ebonite, and the whole of the internal casing was coated with
rubber. The discharge electrodes were of titanium weighted wire format.
These measures were necessary to protect against chlorine gas corrosion.
The collectors were film flow irrigated to remove any deposits carried over
with the chlorine. With the chlorine gas, strict safety procedures were
required for the safe operation of the plant.
13.9.2.3 Ethylene gas cleaning. The type of wet precipitator used for this
duty is virtually identical to that used for high pressure blast furnace gas
cleaning, so the plant construction details are similar to those described in
13.9.1.1.

398

THE WET ELECTROSTATIC PRECIPITATOR

Because the precipitator is usually located within an oil refinery, special


features, particularly those associated with the electrics, have to be incorporated to protect against fire and possible explosion hazards. In all other
aspects, the operation is the same as for other forms of wet precipitator
handling explosive gases. The electrical safety features are fairly standard,
but increase the capital costs of the installation.
13.9.3

Applications following acid gas scrubbers

With the recent interest in air toxics, wet precipitators are being applied as
polishers and demisters following acid gas scrubber installations on many
processes. These include the removal of S03 mist from power station
gypsum FGD plants, ammonia slip and salts from gaseous ammonia or
ammonia liquor injection systems used for DeNOx and DeSOx installations.
13.9.3.1 Power station applications. Because the gases from a gypsum
FGD plant leave the scrubber saturated, the application of wet precipitation
is ideal to remove trace toxic condensed vapours and S03 mist which pass
through the scrubber system. The gas flow rates on this type of application
can be up to 1000 Nm 3/S and, as such, the design and construction details
are of paramount importance, not only to give performance reliability, but
also to produce a long life at the lowest capital and maintenance costs. Most
power plant have half lives in excess of 20 years, with only short maintenance outages every 4 years or so.
The materials of construction for this type of application tend to follow
the guidelines listed in the section on materials (13.7) and are dependent on
the quality of the liquor, mainly its chloride content. Ideally the wash-down
should be of the fully irrigated or film flow type to limit the liquor load on
the common precipitator/wet scrubber system.
On installations which fire high sulphur petroleum coke and other
residues, high acid mist concentrations will be encountered giving rise to a
more hazardous corrosion scenario. The construction materials on one
particular installation have composite FRP film flow collectors with a
Hastelloy discharge electrode system contained within a FRP solid casing.
This approach, having three such series fields to achieve S03 emissions of
less than 5 mg/Nm 3, makes such installations expensive.
13.9.3.2 Gaseous ammonia and ammonia liquor scrubber installations. In
this type of installation the gas flow rates are high and, consequently, to
contain capital costs, the materials used for the internals are selected
depending on the quality of the liquors to be handled. These are normally
high in ammonium salts, sulphites/chlorides, depending on the process being

TYPICAL APPLICATIONS OF WET PRECIPITATORS

399

considered. A general material used is either a carbon fibre coated FRP for
the collectors or, if film flow is being used, straight veiled FRP.
For the former, the electrical resistivity of the deposits of ammonium
salts/water are sufficient to allow intermittent wash-down to be used for
cleaning purposes. However, in both it is necessary to adopt intermittent
wash-down to ensure cleanliness of the discharge system and to remove any
agglomerates from the collector which could lead to flashover damage to
the FRP. If the chloride levels in the wash-down liquor can be controlled,
it is possible to use stainless 317L type steel for both collectors and
discharge electrodes. Normally the precipitator is arranged in the circuit
such that the casing can be of flat-sided construction in order to minimise
cost.
With the large gas volumes being handled from, say, an iron ore sinter
plant or a power station, it is usual to find the plant comprising a number
of multifield parallel precipitators treating the gases from a process unit.
13.9.4

Incineration-type processes

With the growing concern regarding the emISSIOn of heavy metals and
generally PmIo-type particles, the processes giving rise to the highest
concentrations are those associated with municipal, clinical, sewage sludge
and hazardous waste incinerators. The size of these units is generally much
smaller than other industrial processes, but since the feed material can
contain significant quantities of heavy metal compounds, there is a growing
need to 'polish' the outlet gases to comply with legislation. This is particularly so for the gases following wet scrubbers which have been traditionally
used to reduce the acid gas constituents to an acceptable level. The scrubber
in cooling the gases condenses the volatile heavy metal compounds which
are then readily removable by a wet precipitator.
To ease the problem of cleaning the water/liquor used by the scrubbers,
it is normal for the gases to be precleaned before they arrive at the scrubber.
This precleaning usually implies that the gases contain very few particulates
and any subsequent condensation material is too small to be removed by
the installed low pressure wet scrubbers. With ever-reducing emission
legislation, it is now normal practice to find wet precipitator retrofits behind
scrubbers on incineration plant to reduce the heavy metal and Pm lO
particulates to the required emission level.
These wet precipitators are fairly standard in terms of design and
construction. Some, however, employ corrosion resistant type materials if
the precipitator is to use the scrubber blow-down liquor for wash-down
purposes, the materials of construction being dependent on the chemical
analysis of the liquor. The precipitators themselves are fairly small and have
either one or two fields or zones, depending on the required performance
levels, which can vary from 90 to 99 + % in terms of efficiency. The final

400

THE WET ELECTROSTATIC PRECIPITATOR

liquid effluent from these precipitators, being rich in terms of heavy metals,
needs careful processing to avoid increased water pollution, which in many
instances is a more difficult problem than the air discharges.
13.9.5

HAC discharges

There are a number of processes, associated with oil refinery type applications, which fire waste oil products giving rise to a characteristic blue haze
discharge from the stack. This discharge, although small in terms of mass
concentration, is very visible and results from the condensation of hydrocarbon radicals on cooling with atmospheric air. The blue haze is synonymous
with particles in the submicron range, typically 0.15 11m diameter, which can
be readily collected by a standard wet precipitator.
The total gas cleaning plant normally comprises an upstream scrubber/
quench system to control the acid gases, followed by a wet precipitator to
control these HAC emissions.
13.9.6

Glass manufacturing

In the production of glass, the feed materials are smelted in a furnace at a


temperature well in excess of 1200 DC and many impurities and alloying
additions are volatilised from the furnace to be carried forward with the
waste gases. During the production of leaded crystal glass, in particular, a
high proportion of the added lead compounds are volatilised, and with the
legislation regarding lead emissions, wet precipitators are used to polish the
final discharge gases. In addition to the lead, other volatile materials, e.g.
boron, are also captured by this unit.
13.9.7

Other applications

The foregoing sections only indicate some of the more usual applications
and are by no means complete. It is safe to comment that ignoring cost, size
for size, the wet precipitator installation is generally superior performancewise to the dry unit. The wet precipitator has no re-entrainment problems
and resistivity effects are negated by the water film so that achieving low
outlet concentrations is easier. With the reduced operating temperature the
electrical operation is optimised and the gas flow is lower, hence, size is
minimised. Commercially, however, one must consider the higher cost of the
wet plant, particularly if corrosion-resistant materials are employed, the cost
of water and its clean-up, together with the more difficult disposal of a
sludge rather than a dry dust.
The final choice of wet versus dry precipitation must be based on a careful
life-time costing of both systems taking all the above into account. For some

TYPICAL APPLICATIONS OF WET PRECIPITATORS

401

applications it is not possible to consider dry precipitation and in this case


it is important to optimise the costs of the wet unit.
Even so, some other typical applications for which wet precipitators have
been used, as both main cleaning and/or polishers, can be summarised as
follows:

non-ferrous smelters
paper pulp industries
carbon black processing
carbon anode baking furnaces
fibreglass curing oven discharges
vehicular automatic spray paint booths.

14

The mist precipitator: design and applications


K.R. PARKER

14.1

Introduction

The electrostatic precipitator design used for the collection of condensed


liquid particulates is similar to the wet precipitator described in chapter 13.
The main difference lies in the frequency of water washing since the
condensed liquid droplets are self-draining from the electrode system, and
hence, the possibility of troublesome build-up is significantly reduced.
In most industrial processes which give rise to condensed liquid/mist
droplets, there is invariably some solid matter which is carried over to the
precipitator, and for this reason the mist precipitator is usually fitted with
some form of wash-down system. Generally, the quantity of solid material
is only a few percent of the total, so the wash-down is infrequent and only
applied if there is evidence of unstable electrical operation, for example, a
reduction in the operating voltage or corona current, arising because of
deposition of solid material reducing electrode clearances.
For some applications, the carrier gases may be above or only close to
the water saturation temperature, so the precipitator will be collecting
undiluted product in liquid form. In others, the precipitator is preceded by
a scrubber or quench system to ensure the maximum product condensation;
in these circumstances, the mist unit will collect both water and condensation products together; hence, there could be a dilution effect on the
collected products where the liquids are miscible.
Although the operating temperature and dew-point condition does not
effect the precipitation duty, it can impact on the materials of construction;
for example, in the case of acid mist precipitators installed on a contact
process sulphuric acid plant, the acid mist can be oleum, S03 in sulphuric
acid, which is a highly concentrated acid, while on other plants operating at
water-dew point, the acid can be diluted because of the simultaneous
collection of water droplets.
The true mist precipitator collecting condensed liquid particulates, unlike
a wet precipitator used for water demisting following, for example, a gypsum
scrubber on a power station, tends to be relatively small and handles gas
volumes well below 100 Nm 3 /s, usually 10-20 Nm 3 /s. This generally means
that the cost of using sophisticated alloys or pure metals for corrosionresistant precipitator internals is more acceptable as the total weight of
component material is low.

INTRODUCTION

403

Although the mist precipitator only requires infrequent wash-down to


ensure the plant is clean, there are designs, using FRP as a corrosionresistant material, which have film flow collector plates to ensure material
conductivity and also to remove any collected particulates to minimise the
possibility of electrical flashover which may damage the FRP.
In practice, the actual number of true mist precipitator applications is
relatively small, although, since the wet precipitator is used for demisting
(water droplet carryover) on a large number of installations, the distinction
between true mist precipitators and those used for water droplet collection,
is often blurred, as previously mentioned.
On a number of applications the precipitator is of the vertical upward
flow type to facilitate rapid draining of the collected droplets, which are free
flowing at the temperature of operation. Horizontal flow plate type units are
installed on other plants and these work equally well and, having horizontal
flow, enable series precipitation fields to be used for increased collection
efficiency, without the complicated ducting runs associated with the vertical
tube approach. The final selection of upward flow, tubular or plate type, or
horizontal flow plate precipitators, is a matter for the supplier and ultimate
user and all types can be found in practice.
A disadvantage of the upward vertical flow is that, if more than one field
is required to achieve the very low emissions now being demanded for
atmospheric discharge, as against the precleaning of the process gas prior
to further usage, double decking of the units or the ducting for series field
connection become expensive. There is a plate form of design, having liquor
flowing over a fabricated FRP thick collector, where the liquor is collected
from the bottom of each collector and led away to safety. This design, as
illustrated in Figure 14.1, means that it is possible to have multifield series
field operation while still maintaing upward vertical flow.
With the traditional form of tubular unit there is a disadvantage in that
only one side of the tube is used for collection. Although the collecting
efficiency, plate area for plate area, is higher, as a result of the better electric
field distribution, the overall cost tends to be greater than other forms of
plant.
To retain the high strength of the cylindrical casing, synonymous with
the tube-type unit, a concentric ring collector approach was developed and
is widely used. In fact, for small gas flows there is no reason why this form
of plant should not be used more extensively for general wet precipitator
duties. In this design, the collectors take the form of accurately fabricated
concentric cylinders having the discharge electrodes mounted from a concentric ring top and bottom locating frame arrangement.
An alternative to the circular tube is the hexagonal and the plate type of
collector. These and the concentric ring form of collector do not have as
good an electric field distribution as the tube type and hence require slightly
longer contact times to achieve the same efficiency. The cross-section of

404

THE MIST PRECIPITATOR

Figure 14.1 Arrangement of vertical flow precipitator (single field). US Patent No. 4,362,538
1982.

these collectors and the discharge electrode location are illustrated in Figure
14.2.
Gas distribution within the field area is important to maximise the
efficiency, as shown in chapter 5. With the tube form of collector, the
distribution is critical, since once the gas enters the tube it cannot redistribute itself, as can occur to some extent, in plate-type units. As most mist
units have mainly horizontal single inlet and outlet duct connections, their
location and orientation on the unit is important and splitters may be
necessary in the outlet area to prevent preferential one-sided pulling of the
bulk flow.
With the traditional tube unit, illustrated in Figure 14.3, the collectors
are held between an upper header plate and a spacer at the bottom. The gas
inlet connection is usually between the plates and the initial distribution is
achieved by the tubes themselves, although in most instances for ultimate
flow control, there is a perforated plate arrangement in the bottom tube
spacer area. Not having a tube nest in either the concentric ring or

INTRODUCTION

Circuiar tube

405

Concentric ring

X X X X X

X X X X X X
X X X X X X
X X X X X X
X X X X X X
X X X X X X
Hexagonal tube

Parallel plate

Figure 14.2 Cross-section through different collector arrangements (x electrode; 0 stabilizer).

hexagonal tube form of unit in which to initially turn and distribute the
gases, the gas distribution system is more complex and is achieved by
splitters, etc., in the area beneath the collectors.
One type of gas distributor is indicated in Figure 14.4; this comprises a
horizontal tube running across the bottom of the unit. This tube contains
slots of differing widths by which to control the gas flow. Although the mist
unit is relatively small and not much time would be involved in site gas
distribution testing, if the flow requires to be corrected, the materials of
construction mean that modifications are complicated. For this reason a
simple large-scale model can pay dividends, both in terms of cost and time,
particularly, for example, if the precipitator has been in service collecting
acid, where, prior to any access into the unit, it must be thoroughly washed
and all acid removed and neutralised.
As previously indicated, the actual design of the mist precipitator is
similar in many details to that previously outlined for the wet plant. The
mist precipitator design and form will be best described by reference to the
following typical applications.

406

THE MIST PRECIPITATOR

Porcelain styrenated terminal


HTcable
Interlocked inspection
door (far side)
Steam coil
6000
interlocked
manhole

Gas outlet
Discharge electrode
wires

---------===:::nEiiIV
-------~~~~~!li~;::::s~

-------J---lJr'.J

-------LJ=T'-irrrr=TOn

Gas inlet

1500 tubes
(Receiver electrodes)

Perforated
tube plate

6000 interlocked ~
manhole
~

Tar outlet

Seal pot

Figure 14.3 Cross-section of tube-type detarrer.

407

APPLICATIONS OF MIST PRECIPITATORS

Detarrer casing

Collector system

Discharge system
Perforated plate
Slot position
and width from
model test

Gas inlet

Figure 14.4 Slotted tube distribution system.

14.2 Applications of mist precipitators


14.2.1

Collection of sulphuric acid mist

The control of acid mist from metallurgical smelters was probably the first
real attempt to use electrostatic precipitation industrially. The initial commercial installation, following pilot investigations by Dr F. G. Cottrell (see
chapter 2) used a mist precipitator to combat atmospheric pollution from
smelters around the San Francisco basin.
The mean, or d so , particle sizing of sulphuric acid mist is around 0.15 ,urn,
and although relatively soluble in dilute acid, there is the difficulty of
actually contacting the droplets, even in venturi-type scrubbers. The electrostatic precipitator, as indicated earlier, is ideal for this particle size range
and there are numerous mist precipitators in operation around the world
successfully collecting sulphuric acid mists, with efficiencies well in excess of
99%.
In addition to metallurgical smelters, particularly those roasting sulphide
ores, the mist precipitator finds widespread use in the manufacture of
sulphuric acid. Depending on the process design and cleaning need, there
are two situations which arise; the first is where the gases are spray cooled

408

THE MIST PRECIPITATOR

toward their adiabatic saturation temperature and the second where the gas
temperature is further reduced by contact cooling (super cooled) to ensure
maximum condensation. The boiling point of S03 is some 44C, as
compared with sulphuric acid at 360C, but with the high saturation
conditions there is always acid mist present in the cooled gas stream.
As with all mist precipitators, the material collected is usually free flowing
from the collectors and both vertical and horizontal flow units can be found.
The precipitated acid mist, which can vary as a collected material from
dilute to very strong, means that the choice of material for design and
fabrication is very important as regards the final installation costs.
To cater for the wide range of acid concentrations, antimonial lead was
a traditional constructional material and is still used for the internal
components. Nowadays, the problem of exposure to lead fumes arising
during plant fabrication and the dearth of experienced lead burners has
meant that other materials are being more extensively considered for this
application. High nickel and chrome alloys, for example Hastelloys, have
been used, but suffer a cost disadvantage against lead fabrication.
Casings have been constructed from lead-lined A36-type steel, acid brick
construction and more recently straight FRP. Where the precipitator is
installed behind a venturi scrubber, but upstream of a fan, it is imperative
that if lined mild steel is used for the casing, then the lining is pressure relief
vented, to prevent suction damaging or pulling off the lining. The advantage
therefore rests with FRP construction but, even so, it has to be thick and
robust to withstand the process suction, which can exceed 1000 mm w.g.
The electrical energisation of a mist precipitator is typically on a unit
basis, where one rectifier set feeds a single field. Modern installations have
full solid state controls fitted with AVC units, which can be connected to
the plant control room computer, as explained in chapter 8.
The gas composition on most acid mist applications contains a relatively
high moisture content plus sulphur dioxide, together with a large quantity
of submicron fume. This means that it is necessary not only to use high
emission type electrodes, e.g. star lead, but also a higher rated voltage
rectifier unit to overcome space charge effects and provide sufficient field to
enable the precipitator to perform satisfactorily. For a 250 mm diameter
tube unit, having star lead discharge electrodes, a typical operating voltage
would be in the order of 75 kV, giving a field of 6 kVjcm, and a current
density approaching 1.0 mAjm2 of collector surface.
The discharge suspension insulators for an acid mist type precipitator are
similar to those described in chapter 13 for the wet precipitator and
illustrated in Figure 14.5. The main difference lies in the need to maintain a
higher temperature in the surrounding area or an increased purge air
temperature because of the higher dew-point situation.
If acid should become deposited, its conductivity, even at low concentrations, and the high boiling point, usually means that the insulator will

Nitrogen

(b) Oil filled cable feed input, steam coil heating,


double shedded bushing

(c) Oil filled, steam coil heating,


lower end shedded

Figure 14.5 Typical suspension insulator arrangements. Note - components will vary with supplier. (a) Nitrogen filled, steam chest heating,
double shedded bushing. (b) Oil filled, cable feed input, steam coil heating, double shedded bushing. (c) Oil filled, steam coil heating, lower
end shedded.

(a) Nitrogen filled, steam chest heating,


double shedded bushing

[_LP~

,,~--

410

THE MIST PRECIPITATOR

99.9
99.8
99.7
199.5

gal

99

98
97
95

90~~--~---L--~--~--~~

Total contact time in field (seconds)


Figure 14.6 Acid mist typical design curve for tubular-type mist precipitators.

quickly fail electrically, or perhaps mechanically, as a result of power arcing


if the Ave does not step back to prevent it. In either case, the performance and efficiency will fall dramatically, particularly with a single field
configuration.
As indicated earlier, the collection of mists is readily precipitated and is
not subject to rapping re-entrainment. This enables higher bulk velocities to
be used advantageously and units with mean velocities of 2 mls and above
can still obtain efficiencies well in excess of 99%.
For tube-type units, although each manufacturer will have a preferred
modular configuration, the precipitator sizing is based for convenience on a
gas retention or contact time, rather than a plate area or migration velocity.
A typical design curve is shown in Figure 14.6. This must be judiciously
used, since each supplier will have some variation associated with his
particular design.
The acid which is deposited on the collectors flows down and either
drips or is led into the base of the unit, which acts as both storage and
hydraulic seal. On some installations suctions in excess of 1000 mm w.g. can
be experienced. The acid in the base can be used as wash-down liquor if
required, or it can be pumped to another part of the installation for
reprocessing.

APPLICATIONS OF MIST PRECIPITATORS

14.2.2

411

Gas detarring

Coal tar is one of the byproducts resulting from the distillation of carbonaceous fuels, particularly during the production of coke used in iron making,
and from synthetic gas production plants. Tar is a generic term covering a
wide range of organic materials having molecular weights from 100 upwards.
These tars are initially volatilised from the coal at the high temperature
of processing, and leave the coke ovens or retorts as vapour which
condenses following cooling and initial scrubbing of the hot gases. Dependent on the molecular weight, the tar condenses to give particles in the size
range 0.1 to 100 p.m.
The tar, from a typical coke oven installation (Figure 14.7) can, depending on the coal's volatile content, reach concentrations of 15 g/Nm 3 or more
and unless this can be satisfactorily removed, it causes build-up and fouling
problems in downstream equipment, such as ammonia scrubbers, naphthalene recovery, gas purifying plant and the final burners themselves.
The removal of the tar, because of the wide particle size range, is difficult;
high efficiency scrubbers can only achieve some 60% removal and, for this
reason, electrostatic precipitators, called detarrers, are used and can readily
achieve efficiencies in excess of 99%.
The position of the detarrer in a coke oven installation is normally on
the downstream side of the compressor used to drive the gas through the
byproduct plant itself, as indicated in Figure 14.7. Although ideally an
upstream location would prevent fouling of the compressor, the risk of air
ingress forming an explosive mixture within the precipitator means that the
downstream location is preferred. The gas passing the compressor is
normally at a high enough temperature to minimise condensation of the tar,
so little difficulty is encountered by the downstream positioning.
Detarrers have been used for this particular application since the early
1930s following some years of successful operation cleaning 'Towns Gas'. In
spite of its long usage, the basic design approach has not significantly
changed, except for the HT equipment, which is now solid state and is fitted
with an automatic voltage control unit of some form to fully optimise
performance.
One of the most common forms of detarrer to be found is still the vertical
tube and coaxial wire type of unit. As the size of the coke oven has increased,
so has the number of tubes required for a specific contact time. As an
economic alternative to a multiplicity of small units, the very much larger
later designs are fitted with three suspension insulators to support the
discharge electrode frame, as illustrated in Figure 14.8. Having three
insulators, instead of one, not only reduces their specific loadings, but also
helps stabilise the frame. This minimises any tendency for the frame to
swing, which would be detrimental to optimum performance, since any
electrode movement would reduce the electrical operating conditions.

Tar

Waste water

Ammonia
distillation
Ammonia

Sulphur

Holmes
Stretford
plant

H2 S washer

Benzene
Toluene
Xylene
Naphthalene

Benzole
distillation

Figure 14.7 Schematic flow diagram of a coal gas processing plant (courtesy Lodge Sturtevant Ltd).

Tarlliquor
separator

Ammonia
washer

APPLICATIONS OF MIST PRECIPITATORS

413

Generally, the low oxygen concentration of coke oven gas minimises


corrosion risks in spite of the warm humid conditions within the detarrer
itself. This means that ordinary A36, mild steel fabrication can provide an
acceptable life for the unit, particularly as it is normal practice during plant
start-up and purging to pass the gas through the unit for some time, so that
sufficient tar is deposited on the internal faces of the unit to form a
self-protective coating.
Although the risk of normal corrosion is reduced, there have been
incidents of accelerated corrosion being experienced on detarrer precipitators as the result of 'sulphate reducing bacteria' being present in the wash
and scrubbing liquors. It is essential therefore that the liquors be treated
with biocides to eliminate this risk.
The casing is normally fabricated from 10 or 12 mm plate which provides
a corrosion allowance for a long maintenance-free life. To facilitate transportation of the larger site erected installations, the casing is normally split
into several sections. The operating pressure for a detarrer is fairly low at
0.25 bar g, so does not present too much of a design problem and both
cylindrical and flat-sided vessels can be found.
Other mechanically loaded components, such as the top and bottom
discharge electrode frames and gas distribution splitters, etc., are fabricated
from 6 or 8 mm plate, and although subject to gas wetting on both faces,
are thick enough so that corrosion is not a significant concern. The
suspension rods carrying the discharge frame are usually 38- or 45-mm
diameter rods depending on the design and supplier. To avoid tar reaching
and impacting on the insulators, 'Chinamans' Hats' are usually installed on
the suspension rods beneath the bus ring (Figure 14.8).
The collectors in a vertical tube detarrer, are fabricated from 3 or 4 mm
plate with a top header plate 10 or 12 mm thick; this thickness plate is
necessary since most of its mechanical strength has been removed because
of the holes. Concentric ring or conventional horizontal-type collectors are
usually fabricated from 6 or 8 mm plate, which allows sufficient flatness and
mechanical rigidity to be self-supporting and the collectors themselves only
need to be located in position.
The discharge electrodes in a detarrer are normally small in cross-section
to minimise corona suppression effects, and are usually made from ferritic
or austenitic stainless material for long life. Although ordinary mild steel can
be used, these electrodes should then be considered as replaceable items. For
tubular-type units the electrodes are of the weighted wire type and on plate
plants the electrode can be located and tensioned in a frame.
The base of the detarrer, whether it be circular or rectangular, is normally
sloping and the bottom is heated to reduce the viscosity of the tar so that
it is free flowing. The tar is generally evacuated from the vessel through a
hydraulic leg. Any vessel stiffeners required to withstand the design pressure
would be located on the outside faces to ensure unimpeded flow of the
collected products.

Suspension lead
through insulator
EHTcabie
from TR set

'Chinaman's haf

Gas
flow

Inlet
nozzle

Figure 14.8 Large tube-type detarrer - 3 point suspension.

APPLICATIONS OF MIST PRECIPITATORS

415

With the gas being both toxic and inflammable it is imperative that all
access doors, duct connections and casing penetrations are capable of being
sealed 'gas tight'. The insulators used for supporting the top discharge
electrode frame are usually of the bushing or lead through type, shown in
Figure 14.5. To enable the connection of the HT, one of the insulator
support boxes is fitted with an oil or gas filled extension, carrying a
bushing-type insulator, to which the HT cable is connected. Inert nitrogen
containment of the 'none gas' side of the insulator is also possible, where a
change in the nitrogen pressure would indicate that an insulator has failed.
The HT equipment for detarring is similar to that used on other
forms of precipitator and generally operates at around a field strength of
4.5kV/cm and a current density of up to 1.0mA/m2. Again because of the
none re-entrainment scenario, gas velocities can be higher than for dry
precipitators and sizing to meet a specific efficiency can be conveniently
based on contact time, similar to the acid mist curve shown in Figure 14.6.
Because of the potential explosive nature of the coke oven gas when
mixed with air, the start-up and shut-down of the unit is critical. A strict
purging sequence, as described in chapter 11, must be carried out before
energising the plant to ensure that all air has been expelled. With shut-down,
a similar but opposite purging routine is necessary, to ensure there is no gas
remaining inside the unit, particularly if man access is required. Any
remaining tar deposits can be steamed off, or a naphtha solvent can be used
to wash down the unit before air purging commences.
14.2.3

Collection of radioactive particles

In common with all operations in the nuclear industry, all processing is


carried out in a fully contained negative pressure environment to ensure
retention and control of all products. The gases extracted from the containment area have then to be filtered through very high efficiency HEPA filters
before they can be discharged. Although the HEPA filter, being a positive
membrane device, is extremely effective in retaining all particle sizes, its
pressure drop increases with the material collected and, hence, has a definite
useful life before it needs to be replaced to ensure that gas throughout is
maintained.
The main problem facing the nuclear industry, is not so much in terms
of the cost of replacing the filter, as much as what to do with the
contaminated used unit. The only recognised or acceptable method is to
store the exposed filter in a safe containment area which could be better
utilised. To increase the useful life of the HEPA filter and to reduce handling
and storage difficulties when contaminated, mist precipitators have been
fitted upstream to reduce its loading by at least 99%.
In order that the precipitator functions satisfactorily, the inlet gases must
be below their saturation temperature to produce liquid droplets, which

416

THE MIST PRECIPITATOR

then act as condensation nuclei for the radioactive particles. In the field area
of the precipitator the operating corona current must be maintained high
enough to neutralise and recharge any positive ions or particles so they can
be precipitated in the normal manner.
The precipitator used for this application is of a fairly standard approach,
but there are certain differences as a result of the nature of the application.
To avoid the unwanted collection and hence build-up of radioactive
material, there are no horizontal surfaces in the design, the gas typically
entering from beneath the field area. The inlet duct is fitted with a water
seal, not only for isolation purposes, but also to collect the precipitated
products in a concentrated form, i.e. a low volume for 'safe' containment.
To ensure a long maintenance-free life, special attention is paid to the
component fabrication; not only is the material corrosion-resistant, but all
welding is checked for penetration and any areas where crevice corrosion
can be initiated are eliminated in the design.
The HT equipment, although standard, is considered to have a definite
life and to enable this to be changed or maintained, the rectifier is located
outside the active area and HT cabling is used to connect the unit
electrically. The discharge electrodes, because of the need to possibly
neutralise positive particles, are specially designed to have a higher than
normal corona emission. With essentially an air load, the electrical operating conditions are a field voltage requirement of 6 kV/cm and a corona
current of 1.0 mA/m2 of collector surface.
A number of such plants have been in service for some 25 years and have
performed entirely satisfactorily, justifying their instaIlation costs, in terms
of reduced usage of HEPA filters and their ultimate storage demands.
14.2.4

Other mist precipitator applications

Although the foregoing describes some of the more usual and interesting
applications, the mist precipitator can be found in the manufacture of nitric
acid from the oxidation of ammonia at high pressure, and in producing
phosphoric acid from phosphate rock by means of an electric furnace
followed by oxidation of the phosphorous and hydration to form the acid.
Both applications follow condensation/stripping vessels so the precipitator
collects product plus water droplets as dilute acid phase.
Another application is in the collection of oil mists produced during
operations, such as rolling or spinning, or in the production of glass fibre
blocks used for thermal insulation, where oil and phenolic binder fumes are
evaporated from the furnace during the final processing. The alternative
form of two-stage precipitator, as basicaIly used for 'air cleaning', is often
found on this type of application; however, as the two-stage precipitator is
outside the scope of this publication, these processes wiII not be reviewed.

CONCLUSIONS

417

14.3 Conclusions
Compared with other forms of device that might be used for the collection
of submicron particles the mist precipitator otTers the following advantages:
(1) the ability to handle a wide range of inlet conditions, in terms of

(2)
(3)
(4)
(5)
(6)

temperature, particulate loading and particle size


capable of meeting very high efficiencies
low operating costs for energisation and pumping
low pressure loss < 25 mm w.g.
fully automatic in operation
minimum maintenance requirement.

Finally the use of modern engineering materials in the design and construction ensures a long and trouble-free life.

15

Upgrading of existing precipitator efficiencies


K.R. PARKER and H. KRIGMONT

15A

Modifications/changes to existing plant

With the ever-increasing legislative pressures to reduce discharges from all


sources, the owner/operator of a specific process plant fitted with an
electrostatic precipitator, or any form of particulate collection system, is
faced with the dilemma as to how to improve the performance of his
particular plant to comply with these changes of emission, without necessarily completely replacing the collector and with the minimum down-time
or interruption in process production, while changes are made to the
existing installation.
In addition to improvements required for legislative measures, there are
other instances needing precipitator efficiency enhancement, e.g. a change in
production rate leading to higher particulate carryover, higher temperatures, gas flow rates, plus uncontrolled air inleakage producing severe
volume overloading, an alteration in the fuel supplied to a pulverised fuel
installation, or the existing plant is not in a serviceable condition as a result
of corrosion and mechanical damage.
The approach one must adopt, regardless of the need for improvement,
is similar in all instances. Firstly, one must evaluate the current performance
of the existing unit, to establish the precipitability of the particulates, or
migration velocity, and the current operating volumes, temperatures, dust
loadings, etc. These would be determined from a test or series of tests, which
should follow a planned outage, when not only should the precipitator be
brought back into a good mechanical and electrical condition (see chapter
11) but also the gas distribution should be corrected to the standards
outlined in chapter 5. It is assumed that the precipitator electrics will have
been modified earlier, or during the present outage, to include one of the
latest forms of automatic voltage control, so that the performance is fully
optimised electrically and also rapping re-entrainment has been minimised
prior to the tests.
Assuming the existing precipitator has been optimised as above, the next
step is to determine the actual collection efficiency, using one of the methods
given in chapter 10, under what, hopefully, will be representative of the
future operating conditions. These tests, in addition to determining inlet and
outlet particulate concentrations, must also include full gas velocity, temperature and analytical surveys. The sampling of the gases for dust concentration determinations, should also extract sufficient sample to enable

ASSESSMENT OF REQUIRED PERFORMANCE IMPROVEMENT

419

complete chemical and particle size determinations to be made. In the case


of boiler plant, one should also arrange for the fuel to be analysed and a
record compiled of the actual boiler conditions under which the tests were
performed. On other process plants a log of the plant operating conditions
should be made for future reference.
lSA.l

Assessment of required performance improvement

The information from these tests will not only provide the hopefully updated
plant operating conditions, but will enable the precipitability of the particulates to be determined in the form of a migration velocity/modified migration velocity. This will allow a curve to be drawn of efficiency versus specific
collecting area (SCA), similar to that illustrated in Figure 15A.1, from which
the SCA required for the new efficiency can be established, as will be
illustrated in the following procedure.
After establishing the new/required SCA, the total plate area or contact
time can be calculated for the upgraded plant to meet the required efficiency.
From this, one can decide the economics of providing a complete precipitator of the correct size or adding the additional required surface in series or
parallel configuration. Whether the extension is connected in a series, or in
parallel, to the existing installation depends on various factors relating to
the site and plant operating requirements.
In order to obtain a full understanding of the procedure, the following
example illustrates how the performance data are treated for a boiler plant
precipitator which already meets the UK Large Combustion Plant Directive, i.e. an emission of 50 mg/Nm 3 dry at 6% O 2 , but for economic reasons,
the client wishes to increase the generated output by some 8%. Although in
99.99
99.8
99.6

199.4
>. 99.2
g 99
Q)

'13

ffi

98
96

94
92
90~~~--~--~--~~~~

Figure lSA.l Precipitator performance line. Constant "'k versus specific collecting area.

420

UPGRADING OF EXISTING PRECIPITATOR EFFICIENCIES

the example the existing precipitator meets the Large Combustion Plant
Directive, the same procedure and methodology would be adopted for any
precipitator requiring upgrading for any reason.
For this worked example, it is assumed that the existing fuel supplies will
be maintained over the next 5 years or so, having the following typical
analysis.
Moisture
Ash
Sulphur in coal
Na 2 0 in ash
Vol Matter
Cal Value

9.0%
20.0%
1.42%
0.42%
24.0%
24.0mJjkg

The existing precipitator configuration is as follows: two parallel flows,


with conventional straight inlet and outlet duct connections. Each flow has
three series fields, containing collector plates of 5000 x 11 000 mm, and each
field has 40 ducts at 300-mm centres. The total plate area as installed is
26400m 2 .
For this application, two series of efficiency tests have been carried out,
the first to re-establish the design performance and the second to determine
the precipitator operating conditions under the higher plant output.

Table lSA.l Measured test data


100% MCR rating
Gas flow (am 3 /s)
Gas temp. eC)
Duct pressure. mm w.g. (suction)
Oxygen at inlet (% d v/v)
Moisture at inlet (% d v/v)
Inlet dust cone:
Emission'
Efficiency (%)
Gas velocity (m/s)
Con tact time (s)
Deutsch EMV (cm/s)
Modified Deutsch EMV (cm/s)
SCA (m 2 /m 3 js)

400
140
300

5
8
16.0
50
99.67
1.51
9.9
8.74
5041
66.0

MCR

+ 8% rating
440
150
350
5

18.0
97
99.46
1.67
9.0
8.70
4543
60.0

'Inlet dust loading and emission corrected to mg/Nm 3 d 6% O 2 ,

From the derived modified Deutsch EMVs, one can evaluate the anticipated
efficiency for only two of the three fields in service. This gives the following

PERFORMANCE IMPROVEMENT MODIFICATION OPTIONS

421

figures: MeR operation with two fields only, efficiency 99.1 %, and for the
increased duty, 98.6%. These values have been used to construct the curves
shown in Figure 15A.l, from which the SeA required for the new efficiency
of 99.72% can be read off using the derived modified Deutsch number of
45.43 cm/s. A reason for the slightly lower figure under the proposed future
operating conditions is attributed to the higher operating temperature at the
increased boiler loading.
lSA.2

Performance improvement modification options

Applying the modified Deutsch number, 45.43 cm/s, and new efficiency of
99.72%, directly into the Matts-Ohnfeldt equation [1] produces an SeA of
76.1 m 2 /m 3 /s. This gives a total plate area of 33487 m 2 , i.e. an increase over
the existing area of 7087 m 2 , based on a collector separation of 300 mm.
As a series field extension to the existing precipitator, the additional field
length required would be 4000 mm. The precipitator would then be
19000 mm overall and have a contact time of 11.4 s. Although not essential
if the extension is directly coupled and maintains the same collector height
and spacing, one can assume that the gas distribution will remain satisfactory and the general outline of the precipitator will be unaltered.
Although changing the internals to fit higher collectors is probably not
justified in this case, because of the excellent test results, the required total
plate area could be achieved by removing the roof and fitting collectors
having a height of 13 950 mm. This would reduce the gas velocity to 1.31 m/s
and raise the contact time to 11.4 s. If such an approach was to be
considered, it might be more economic to install only 30 ducts per flow at
400-mm centres, while maintaining the 11.4 s contact time, but in either case,
one must reconsider the gas distribution to fully utilise the taller plant
internals.
The additional plate area could take the form of a parallel unit, which
keeping the same size of collector and number of fields would have 22 ducts
at 300-mm centres, or 17 ducts at 400-mm centres. To meet the overall
efficiency/emission, the flow through the existing unit would have to be
reduced to 347 am 3 /s, which would reduce the gas velocity through the
existing unit to 1.31 mis, with the remainder passing through the parallel
unit again with a contact time of 11.4 s.
A different approach might be to reduce the flow through the existing
precipitator to some 347 am 3 /s and treat the remainder through a parallel
bag filter. Assuming a filtering rate of say 1.4 m/min on a gas flow of
93 am 3 /s would imply a filter media area of some 4000m 2 . This could be
arranged in four sections and pulse jet cleaned. Because of the increased
pressure drop across the bag filter of say 200 mm w.g. total, it would require

422

UPGRADING OF EXISTING PRECIPITATOR EFFICIENCIES

an addition fan of 250 kV A rating to pass the gas through the bag filter and
return the cleaned gas to the upstream side of the main plant ID fans.
In practice, as the bag filter should give a lower emission than 50 mg/
Nm 3 dry at 6% O 2 , the gas volume passing through the bag filter may be
reduced with cost savings in terms of fan power, and the existing precipitator allowed to accept a higher gas volume than 347 am 3 /s, provided the total
emission remains within design 50 mg/Nm 3 limit.
While one might consider replacing the internals by a bag filter, which is
unlikely because of the satisfactory performance of the existing precipitator,
on this unit the available area of the top plates would be insufficient to
support/accommodate around 19000 m 2 of filter material (assuming a
filtering rate of 1.4 m/min), and therefore in this instance such an approach
would be totally uneconomic.
After completing the foregoing analysis and deciding the various options,
one then needs to assess the most economic approach to modify the plant
to comply with legislation. The economics of such an appraisal is, however,
very site specific, so will not be further explored. There are, however, various
scenarios which impact on the economics and need to be considered.

o
Figure 15A.2 Parallel units.

Figure 15A.3 Series units.

ALTERNATIVE SOLUTIONS

423

There are advantages with the parallel arrangement (see Figure 15A.2) in
that the final operating gas velocity through the existing unit is reduced and
the new precipitator, or bag filter, can normally be fully installed without
the main plant being off-line, except for the final break-in, so process
production is maintained.
The series connection (Figure 15A.3), on the other hand, offers an in-line
arrangement, so there is less problem in balancing gas flows and possibly
simpler connection to the existing dedusting system, but its main problem
lies in the down-time for building the extension in most instances, unless it
is erected as a separate downstream unit, when the only outage necessary is
for the final duct connections to be made.
15A.3

15A.3.1

Alternative solutions

Electrical

Reference has previously been made in chapter 8, dealing with electrical


operation of precipitators, as to the effect of applying intermittent energisation or pulse charging to precipitators handling 'difficult' fly ashes. In these
situations the alternative energisation serves to modify the particle charging
to minimise the impact of reverse or back-corona.
For intermittent energisation, as most modern TR controller equipment
incorporates this facility, or can be easily modified, the application of this
approach can be most beneficial in improving the performance under
difficult conditions; again whether or not the final emission will satisfy
present regulations depends on the precipitator design and operating
conditions.
Pulse energisation can have remarkable effects on precipitator improvements [2]; however, not only is the equipment expensive to install, but the
design of the precipitator and operating conditions may limit its general
application. For an operating plant, it would be pertinent to install the
equipment as a single field trial, rather than commit the system on a
permanent basis, to establish if the expenditure on pulse charging equipment
will bring about the required degree of improvement for this particular
application and duty.

15A.3.2 Mechanical changes


For some designs of precipitator installed prior to 1970, it has been feasible
to increase the SCA by up to 15%, but generally 10%, by fitting extended
collector plates and discharge systems into the inter-field area. On others, a
larger SCA increase has been possible by lifting the roof and fitting taller
collector plates within the existing casing. These approaches were normally

424

UPGRADING OF EXISTING PRECIPITATOR EFFICIENCIES

o 250 mm spacing
400 mm spacing

t;.

99.7

l!><.l
c:

;gw 99.5
Q)

99.0

10

11

12 13 14 15
Treatment time (s)

16

17

Figure 15A.4 Improvements in performance as a result of change to 400 mm collector spacing.

associated with a planned internal replacement programme, during which


the latest precipitator improvements, such as gas distribution, rapping and
the latest TR controls, should also be included. Although successful in
improving the overall performance, whether or not the final plant will meet
the required emission level is dependent on whether the total plate area is
sufficient to meet the uprated efficiency.
As both the collector plate height increase or lengthening implies a
complete internal replacement, it may be more economic to consider fitting
out the plant with a collector spacing of 400 mm and save on component
costs. The change to 400 mm spacing needs careful consideration of clearances outside as well as within the field area. Some plants have experienced
voltage limitations in adopting the wide collector spacing approach, because
of insufficient electrical clearance in bus ducts, bus rings, support insulators
and discharge electrode rapping shaft areas, all creating lower than optimum voltage operation. Where the application has been successfully carried
out significant improvements in performance have been achieved as illustrated in Figure 15A.4, from Darby [3], when comparing before and after
effects for the same contact time as a result of the general improvements in
electric field and current distribution.

15B

ISB.l

Precipitator improvements achieved by changing


the electrical resistivity of the particulates
Change of temperature/relative humidity of the gases

The use of water injection ahead of dry precipitators in the cement,


metallurgical and other process fields has been mentioned in chapter 12. The
injection is not only to reduce the temperature and volume of the gases, but
more importantly to increase the relative humidity of the gases, such that
the electrical resistivity of the particles is reduced, as illustrated in Figure
6.10. Theory, coupled with operational experience, has shown that generally
the higher the relative humidity, the better the performance of the precipitator. Generally the quantity of water that can be evaporated is limited
by the incoming temperature, liquid droplet size and available contact
time.
In the collection of particles and fume from a comprehensive sinter strand
dedusting system, precipitator improvements were obtained by increasing
the gas moisture level by only 1.2% vIv. The emissions decreased from
70mg/Nm 3 down to 40mg/Nm 3 . The increase in moisture, on what is
basically a low temperature/entrained air application, was achieved by
spraying water into the hot discharge duct leading to the sinter breaker at
the discharge end of the strand. The effect of the added moisture was to
increase both the operating voltage and current into the precipitator, which
resulted in the efficiency increasing from 99.63% to 99.79%. In this case, the
performance change was achieved without significantly reducing the gas
volume or temperature of the incoming gases.
As an alternative to water injection, to increase the relative humidity, the
same effect can be achieved by cooling the gases by means of an upstream
heat exchanger. The effect of temperature on electrostatic precipitator
performance is, as previously illustrated in Figure 12.1. One advantage of
such an approach is that the gas volume is automatically reduced serving to
increase the specific collection area (SCA).
Investigations have been carried out in Japan, where the gases immediately upstream of a pilot precipitator have been further cooled by means of
a water/gas heat exchanger, rather than a conventional gas/gas exchanger,
to limit the amount of air inleakage (usually around 10% for a conventional
gas/gas exchanger) [4]. The application can be limited in its general
application, however, by the potential re-entrainment of low resistivity
particles and hopper dust handling problems, plus possible acid dew-point
difficulties unless associated with an acid gas absorption system.

426

IMPROVEMENTS BY CHANGING PARTICULATE ELECTRICAL RESISTIVITY

10 14

Advanced system
~

Conventional system
~

Mark
0
t:.

...

90

Coal
C

0
F
G

%S
0.21
0.48
0.27
0.56

%N~O

0.26
0.06
0.28
0.05

100 110 120 130 140 150 160


Gas temperature (0C)

Figure ISB.l In-situ dust resistivity of ESP inlet fly ash.

The results of this investigation, carried out on very difficult imported


fuels, has shown some significant performance improvements when operating at 90C, compared with the conventional temperature of 130 0c. Figure
15B.l, from Tanaka et al. [4], indicates that the electrical resistivity of the
particles is lowered by a factor of two by operating at the reduced
temperature for the particular coals investigated.
15B.2

Flue gas additives to improve performance

For boiler plants, which fire low sulphur fuels, not only is it possible to
increase the particulate collection efficiency by the use of conditioning
agents, such as sulphur trioxide gas injection, but by the use of intermittent
pulse mode electrical energisation plus S03' it may be possible to further

FLUE GAS ADDITIVES TO IMPROVE PERFORMANCE

427

3
S03 Australia

>

::E

.5

312
ell
~

.s
~

~~;;;;;;;;~~~~;;~~~~~~:=====:l~~~~

io 1 l....
a:

Burning phosphorous UK
Ammonia UK
Sulphur dioxide
Injection rate of water (% volume)

2 3 4

10
15
Injection rate of conditioning agent (ppm)

20

25

Figure 15B.2 Relative increase in migration velocity as a result of conditioning.

improve the performance of an eXIstIng preCIpItator. In this situation a


planned testing programme could identify this possibility. The effect of S03
gas conditioning with high sulphur coals would not be expected to result in
a significant performance improvement, regardless of the method of energisation; in fact additional sulphur trioxide injection may bring about an
increase in opacity due to excess S03 passing through the precipitators, to
condense as H 2 S0 4 in the cooler back end regions of the plant.
The application of flue gas conditioning systems is not new and
was recognised and used in the early days of precipitation. The modern
approach dates back to the early 1960s to improve the collection efficiency
of precipitators handling fly ash arising from low sulphur fuels [5]. The
initial installations were pseudo pilot plant evaporating stabilised liquid
sulphur trioxide and injecting it with completely dry air, having a dew-point
of - 40C to prevent the formation of sulphuric acid in the injection
pipework.
These conditioning plants, although performing reasonably well, were
not acceptable commercial installations, since they required unacceptable
levels of maintenance. The results of the conditioning were remarkable,
however, as illustrated in Figure 15B.2 from Busby and Darby [5]. These
were obtained from full-scale testing of power plant precipitators in several
countries, all firing low sulphur coals producing various degrees of backionisation which limited precipitator performance to an unacceptable level.
At this stage there was less pressure on environmental control and the
conditioning units were allowed to fall into disrepair through lack of
necessary maintenance.
It was not until the early 1970s when various countries enacted legislation controlling emissions, that flue gas conditioning was 'rediscovered' and

428

IMPROVEMENTS BY CHANGING PARTICULATE ELECTRICAL RESISTIVITY

led to the development of conditioning plants, based on those used for


sulphuric acid manufacture but on a much smaller scale, and these have now
become part of the overall plant air pollution control system in meeting
legislation. These conditioning plants are now used worldwide for improving precipitator performance when handling low sulphur coals giving rise to
high resistivity fly ash.
In addition to sulphur trioxide flue gas conditioning, some plants need
to employ dual conditioning of sulphur trioxide plus gaseous ammonia to
not only reduce resistivity but also to increase particle cohesivity when
handling fly ash comprising mainly 90% plus silica and alumina compounds. Such fly ash was found to be subject to severe re-entrainment losses,
which could be minimised by the addition of gaseous ammonia to produce
ammonium salts, which increased the general cohesivity of the collected
material.
With the more recent demand for NOx reduction, units fitted with low
NOx burners have experienced high carbon carryover from the furnace;
these carbon particles, particularly if in the form of 'coke' having a low
resistivity but large surface area, tend to be easily re-entrained and impact
on the overall emission from the precipitator. In these cases, the use of
ammonia gas conditioning has been used to increase the particle cohesivity
and thereby reduce re-entrainment such that the emissions now comply with
legislation.
Because flue gas conditioning is now very much associated with electrostatic precipitation engineering, whether it be used to improve the performance of existing units, or initially installed to reduce the size and capital cost
of the plant by having a smaller precipitator; the methodology is important
and the following section of this chapter is devoted to flue gas conditioning
in its entirety. It has been prepared by Dr Henry Krigmont, who spent many
years with one of the main conditioning plant suppliers and has tackled the
subject, considering both the theory and the practical side of the equipment,
plus how the injection rates can be optimised to minimise operational costs.
References (15A and 15B)
1. Matts, S. and Ohnfeldt, P.O. (1963-1964) Efficient gas cleaning with SF electrostatic
precipitators. Flakt Ref., 6,7, 105-22.
2. Parker, K.R. (1987) The impact of pulsing on electrostatic precipitator performance. Inst. of
Physics Symposium on Electrostatics, Oxford, Institute of Physics, London, UK.
3. Darby, K. and Novogaratz, D. (1990) Increased plate spacing in electrostatic precipitators.
8th EPRIIEP Symposium on the Transfer and Utilisation of Particulate Control Technology,
San Diego, USA, Session 6A EPRI, Palo Alto, CA, USA.
4. Tanaka, T., Fujishima, H. and Tsuchiya, Y. (1993) Development of advanced dust collecting
system for coal-fired power plant. 5th International Conference of the Electrostatic Precipitation Society, Washington, USA, April, pp. 35.1-17, EPRI TR 103048 2, Palo Alto, CA, USA.
5. Busby, H.G.T. and Darby, K. (1963) Efficiency of electrostatic precipitators as affected by
the properties and combustion of coal. J. Inst. Fuel, May, 36, 184-97.

15C

lSC.l

Theory, principles of operation, equipment and


applications of flue gas conditioning

Introduction

Acceptance of the process called flue gas conditioning (FGC) as an engineering solution to a common environmental problem has come slowly. To
start with, the process is mis-named. It is not the gas which is conditioned,
but the particles of dust in the gas or, in the most common application, the
fly ash. Adjusting the resistivity of fly ash particles by application of a dilute
acid solution to their surfaces is an easily understood and straightforward
task. The very idea that the behavior of a gas stream might be adjusted in
any useful way by the addition of a few parts per million of anything boggles
the mind.
Some of the earliest applications of the electrostatic precipitation process
were in the collection of dust and fume from non-ferrous smelters in the
western part of the United States and Canada [1]. In 1912 it was found that
dust from the copper-converter furnaces at Garfield, Utah, was easily
collected when sulfitic ores were processed, but that an insulating layer of
dust formed on the collecting electrodes and the collection was effectively
shut-off when the sulfur content of the ores was low. It was reasoned that
the absence of a minimum quantity of sulfur trioxide in the flue gases made
the dust non-conductive, causing the 'shut-off' effect. This theory was
subsequently verified by evaporating sulfuric acid in the high-temperature
gases and observing that normal precipitator operation was restored. This
appears to be the earliest documented recognition of the importance of dust
resistivity in ESP operation, as well as the first use of flue gas conditioning.
It is interesting that control of resistivity by sulfur trioxide was also
recognized at this time.
Electrical resistivity is one of the critical parameters influencing fly ash
collection by electrostatic precipitators. The electrical resistivity of fly ash
depends on the chemical composition of the ash, the constituents of flue
gases, and the temperature. Fly ash composition is largely determined by
the type and composition of the coal being burned, and the furnace
operating conditions. At lower temperatures, fly ash resistivity is determined
by electrical conduction over the surface of the particles. The latter is
produced by the movement of ions in molecular thickness coatings on the
particles, and is termed surface conductivity. High temperature resistivity,
when plotted against inverse absolute temperature, is a straight line,

430

THEORY, PRINCIPLES, APPLICATIONS OF FLUE GAS CONDITIONING

illustrating conduction through the bodies of the particles, called 'bulk' or


volume conductivity.
Surface conductivity is dependent on interaction between the flue gas and
the ash. Environmental factors include temperature and the concentration
of gaseous and condensed phases in contact with ash. Flue gas temperature
influences the concentration of water vapor, the existence of a condensed
phase, and the reactivity between the ash and the environment. The surface
conductivity is also sensitive to the chemical composition of the fly ash,
because alkali metals serve as charge carriers. Alkali metal ions have been
identified as the positive charge carriers. Lithium and sodium are particularly effective. However, sodium is the principal migrating ion due to its
larger concentration. Although the mechanism of charge transfer is nearly
the same for surface and volume conductivity, the former dominates at low
temperatures because of acid and moisture adsorption, while the latter
dominates at high temperatures. Conduction may be considered as taking
place through two parallel paths, one dependent on the volume conductivity
and the other on surface conductivity.

lSC.2

Electrical resistivity

The resistivity of the collected particulate layer has a significant importance


on the operation of electrostatic precipitators. The amount of useful electrical power that can be supplied to a precipitator is limited for the most part
by the electrical resistivity of the fly ash layer of the collecting plates.
Corona current from the discharge electrodes must pass through the
collected dust layer on the plates to reach grounded collecting plate surfaces.
Passage of the current builds up a voltage across a dust layer in accordance
with Ohm's law. Theory and experience indicate that when the dust
resistivity exceeds a critical value of about mid-lOlO Q-cm, corona current is
limited by electrical breakdown in the collected dust layer. This, in turn,
limits useful operating voltage and reduces precipitator efficiency. The loss
in performance increases quite rapidly for resistivity greater than 1010 Q-cm,
and resistivity is therefore a major factor in precipitator design [2].
The following is the generally accepted dust resistivity classification in
Q-cm:
104 _10 8
10 8 _10 10
101_1011
10 11 _10 13

Conductive
Normal
Moderate
High

There are two conduction mechanisms which determine the resistivity of the
collected dust layer. These mechanisms are termed volume and surface
conduction. The former relates to the bulk composition of the material,

ELECTRICAL RESISTIVITY

431

3.2 3.0 2.8 2.6 2.4 2.2 2.0 1.8 1.6 1.4 1.2
\

\
\

I Volume resistivity
\

\1

Composite of
surface and
volume
resistivity

109~~--~~--~~-L~~~~~~~__~

70 100
(21)(38)

150 100 200


300 400
600 800 1 000
(66) (38) (93) (149) (204) (316) (417)(538)
Temperature, DF (DC)

Figure lSC.l Typical

temperature~resistivity

relationship for fly ash.

while the latter depends primarily on a film adsorbed on the particle surface
and is related to the composition of the gas and dust surface (Figure 15C.1).
Fly ash resistivity depends primarily on the chemical composition of the
ash, the ambient flue gas temperature, the water vapor and sulfur trioxide
content of the flue gas. At air pre heater outlet gas temperatures common for
cold-side utility applications 120-175C (248-347 OF), surface conduction
over the fly ash particles is the prevailing mode and the conductivity
depends mainly on the amounts of S03 and water vapor adsorbed on the
particles. At higher gas temperatures ahead of the air preheater, so-called
hot-side installations 316-427C (600-800 OF), volume conduction through
the particles prevails.
In volume conduction, electric charge is transferred through the bulk
material that comprises the collected dust layer. In most materials volume
conduction occurs by means of electron carriers within the materials and is
dependent upon the thermal excitation of the electrons in the molecular
structure of the materials.
The voltage drop across the precipitated dust layer is dependent upon the
corona current density, the electrical resistivity of the dust, and the thickness
of the dust deposit. For high resistivity dusts, the voltage drop across the
collected dust layer can be significant. Obviously, the electrical energization

432

THEORY, PRINCIPLES, APPLICATIONS OF FLUE GAS CONDITIONING

equipment must be capable of providing sufficient voltage to accommodate


this voltage drop, while maintaining adequate voltage across the interelectrode spaces.
Another important consideration of the effects of dust on corona generation concerns the deposits formed on both collection and discharge
electrodes. On the collection electrode, dust deposits alter the electric field and
sparking conditions as a result of the voltage drop within the dust layer. This
effect limits the voltage and current at which the precipitator can operate.
The effect of the resistivity of the collected dust layer or performance,
however, is more severe than its influence on power supply voltage. The
electric field in the dust layer can be quite high for high resistivity dust
leading to spark over of the precipitator at lower applied voltage and current
density. If dust resistivity is further increased, a diffuse corona glow may
appear over large areas of the dust surface. Under these conditions the
positive ion production by the reverse corona may be sufficient to completely disrupt the charging process, and effective precipitation is not
possible under these conditions.
Dust deposits also form on the discharge electrodes of operating precipitators. These deposits can be quite heavy in the case of some types of dust.
The effect on corona can be considerable depending on the nature of the
deposit and the electrical properties of the dust [4]. If the dust resistivity is
reasonably low, the effect of the deposits will be to effectively increase the
diameter of the discharge electrode. This results in higher voltage required
for corona initiation or reduced corona current for a given voltage. If the
dust deposits are uneven, an uneven distribution of the corona along the
length of the wire may result. If the dust resistivity is high, the effect
generally would be to reduce corona current for a given voltage. However,
if the deposits are somewhat porous, breakdown of the gas in the interstitial
region can occur and the effect of the deposits may be reduced.

15 C.2.1

Prediction of fly ash resistivity

It is frequently necessary to estimate the resistivity to be expected from fly


ash from coals which may not be available for such measurements until the
ESP and/or FGC system have been sized, designed and installed. The US
EPA sponsored a development by Dr Roy E. Bickelhaupt at the Southern
Research Institute (SoRI) to produce a method for fly ash resistivity
prediction using detailed analysis of the coal and ash [5]. Resistivities
calculated by this method are in good agreement with actual field measurements, and form the basis for estimation of ESP performance. Figure 15C.2
illustrates a typical result of this procedure, showing predicted resistivity as
a function of temperature.
The right-hand portion of the trace in Figure 15C.2 is essentially a
straight line descending from left to right, and represents the effect of volume

433

ELECTRICAL RESISTIVITY

1012

r-----------------------,

1012L-~__~__~~~~~~~

1 OOO/T (K) 3.0

1.6

C 60

349
Temperature

Figure lSC.2 Typical fly ash resistivity curves.

conduction through the body of the fly ash material. Three lines are shown
which descend from right to left. The left-most line shows the effect of water
vapor present in the flue gas, depicted as though no S03 were present. The
inner lines illustrate the added effect of S03 for two different concentrations.
It is seen that the chemical composition of the ash and the water vapor
content of the flue gas act to determine an upper boundary for the resistivity
values, and the S03 quantities present reduce the resistivity below the
boundary values. Thus, if the resistivity, resulting from the combination of
flue gas and ash properties with the amount of S03 naturally produced by
combustion, is higher than desired or best ESP operation at a particular
temperature, a suitable lower value can be obtained by the introduction of
additional S03.
J5C.2.2

Resistivity effects in ESPs

Back-corona is one of the major problems that leads to non-ideal behavior


in electrostatic precipitators. Back-corona describes the condition where a
local electrical breakdown is taking place in the high resistivity dust layer.
This condition initiates at the top surface of the dust layer and leads to the
formation of negative ions that are driven to the positive collection elec-

434

THEORY, PRINCIPLES, APPLICATIONS OF FLUE GAS CONDITIONING

20r---------------------------,
Good performance

'0
Q)

~15C-----~-----

~
~

ai 10

e!

Co

'0

!!!

a..

Poor performance
OL---~--~----~--~----~--~

107

108

109

1010

10 11

1012

10 13

Resistivity (ohm-cm)

Figure lSC.3 Precipitation rate (migration velocity) for laboratory pilot ESP.

trode. A corona here generates positive ions that are projected into the
interelectrode space and driven toward the corona electrode.
As positive ions flow into the interelectrode space, they encounter
negatively charged particles and negative ions. The electric field from the
charged particle exceeds that from an ion at reasonable distances; therefore,
the majority of the positive ions will flow to the negatively charged dust
particles, neutralizing their charge. This causes a proportionate reduction in
the electrical force acting to collect these particles. Additionally, backcorona disrupts the space charge that contributes to the electric field
adjacent to the collection electrode.
Figure lSC.3 represents the effect of resistivity on dust collection for a
small laboratory precipitator [6]. The vertical scale of this figure displays
the migration velocity, a measure of the rate of dust collection, and the
horizontal scale shows the apparent bulk resistivity of the collected dust.
Because of the gas, dust, and electrical conditions are relatively uniform
throughout such a unit; the transition from high collection rates on the low
resistivity side of the chart to significantly lower collection rates at high
resistivity giving the abrupt change between the two operating states
becomes readily evident.
A full-scale precipitator may be considered to be made up of a number
of laboratory-scale units arranged in series and parallel. The exact operating

435

ELECTRICAL RESISTIVITY

16r-----------------------------------------~

Good performance

.....
..: .
'

.'

'

OL-----~

106

107

____

-----Poor performance

____-L_____ L_ _ _ _

____

108
109
1010
10"
Average resistivity at ESP inlet (ohm-cm)

____

1012

Figure lSC.4 Typical resistivity effect. Fullscale ESP.

conditions in each of these hypothetical small portions differs from the


others so that the drop of performance, as the average resistivity of the dust
at the precipitator inlet increases, occurs at different times, giving a less
abrupt transition from good to poor performance as shown by Figure 15CA
[7]. It is clear that obtaining the maximum dust collection rate from the
ESP requires that the dust resistivity fall in an optimum range, and it is also
apparent that improvement in performance is available from proper adjustment of the resistivity if the unit is operating at resistivity levels above the
optimum range.
Figure 15CA also indicates that reduced dust collection rates are attained
at low values of resistivity. The electrical forces holding the dust to the
collecting plates decreases as resistivity is reduced and this causes increased
losses due to rapping and re-entrainment by the gas stream.
The presence of carbon particles (very low resistivity) in the fly ash, in
amounts greater than 10%, usually creates excessive re-entrainment and
particle losses (Figure 15C.5) [8]. Those losses are predominantly due to a
so-called 'dancing particles' effect caused by low electrical resistivity (less
than 104 Q-cm), where, as the particles approach the collector they are
repelled back into the gas stream to be recharged and the process repeats.
As they reach the outlet they can escape as uncharged particles with the
outlet gases.
On the other hand, when the carbon particles are retained in the dust
layer (with the aid of dual flue gas conditioning, which will be discussed
later), the resultant effect is similar to lowering the dust cake resistivity. The
carbon acts by a physical admixture effect in which the current is carried by
the carbonaceous particles in the matrix of the fly ash and unburnt carbon
as a whole.

436

THEORY, PRINCIPLES, APPLICATIONS OF FLUE GAS CONDITIONING

12

10

E
0

E8

.s::::
~

">

t1 6

iii
~

OJ

.2

2~~~~--L-~~

__~~-L~__~~~~

10 12 14 16 18 20 22 24 26 28
Carbon content (wt"lo)

Figure 15C.5 Fly ash resistivity versus carbon content of ash.

15C.3

Flue gas conditioning

The use of additives to ESP inlet gas streams to improve particulate


collection rates originated shortly after the first commercial use of ESPs was
begun early in this century. Evaporation of acid in smelter gases to reduce
dust resistivity occurred in 1912 [IJ; water injection into gases from cement
kilns and steel refining vessels has been used for many years, and ammonia
treatment of catalyst dust in petroleum refineries was begun in the 1940s.
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 [1,7]. Commercial conditioning of fly ash usually involves the injection of a chemical reagent
into the flue gas stream to control the electrical resistivity of a dust and
thus improve its collection. Currently, however, this term is assuming a
much broader meaning, for conditioning involves processes other than
resistivity modification (i.e. coagulation, agglomeration, cohesivity improvement, etc.).
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.

FLUE GAS CONDITIONING

15 C3.1

437

Conditioning by sulfur trioxide

The most common conditioning agent used for modifying high fly ash
resistivity is sulfur trioxide [7]. Two major factors control ash resistivity: the
sulfur content of the coal, and the overall elemental composition of the ash.
Sulfur occurs in coal as organic and inorganic compounds-notably
pyrites and sulfate salts. When coal is burned, more than 95% of the pyrite
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.
When the temperature of flue gas drops to around 300 C (about 572 OF),
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 150C (about 300 OF), where electrostatic precipitators normally operate [9]. 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
resistivity modification results from sulfuric acid, rather than sulfur trioxide,
effects.
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.
Thus, sulfur trioxide conditioning is effective at flue gas temperatures
below about 204C (400 OF), where surface conduction prevails. At higher
temperatures, the 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.
The most commonly used method for generating sulfur trioxide is
catalytic conversion from sulfur dioxide [7]. 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

438

THEORY, PRINCIPLES, APPLICATIONS OF FLUE GAS CONDITIONING

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.
lSC3.1.1

S03 injection rate prediction

(a) Discussion. Since the electrical resistivity has a pronounced effect


on the electrostatic collect ability of fly ash, it is desirable to have advanced
knowledge regarding the magnitude of fly ash resistivity that one might
expect from a given coal. Obviously the best source of this information
would be in-situ resistivity measurements made during the burning of the
subject coal in a commercial boiler. If the coal has not been used commercially, one has the option of burning the coal in small-scale pilot furnace and
measuring the resistivity in-situ or in the laboratory, or one can utilize one
of the methods for predicting fly ash resistivity.
The methods for predicting resistivity are based on correlations that have
been established between resistivity and fly ash compositions for specific
laboratory test conditions.
The key to better understanding and more accurate estimation of flue gas
conditioning by S03 was developed by Dr R. E. Bickelhaupt at the
Southern Research Institute under the sponsorship of the US Environmental Protection Agency [5]. Application of the findings of these studies results
in resistivity versus temperature plots typified by Figure 15C.2. As discussed,
the uppermost line on this figure represents the expected resistivity with no
S03 present in the gas. A maximum resistivity value occurs at approximately 140C in this example, dropping to lower values at higher temperature due to increased electrical conduction through the body of the ash
particles, and similarly dropping at lower temperatures from the effect of
conduction through water attached to the ash surface. Two lines emanating
from the upper curve show the effects to be expected from the presence of
different amounts of S03 in the gas.
Since it is possible to make empirical estimates of the quantity of S03
naturally generated from fuel sulfur, the expected resistivity of the ash
without the addition of a conditioning agent can be estimated, and a
decision can be made regarding the necessity for reducing resistivity to
improve collection.
The calculation methods developed by Dr Bickelhaupt were published
in a form which estimates the resistivity of fly ash in the presence of a
given concentration of sulfuric acid. Simple mathematical manipulation
of the Bickelhaupt formulae makes it possible to calculate the acid concentration required to bring about a desired change in resistivity. This

FLUE GAS CONDITIONING

439

amount of acid has proven to be an important factor in the injection rate


equation [10].
The acid values in the Bickelhaupt equations are equilibrium concentrations determined by laboratory experiments which maintain a fixed volume
concentration of acid in the gas in contact with the fly ash sample over a
relatively long period of time at a fixed temperature. This obviously differs
from the situation in an operating boiler plant where the fly ash is exposed
to injected S03 at concentrations which vary with respect to time, temperature and physical location, and where the time of exposure is quite short.
Thus, the acid figures derived from the Bickelhaupt formulae are indicative
of, but not directly convertible to, the required injection quantities.
Since the conditioning process is concerned with the application of an
acid film to the surface of the fly ash particles, it is clear that the coal ash
content will affect the amount of acid to be injected. A degree of compensation for ash in the fuel is contained in the Bickelhaupt calculation methods,
but it is considered to be unlikely that this fully accounts for injection
requirements arising from this factor. Thus, it appeared initially that the acid
requirement <;alculated by the Bickelhaupt methods and the ash content of
the coal would be two of the variables to be examined for inclusion in any
injection rate equation.
Acquisition of an acid film by the fly ash particles is presumed to take
place by a condensation/partial pressure deposition process and would
therefore be expected to be temperature sensitive. Dr Frisch has shown that
the surface composition of the ash particles can affect acid attachment [9].
Dr Bickelhaupt also showed that a 'threshold' effect was exhibited by certain
ashes, evidenced by a lack of change in resistivity at ambient acid concentrations below the threshold level [5]. These phenomena are clearly related
to ash chemistry, and call for inclusion of all possible combinations of ash
chemical properties as variables to be tested for significance.
Intuitively, one would expect that ash particle size distribution, by
determining ash surface area as a function of weight, would affect acid
requirements. It has been indicated that the presence of elements which
volatilize at flame temperature and the viscosity of the fluid ash in the
furnace affect the quantity of fine fly ash particles generated [8]. To take the
latter effect into account the hemi-spherical ash fusion temperatures for
oxidizing and reducing conditions were calculated from ash chemistry and
included in the list of trial variables. Other items were speculatively
included, on the chance, however remote, that they might prove to have
significant injection influences.
Over 20 legitimate data points (sites with known coal supplies, injection
rates and precipitator performance data) were used in the development of
the injection rate equation. They include coals ranging in rank from lignite
to low-volatile bituminous, taken from mines in eastern and western USA
locations, as well as South Africa and Australia. Ash contents range from 4

440

THEORY, PRINCIPLES, APPLICATIONS OF FLUE GAS CONDITIONING

to 26% and flue gas temperatures from 140 to 216C. Nearly every type of
pulverized coal firing arrangement except cyclone furnaces was represented
[10].
(b) S03 injection rate equation. The Bickelhaupt method cannot be
directly applied to determine S03 injection rates, because the S03 quantities
used in the method are amounts in equilibrium with the gas-and-ash
environment in a laboratory situation, and these differ from injection
requirements for flue gases containing suspended fly ash. Having an estimate
of the ash resistivity without conditioning and knowing from experience the
desired final resistivity to obtain best ESP performance, both at the intended
operating temperature, simple subtraction gives the value by which the
resistivity has to be changed, and common sense says that the injection rate
will be a function of this change. Similarly, the required injection will be a
direct, but not necessarily linear, function of the quantity of ash to be
treated. Other factors than these two will enter into the determination of the
injection rates. As it was noted earlier, Dr N. W. Frisch pointed out that the
surface condition of the ash will have a bearing on the ability of S03 to
attach to the particles [9]. Computerized regression analysis was used to
sort through a variety of possible factors and to choose those having
significance in determining required S03 injection rates for a wide variety
of coals. [10]. Ultimate analysis of the coal and chemical analysis of the ash
are required as input data to the resulting equation for the calculation of
injection rates.
The significance of the trial variables was tested by computerized stepwise
regression analysis, first taking a single independent variable in the form

(15C.1)
where y is the dependent variable (injected S03 + natural S03 from coal
sulfur), X 1 is the independent variable, and K l' K 2 , a are constants.
Regression analysis was applied to determine the value of the constants
in the equation, while substituting for the independent variable in turn, each
of the trial variables listed in Table 15C.1, plus their logarithmic and
exponential values, their squares and square roots, and the inverse of all of
these. The variable and its form having the highest correlation coefficient
(R2) was then chosen to replace the independent variable in the term tested,
and another term was added to the equation in the same form as those
preceding. All the trial variables were then tested in the new term, and the
process continued. At each term addition, the new term and all those
preceding were tested for statistical significance.
The detailed description and the final form of the derived injection rate
equation is described in [10] and a typical S03 injection rate us. temperature curve is presented in the Figure 15C.6.
It is seen that the required rate of S03 injection at temperatures below
about 165C is essentially independent of temperature. Resistivity curves

FLUE GAS CONDITIONING

441

Table ISC.l Trial variables


Category

Symbol

Ash

LI
NA
K
MG
CA
FE
AL
SI
TI
P
S

Description
Lithium oxide, percent weight
Sodium oxide, percent weight
Potassium oxide, percent weight
Magnesium oxide, percent weight
Calcium oxide, percent weight
Iron oxide, percent weight
Aluminium oxide, percent weight
Silicon oxide, percent weight
Titanium oxide, percent weight
Phosphorus pentoxide, percent weight
Sulfur trioxide, percent weight
(Plus all of the above taken as molecular percentages,
plus all of the above taken as cation percentages)

Ash-derived

SL
CM
ALSIM
HTO
HTR
BAR
DR
OXF
ALK

Sodium plus lithium, cation and molecular percentages


Calcium plus magnesium, cation percent
Alumina plus silica, molecular percentage
Hemispherical temperature, oxidizing
Hemispherical temperature, reducing
Base/acid ratio, both weight and molecular percentage
Dolomite ratio
Oxide factor
Alkali ratio

Coal

SUL
ASH
HV

Sulfur, percent weight


Ash, percent weight
Heating value

Combustion

TEMP
ASHG
ASHC
ACIDDP

Flue gas temperature


Total ash generated, wt./unit gas volume
Ash carried over, weight/unit gas volume
Flue acid dewpoint temp., coal SO}

Other

ACIDB

Ppm v SO} to change ash resistivity at flue gas temperature from the no-injection value to the desired value as
computed by reference [1] methods

DPACID
TVS

Total acid for dew point at flue gas temperature, ppm v


Temperature of interaction of volume and surface resistivity curves

TVSA

Temperature of intersection of volume-plus-surface and


acid resistivity curves

calculated by the Bickelhaupt method, such as those shown in Figure 15C.7


for a Western USA Powder River Basin (Wyoming) coal, have relatively
steep slopes of resistivity with respect to temperature for a given equilibrium
acid concentration.
(c) Application of the theoretical approach. Figure 15C.6 shows predicted
injection rates for two coals of widely differing characteristics and origins.
Each of the two plots shows a low temperature region which is essentially
a reflection of the ash resistivity curve. As temperature is increased, both
curves come to an inflection point at which transition is made to a rapidly

442

THEORY, PRINCIPLES, APPLICATIONS OF FLUE GAS CONDITIONING

40r---------------------~

O~~L-~~--~~--~~

75

100 125 150 175 200 225 250


Flue gas temperature (0C)

Figure lSC.6 Predicted S03 injection rates.

rising characteristic which approximates a straight line. At temperatures


below the inflection point, virtually all of the injected S03 attaches to the
ash. At higher temperatures, the amount of S03 injected needs to be in
excess of the S03 level at the inflection point to provide the partial pressure
for attachment of sufficient S03 to the ash to obtain the desired resisivity.
The excess S03 above the inflection point level remains in the gas stream
and, if the excess is about ten parts per million or more, may be visible as
a blue-white condensation plume from the stack.
The trace on Figure 15C.6 having an inflection point at lower temperatures is representative of coal ashes having very low base-to-acid ratios, such
as those from the Sydney Basin of Australia. The characteristic having the
higher inflection point is for coals such as those from the Powder River
Basin in the USA which have very high base-to-acid ratios. These are two
extremes-nearly all the world's coals have inflection point temperatures
intermediate to these two. Although it is convenient to refer to these effects
in terms of base-to-acid ratio, it should be remembered that this is only a
rough general index to the inflection point temperature for discussion
purposes; other factors appearing in the injection rate equation must be
included for accurate determinations.
At temperatures from about 165 to 190 DC, depending on ash analysis, an
inflection point occurs in the injection rate curves, with the rate increasing
with temperature above this point. It seems logical that the temperature at
which inflection occurs is a function of ash surface conditions-a measure of
susceptibility to attachment of acid, perhaps, in as much as the lower

443

FLUE GAS CONDITIONING

1.0E+ 12

1.0E+11
~

./

-1'-

-m

PI'-

rI /

,;j/l"
~

+T

l"

/ d

/ /

J.I tl
"*

ril.f

jJ

1.0E+08

==
-

-+-

0.4

-+-

=
=
==
-

--B-- 4

--*1.0E +07
50

100

10 PPM 803

150
200
Temperature (0C)

I
250

300

Figure lSC.7 Fly ash resistivity versus temperature for PRB coal.

inflection temperatures occur for ashes having low base-to-acid ratios and
the higher points correspond to higher ratios.
The inflection point effect explains some of the difficulties various
experimenters have encountered in attempting adjustment of ash resistivity
with sulfur trioxide alone. 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%). Here, the injection of sulfur trioxide alone to reduce
resistivity loses some of its effectiveness 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. Furthermore, if the
operating temperature is much higher than the inflection point, no reasonable amount of injected S03 will be effective in reducing the resistivity.
The increased requirement for sulfur trioxide at temperatures above the
inflection point indicates that an increase in the required partial pressure of

444

THEORY, PRINCIPLES, APPLICATIONS OF FLUE GAS CONDITIONING

30r---------------------------~

25

140

160

180

200

Temperature (0C)
Point is field data- Curves calculated

220

Figure ISC.S Effect of temperature on injection rates and equilibrium curves.

sulfur trioxide is needed to attach sufficient 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.
The temperatures for the constant target resistivity (for example, of
S x 10 9 Q-cm) may be plotted as an equilibrium (iso-resistivity) curve with
the predicted injection rate curve for the same ash (Figure lSC.8). For
comparable conditions of acid deposition on the ash it would be expected
that the equilibrium curve, representing the acid remaining in the gas stream
after deposition, would always fall below the injection curve. However, it is
seen that the two curves intersect, and at higher temperatures the equilibrium value exceeds the injected amount. The difference cannot be explained
as representing the S03 originating from combustion of coal sulfur because
this would imply no deposition of acid on the ash, and it is known from the
operation of actual boiler units in the higher portion of the temperature
range that the ash resistivity is modified by injection, thereby verifying the
attachment of acid to the ash. It must be concluded that the differences
between constant-temperature laboratory deposition and the injection process tend to promote more effective transfer of the acid to the ash particles
in the injection case.
It is thought that the higher transfer efficiency with injection results from
the relatively high acid concentration of the injected stream. In the process

FLUE GAS CONDITIONING

445

of mixing the injected stream with the flue gas, at least some of the cooler
ash particles will penetrate the injected stream and, thus, be subjected to a
high driving force for acid attachment. This will result in the deposition of
more acid than would be calulated from equilibrium (completely mixed)
conditions.
Accurate prediction of injection rates for S03 flue gas conditioning
systems is now possible for the majority of steam coals in use. The described
procedure may be the first step in any FGC system optimization. Coal and
ash analyses are required for input to the calculation procedure. Equipment
size and feedstock requirements can be closely determined, given knowledge
of the range of fuel properties to be accommodated.
As with any mathematical function derived by regression analysis, care
must be taken in extending the results to areas beyond the limits of the
original database. No cyclone boilers were included in the database, and it
is therefore expected that application of rate prediction to that type of unit
would require some compensation for the reduced carryover of ash to the
precipitator, and possibly for changes in the ash analysis in the fraction
carried over as compared with the bulk analysis. Also, some ashes are
composed almost entirely of silicon and aluminum oxides, with very minor
amounts of other elements. No ashes representing the extremes of this
situation were included in the study so application of this theoretical
approach to those types may be misleading.
J5C3.2

Ammonia conditioning

Numerous investigators have reported the use of ammonia conditioning


with various degrees of success. Ammonia conditioning usually works best
with the acidic fly ash. If injected ahead of the air heater (unless air heater
temperatures were unusually high - > 400 OF) the air heater was reported to
have a tendency to plug as an additional, but unwanted, result. Ammonia
injection downstream of the air heater was reported to be unpredictable
[13].
The use of ammonia (NH 3) for modification of dust resistivity has not
been developed to the same degree as S03' Some experiments have shown
that ammonia injection improved the collection of fly ash in ESPs, but it is
not clear whether the improvement resulted from resistivity modification or
other effects [11]. Other trials have shown little or no change in fly ash
resistivity or ESP performance [12].
No effect of the ammonia on fly ash resistivity has been reported with
any degree of certainty. There has been much debate by many authors
regarding the use of ammonia alone for modifying the electrical resistivity
of fly ash. One argument suggests that for high resistivity ashes a reduction
in resistivity might be produced, permitting higher voltages and currents to
be maintained without sparking or back-corona. On the other hand, for

446

THEORY, PRINCIPLES, APPLICATIONS OF FLUE GAS CONDITIONING

lower resistivity ash an increase in resistivity, due to neutralization of excess


acid, might permit a sufficient electric field to exist across the ash on the
precipitator collection electrodes to overcome losses from rapping re-reentrainment. The effect of ammonia appeared to be an enhancement of the
space-charge component of the electric field. In addition, the increase of
the cohesiveness of collected ash and a reduction in re-entrainment appears
to be another 'major player' in the successful application of ammonia
conditioning.
While successes have been reported using ammonia as the sole conditioning agent for some low sulfur Australian coals, the US experience has been
improved precipitator performance for some low sulfur coals and not for
others. It has been theorized that ammonia conditioning is more effective for
acidic ashes, in terms of the relative concentrations of acidic and basic
oxides present (acid/base ratio), but this has not been true in every case
[10]. Ashes having high acid/base ratios with small to modt-rate 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.
15 C3.2.1 Background. The value of ammonia as a conditioning agent
was reportedly discovered in 1942 in efforts to deal with the high resistivity
of catalyst dust in the petroleum industry. The use of the ammonia for
treating fly ash in coal-fired power plants, on the other hand, is 'comparatively' new. First reports of the intermittently successful use of the ammonia
as a conditioning agent came from Australia, by Watson and Blecher in
1966 [14]. The results of this study showed good results while using the
ammonia as a conditioning agent. There the S03 flue gas conditioning was
proved to be ineffective due to highly acidic ash (pH of 3.5-4) and, thus,
speculations were that conditioning with a basic rather a thin acidic agent
may prove more effective. The initial pilot plant trials with ammonia
conditioning conducted at the Tallawarra plant in 1959 were found to be
very successful. Similar promising results were obtained at the No. 5
Tallawarra precipitator. However, subsequent ammonia conditioning at the
No.6 precipitator at the Tallawarra did not produce the same encouraging
results that were obtained on No.5.
The electrical behavior of the precipitators with ammonia conditioning
changed considerable within minutes after the introduction of ammonia into
the flue gas. Increases of 50 to 100% in electrode potential and corresponding decreases in electrode current were maintained consistently and tended
to persist for several hours after ammonia injection was stopped.
Resistivity measurements were also carried out; however, these presented
rather perplexing results when ammonia-conditioned dust was evaluated. It
is generally accepted that the increase in precipitation efficiency brought
about by flue gas conditioning is caused by a reduction in the particle

FLUE GAS CONDITIONING

447

surface resIstIvIty and the elimination of back-corona, a phenomenon


reported to be quite common with New South Wales fly ash. The resistivity
of the ammonia-conditioned ash, however, showed no appreciable difference
from that of non-conditioned fly ash. Both were in the range of 10 13 to
10 14 a-cm which is certainly well above the critical level of 10 10 a-cm
suggested by Dr White [2].
Watson and Blecher [14] proposed a hypothesis that perhaps ammonia
gas conditioning changed flue gas characteristics sufficiently to eliminate
back-corona and/or the forces of particle adhesion and cohesion sufficiently
to bring about cleaner electrodes and, hence, better performance.
Shortly thereafter, Baxter [15] reported success with ammonia conditioning of several plants in the United States, where the apparent effect was to
lower ash resistivity in some instances and to increase it in others. Initially,
he injected sulfur trioxide with unsatisfactory results. Performance improvement was attained only by the use of excessive conditioning which allowed
an objectionable acid vapor plume to escape from the stack. The presence
of the acid plume indicates that the necessary adsorption on the dust
particles was not occurring. Where sulfur trioxide conditioning has been
effective, stack measurements show no increase in effiuent S03 concentration; rather the injected gaseous agent attaches to and is discharged with the
dust collected in the precipitator, as reported by Darby and Heinrich [16].
Analyses of the collected ash indicated that S03 flue gas conditioning
reduced laboratory measured resistivity from 10 12 to 1011 a-cm and increased acidity of the dust from a pH of 5.5 to 3.5.
Subsequent trials with ammonia injection reportedly produced marked
precipitator performance improvement. Dust analyses showed a decrease in
resistivity, similar to that obtained with sulfur trioxide conditioning; however, the pH of the ash increased to a neutral condition. Although Baxter
[14] reported varying success with precipitator performance, improvement
has only been accompanied by air heater plugguage at each test location.
Finally, tests reported from a power plant with two identical cyclone fired
boilers produced still unexplained results, i.e. the units reacted in a very
explicitly different manner. One of the units responded well to the ammonia
injection, while the other reacted rather poorly. Moreover, chemical and
physical analyses of the dust produced no reasons for such a difference. The
resistivity was measured and found to have actually increased. Dust pH also
increased but remained quite acidic. Unfortunately the authors did not
publish detailed analyses. Furthermore, it is worth noting that, according to
the authors, 'No correlation can be established between effectiveness and
initial ash resistivity, nor between ammonia injection and treated ash
resistivity'.
In 1968, Reese and Greco [17] were successful in the use of ammonia to
treat the fly ash from high sulfur coal and concluded that the effect of
ammonia was to overcome an unfavorably low resistivity stemming from an
excess of naturally occurring sulfur trioxide.

448

THEORY, PRINCIPLES, APPLICATIONS OF FLUE GAS CONDITIONING

Some of the trials with ammonia flue gas conditioning have been
successful, but others have not produced the desired results. One source of
difficulty in the use of the ammonia conditioning has been the absence of
basic information on mechanisms of conditioning with this compound and
the circumstances in the power plant that favor its use.

J5C3.2.2 Chemical reactions involved in ammonia conditioning. 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. There is no way to confirm
the reaction of ammonia gas with sulfuric acid gas experimentally. However,
if the theoretical tendency is considered qualitatively, the following changes
can be visualized.
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; finally,
the solution is exhausted and solidifies.
Elementary explanations of ammonia conditioning 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 NH 4 HS0 4 or, if the stoichiometric ratio is appropriate,
the normal sulfate (NH 4)zS04' Besides these, the existence of intermediate
chemical compounds is possible. The melting points change continuously
according to the H Z S0 4 /NH 3 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
150C (300 OF) are those with stoichiometric ratios 1.2 and above.
The chemical reaction postulated by Rendle and Wildson [18J indicates
that ammonia will react with any natural sulfur trioxide present plus
moisture to produce ammonium bi-sulfate:

S03 gas + NH3gas + H 2 0 gas => NH4HS04IiQUid,aCidicsulfuricammOflla(ammOniumbiSulfate)


(1SC.2)

FLUE GAS CONDITIONING

449

NH 4HS0 4 has a melting point of 146.9C (296.4 OF), and deliquescent


characteristics, i.e. it absorbs water readily. When the flue gas temperature
is about 140 DC (284 OF), NH 4HS0 4 begins to solidify.
In the presence of excess ammonia, this reaction will slowly continue to
produce ammonium sulfate:
(ISC.3)

or
NH 3 gas + NH41iQUidivapor + HSO 41iquidlvapor <=> ( NH 4) 2 S0 4 so lid sulfur ammonia(ammonlum sulfate)
(1SC.4)

or
(1SC.S)

The melting point of (NH4)2S04 is S13 C (9SS OF). In the above equation
(1SC.2), (NH4)2S04 is dissociated at temperatures higher than 200C and
decomposes into NH 4HS0 4.
Lacking excess ammonia, or as cooling occurs, the liquid bi-sulfate will
freeze at temperatures below 147C (297 OF) into a sticky solid. It is this
adherent nature of the resultant that has been a 'suspect' to playa major
role in the precipitator performance improvement.
A variety of interactions of ammonia with sulfuric acid are possible as
follows:
(1SC.6)

+ nucleus => NH4HS04pac"cle


(NH4)2S04s0lid + nucleus => (NH 4}zS04pac tkle

(1SC.8)

+ NH4HS04pa"ic1e => (NH4}zS04pachc1e


H2S04gas + (NH4)2S04pa"ic1e => NH4HS04pa"ic1e

(1SC.10)

NH4HS04s0hd

NH3gas

(1SC.9)

(ISC.ll)

The reactions in (1SC.6) and (1SC.7) are collision reactions. Due to the low
vapor pressures of NH 4HS0 4 and (NH4)2S04' these compounds usually
solidify easily. The quantity of (NH4)2S04 formed increases as the quantity
of ammonia increases. The reactions in (1SC.1 0) and (1SC.11) are subject to
the effects of the diffusion of the gas currents of NH31gas) and H 2 S0 4(gas)' The
respective diffusion constants are 0.37 cm 2 js and O.1S cm 2 js in 140 DC
(284 OF) air; therefore, the reaction in (1SC.10) should be somewhat faster
than in (1SC.ll).

J5C3.2.3 Mechanisms of the ammonia conditioning. Ammonia in the


presence of sulfur trioxide may work in a variety of ways. Although the

450

THEORY, PRINCIPLES, APPLICATIONS OF FLUE GAS CONDITIONING

exact mechanisms responsible for improved precipitator performance are


not completely defined, they appear to include such phenomena as an
improved acid attachment, space-charge effects, agglomeration and increased cohesiveness of the fly ash,

(a) Effects of ammonia on the electrical resistivity. A study of the effects


of the ammonia on the electrical resistivity of the silica-alumina catalysts
may shed some light on the possible mechanism of the conductivity
improvement on the one hand and, perhaps, answer numerous questions
why ammonia is very unpredictable in its conditioning nature.
It has been demonstrated that sodium and potassium increase the
conductivity of silica-alumina materials (note: fly ash is predominantly
silica-alumina compound) at about 204 DC (400 OF) or higher. It was further
speculated that the increased conductivity results from ionic conduction of
sodium and potassium ions. Ammonia and amines are reported to markedly
increase the resistivity of silica-alumina catalysts below 93C (200 OF). The
increased conductivity is closely associated with large amounts of physical
adsorption, which probably increases the mobilities of the current carriers.
The irreversible adsorption of ammonia below 93C (200 OF) is independent
of temperature and is a measure of the 'acid sites' on the particulate. The
reversible adsorption increases with decreasing temperatures. The increase
in conductivity depends on the properties of both adsorbate and the
particulate.
The chemisorption of ammonia on silica-alumina compounds, like fly ash
particles, results in the formation of ammonium ions from the reaction of
ammonia with protons on the ash particle surface. This, perhaps, could be
responsible for the lower resistivity over the temperature range of 150260C (300-500 OF). With ammonia, it seems reasonable to conclude that
most of the conductivity occurs on the surface. The physically adsorbed
ammonia offers an excellent medium for the passage of ions. In addition, the
relative size of the ammonia ion, the internal structure of silica-alumina and
the temperature regions where the precipitators are operating are not
favorable for ionic conduction (by ammonium ions) in the bulk. The exact
mechanism of the increased surface conductivity resulting from ammonia
adsorption is difficult to deduce, but one could speculate that the presence
of traces of alkali metals would contribute appreciably to the conductivity
of the fly ash particles.
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. It creates a thin conductive film on the ash
surface, believed to consist of low melting point ammonium sulfate products
which contribute to the resistivity modification. It has been suggested that
the addition of ammonia improves upon the efficiency of the sulfur trioxide
conditioning by providing an additional charge carrier, the NH; ion [19].

FLUE GAS CONDITIONING

451

Since the conditioning process is concerned with the application of an


acid film to the surface of the fly ash particles, it is clear that the coal ash
content will affect the amount of acid to be injected. Acquisition of an acid
film by the fly ash particles is presumed to take place by a condensation/partial pressure deposition process and would, therefore, be expected to
be temperature sensitive.
Those who have studied factors affecting the resistivity of fly ash in a flue
gas atmosphere are very familiar with the hump-backed characteristic of
resistivity as a function of temperature. Generally, resistivity has a maximum
value at a temperature around 150C (300 OF) with decreasing values above
and below the maximum point. In determining the amount of S03 to be
required to reduce the ash resistivity to a desired lower value, one would
think that the maximum rate would be required at the maximum unconditioned resistivity point, with decreasing amounts at higher and lower
temperatures; that is, a hump-backed curve of injection rate reflecting the
unconditioned resistivity characteristic would be expected. An interesting
discovery made in the development of S03 flue gas conditioning is that the
amount of S03 required to attain a desired level of resistivity follows the
expected dome-shaped curve with respect to temperature only up to a point,
after which it breaks off to a rapidly rising characteristic [20].
It has been found that the simultaneous injection of ammonia (NH3) with
the S03 will condition the ash surface to accept resistivity adjustment
without excess S03 being required. Ammonia FGC tends to overcome this
problem by allowing greater adhesion of the residual sulfur trioxide (again,
one has to presume that there is a sufficient supply of natural S03 present)
to the acidic ash particle surfaces to optimize its resistivity. This allows for
proper fly ash resistivity adjustment without using excess sulfur trioxide. It
seems logical that the temperature at which inflection occurs is a function
of ash surface conditions - a measure of susceptibility to attachment of acid,
perhaps, in as much as the lower inflection temperatures occur for ashes
having low base-to-acid ratios and the higher points correspond to higher
ratios.
(b) Space-charge effect. When ammonia is injected into a gas stream in
the presence of the sulfur trioxide, a fine fume consisting of a variety of
ammonium sulfate particles, all much less than 111m in size, is produced.
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-documented effect arises when fine fume
is charged in the precipitator and the electric field is thereby increased.
Moderate, strictly controlled quantities 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

452

THEORY, PRINCIPLES, APPLICATIONS OF FLUE GAS CONDITIONING

happens because a large proportion of the particles are collected in the front
fields and the remaining small quantity is not sufficient to produce a
significant space charge,
( c) Fly ash cohesivity control. Ammonia can generate low melting point
substances depending on the H Z S0 4 /NH 3 stoichiometric ratio; at low
stoichiometric ratios low melting point substances having high moisture
absorbency are generated which 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 re-re-entrainment. To
simplify, the ammonia combines with sulfur trioxide to form ammonium
bisulfate. The melting point for this material, as discussed earlier, is about
147C (297 OF), 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 larger and 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
resistivity 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. Good cohesion has resulted in high
efficiency due to reduced rapping losses and re-re-entrainment, or more
effective initial capture of particles. The electrical forces of attraction
between particles in an ash layer on the collecting electrode in an energized
precipitator are the predominant 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, and under
these circumstances the mechanical forces causing re-re-entrainment become
important.
15 C.3.3

Dual flue gas conditioning

Simultaneous injection of S03 and NH 3, arbitrarily called dual conditioning, was mentioned earlier as assisting in obtaining attachment of S03 to fly
ash at temperatures above the inflection point of the injection rate curve.
Therefore, one may define a dual FGC as a system which 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. It has been demonstrated that it may be used to control
rapping and re-entrainment losses from ESPs. It appears that the process
can be crudely visualized as reacting the two conditioning gases to form
ammonium bisulfate which has a melting point close to typical ESP
operating temperatures. This is thought to act as a binding agent (liquid

FLUE GAS CONDITIONING

453

glue?), increasing the cohesivity of the ash particles so that they adhere to
each other and fall into the hopper with reduced dispersion into the gas
stream.
A striking demonstration of this effect occurred at the Monroe Station of
Detroit Edison where S03 alone permitted operation only up to 450 MW
before the compliance limit on particulate emission was reached, but dual
conditioning allowed the full 750 MW unit rating to be attained. In tests
conducted by Ontario Hydro it was found that dual conditioning was
sufficiently effective to obtain significant reduction of stack losses, caused by
high unburned carbon levels in the ash.
From the preceding description of the inflection point, which appears in
the injection-rate-versus-temperature curves, it is clear that one criterion for
needing dual conditioning is flue gas temperature exceeding the inflection
point temperature. Criteria for advance determination of a need for dual
conditioning to suppress excess rapping and re-entrainment losses by means
of fuel and ESP data have not yet been developed, partly because the factors
affecting ash cohesivity in the ESP have not been explored or defined, and
partly because the mechanical and fluid dynamic characteristics of the ESP,
which affect the magnitude of rapping and re-entrainment losses, may differ
radically from one installation to the next and have not been subjected to
any investigative procedures capable of developing applicable loss factors.
It seems odd that such a significant ESP loss mechanism should be so
poorly characterized, but it is expected that continued research will help to
clarify this situation in the future. In the meantime, the only recourse is the
observation of trials with fuel, S03 and ESP conditions as nearly as possible
the same as expected in service. Then, observation of the difference between
recordings of flue gas opacity, with and without rapping, will immediately
indicate, if the difference is more than a few percent and large rapping spikes
appear on the instantaneous trace, that dual conditioning should be
considered.
15 C.3.4

Balance of plant impact

The addition of S03 alone to boiler flue gas streams is only necessary if the
coal is low in sulfur, as otherwise the ash resistivity will be low enough not
to require conditioning. Nearly all of the injected S03 is attached to the ash
particles and is captured with them in the ESP. Without conditioning,
1-1.5 % sulfur content in the coal would typically give the same ash
resistivity as that obtained by injection of a low sulfur coal, meaning that
the same amount of S03 is on the ash surface; however, an additional
amount would still be present in the gas stream and would be emitted from
the stack. This latter amount, typically about one-tenth percent of the SOz
content of the gas stream, is not present in the emissions from the
conditioned unit, unless operation at temperatures above the inflection

454

THEORY, PRINCIPLES, APPLICATIONS OF FLUE GAS CONDITIONING

point of the injection rate curve is attempted without the simultaneous


injection of ammonia.
Fly ash from a conditioned system carrying acid on its surface, in about
the same concentration as that from unconditioned combustion of one-plus
percent sulfur coal, is not significantly different in handling requirements
and environmental effects from the ashes dealt with in the past.
Ash handling systems which sluice S03-conditioned ash to a pond and
recirculate the water may find that the pH of the pond will be depressed
below acceptable levels for discharge to the environment, making neutralization treatment necessary.
Ammonia treatment of fly ash may cause release of objectionable odors
from concrete made with cement in which the fly ash has been used as an
additive. Determination of the maximum acceptable ammonia content for
fly ash for this use is a trial-and-error process. The level has been found to
be greater than zero in all cases to date, but the maximum appears to be
influenced by cement and ash chemistry and ash handling conditions.
Ammonia and sulfur, as used by conditioning systems, are essentially
pure chemicals which neither introduce nor generate any hazardous or toxic
pollutants. Except for improved control of particulate emissions, the overall
impact of the boiler system on the environment is essentially unchanged by
the use of flue gas conditioning.
Some experimenters in the past found that the injection of NH3 into flue
gas streams containing appreciable quantities of S03 caused the ash to
become 'sticky' enough to cause problems in ash handling systems. Experience with dual conditioning to date indicates that this problem arises when
reatively large quantities of NH3 and S03 are present as compared with the
amount of ash. The quantities required to provide effective prevention of
excessive rapping and re-entrainment losses are controlled to much lower
values. None of the dual conditioning installations thus far (14854 MW)
have had problems with ash handling from increased ash cohesivity.

15C.4 Flue gas conditioning equipment


15C4.1

S03 FGC systems

15 C 4.1.1 Conventional sulfur trioxide generators. With very few exceptions, all commercial S03 FGC systems installed to date are based on
catalytic conversion of sulfur dioxide (S02) which is either supplied in liquid
form or obtained by burning elemental sulfur. Elemental sulfur is the
preferred feedstock for long-term permanent operations because its costs are
lower, but liquid SOl is used for trials and small or short-term situations,
where reduced capital costs can offset increased feedstock expense. The

FLUE GAS CONDITIONING EQUIPMENT

455

catalytic conversion design was chosen over other methods by which S03
may be made available, partly because it is flexible and easily controllable,
but mainly because it minimizes the quantity and the difficulty of handling
hazardous materials resident in the system.
Figure 15C.9 is a diagrammatic representation of a typical sulfur-burning
system for S03 conditioning. Molten sulfur, a common trade item in the
chemical industry available throughout the US, is delivered by thermally
insulated tank trucks fitted with steam coils for melt-out. In locations where
sulfur cannot be delivered in molten form, bagged or bulk solid sulfur may
be supplied for melting on site. The sulfur grade is designated as 'Bright
Yellow' which contains very low levels of contaminating materials. The
exact analysis varies slightly among suppliers, but completely lacks chemicals which could act as catalyst poisons and contains only very minute
quantities of hydrocarbons. Storage in insulated steel tanks with steamblanketing provisions for fire suppression is standard, but concrete-lined pits
are sometimes used. Tanks and molten sulfur piping are heated by steam
controlled to a saturation temperature of approximately 145C (290 OF) at
which the sulfur has ideal flow characteristics. Steam tracing is strongly
preferred because of the ease with which controllable highly uniform
temperatures can be maintained throughout the system. Sulfur metering
pumps are supplied in duplicate so that one may be serviced while the other
is in operation. Even though extensive precautions are taken to maintain
cleanliness of the sulfur and piping, it is impossible to eliminate all possible
chances of contamination and, therefore, impossible to guarantee continuous unimpeded pump operation over long periods of time. In as much as
emission compliance often depends upon continuous operation of the flue
gas conditioning system, it is usually considered that the cost of an installed
spare pump is justified.
Combustion of the sulfur to generate S02 is obtained by introducing
sulfur into an air stream which has been electrically preheated for start-up
purposes to the temperature at which the catalyst becomes active. Since this
temperature exceeds the autoignition point of molten sulfur, burning is
initiated immediately upon the introduction of sulfur and S02 is delivered
to the catalyst for conversion to S03. Roughly 4000 BTUs of heat are
generated per pound (2222 kcal/kg) of sulfur burned. This replaces a portion
of the start-up electrical heat input, and at full system rating all the required
heat to maintain the catalyst at operating temperature is supplied by sulfur
combustion. Clearly, operation of the system in this manner allows the
generation of any quantity of S03 from zero to full system rating as a
function of any selected control signal used to determine the rate at which
sulfur is delivered.
15C.4.1.2 Control systems. In the past, controls for S03 FGC installations consisted of safety and sequencing devices, plus load-following

Air in

Main air
blower

Saturated steam
at35-40PSIG

125-900F

Controlled
to 800F

=H2S04

Figure lSC.9 Sulfur burner typical conditioning system.

gas

Flue

Over temp. Low temp.


alarm set alarm set
to 1100F to 700F

Converter
S + 1/2 O2 =S03

800-1100F

S03 to probes

S03 + H20

Over temp.
alarm set
to 1000F

Sulfur burner
S +02 =S02

Over temp. Low temp.


alarm set alarm set
to 1000F to 775F

Air
heaters

.1

D
D
ESr'~t'r-~.~I+II~I~II~IWIILIU,,~.IWI.I

High
temp.
II
II
alarm set I
I
to 290F Unloading
pump

Precipitator

FLUE GAS CONDITIONING EQUIPMENT

457

apparatus to modulate the rate of delivery of S03 in proportion to boiler


load. The level of S03 to suit a given coal was determined experimentally
by the operator by manual adjustment of a proportioning factor between
the boiler load signal and the S03 generation rate while observing the effect
on ESP operation and emissions. It was generally found that the setting for
a given coal, once determined, was seldom subject to much variation, so it
was possible for most stations to make initial determinations for the various
coals they fired and use those values with little need for subsequent
adjustments. These control functions were implemented with relay logic and
analog controls, and permitted operation with a minimum of attention from
station personnel. They were reasonably satisfactory for most applications.
There were, however, instances where the coal being burned at any given
time might be unidentified or mis-identified, so that previous history was of
no value in determining the injection rate setting and there were some
stations burning 'spot' coals received in small quantities where constant
determination of new settings was required. There were also locations where
difficulty was encountered in determining optimum settings through unfamiliarity with ESP operation. For these reasons, as well as the desirability of
freeing the operators from the task of identifying which coal was being fed
to the burners, and especially the times at which a particular coal started
and ended, effort was directed to automatic determination of the optimum
S03 rate. An initial attempt was based on the idea that the rate would be a
function of the sulfur content of the coal, and that quantity could be inferred
from measurement of flue gas SOz. This proved to be partially workable,
but did not adequately account for ash chemistry or ash quantity differences
between coals with the same sulfur contents. Another aborted attempt relied
on the increase in ESP power input accompanying S03 injection, but
seemed to have an unbreakable habit of running to maximum system rating,
whether that injection rate was optimum or not. Excess injection of S03 will
tend to make the ash resistivity in the ESP too low, decreasing its collection
efficiency, causing severe rapping puffs and an increase in stack opacity
because of S03 carryover. If S03 injection is ahead of the air pre heater, acid
condensation and corrosion may occur in the cold end baskets.
Controls are now available, implemented with solid state apparatus,
which digitally store the operating characteristics of the ESP as a function
of the injection rate setting and, for any coal being fired, determine the
characteristic in use and proceed to the optimum point of that curve.
Periodic perturbations of the setting are made to determine that the set
point remains optimum. If not, a new optimum is determined, and the new
characteristic is stored for future reference. Besides freeing the operator from
the necessity of closely monitoring the fuel and adjusting the set point, these
controls retain a history of operation which can be retrieved for information
on control and system performance.

458

THEORY, PRINCIPLES, APPLICATIONS OF FLUE GAS CONDITIONING

15C.4.l.3 Catalyst for S03 flue gas conditioning applications. The catalyst
for the conversion of SOl to S03 may be chosen from any of the types
developed for the manufacture of sulfuric acid and similar applications.
Vanadium pentoxide is the active ingredient in most of these, and is
classified as a hazardous material. It appears that catalyst life in this service
is on the order of 10 years, so that frequent handling is not required, but
prevention of dust dispersion ingestion of, or contact with, the material,
retention of the material in sealed containers, and disposal by qualified
handlers, are necessary. In spite of its hazardous classification, servicing of
the catalyst bed is neither particularly difficult nor expensive, but it is
nevertheless desirable to do it as infrequently as possible. To this end care
should be taken to exclude dust and water from the air intake to the
maximum extent possible, and to service the air intake filter on a regular
basis. Dust entering the system will tend to plug the inflow end of the
catalyst bed and decrease the available air flow, eventually causing over
temperature tripouts. Water has the effect of breaking down the physical
structure of the catalyst pellets, so that interstices in the bed become plugged
and air flow is restricted. Air intake locations which might be subject to
overspray from cleaning hoses or windblown rain, for instance, should be
suitably shielded or relocated to an absolutely dry position.
15C.4.1.4 S03 delivery to flue gas. The combustion air stream, bearing
the generated S03' will exit from the catalyst at temperatures from 400C
(750 OF) to about 540 C (1000 OF), depending on the rate of S03 production. It is essential that this stream be held above its acid dew-point
temperature all the way through the delivery manifold and injection probes
until it actually issues into the flue gas. If the temperature falls below the
dew-point, the acid will condense out in the manifold or the injection
probes. This is undesirable for two reasons: firstly, none of the acid will
reach the flue gas to do the intended conditioning job, and, secondly, the
condensed acid will corrode the piping and nozzles. Since the S03 concentration is a few percent in this stream, the acid dew-point will be on the
order of 240C (460 OF), but maintaining the delivery end of the system in
a condensation-free state requires that the gas temperature, as it issues from
the injection probe nozzles, be not less than 260C (500 OF). For this reason
the distribution manifolds are heavily insulated, and in addition the injection probes, if installed on the cold side of the air preheater, are thermally
insulated from the flue gas. Also, calculations of the expected temperatures
at the most distant (or coldest) injection nozzle are made for every cold side
installation.
It is probably obvious that avoidance of acid condensation in the
manifold and probe areas requires preheating of the system with hot air
when being started up from cold. It is not quite so obvious that a hot purge
of the system during shut-down is required to remove sulfur products from

FLUE GAS CONDITIONING EQUIPMENT

459

the catalyst, so that acid will not be present in the warm-up gases at the next
start-up. When proper start-up, operating and shut-down procedures are
observed as described here, long manifold and probe life is obtained. Some
early systems are approaching 20 years on-line with the original probe and
manifold parts still in service. Steel sheathing over the insulation of cold-side
probes is subject to the same erosive fly ash attack as other internal parts
of the ductwork, and requires similar periodic maintenance and repair.
15C.4.1.5 Hot-side injection vs. cold-side. Because installation of injection
probes is ordinarily more easily accomplished in the ESP-type ductwork on
the cold side of the air preheater than in the boiler-type construction on the
hot side, most of the present FGC installations inject on the cold side. It
should be noted, however, that hot-side installation has the advantage of
lacking any close approach to acid condensation temperatures in the probes,
and provides excellent mixing and contact between the S03 and the fly ash
as it passes through the air preheater. Hence, cases where long manifold
runs may make it difficult to maintain sufficiently high probe temperatures,
or where there is little cold-side gas travel length for mixing of the injected
S03 with the flue gas, favor location of the injection probes on the hot side.
In determining the desirability of putting the probes on the hot side, possible
S03 condensation in the cold side of the air preheater must be considered.
This is not usually a problem because the injected S03 quantity establishes
air preheater conditions roughly equivalent to those expected from an
unconditioned boiler firing coal containing 1.5% sulfur. If the average cold
end temperature and basket material of the air preheater are compatible
with that assumption, hot-side injection would ordinarily be acceptable.
Probes for hot-side installation are less expensive than cold side because
no thermal insulation is required and the erosion-protective outer sheathing
can sometimes be eliminated. The probe insertion point into the boiler on
the hot side may expand/distort more than a cold-side point when the boiler
is started up and shut down, and this can add to the difficulty of design and
the expense of the manifold system, possibly offsetting savings in probe
costs. As far as operation is concerned, every existing hot-side installation
works as well as, or better than, equivalent cold-side units.
Note that the foregoing applies to injection of S03. Injection of ammonia, to be discussed later, is ordinarily not allowed on the hot side of the air
preheater because of the tendency to plug the preheater with ammoniumsulfate compounds, unless there is little or no S03 in the boiler gas.
15C.4.1.6 One FGC system for multiple boilers. On the face of it, it seems
like a good idea, considering the relatively high cost of the S03 generators,
to use a single generator to feed conditioning gas to several boilers. The
primary objection to such an arrangement is that, assuming operation of the
FGC system is necessary for compliance with emission limits, all the boilers

460

THEORY, PRINCIPLES, APPLICATIONS OF FLUE GAS CONDITIONING

served by whatever portion of the system is common must be taken otT-line


simultaneously before the common portion can be shut down for service.
The high availability figures compiled by existing installations are based on
periodic (at least, annual) boiler outages during which accumulated maintenance requirements of the conditioner system can be met.
If, however, measures can be taken to maintain emission compliance with
the FGC system otT-line, or if simultaneous removal from service of all
boilers on the common system is not objectionable, then a proportion or all
of the S03 generator may be made common to more than one boiler. The
most straightforward way to do this is by making the blower and air heating
apparatus common and providing individual burners (if used), catalytic
converters, manifolds and injection systems for each boiler. Control of such
an arrangement is uncomplicated, and manifolding and injection requirements are identical to those for single boilers. The possible alternative
arrangement where the burner and catalyst are also made common appears
to otTer cost savings, but is actually difficult to implement unless simplifying
assumptions, such as operating all common boilers at the same output level,
etc., can be made. To retain full flexibility in boiler operation it is necessary
to provide valving in the manifold system. Careful heat tracing of the valves
and the portions of the piping which may become dead-ended while still
exposed to S03-containing gases is required to prevent acid condensation.
Cold spots in valve bonnets are not easily avoided. Acid-resistant valves are
expensive. Another less severe problem with this system concept is that fully
flexible operation of the individual boilers requires a relatively complicated
control system. In general, this approach is not recommended.

15C.4.1.7 Patterned or 'tailored' injection. Ideally, the amount of conditioning gas delivered to any single injection nozzle in the flue gas stream
should be proportioned to the amount of ash passing through that nozzle's
treatment area and adjusted for the flue gas temperature at that point. It is
entirely possible to make comprehensive sets of measurements which determine these values, at least for one set of operating conditions, and it is also
possible to divide the delivery of conditioning gas among the injection
nozzles in accordance with such measurements. In practice, a full scientific
treatment of this type is never done, and is rarely even approximated. The
reason, aside from the variability of the target conditions, is that ESPs have
a considerable degree of built-in tolerance for random variations. Witness
to this fact is provided by many existing installations where flue gas
conditioning was neither needed nor used. It is common in these units to
find that a wide spread of temperatures across the air preheater outlets is
passed through to the ESP inlet along with non-uniform dust concentrations, in spite of which the ESP exhibits excellent operation. In general, inlet
conditions which meet industry criteria for satisfactory ESP operation allow

FLUE GAS CONDITIONING EQUIPMENT

461

treatment of a flue gas conditioning system on the basis of uniform


temperature and flow distribution. Another way to state this is that
situations which could require special precautions for distribution of conditining gases usually also require modifications to permit satisfactory ESP
operation, and incorporation of the latter will usually eliminate the former.
15 C 4.1. 8 Load cycling. The burner and catalyst chambers of FGC
systems are lined with refractory materials which can be damaged by
thermal shocks. Recommended procedures for starting a system from cold
conditions require a gradual warm-up procedure over several hours. Some
modern boilers are capable of cycling from stand-by to full load in as little
as 45 min. To permit the FGC system to follow this type of load change, it
is held in stand-by condition with internal temperatures maintained at
operating levels by the start-up heat source. For economy, air flow rates in
stand-by mode may be reduced from normal operation, thereby reducing
heat input. The FGC unit is capable of going from stand-by to full output
in 30 min, and can therefore accommodate the requirements of boiler
cycling. Because of the length of time required for a cold start it is normal
practice to hold the FGC system in stand-by when the boiler is in a stand-by
or banked-fire condition, and not go to a cold shut-down unless the boiler
will be off-line for at least a few days.

15C4.2 Ammonia FGC systems-design features


15C4.2.1 Gaseous or anhydrous ammonia systems. The anhydrous ammonia generating systems generally consist of a liquid ammonia unloading
station, combined storage/vaporizer tank, ammonia system skid and a
manifolded injection system.
Liquid anhydrous ammonia is received in a tank truck. The ammonia is
fed into the storage/vaporizer tank by means of a flexible transfer line. The
anhydrous ammonia storage/vaporizer tank produces a supply of ammonia
vapor at a constant pressure by maintaining boiling liquid ammonia at a
preselected pressure in the tank. External electric immersion heaters with
solid-state controls maintain a constant pressure in the tank as ammonia is
vaporized. Upon withdrawing gaseous ammonia, the pressure and temperature will rapidly shift to the control point conditions of 4 Barg (60 psig) and
4.4 C (40 OF). High and low pressure or liquid level will cause alarms. The
vaporizer tank is fully automatic in operation and has a control range from
o to 100% capacity.
The system skid includes a blower to provide the necessary filtered
process air which mixes with ammonia gas from the vaporizer. Air discharging from the blower passes through an air flow indicator before mixing with
the gaseous ammonia. Ammonia flow rate is adjusted automatically depending upon boiler load.

462

THEORY, PRINCIPLES, APPLICATIONS OF FLUE GAS CONDITIONING

The ammonia/air mixture is fed to a separate manifold system with


probes for distribution into the gas stream. Each probe will contain properly
sized and spaced nozzles, which, in turn, inject the mixture at high velocity
to ensure good mixing of ammonia and flue gas.

15C4.2.2 Aqueous ammonia systems. Basically, the aqueous ammonia


generating systems include an air blower, an air heater, controls and mixing
chamber.
Liquid aqueous ammonia is supplied from the storage tank to the system
skid by a metering pump. The flow of ammonia will then be controlled and
delivered to the vaporizer vessel. A centrifugal blower, located on the system
skid, provides the necessary filtered process air which is heated through an
air heater box assembly, and injected into the mixing chamber with
ammonia spray from the nozzles. Air discharging from the blower passes
through an air flow indicator before entering the heaters. Ammonia flow
rate is adjusted automatically depending upon a boiler load signal.
The ammonia/air mixture is fed to a manifold system consisting of probes
for distribution into the gas stream. Each probe contains nozzles, which
inject the mixture at high velocity to ensure good mixing of ammonia and
flue gas.
15C4.3

ESP power consumption

During most of the history of ESP collection of fly ash, it was considered
that use of the full available output of the high voltage supplies indicated
good operation and high efficiency collection, and this was highly desirable.
In later years attempts to collect high resistivity dust, without resort to flue
gas conditioning, lead to the construction of ESPs of monumental size,
complete with row on row of hoppers, high voltage power supplies and ash
handling apparatus. Whether or not these passed their guarantee test
requirements, quite a few were eventually fitted with FGC which, among
other things, promptly caused all the power supplies to go to full rated
output. What once had been considered good ESP operation was now
rightly considered outrageous consumption of station auxiliary power.
Fortunately there is a cure for this situation.
Control units for the high voltage power supplies are now available, from
most ESP manufacturers, which provide a feature called 'intermittent
energization' or 'skip-cycle control'. Although the claim that this type of
control materially improves collection of high resistivity dust has proved to
be somewhat optimistic, there is a definite economic justification for its use
in conjunction with FGC. Once the fly ash resistivity has been adjusted by
FGC to an optimum value, high collection rates may be maintained while
ESP power consumption is greatly reduced by use of the 'skip-cycle' feature.
Power reductions to levels on the order of 20~30% of the unmodified power
input have been obtained without noticeable increases in emission.

FLUE GAS CONDITIONING IN HOT- TO COLD-SIDE CONVERSION

15 C. 4. 4

463

Mixing requirements

The length of travel of the flue gas in the ductwork after the injection point,
required to provide essentially complete mixing with the S03' is often glibly
cited as 'one second mixing time' or 'ten times the nozzle spacing'. These
rules of thumb are derived from experiments showing that in turbulent flue
gas flow, complete mixing with another gas, injected through a bank of
nozzles arrayed as a uniformly spaced grid, occurred at a distance downstream of the grid equal to about eight to ten times the nozzle spacing. In
FGC systems a nominal grid spacing of 3 ft (94 mm) has been found to be
a reasonable compromise. Ten times this spacing gives a mixing distance of
30 ft (9.14 m), which at typical duct velocities of 60 ft/s (18.3 m) allows only
0.5 s mixing time, and this is adequate provided you have a straight run of
uniform size that long. Ordinarily you don't, hence the conservative quest
for 1 s. Expansions, contractions and bends in the ductwork interfere with
net mixing patterns. If the ESP ductwork is being modeled for flow
visualization, it is desirable to include an examination of the mixing
behavior which may be expected. Modification of the spacing, location and
gas delivery rate of the nozzles are steps which may be taken to cope with
less-than-optimum mixing situations.
lSC.S Application of flue gas conditioning in converting hot-side fly ash
precipitators to cold-side operation

In some existing precipitator installations the difficulties associated with


hot-side operation have become so great that complete revision of the
emission control system is necessary to permit economical operation in
compliance with emission limits. Studies have shown that conversion of the
ductwork arrangement to reposition the precipitator to the cold side of the
air preaheater, coupled with the use of sulfur trioxide injection to solve the
high resistivity problem, is often the most economical alternative.
Increases in precipitator ash collection and boiler thermal efficiencies
over the hot-side design basis becomes available. Net economic gains from
such a conversion can provide payback of the conversion over reasonable
time periods, especially, if the hot-side installation incurs generation losses
from outage time for precipitator mechanical problems or from derating of
output to meet emission limitations.
15 C. 5.1

Introduction

The ills which can afflict hot-side precipitators are well known. With all the
recited problems, why in the world do we have so many units in our boiler
plants? The underlying reason can be succinctly stated in two words-ash
resistivity. In the 1970s the hot-side concept became temporarily popular

464

THEORY, PRINCIPLES, APPLICATIONS OF FLUE GAS CONDITIONING

because burning low-sulfur coals became common, either as a way to meet


restrictions on sulfur dioxide emissions or because they were the only
economical fuels available to the plant. There were at that time only two
ways proven in utility service to successfully collect the resulting high
resistivity fly ash: big cold-side precipitators, or smaller ones on the hot side.
Figure 15CA depicts the typical form of the response of precipitator ash
collection rates to variations in ash resistivity. The dashed portion on the
left indicates reduced collection due to rapping and re-entrainment losses
when the resistivity is quite low. The position at which this etTect appears
depends on gas velocity and rapping practice. As resistivity increases a
region of maximum collection rate is found, usually around 5 x 10 10 Q-cm.
As resistivity increases further the collection rate drops otT, until finally at
high resistivities a level of low collection rate is found.
Figure 15C.2 is a typical plot of resistivity versus temperature for fly ash
in the form developed by Dr R. E. Bickelhaupt [4]. It is clear that operation
at cold-side temperatures will produce resistivities with this ash that are
considerably higher than the 5 x 1010 Q-cm value at which (for this set of
curves) best collection rates would be obtained. One possible choice for
collecting this ash, then, would be to build a precipitator large enough to
achieve the desired efficiency at the reduced collection rate. Unfortunately,
in the 1970s (and maybe now) selection of the correct precipitator size in
this operating mode was somewhat uncertain and vendors' sizing margins
for guaranteeability made for excessively large and expensive units. However, examination of Figure 15C.2 shows that the resistivity comes down
very nicely to the optimum level if the temperature is raised above about
350 DC (650 OF). With care taken to postpone the onset of rapping and
re-entrainment losses at low resistivities as much as possible, this was
expected to be an area of high collection rates, handicapped only by the
necessity of sizing the precipitator to accommodate the gas volume generated at the chosen temperature.
Not all hot-side precipitators exhibit severe problems. Some worked well
as-built, and others as-modified. But some didn't. The assorted structural
problems arise from improper execution of high temperature designs, not
from the concept of hot-side operation. However, two major problems with
the concept itself have surfaced. First, the 'sodium depletion' phenomenon
was unknown when these units were designed. It was only after ash
resistivities which were supposed to be low turned out to be drifting high as
operating time was accumulated, that Dr Bickelhaupt and the Southern
Research Institute proposed and documented this nasty bit of ash behavior
in precipitators [21]. Secondly, the procurement specifications for these
precipitators almost invariably spelled out performance requirements at full
load, but were silent on the subject of reduced loads. When load drops,
temperature decreases, resistivity increases, and the units tend to run out of
the window of good operation and produce excess emissions at lower loads.

FLUE GAS CONDITIONING IN HOT- TO COLD-SIDE CONVERSION

465

As the units age, cycling often becomes common, adding to the owners' woes
[22].
In more than a few cases these problems with the hot-side concept, added
to those of unsatisfactory execution, have led to consideration of the
possibility of switching problem units to cold-side operation. If this is done,
there are two basic approaches to attaining adequate collection of the fly
ash. One is to accept the ash resistivity as produced by the boiler-generally
high, the reason hot-side was chosen in the first place. Sufficient fly ash
collection with this approach can be obtained in three ways: by building a
new baghouse which is insensitive to resistivity; or by extending the existing
precipitator, again hoping to make it big enough. In the 400 to 600 MW size
range, current installed costs for baghouses or new precipitators are $25
million and up, mostly up.
The second approach for cold-side operation is to adjust the ash
resistivity to an optimum value by sulfur trioxide flue gas conditioning
(FGC). Current costs for complete conversion including FGC, again in the
400-500 MW size range, appear to be from $9 to $12 million per boiler unit.
Factors involved in this choice are the subject of this chapter.
In some cases the existing precipitator, or that precipitator with an
addition, may be large enough for cold-side collection of a particular ash
without FGC. This occurs when the expected cold-side ash resistivity is low
enough to be in the good operating range. Such situations should be
thoroughly examined to be sure that the expected resistivity will be achieved
over the full range of coal and ash analyses, and that reduction of resistivity
to the desired point is not heavily dependent on the sodium content of the
ash. The reason for concern with reliance on sodium content to produce the
desired resistivity is that sodium in coal is believed to result from the salt
content of water to which the coal was exposed during formation. Considerable variation often occurs as mining progresses. Also, the sodium depletion
phenomenon is at work at low temperatures as well as high, and the familiar
drift of resistivity with operating time can occur, resulting in high apparent
ash resistivity. However, if these factors have been taken into account and,
if flexibility with respect to coal sources is not a major consideration, the
existing or enlarged precipitator may be adequate without FGC. Costs for
this option are, of course, highly site specific.
15 C. 5.2

Flue gas conditioning

Table 15C.2 shows design figures typical of hot-side precipitators, compared


with the expected performance on a very conservative basis for the same
unit converted to cold-side with FGC. Because the decreased gas volume
increases the operating SCA, and the increased gas density allows higher
operating voltages, a significant reduction in emission occurs with the same
ash loading at the precipitator inlet. Unit 1 at the Pleasant Prairie station

466

THEORY, PRINCIPLES, APPLICATIONS OF FLUE GAS CONDITIONING

Table 15C.2 Typical factors- hot-to-cold conversion

Flue gas temperature eF) Cc)


Flue gas volume (%)
SCA Ft 2/lOOO ACFM (m2/m3/s)
ESP migration velocity
wl.O (cm/s)
wO.5 (cm/s)
ESP efficiency (%)

Hot

Cold w/FGC

800 (427)
100
300-350 (59-69)

300 (149)
60
500-580 (99-115)

8.5
44
99.5

50
99.9

6.5

of Wisconsin Electric Power Co. has operated for nearly 7 years in this
mode (converted to cold-side before start-up) and has reliably and continuously performed at higher migration velocities and lower emissions than
shown by Table 15C.2 [23]. Unit 2 at the same plant has operated similarly
for more than 1 year. Both units burn coal from the Powder River Basin of
Wyoming, and are deep-cycled on a daily basis. FGC has made hot-to-cold
conversion possible with a major gain in compliance margin for a fixed
precipitator size.

15 C. 5. 3 Installation design
15 C. 5. 3.1 Probe location. A first consideration for FGC installations is
location of the injection probes. In a hot-to-cold conversion there is some
freedom of design of the new ductwork connections from the economizer to
the air preheater and from the APH to the precipitator inlet. Advantage
should be taken of this flexibility to avoid handicapping FGC operation
through lack of reasonably satisfactory probe positions. As discussed, it is
desirable that there be at least 1 s residence time of the injected S03 in the
flue gas ahead of the precipitator face. At least half of this should be in
non-expanding ducts ahead of any divisions. Vanes should be at least one
equivalent duct diameter distant from the probes, both up and downstream.
A sizeable temperature spread usually exists across the gas outlet of most
APHs, so, where possible, mixing of streams from multiple APHs should be
done so that the individual temperature distributions tend to cancel out and
produce final streams of relatively uniform temperature; this is important
since ash resistivity and S03 injection rates are temperature-related and
more uniform temperature promotes more uniform precipitator operation.
In plants where the primary APH(s) is not banked with the secondaries,
similar precautions should be taken in positioning the ductwork connections which join its gas stream into the main gas flow.
In some cases it is impossible to establish a reasonably satisfactory probe
location downstream of the APH. In these cases locating the probes on the

FLUE GAS CONDITIONING IN HOT- TO COLD-SIDE CONVERSION

467

hot side of the APH is an alternative. There are numerous existing


installations where this arrangement has operated for a number of years
without problems. Average cold end temperature considerations and soot
blowing requirements for the APH then become roughly equivalent to those
experienced by boilers burning coals having sulfur contents in the 1.5-2%
range.
The preferred orientation for probe insertion is vertically downward.
From the standpoint of system operation, there is no difference between
vertical, horizontal or slanted orientations, and all have been used successfully. However, probe inspection and maintenance is easier with a topmounted vertical arrangement.
Few, if any, of the FGC installations now in operation fully comply with
all the probe location requirements set forth here. There is, therefore, some
latitude available to accommodate the physical restrictions inherent in
existing plants. Following these guidelines to the greatest reasonable degree
will help in obtaining best operation and simplest maintenance.

J5C5.3.2 Burner-converter. Having fixed the probe location, the next


step is locating the burner converter unit. This is typically about 8 ft wide
by 8 ft (2.44 m) high by 20 ft (6.1 m) long, weighs about 13 tonnes, and is for
indoor service. Three feet of space on all four sides should be provided for
access. The first consideration in locating this unit is that it be as close as
possible to the probe location, preferably within 17 to 24 m (50 to 80 ft) to
minimize the installation expense and heat loss from the S03 hot gas piping,
and, secondly, that the station operators should have easy access to the
chosen position.
J5C5.3.3 Tank and pumps. The sulfur storage tank and the pumping
unit are relatively free of restrictions on their location. They will be at grade,
and the pumping unit requires weather protection. Truck access for sulfur
delivery is, of course, required.
15C5.3.4 Piping. The hot gas piping which conveys the S03 gas from
the converter outlet to the injection probes is stainless steel, and is heavily
insulated to prevent the S03 gas temperature from dropping below its aid
dew-point before injection into the flue gas. This pipe run must be designed
and supported to accommodate thermal expansion effects arising from
temperature changes from ambient to 540C (1000 OF). Also, pipe routing
and the positioning of anchor points and supports must be chosen to
prevent overstressing due to ductwork movement, wind and seismic loadmgs.
Pipe work for molten sulfur is steam-jacketed throughout its length. Its
design is less demanding than hot gas piping, since its operating temperature
is approximately 140C (285 OF), but care must be taken to provide for

468

THEORY, PRINCIPLES, APPLICATIONS OF FLUE GAS CONDITIONING

proper condensate drainage from the jacket. Concentric sulfur-inside-steam


arrangements and straight-through valve jackets are used to minimize
pressure drop in the steam system and to avoid the cold spots and steam
flow restrictions which appear at flange-and-jumper joints. Selection of
proper steam flow patterns and points of steam supply to the system is
important so that freezing and thawing of the sulfur does not develop
excessive internal pressures at pumps or instruments.
J5C.5.3.5 Steam supply. A single 600 MW boiler installation will require
about 70 kg/h (150 lb/h) of saturated steam at 2.4 bar (35 psig) for cold start
conditions (frozen tank and lines) and about 23 kg/h (50Ib/h) during
normal operation. Naturally these figures vary with tank size, pipe length,
quality and type of insulation, etc. This is usually taken from a low-pressure
station header. Desuperheating may be required, depending on source
conditions. For these relatively small quantities saturator-type de superheaters appear to be inexpensive and reliable. Small auxiliary boilers are
sometimes supplied for start-up of FOC systems in isolated plants.
15 C. 5.4 FGC Operation

An objective of the design of S03 FOC systems has, from the start, been
minimization of the amount of operator attention required. Most of the
operator input for normal running takes place at the burner-converter
control panel, with occasional operations at the pump unit control panel.
However, a cold start which occurs only after an extended outage
requires that the steam supply system valves be opened to initiate thawing
of sulfur about 4 days before the system is to be put into service. Then,
about 2 days in advance of boiler operation, warm-up of the burnerconverter unit is commenced by operating the 'Standby' push-button at the
control panel. This starts an automatic program which gradually heats the
refractory materials and catalyst chamber to operating temperature. When
this is complete, S03 delivery may be begun by pushing the 'Run' button,
and the system will go on-line. Adjustment of the S03 injection rate is made
by trial-and-error variations, starting from an estimated value or from a
setting determined by previous experience, and using opacity measurements
or precipitator power input as a criterion for evaluating the effect of ongoing
rate changes. To stop injection for a short outage the 'Standby' button is
pushed and the unit reverts to a hot ready-to-run condition. For a complete
shutdown the 'Purge' button is pushed and the FOC operation is terminated
after automatically stopping sulfur flow and blowing all sulfur products out
of the hot gas pipe and probes.
When fuel and ash analysis changes, a repeat of the trial-and-error
procedure for determining the best injection rate is again necessary. In most
installations there is a fairly broad range of injection rates at which ash

FLUE GAS CONDITIONING IN HOT- TO COLD-SIDE CONVERSION

469

resistivity and precipitator performance are near optimum, and determination of a satisfactory injection rate is easily and quickly accomplished. For
stations burning a wide variety of coals, additional provisions for automatic
adjustment of injection rate may be desirable.

15 C. 5. 5 Economics
Experience indicates that the cost of a single-boiler FGC system is about
10-15% of the cost of converting the precipitator from hot to cold. The
percentage decreases for multiple boiler systems because of shared pump
skid, tank, and engineering costs. Where the existing air preheater can be
used in the new system without relocation, the outage time for conversion
of the existing precipitator with FGC is almost invariably less than that of
any other option.
When the air preheater must be moved and/or replaced, longer outage
times may be required, perhaps making other options competitive insofar as
outage time to convert is concerned. In any case, minimum outage time
requires careful and knowledgeable scheduling of prefabrication and installation operations.

Table lSC.3 Factors for preliminary FGC cost estimation sulfur-burning systems'

Equipment cost b
Cost to instaW
Sulfur usage lb/h (kg/h)
Maximum d
Typical e
Electrical power (kW)
Connected
Operating
Steam usage f lb/h (kg/h)
Max., cold start
Operating
Operator (man-h/year)
Maintenance (man-h/year)
Spare parts cost/year ($)

300MW

600MW

1,200,000
See note

1,550,000
See note

101 (46)
51 (23)

205 (93)
103 (47)

170
97

350
195

117 (53)
33 (25)
300
300
5000

176 (80)
48 (22)
300
300
5000

'Single boiler, 0.4% sulfur coal, 10-20% ash, 20 ppm S03'


b1987 $ F AS East Coast port, includes supply of storage tank and preparation of preliminary
installation design plus services to inspect construction and to supply start-up assistance,
operator training, and instruction manuals. For additional boilers on same system add 65% of
single boiler cost per added boiler.
CInstallation cost for material and labor to install system and make piping and wiring
connections between factory units and to plant utilities is approximately 35% of equipment
cost.
dAt full output capacity (20 ppm).
eAt 10 ppm output.
fSteam at 35 psi g, saturated (2.4 bar).

470

THEORY, PRINCIPLES, APPLICATIONS OF FLUE GAS CONDITIONING

A preliminary estimate of FGC system costs can be made from the figures
in Table 15C.3, Note that the figures given include services to perform
preliminary site surveys, prepare recommendations for equipment locations,
and furnish piping support and routing information, intermittent construction advice on site, start-up assistance and operator training, instruction
books and spares recommendations, along with on-site vendor-owned
spares to support warranty requirements. Other services available include
estimation of cold-side precipitator performance and emission opacity, and
extra costs after start-up services, ranging from periodic inspection and/or
maintenance, with or without parts supply, to continuous on-site operation.
S03 FGC does only one thing - it eliminates problems due to high ash
resistivity. The conversion program should include funds to correct all the
accumulated precipitator faults which have been recognized in previous
operation, or have been found in internal inspections, so that trouble-free
performance will be obtained on a long-term basis. Also, provision should
be made for insulator venting and heating, and for hopper heating and level
control- items which were often minimally present or omitted from hotside designs. Some ash-handling systems have been based on the use of
precipitator hoppers as ash storage bins. This has been found to be poor
practice, and modifications to allow constant ash removal should be
installed [24].
Precipitator power controllers should be state-of-the-art in every respect,
especially in their ability to control sparking and to follow variations in
operating conditions. Obsolete or marginal controllers should be replaced,
and power savings possible through the use of precipitator energy management systems or 'intermittent energization' (IE) control units should be
evaluated. With FGC all the power supplies downstream of the first two
fields tend to go to maximum rated power input. Use of IE controllers,
which interrupt conduction of power to a precipitator field during a fraction
of the line-frequency cycles, permits a power saving of one-half or more of
the total rated kilowatts of all the transformer rectifier units on the
precipitator without sacrificing collection efficiency [25]. This is sufficient,
in many cases, to payoff the investment in new or modified power
controllers in an attractively short period of time.
Factors which contribute to amortization of the hot-to-cold conversion
expense are the increase in thermal efficiency from decreased heat losses
ahead of the air preheater (to the extent that heat can be recovered in the
APH), possible decreased ID fan horsepower from decreased air in leakage
(often high on hot-side units), decreased lost generation from derating for
emission compliance or from wash-down outages, and, for some, elimination
of penalties imposed for excess emissions. Other positive factors more
difficult to quantify include decreased precipitator maintenance, better stack
appearance, greater choice of fuel sources, and freedom to make load
reductions or to cycle load without emission excursions.

FGC SYSTEMS-OPTIMIZATION

471

15 C. 5. 6 Conclusions
It is concluded that S03 FGC provides a solution to the ash resistivity
problem which gave rise to the use of hot-side precipitators in the first place.
Use of the existing precipitators converted to cold-side operation is, in many
cases, more economical than possible alternative cold-side conversion options, as well as being more economical than continuing with the existing
hot-side operation. Experience with more than 200 retrofitted boiler units
shows that cold-side FGC can be relied upon. In summary:
1. FGC allows maximizing precipitator performance, independent of natural ash resistivity.
2. FGC allows a precipitator switched to the cold side to outperform its
hot-side design basis.
3. FGC plus hot-to-cold conversion is often far more economical than other
options.
4. Theory and operation of FGC has been repeatedly proven in both new
and retrofit installations, and is both predictable and reliable.

15C.6 FGC systems - optimization


The main objective of the optimization methods is how to extract the
maximum amount of unbiased information regrading the factors affecting
an 'injection rates - opacity' relationship from as few tests as possible. A
term 'optimization' for the purpose of this discussion is defined as a search
for conditioning agent(s) injection rate(s) to provide the best ESP performance, i.e. lowest stack emissions and opacity.
Typically, the FGC systems allow for an independent adjustment of the
S03 and/or NH3 injection rates to evaluate their impact on the precipitator
performance (stack opacity). Therefore, one could adjust the injection rates
in a systematic manner to learn which factors have greatest impact on the
resultant opacity. Using this information, the injection rates can be constantly improved until optimum performance is obtained.
There are three fundamentally different approaches to the FGC system
optimization:
1. theoretical approach, based on the computer models for estimating a S03
vs. flue gas temperature relationship;
2. procedural approach to adjustment of one variable at a time; and
3. statistical methods of experimental design while working with more than
one independent variable:
(a) simplex method
(b) central composite design

472

THEORY, PRINCIPLES, APPLICATIONS OF FLUE GAS CONDITIONING

15 C 6.1

General

One assigned person should be responsible for all the injection rates at any
given site. The reason for this is that experience will make it possible to
recognize the effects of change and the proper steps to be taken with the
leas't amount of delay and experimentation.
It is important to make sure that the meters on the precipitator power
supply controls are in good operating order. All power supply controls
should be in good operating condition with current limits set to the primary
current rating of the transformer-rectifier sets, and they should run under
'automatic' control. Also, opacity monitors should be in calibration and
operating correctly. All this is necessary because these are the devices to be
used for references in making injection rates adjustments.

15C6.l.1 Definitions. Figure ISC.IO represents a specific case where two


opacity meters register instantaneous opacity side-by-side.

Rapping spike
Baseline
Baseline trend

15 C 6.2

A single peak registered on the opacity chart. The


rapping spikes are usually the best indication of rapping
mechanism impacts.
The line approximating the lowest portion of the opacity
trace.
The direction of the baseline (either downwards when
opacity is decreasing or upwards when opacity is increasing).

Theoretical approach

Discussion. The calculation methods developed by Dr Bickelhaupt were published in a form which estimates the resistivity of fly ash in
the presence of a given concentration of sulfuric acid. Simple mathematical
manipulation of the Bickelhaupt formulae makes it possible to calculate the
acid concentration required to bring about a desired change in resistivity.
Using the latter as a starting point, Coe and Krigmont developed a method
for S03 injection rate estimation [10]. Accurate prediction of injection rates
for S03 flue gas conditioning systems is now possible for the majority of
stem coals in use. The described procedure may be the first step in any FGC
system optimization. Coal and ash analyses are required for input to the
calculation procedure. Equipment size and feedstock requirements can be
closely determined, given knowledge of the range of fuel properties to be
accommodated.
15 C 6.2.1

(a) 0500

0600

r-

0700 1

0800 t-

:]

1100

1200

1300

--=-

..-

(b)

0500

0600

0700

0800

0900

1000

1100

1200

....

Rapping spikes

Figure 15C.1 0 Two opacity meters registering instantaneous opacity.

-C

......

~
....:J!!I!II'-

-'=......

Rapping spikes

..

474

THEORY, PRINCIPLES, APPLICATIONS OF FLUE GAS CONDITIONING

15 C 6. 3 Procedural approach
15C6.3.1 S03 injection optimization. A need for S03 injection is usually
shown by increasing opacity and decreasing power input to the precipitator,
as compared with clean start-up conditions. The initial amount of S03
required may be obtained either from the computer models or assumed
based on previous experience of the 'optimizer'. The former will be estimated
based on the most recent coal and ash data and is not an exact value but
should be approximately correct. The correct value is the one which gives
the least emissions, to be determined during optimization trials. The ppm
amount of S03 being injected is typically set by the ratio station.
Generally, fly ash resistivity increases with increasing temperature and
decreases with a drop in temperature. Resistivity too high for best operation
causes reduced precipitator power input and increased opacity. Resistivity
which is too low can lead to excessive rapping losses. A proper S03 injection
rate will eliminate the high resistivity effects without bringing about excessive rapping losses. Rapping losses show up as 'spikes' on the opacity chart.
Some spiking may be present at the best injection setting, but should not be
so great as to raise the average opacity more than a few points above the
opacity base line.
To find this balance point it is suggested to start with the recommended
S03 injection rate. This should begin to bring down the opacity and
increase the current delivered by the power supplies. Where a layer of high
resistivity ash has been laid down on the plates, there may be a considerable
time delay before the S03 penetrates into the layer and reduces its
resistivity. There have been delays as long as 5 days, but a change is usually
seen within 8 h. If there is no change after 24 h and the opacity trace is not
showing high rapping spikes, the injection rate should be increased in steps
of about 20% of the previous rate, waiting 8 h after each change, until
operation starts to improve or the rapping losses become excessive.
From here on, trial-and-error adjustments of the ppm rate are made, first
upwards, and then, when total emissions (by stack opacity monitor) start to
increase, start to back down the ppm rate until the total emissions start to
rise. At this stage, each step of adjustment should be about 10% of the rate,
or 1 ppm, whichever is smaller. On the first few steps watch the opacity
monitors and see how long it takes after a step change for the precipitator
operation to settle out to a constant emission level. Allow this length of time
plus 1 h between changes of the rate. There is nothing wrong with a longer
waiting time between steps, just be sure to wait long enough to see the full
effects of the previous change. Usually 8 h per step is enough, but go by the
monitors. The final result is to determine the lowest injection rate at which
minimum emission is obtained.
15C6.3.2 NH3 injection optimization. A need for NH3 injection is typically shown by excessive rapping losses (high rapping spikes on an opacity

475

FGC SYSTEMS-OPTIMIZATION

chart) remaining after S03 injection rate has been optimized (in the case of
the dual FGC system), or existing due to the 'naturally' high amounts of the
S03 present.
To find the best combination it is suggested to begin to increase the NH3
injection rate in 1 ppm steps beginning with 1 ppm.
When NH3 is introduced with S03 present, a fine fume consisting of
different ammonia sulfate products is formed. The effect of these products
can be observed in the increased space charge and improved ash coagulation
and agglomeration characteristics. The former may be detected by increased
voltage (in the first field predominately), while the latter should command
reduction in the rapping losses (reduced rapping spikes on opacity charts).
The S03 - NH3 chemical reaction obviously will reduce the amount of
S03 available for the resistivity modification, which may lead to reduction
of precipitator power consumption and increase of the stack opacity. That
will be a signal to increase the S03 injection rate by the same amount as
the latest change of NH 3.
Further, continue to make trial-and-error adjustments, first to the NH3
injection rate followed by the same ppm of S03 should the precipitator
power consumption begin to decrease and/or stack opacity begin to
increase. Follow the guidelines as described in 15C.6.3.1. The final result is
to determine the best combination of S03 and NH3 to obtain lowest
possible average stack opacity by eliminating or significantly reducing
rapping losses.
.
As long as the coal supply and boiler operating conditions remain
relatively constant, the ppm rate should stay the same over a long period of
time. However, occasional trials by raising and lowering the rate one step
should be made to make sure operational changes have not occurred. For
all such tests, both initially and later on, record boiler operating data and
precipitator power supply readings at each ppm rate for future reference.
Opacity monitors are the most convenient and informative references for
this kind of adjustment, but precipitator power data can serve the same
purpose.
15 C. 6. 4 Statistical approach
The first step in finding the optimal combination of S03 and NH3 injection
rates is to specify the goal function. It must take into account the effects of
all the important real life variables, with the exception of the independently
controlled injection rates. As the first approximation it may be the stack
average opacity itself. As for the second approximation, one may additionally incorporate boiler load change and specify the goal function as:
.
Opaclty factor

Constant

%)

= --------;-----------;-

Bl
1 d 1 (Opacity,
01 er oa x n
100

(lSC.12)

476

THEORY, PRINCIPLES, APPLICATIONS OF FLUE GAS CONDITIONING

Low

Figure 15C.ll Simplex method layout.

The boiler load and the stack opacity could be further normalized to
account for the boiler maximum continuous rating (MeR) and established
opacity limit. After the goal function is established, the optimization task is
to minimize this function in the two-dimensional matrix (n = 2) where the
coordinates are the S03 and NH3 injection rates in ppm,
J5C6,4, J Simplex method. In the most simplified form, the Simplex
method for two independent variables (S03 and NH3 rates, ppm) might be
realized on the two-dimensional matrix (Figure 15Cll), At the starting
position, one has first to find the value of the goal function (a function that
has to be minimized - opacity or opacity factor) in two randomly selected
points on the matrix, The difference between the appropriate injection rates
for these points is usually chosen in the range 2-4 ppm. Next, one has to
find the value of the opacity in the third point, which is located on the
matrix so that it forms a nearly normal triangle with the two previous
points. That is, it is equally distanced from the above-mentioned points. The
three values of goal function (opacity) are then compared and the highest
one is chosen. The point where this value was observed has to be excluded
from the optimization process. To find the next, or 'third' point for the
triangle (simplex) on the matrix, one moves in the direction opposite to the
excluded point. The position of this new point may be easily found by
reflecting the rejected point symmetrically with respect to the line connecting the two remaining points. When the value of the opacity in this point is
found, the values of opacity for the three recent points are compared. Again
the 'worst' point is located and rejected. This procedure is repeated continuously until the triangle will start to rotate on the same zone of the matrix.
Then one may decrease the distance between the points (i.e. the dimensions
of the triangle) to locate the optimum point with satisfactory accuracy.

FGC SYSTEMS-OPTIMIZATION

477

One of the most efficient and reliable methods for the search of the
optimal combination of these rates is the modified Simplex method (NelderMead method). The idea of this method is to compare the values of the goal
function in n + 1 (three in our case) points and to move the simplex
(triangle in this case) in the direction of the optimal point. During this
process the simplex (triangle) changes its form to reflect the specifics of the
response surface shape - it extends in the direction of surface slope, rotates
near the surface cavities and compresses itself near the optimum point.
In each iteration step (while using this method), one has to find a new
triangle on the plane (S03' ppm-NH 3, ppm). Following the above procedure, at the start of the process the goal function values at three points
(low, medium and high at Figure 15C.ll) are determined. Next the coordinates of the reflection point are calculated and the value of the goal function
in it is found. If this value (opacity reflection) is less than the lowest value
for the three points used (opacity low), the coordinates of the extension
point are calculated and in the next experiment the goal function value
(opacity extension) is found.
In either case the opacity reflection is compared with the opacity medium
and based on the results of this comparison we use this reference point, or
go to a new point, called compression. The position of this compression
point is calculated based on the sign of inequality between opacity factors
in the reflection point and the 'highest' initial point. The next step gives us
the next three points that are the basis for the following step. This iterative
process is going on continuously with the duration of these steps equal to
the time required for unit opacity stabilization on any new point.
Theoretically, the optimization process would stop itself when for a
recently found triangle highest and lowest values of opacity factor became
close enough (for a predetermined accuracy). But in real life, due to the
possible changes in fuel, or in unit operating parameters, this process will
continue to look for the current position of the optimum point.
J5C6.4.2

Central composite design method

(a) Overview. Basically, the yield of the dual FOC system optimization
is a function of precipitator performance and, respectively, the stack opacity.
Unfortunately, these two variables often do not affect the resultant opacity
in linear fashion. In other words, it is not 'the more, the better'. Rather, both
of these conditioning agents are related in a curvilinear fashion to the
resultant stack opacity. Thus, our goal would be to optimize the yield surface
that is created by the two independent variables: S03 and NH3 conditioning
agents, and the opacity as a dependent variable.
Figure 15C.12 represents an example of such a surface. Specifically, it is
recommended to use a 'second-order central composite design' [26] to
design an experiment that would allow one to estimate this surface. The two

478

THEORY, PRINCIPLES, APPLICATIONS OF FLUE GAS CONDITIONING

30
28

26

24

0..

22

o 20
18
16

o0
Figure lSC.12 Second-order central composite surface.

horizontal axes in the plot are the two process independent variables (S03
and NH3 injection rates). The vertical axis shows the expected opacity.
Clearly, resultant opacity is not a simple linear function of the two variables.
However, given the information on the plot we would have a fairly good
idea in which direction to proceed, i.e. how to change the injection rates in
order to optimize the stack opacity.
(b) Basic description. Intuitively, it should be clear that in order to detect
curvature in the relationship between the input factor (such as S03 injection
rate) and the outcome-dependent variable of interest (stack opacity), one
would need at least three levels for the respective factor (or, what is called
in statistical science, three design points). Three points are: zero (0) or the
middle point of the injection rate range, for example 10 ppm for a 20 ppm
system; negative one (-1) or one-half of the range under evaluation, for
example it could be 5 ppm for a 20 ppm system; and a positive one ( + 1),

FGC SYSTEMS-OPTIMIZATION

479

for example keeping the same 20 ppm system, it would be 15 ppm value.
Only then can it be decided whether, for example, the change from point
- 1 to 0 and from 0 to + 1 is linear, resulting in proportional changes in the
outcome variable, or whether the relationship is curvilinear, resulting in
disproportional changes in the outcome variable. In the simplest experiment
this is exactly what one would do, i.e. choose three design points (levels of
the independent variables), and explicitly test the significance of the linear
relationship and the curvilinear relationship.
This logic easily extends to multiple factors; however, in that case we
would not have a simple (additive) relationship between two variables (S03
and NH3 injection rates), but the outcome variable of interest (stack
opacity) rather forms a surface over the ranges of the injection rates.
The surface in question is constructed (estimated) entirely from the points
in the design (combinations of setting for different factors). There is the
necessity to include at least three levels (settings for each factor in the study),
otherwise the curvature of the surface could not be detected.
Therefore, to the design points forming square (only two dimensions) or
'cube' points, we have to add so-called 'center points' (coordinates 0,0) to
allow for an explicit test for curvature.
Furthermore, in a second-order central composite design, the 'cube' is
enhanced by a 'star', the design points required to estimate the nature of the
curvature; specifically, to allow estimation of the second-degree polynomial
or quadratic components of the relationship between the factors and the
dependent variable.
The levels for the factors that make up the design (experimental) points
may appear as arbitrary. In fact, these numbers are not at all arbitrary but
rather are carefully chosen so that the 'star' points are arranged to form a
circle, together with the 'cube' points around the center point. Therefore, the
information functions, that is, the amount of information we can obtain
from the design about the surface, is symmetrical about the center point.
Consequently, the design can be rotated (around the center point) in any
direction without gaining or losing any information.
Computationally, this type of the experiment can be analyzed via multiple
regression procedures. The data is fitted in the following model:

Thus we estimate the following parameters:

Po, a constant (intercept);


PI' the parameter describing the linear relationship of y (the dependent
variable opacity) with variable XI (S03 injection rate);
P2' the parameter describing the linear relationship of y (opacity) variable
with X 2 (NH3 injection rate);

480

THEORY, PRINCIPLES, APPLICATIONS OF FLUE GAS CONDITIONING

f311' the parameter describing the curvilinear (quadratic) relationship of y


with Xl (this parameter is only estimated when analyzing a second-order
central composite design);
f322' the parameter describing the curvilinear (quadratic) relationship of y
with X z (this param,eter is only esimated when analyzing a second-order
central composite design);
f312' the parameter describing the (linear) interaction between Xl and X z in
their relation to y.

15C6.5 Rapper adjustments


Once optimum injection rates are found, further reduction of emissions may
sometimes be achieved by optimization of the timing and impact setting of
the rapping system. Here again the opacity monitors will serve as a guide.
There is no clear rule of thumb with respect to the rapping frequencies.
A factor of four (4) may be a good starting point. Using the 15 min first
fields frequency extend each subsequent field four times longer than the

Table lSC.4 FGC system optimization data acquisition


FGC system information
Boiler data

Precipitator

SO, skid

NH3 skid

Boiler load (MW)

Primary power
consumption (A, V)
Secondary power
consumption

S03 ratio
station setting
Boiler load
signal (%)

NH3 ratio
station setting
Boiler load
signal (%)

Air heater control


setpoint (' F) cc
Converter inlet
temperature CF) cc
Converter outlet
temperature (' F) 'C
S03 skid air Fl.
(sefm) sm 3/s
Sulfur flow
(%, lb/h) kg/h

Ammonia tank
pressure (psi) bar
Air flow
(inch w.g.) mb

Coal flow (24 h


period)

(rnA, kV)

Coal proximate
analyses
Economizer outlet
temperature ('F) cc
Air heater outlet
temperature C'F) cc
Precip. outlet temp.
(,F)T

Steam flow (lb/hr)


Oxygen (%)
NO" ppm, lb/h, kg/h
lb/MBTU
S02 (ppm, lb/h, kg/h
lb/MBTU) kg/MJ
Stack opacity (%)
Ambient temp (,F)ec
Barometric pressure
(mmHg)mb

Spark rate
Rapping cycle
schedule
Number power
supplies in/or out
of service

REFERENCES

481

preceding fields. The spiking on the opacity trace will be an overall


indication of success. Next, double those values could be evaluated.
Plenty of time should be allowed between changes of rapper settings to
allow the dust collection process to come to equilibrium. Several days is
desirable because this can sometimes be a very slow process.
15C.6.6 Data collection
To assist the optimization efforts, system data should be collected at regular
intervals. Initially, the data intervals are short (l/h typ.). This is due to the
unknown response time and equilibrium times of the system. Once we have
identified the reaction times, data acquisition can revert back to boiler house
standards.
Table 15CA shows a typical data request. Boiler data and ESP data
should be recorded on an hourly basis. FGC system data should be taken
once per shift.

15C.7

Conclusions

In any ESP operation, dust collecting performance is often adversely


affected by resistivity effects or by rapping and re-entrainment. Problems of
this type reduce the performance level of the equipment compared with that
which it is capable of attaining if the dust to be collected is treated to permit
optimum operation of the collecting mechanisms. In many cases maximum
economy may be attained by application of flue gas conditioning to an
appropriately sized dust collector having fewer hoppers, less acreage, less
power and much less maintenance than conventional units handicapped by
being required to collect dusts which have not been properly prepared for
efficient collection. Advantages obtainable by the use of flue gas conditioning will permit better emissions control with smaller, more reliable, less
expensive ESP installations. There is a possibility, yet to be fully confirmed,
that fabric filter performance may be similarly subject to improvement in
certain cases.

References (15C)
1. White, H.I. (1963) Industrial Electrostatic Precipitation. Addison-Wesley Publishing Co ..
Reading, MA.
2. White, H.1. (1974) Resistivity problems in electrostatic precipitation. J. APCA, April,
313-38.
3. Oglesby, S. et al. (1970) A Manual oj Electrostatic Precipitation Technology, Southern
Research Institute, 25 August, SoRI Publishing, Birmingham, AL, USA.

482

THEORY, PRINCIPLES, APPLICATIONS OF FLUE GAS CONDITIONING

4. Gooch, J.R. et al. (1973) A Mathematical Model of Characteristic Precipitation. Southern


Research Institute Publication, Birmingham, AL, USA.
5. Bickelhaupt, R.E. (1979) A Technique for Predicting Fly Ash Resistivity. EPA Report
60017-79-204, August.
6. Sproul, W.T. (1954) Laboratory performance of a special two-stage precipitator for
collecting high-resistivity dust and fume. Presented at American Chemical Society Meeting,
New York, NY, September.
7. Krigmont, H.Y. and Coe, E.L. Jr. (1990) Experience in conditioning electrostatic precipitators. Presented at the 4th International Conference On Electrostatic Precipitation, Beijing,
14-17 September, International Academic Publishers, Beijing, 1993.
8. Lee, G.K. and Whaley, H. (1983) Modification of combustion and fly ash characteristics
by coal blending. J. Inst. Energy, December.
9. Frish, N.W. (1984) 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, EPRI, Palo Alto, CA, USA.
10. Coe, E.L., Jr and Krigmont, H.Y. (1987) Prediction of S03 injection rates for fly ash
conditioning systems. In Proceedings of the Third International Conference on Electrostatic
Precipitation, Abano, University of Padova, italy, 28 October.
11. Dismukes, E. (1975) Conditioning of fly ash with ammonia. J APCA, 25(2), 152-6.
12. Dismukes, E.B. (1976) Fly ash conditioning. Conference on Particulate Collection Problems
in Converting to Low-Sulfur Coals, Research Triangle Park, NC. 9-10 June, CPA.
13. Busby, H.G. and Darby, K. (1963) Efficiency of electrostatic precipitators as affected by the
properties and combustion of coal. J. Inst. Fuel, 36, 184-97.
14. Watson, K.S. and Blecher, K.J. (1963) Further investigation of electrostatic precipitators
for large pulverized fuel-fired boilers. Air Water Pol/ut. Int. J., 10; 573-83.
15. Darby, K. and Heinrich D.O. (1966) Conditioning of boiler flue gases for improving
efficiency of electrofilters. Staub 26 (11). 12-17, Nov.
16. Baxter, W.A. (1968) Recent electrostatic precipitators experience with ammonia conditioning of power boiler flue gases, JAPCA, 18 (12), 817-24.
17. Reese, J.T. and Greco, J. (1967) Experience with electrostatic fly-ash collecting equipment
serving steam-electric generating plants. ASME 67- WAIA PC-3 (November).
18. Rendle, L.K. and Wilsdon, RD. (1956) The prevention of acid condensation in oil fired
boilers. Combustion Mag. luly, 39-46.
19. Dahlin, R.S. et al. (1984) A field study of a combined NH 3 /S0 3 conditioning 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, lune 1984, APCA, Pittsburgh, USA.
20. Krigmont, H.Y. and Coe, E.L. lr (1990) Dual flue gas conditioning, The Eighth EPAIEPRI
Symposium on Transfer and Utili::ation of Particulate Control Technology, San Diego,
California, 20-23 March, EPRI, Palo Alto, CA, USA.
21. Bickelhaupt, R.E. (1980) An interpretation of the deteriorative performance of hot side
precipitators. J APCA, 30, 812.
22. Can dele ria, R.B. and 10hnson, 1.S. (1983) Operating and maintenance of electrostatic
precipitators at the Navajo generating station. EPRI CS-2908 Electric Power Research
Institute, Palo Alto, CA, April, pp. 5-65.
23. NeIder, 1.H. (1983) Experience with cold side precipitator on western coal at Pleasant
Prarie power plant. Proc. American Power Conference, Chicago, IL, April.
24. Singer, 1.G., Mellinger, G.A. and Cozza, A.1. (1979) Design for continuous ash removal:
alternative concepts in ash handling. Proc. American Power Conference, Chicago, IL, April.
25. Crynack, R.R. (1985) Improving precipitator performance and saving energy with intermittent energisation - a case study. Paper 85-53.2, Presented at the Annual APC A Meeting,
Detroit, MI, June, APCA, Pittsburgh, USA.
26. Box, G.E.R. et al. (1978) Statistics for Experiments. Chap. 15, pp. 510-39, 10hn Wiley and
Sons, New York, USA.

16

Possible future developments in the field of


electrostatic precipitation
K.R. PARKER, C. RIEHLE and H. KRIGMONT

An attempt has been made in the previous chapters to review the major
applications and practice of electrostatic precipitation engineering, such that
it provides an update to the current state-of-the-art, which will provide the
non-specialist and specialist alike with sufficient data and information to
understand some of the finer points, of what in the past has been considered
as the art, rather than the science, of precipitation.
The information should enable the non-specialist, who has electrostatic
precipitators in his process clean-up train, to understand, not only the basics
but, more important, some of the operational factors which will impact on
the performance of the precipitator, which could result in emissions falling
outside those demanded by regulation or legislation.
The specialist will hopefully find the contents useful in terms of a starting
point for further investigations and measures to improve the performance,
reliability and availability of the precipitator itself.
The precipitation industry has responded to the challenge of environmental pressures not only in controlling emissions using conventional
precipitator designs and equipment, but like all modern technologies, is
carrying out further research and development activities to further improve
the precipitator's capability, in terms of performance and reliability of the
equipment without hopefully increasing capital and operating costs.
The electrostatic precipitator, although mechanically not significantly
different in basic design to that originally employed, has closely tracked the
latest developments in a number of fields. The latest material science has led
to improved fabrication techniques and materials, such that the mechanical
reliability of the precipitator is significantly increased. It is not uncommon
to find plants installed 20 years ago, still meeting the original performance
figures with only normal routine maintenance activities being necessary.
As indicated in chapter 2, the electrostatic precipitation industry has
always closely followed advances in the electrical field. The present developments in fast switching silicon technology and microelectronics, has not only
led to improved control of the electrics, but the latest microprocessor-based
controllers have simplified performance optimisation and will eventually
lead to complete 'hands off' operation of the precipitator.

484

POSSIBLE FUTURE DEVELOPMENTS IN THE FJELD OF ESP

The latest programming now includes a self-diagnostic 'menu'-type


window, which should assist the operators in quickly identifying any
difficulty or problem effecting precipitator performance, whether it be
electrical or mechanical in origin (see chapter II-Commissioning). These
programs will also have 'intelligent' self-seeking optimisation memory programs, as reviewed in chapter 8. The Electrical Operation of Precipitators.
Many of these programs are now available and should simplify the optimisation and reliability of the unit.
While it is not possible, because of commercial sensitivity, to include too
many details of ongoing investigations and research, the following outlines
some of the activities and developments that are being undertaken and some
research topics which need, or are, being investigated by the various workers
in the field.
16A

Electrical developments
(1) High frequency power converters - switch mode power supplies.
(2) Nanosecond pulsing systems.
16B Use of natural sulphur dioxide in the flue gas as a feed stock for flue
gas conditioning.
16C The application of high temperature and high pressure precipitator
technology to advanced power generation particulate clean-up systems.
16D Precipitator sizing techniques based on computer-derived data.

In addition to the above items, which will be covered later in some detail,
the following items are some personal ideas which are suggested for further
investigation.
(a) While much has been accomplished in both the mechanical and
electrical aspects of improving the performance, availability and reliability
of electrostatic precipitators of the single type, where both charging and
collection is achieved by the same set of electrodes, some useful investigations have been carried out, both in the laboratory and full-scale, to develop
a hybrid precipitator having both single and part two-stage precipitation
within the same casing. These trials were successful in minimising the
problems of reverse ionisation, but suffered from rapping re-entrainment
from the precipitation part of the two-stage section. Further work will no
doubt be carried out where the precipitator will be further subdivided to
include several hybrid stages in series, such that any uncharged particles
leaving the precipitation stage will be recharged by a downstream singlestage section and thus minimise re-entrainment losses.
(b) As indicated elsewhere, if particulate re-entrainment could be eliminated from dry precipitators, the economics would significantly change. Even
with a well-designed and controlled precipitator, measurements have shown
that at least 30% of the emission arises from re-entrainment losses. Several
different systems have been adopted, from total flow isolation during

POSSIBLE FUTURE DEVELOPMENTS IN THE FIELD OF ESP

485

rapping to the use of internal dampers or air curtains to blank-off only the
ducts being rapped. These approaches work in reducing the re-entrainment
losses, but whether the additional complications, operating difficulties and
costs can really be justified over a somewhat larger conventional plant, with
fully optimised controlled rapping, minimising re-entrainment, is debatable.
Most dry precipitator applications, even when the rapping is fully
optimised, tend to exhibit a re-entrainment factor of some 30%. In the past
when higher emissions were acceptable and precipitators had only two series
fields the re-entrainment losses favoured precipitator designs having moving
belt-type collectors with scrapper mechanisms for removing the dust, but the
mechanical integrity and high maintenance of these prevented general usage.
Although the wet film flow collector design of precipitators offers the
advantage of virtually zero re-entrainment, possible corrosion and water
treatment costs restricted their application to special processes as indicated
in chapter 13.
The importance of minimising re-entrainment was accepted by the
precipitation industry as a challenge and on modern dry precipitators the
rapping frequency is microprocessor or PLC controlled, such that no
consecutive areas are simultaneously rapped, thus limiting the impact of
re-entrainment on visible emission. Other controllers monitor and use the
change in electrical operating conditions to initiate rapping on plant which
have fairly consistent inlet conditions. Investigations into weighing of the
collector plates using load cells and only rapping when a certain thickness/weight has been achieved, has proved promising, but again the high
costs of monitoring instrumentation has restricted general usage.
(c) The use of CFD and associated systems will be developed to become
fully acceptable as a means of achieving a uniform gas distribution within
the precipitation field. This should produce data which are more representative of real plant operating conditions, since it should be possible to
include space charge, ionic wind and other effects, which are impossible to
simulate using a modelling approach, which cannot readily be energised
with dust passing through.
(d) With the larger and faster computers which are becoming available,
the field and current distributions within a precipitator duct should be
readily determined, so that it should be possible to identify the best electrode
arrangement for a particular application. The computer approach should
also enable performance and sizing evaluations to be made with respect to
operation under the various energisation systems that are becoming available as proven operational units.
(e) With the increasing concern over Pm 10 and smaller particles,
particularly heavy metals, and their cumulative effect on health, there is a
need to explore the collection of these particles to a much greater extent.
While very low emissions have been determined from precipitators whose
design and sizing have been based on total mass collection, in future, it may

486

POSSIBLE FUTURE DEVELOPMENTS IN THE FIELD OF ESP

be necessary to base the design specifically addressed to Pm 10 and smaller


particles.
Obviously one approach is to cool the gases to ensure the volatile metal
components are all condensed at the operating temperature; this usually
implies wet-type precipitation (chapter 13). The addition of a wet precipitator would, however, significantly add to the cost of the overall plant, even
if the water treatment side could be incorporated into a wet acid gas
scrubbing system; even then, one would need to precipitate the heavy metals
separately from the general sludge effluent for 'safe disposal'; otherwise there
is a risk of recycling and loss of the materials one is attempting to remove.
Even with wet precipitators there could be gaseous heavy metal elements
still escaping, so one can envisage some form of target adsorption system
being installed upstream of a dry precipitator. In this scenario it would be
possible to avoid reducing the gas below saturation temperature, which
should simplify collection and disposal methods.
The use of a target adsorption injection, upstream of the precipitator, has
been successfully used in the incineration industry for the removal of
numerous condensible phase materials. One advantage of this approach is
that the target should remove predominantly the finer particle sizes, which
both theoretically and from practical determinations show a higher penetration window on the fractional efficiency curve in the 0.5,um area (see
chapters 3 and 9). The addition of a target material would mean that a
higher overall precipitator efficiency would be required, but the size could
be less than the size increase necessary to capture a higher proportion of the
0.5,um size particles.

16A

16A.l

Electrical developments

High frequency power conversion or switched mode power supplies

For optimum performance, i.e. maximum collection efficiency, it is important to electrically operate the precipitator as near to breakdown/flashover
condition as possible. This condition provides the maximum voltage and
hence field strength, so that performance, which is theoretically proportional
to the field strength squared, is maximised. Unfortunately, even on wellregulated and operated processes, changes in the inlet conditions invariably
lead to flashover and hence discharging of the capacitive component of the
precipitator, such that, during this time, the performance is compromised.
With conventional DC energisation, while one raises the operating
voltage to the point of breakdown, as a resistance/capacitance load there is
a significant ripple voltage which reduces the maximum field strength.
Under flashover conditions, the minimum time frame to switch off the
thyristor is half a cycle or 10 ms on 50 Hz supply (8.6 ms on 60 Hz), and at
least a further half cycle to commence recharging the precipitator. On plants
operating with multiple flashovers, the overall time lost can be very
significant. This was recognised at an early date and attempts using
three-phase energisation led to unresolvable problems in arc quenching
within the field and three-phase investigations were discontinued.
Over the past two decades, the general availability of very fast switching
devices in the form of high frequency power conversion devices widely used
in PCs, welding machines and electroplating, etc., together with improved
ferrite materials having low losses, has led to the development of a new
approach to precipitation electrical energisation, by using high frequency
power conversion (HFPC) or switch mode power supplies (SMPS).
The block diagram of such a system is shown in Figure 16A.l; in this the
incoming three-phase supply feeds a power converter, an insulated gate
bipolar transistor package (IGBT) operating at a frequency of some 10 kHz.
This provides an initial DC link, which can be modulated, or 'chopped', by
semi-conductor power switches, to any desired frequency, before feeding the
ferrite cored HT transformer, whose output after rectification supplies the
precipitator. The output from the HT transformer is controlled by varying
the on/off times of the power switches using a feedback system which
continuously monitors the output voltage and current levels to achieve
optimum conditions on the precipitator.

488

ELECTRICAL DEVELOPMENTS
Input
rectifier
AC
AC

Switching
network Transformer

))
I I

111[

Output
rectifier
AC
DC

Output
filter

]*

DC

Figure 16A.l General converter AC to DC (mains in. DC out).

The operational problems of interrupting the power supply, to quench an


arc/flashover and re-establishing current to the precipitator, are considerably reduced by operation at high frequency, since at 10 kHz, the minimum
switching time is 0.05 ms, compared with 10 ms at 50 Hz, and the recovery
time, to recharge the precipitator, can be more controlled with the smaller
time step increments, such as to rapidly re-establish precipitator performance.
Initial trials using SMPS units on a precipitator handling particulates
from a wood chip installation as a load have been carried out and the results

-55kV

Conventional
power supply

okV

- - - - -:-, - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - --

Figure 16A.2 ESP voltage, conventional and HFPC power supplies, same current.

HF POWER CONVERSION OR SWITCHED MODE POWER SUPPLIES

489

Conventional
power supply

OkV --------------------------------------------------------

Figure 16A.3 ESP voltage, conventional and HFPC power supplies, same peak voltage.

compared with the same unit under conventional energisation. The results
of these trials [1] are illustrated in the form of voltage oscillograms; Figure
16A.2 was obtained with both power supplies adjusted, without flashover
arising, to give the same secondary currents of 200 rnA. Under this condition, the conventional supply produced a peak voltage some 10 kV higher
than the SMPS system, but at a lower mean voltage level. Figure 16A.3 was
produced with the peak voltage on both adjusted to 55 kV; in this case, the
currents for the conventional and SMPS systems were 110 and 230 rnA,
respectively. In Figure 16A.4, the power supplies were free running, with the
current limited only by flashover/spark over conditions; here the corresponding currents were 120 and 230 rnA, respectively.
The above results indicate that SMPS energisation is capable of operating at a higher mean voltage and secondary current level, thereby providing
significantly increased power inputs of up to three times that of a conventional DC system and consequently a much higher performance for a given
size of precipitator.
This SMPS or HFPC development will probably have a greater impact
on future precipitator technology than the pulse charger or flue gas
conditioning systems, which, although ideal for improving precipitator
performance for 'difficult dust' situations, does little for ordinary dusts
typically met. The SMPS approach, on the other hand, appears to improve

490

ELECTRICAL DEVELOPMENTS

"""'fWW"- HFPC

OkV

----~-----------------------------------

_-t----+'

Conventional
power supply

Figure 16A.4 ESP voltage, conventional and HFPC power supplies during normal operation,
sparking.

all particulate collection efficiencies, as a result of the much higher power


input and, for a given duty, should lead to a smaller size and hence lower
cost precipitator.
In addition to the performance improvement, for the same power input,
the primary current from each phase of the supply is lower than from a
conventional single phase system. This has the advantage that all primary
components, such as switch gear, control unit and all incoming cabling, are
much smaller and hence there is a saving both in cost and in floor space. In
some designs, the SMPS unit is a complete free-standing module, requiring
only an incoming three-phase supply connection. With the much smaller
and lighter power supply units, it is possible to consider dividing the
precipitator into smaller bus sections, each with its own power supply,
which would further improve the overall particulate collection efficiency, or
lead to a reduction in size for a new unit.
16A,2

Nanosecond pulse operation and acid gas control

The initial object of operating a precipitator with nanosecond, rather than


millisecond, pulse widths was to improve the performance of precipitators
when handling 'difficult' fly ash.

REFERENCES

491

Although such a system had merit in improving the collection efficiency


of precipitators, there has been a move toward using this approach, under
controlled precipitator operating conditions, to promote the formation of a
plasma rather than a corona, such that the plasma gives rise to highly
reactive gaseous radicles, for example 0 +, which are then available for
attachment to S02 molecules, to form initially S03 and then H 2S0 4 , which
can then be removed by the precipitator [2,3]. A similar reaction would
occur with NOx molecules, so that the system is capable of controlling acid
gas emissions in addition to fly ash from power stations.
Recently the system has been extended outside the power station field
into waste incineration, where the plasma is being applied to break down
VOCs and undesirable substances, such as dioxins and furans, and also to
react with elemental mercury to simplify its collection as a condensed, rather
than a gaseous, material.
This work is still in its experimental stage, but if successful, as indicated
from laboratory and pilot field work, could change the present back end
clean-up systems on a number of industrial process plant.
References (16A)
1. Ranstead, P. and Porie, K. High frequency power conversion: a new technique for ESP
energisation, EPRI/DOE International Conference on Managing Hazardous and Particulate Air Pollutants, Aug. 15-17th, Toronto, Canada, in print.
2. Dinelli, G. (1987) Electrostatic precipitator performance by means of narrow pulse power
energisation. 3rd International Conference on Electrostatic Precipitation. Abano, italy,
University of Padova, October.
3. Mazuda S. et al. (1993) High voltage nanosecond pulse discharge system for removing S02
and NO, in flue gas. 10th EPRI Particulate Control Symposium and the 5th International
Conference on Electrostatic Precipitation, Washington, USA, October, Paper 32.1-14,
EPR! TR 103048 2, Palo Alto, CA, USA.

16B

Use of natural sulphur dioxide as a feed stock


for flue gas conditioning systems: flue gas
conditioning today and tomorrow

The present status of flue gas conditioning systems is to use either elemental
sulphur or liquid sulphur dioxide as a feed stock, which is converted to
sulphur trioxide as the conditioning agent. This approach, while being
completely satisfactory, has a cost element existing over the life of the plant.
In many applications, particularly those firing carbonaceous fuels, there is
usually sufficient sulphur dioxide present in the flue gas which, if converted,
can provide the sulphur trioxide necessary for flue gas conditioning duties.
The following section, by Dr Henry Krigmont, explores and develops the
various approaches that can be used to convert the naturally occurring
sulphur dioxide as the feed stock for conditioning.

16B.l

Background

As discussed, typically, S03 is formed by a catalytic conversion of gaseous


sulfur dioxide (SOz) [1]. The SOz usually comes from an 'external' source
(feed stock) by either evaporating liquid sulfur dioxide or by burning molten
or solid sulfur. Conventional flue gas conditioning systems have two major
drawbacks: they require an external continuous supply of feed stock and in
the process of operation these systems slightly increase SOz emissions.
Another approach to create S03 is to utilize the 'native' SOz formed
during combustion of sulfur contained in fossil fuels for a subsequent
conversion to S03. The rate of such a process depends on temperature,
concentrations of SOz, 0z and water vapor as well as the catalyst's
properties. Various systems implementing this technique were proposed, but
none were free of major deficiencies.
Figure 16B.1 depicts a summary of the S03 FGC technology [2]. A
variety of designs could be offered to satisfy the preference and applications
specifics of any customer. The W AHLCO designs substantially improve
in-situ flue gas conditioning technology by providing simple yet reliable
systems to furnish the required amount of sulfur trioxide for efficient
electrostatic precipitator operation, without adding more sulfur oxides to
the system and in a fully controlled manner. On the contrary, the amount
of sulfur dioxide in the system is reduced.

Using solid
sulfur

Using molten
sulfur

Variable
catalyst
temperature

Variable
exposed area

Variable flow
through
catalyst

With gas
cleaning

I
Without gas
cleaning

Slip-stream
systems

Fixed amount
of catalyst
in duct

Figure 168.1 FGC systems - classification.

Variable flow
amount of
catalyst in duct

In-duct
systems

Sulfur based
systems

S02 based
systems

I
Internal
feedstock

External
feedstock

1
Hot side
(before air
heater)

Cold side
(after air
heater)

Flue gas
conditioning
with S03

494

16B.2

FLUE GAS CONDITIONING TODAY AND TOMORROW

'Native' or 'internal' feed stock FGC technologies

Conventional FGC systems work well, and are widely used, but in some
instances there are drawbacks. The catalytic conversion of sulfur dioxide to
sulfur trioxide is not completely efficient, and additional sulfur dioxide is
added to the flue gas flow. A constant supply of sulfur feed stock is required,
and this feed stock must be safely handled. The components for burning,
catalyzing, and injecting systems must be kept in good working order, and
there is a power consumption associated with the process.
Another approach to create S03 is to utilize the 'native' S02 formed
during combustion of sulfur contained in fossil fuels as a feed stock for a
subsequent conversion to S03' Extensive experimentation at the Lehigh
University [3] was conducted to investigate a new approach to fly ash
conditioning based on conversion to S03 of a portion of the S02 normally
present in flue gas even when low sulfur coals were burned.
The chemical equilibrium conditions predict that when the gas contains
about 5% 2, about 99% of the S02 can be oxidized to S03 at about 399 DC
(750F) and about 90% at 510C (950 F), with respectively more S03
being generated at lower temperatures and less at higher temperatures.
Typically, however, not more than about 1-3% of the 'natural' S02 is being
oxidized in S03' the rate of such reaction usually depending on the gas
temperatures, the concentration of S02' 02 and water vapors, as well as the
possible catalytic properties of the boiler convection surfaces.
Kanowski and Coughlin [3] proved that it is possible to oxidize
significant quantities of S02 present in flue gas at concentrations typical for
commercial coal-fired boilers burning low sulfur coal, by inserting commercial catalysts in the S02-containing gas. Several systems implementing this
technique were subsequently proposed, but none were free of major defiCienCIes.

16B.2.1

'Slip-stream' FGC systems

One earlier published approach suggested using a fan to draw a portion of


the clean gas (taken from downstream of the precipitator) into a slip stream
which is subsequently passed through a chamber with the catalyst, but gives
no further details as to the amount of the flue gases, temperature and reheat
requirements and how the amount of sulfur trioxide can be controlled. One
can imagine that valving could be added to the slip stream to control its
total flow, but such valves are complex, expensive, and difficult to fabricate
and maintain.
Recently, a different approach was proposed and is being investigated as
a possible way to avoid the purchase of sulfur or liquid S02 feed stock. A
so-called 'EPRICON-Agentless Flue Gas Conditioning System' offers an
approach in which a slip stream (about 3-5%) of the entire flue gas flow is

'NATIVE' OR 'INTERNAL' FEED STOCK FGC TECHNOLOGIES

495

drawn from the main flow ahead of the economizer and passed through the
catalyst bed [4]. A portion of the sulfur dioxide in the slip stream is oxidized
to sulfur trioxide, and the slip stream is merged back into the main flue gas
flow downstream of the air preheater. Due to the relatively small portion of
the flue gases used, very high efficiency catalytic conversion is required,
which means strict control of the gas temperature and other operating
parameters. While of interest, this approach has major drawbacks when
implemented, such as reduced boiler thermal efficiency, because less heat is
recovered, and there is typically insufficient mixing of the slip stream and
the main flow at the point where they rejoin, due to a small pressure
differential.
Furthermore, questions yet to be thoroughly investigated include control
of S03 quantities to proper levels when a variety of different coals are
burned producing ashes with differing conditioning requirements. Also,
variation of flue gas temperatures as boiler loads change may lead to
insufficient production of S03 at low loads, causing fouling of the ESP with
high resistivity ash, which may interfere with emission compliance upon
return to full load operation.
For economy of installation and simplicity of operation it is preferred
that booster fans are not used, which means that the system should rely only
on the relatively small pressure drop in the main system to provide the
driving force for the side loop in which the catalytic conversion of SOz to
S03 occurs. This may lead to big ducts, large volumes of catalysts, big
injection probe sizes and difficulty in obtaining uniform distribution and
mixing of S03 in the downstream gas. The expense of custom ductwork
design and installation for such a system is not negligible. It is expected that
a considerable amount of testing and demonstration work will be required
to accurately assess the suitability of this system for commercial operation.
16B.2.2 In-duct FGC systems

In another design the catalytic converter is located in the boiler economizer


outlet duct prior to the air heater [5]. By exposing most of the native sulfur
dioxide in the gas stream to the catalyst, very low conversion efficiencies for
the converter may suffice to produce S03 in the required quantities. The rate
of conversion is proposed to be controlled by inserting or retracting the
catalyst bed into the ductwork.
This approach also has several obvious deficiencies. Since the gas flow
follows the path of least resistance, it could be difficult to force the gas into
the catalyst bed presenting substantial resistance to the moving gases. This
approach offers very little means for strict control of the S03 generation, for
the catalyst bed will have to be in a continuous state of motion to provide
the required S03 amounts in response to changes of boiler load, excess air
(0 2 ), various boiler operating characteristics, etc. Therefore, this design

496

FLUE GAS CONDITIONING TODA Y AND TOMORROW

offers only 'ON/ OFF' type operation. Furthermore, this design will introduce yet another source of the possible air inleakage through the ductwork
openings and create significant flow maldistribution.
16B.3

In-situ gas conditioning (lGC) approach

Recently, new solutions to these problems have been developed. These designs
allow selective control of in-situ catalytic S03 generation. A sulfur trioxide
conditioning system includes a catalytic converter that converts a portion of
'native' or naturally formed sulfur dioxide in a flow of flue gas to sulfur
trioxide. In one design, the amount of the catalyzed surface exposed to the
flow of flue gas is selectively varied to control the conversion of S02 to S03'
Another approach provides the means to vary the catalyst bed temperature to
selectively enhance or suppress the sulfur dioxide oxidation. In another
design, variable amounts of flue gas are drawn through a fixed catalyst bed.

16B.3.1

Variable exposed area ICC system

Figure 16B.2 presents an artist's view of the Wahlco variable exposed area
IGC system. This IGC system includes a catalytic converter consisting of a

Figure 168.2 The Wahlco variable exposed a rea IGC system.

IN-SITU GAS CONDITIONING (IGC) APPROACH

497

bed of catalyst substrates distributed across the main duct, with the catalyst,
for the in-situ oxidation of sulfur dioxide to sulfur trioxide. The bed is
positioned in the flue gas stream at a location having the correct temperature to accomplish the catalyzed oxidation. The amount of sulfur trioxide
produced is controlled by selectively adjusting the amount of catalystcovered surface that is exposed to the flue gas stream.
To control the amount of sulfur trioxide generated, a shield is placed over
a portion of the catalyst-covered surface. The shield prevents active contact
between the flue gas flow and the covered portion of the catalyzed surface,
so limiting the oxidation of sulfur dioxide. The greater the exposed surface
area of catalyst, the greater the conversion of sulfur dioxide to sulfur
trioxide. The shield also serves the important function of reducing erosion
damage to the catalyst, and of permitting compensation for the effects of
decreasing efficiency through ageing of the catalyst. This design can compensate for this change in the properties of the catalyst simply by exposing
the required additional amount of catalyzed surface.
This design is easily automated. The operating parameters of the system
are measured, such as, for example, boiler load, flue gas composition,
electrostatic precipitator power consumption, fly ash properties, fly ash
resistivity, and/or stack gas opacity. These parameters are provided to a
computer, which determines whether any change in the sulfur trioxide level
is required, and, if so, the nature of the change required. The shield is moved,
if necessary, to change the amount of exposed catalyst surface area responsive to these measured operating parameters.
The gas flow distribution and sulfur trioxide distribution in the flue gas
duct can even be improved by selection of the nature and distribution of the
catalyst supports. For example, if there is an uneven temperature naturally
in the duct, then the distribution of catalyst, or its exposed area, can be
selected so that the production of sulfur trioxide matches the spatially
varying requirements of the flue gas.
16B.3.2

Variable catalyst temperature IGC system

This sulfur trioxide flue gas conditioning system also includes a catalytic
converter for converting a portion of the sulfur dioxide in the flue gas
to sulfur trioxide. The catalytic converter consists of a bed with catalyst
support distributed across the boiler back pass. In this design, the catalyst bed is furnished with a temperature controller to allow for the catalyst
support heating or cooling to vary the conversion efficiency of the catalyst.
The catalyst support temperature is maintained independently of the temperature of the flue gases, e.g. it could be electrically heated if the catalyst
requires heating, or a cooling system, should the catalyst need to be cooled.
The main benefit of this design is that the catalyst bed could be located
independent of the flue gas temperature, with the temperature controller

498

FLUE GAS CONDITIONING TODAY AND TOMORROW

providing optimum conditions for the IGC system operation. As an option,


the catalyst bed could be positioned at the location where the gas temperature is always higher than that required; then the temperature controller
may consist of only a cooler. Conversely, if the catalyst bed is positioned at
the location where the gas temperature is always lower than the optimum
the temperature controller may consist of only a heater.
The heating and/or cooling of the catalyst provides for a highly effective
way to precisely control the extent of the S02 to S03 conversion.

16B.4 Variable flow IGC system


In this design the catalyst is enclosed in modules that are installed in a
grid-like fashion in the flue duct (Figure 16B.3). Each module has an
individual controllable aspirating device. Within the aspirator compressed
air or stream is employed to induce the flue gas flow through the catalyst.
The modules are positioned in the gas stream at a location having the
proper temperature range to accomplish efficient catalytic oxidation of S02
to S03.

Compressed air
distribution

Duct wall

Catalyst element
Tubular support
structure

Figure 168.3 Modular design for variable flow IGC.

CAT AL YST SELECTION

499

The amount of S03 generated is controlled by varying the volume of flue


gas that passes through the catalyst. By controlling the compressed air
pressure or steam flow and, therefore, the velocity of the gas flow through
the module, the design allows for precise control of the S03 generated and
also limits catalyst wear. Only the minimal flow required for S03 generation
is introduced into the catalyst module. The catalyst is shielded from direct
erosion by the fly ash by a deflector located at the entrance to the module.
This design provides a method for a controllable amount of sulfur
trioxide to be created and then uniformly distributed in the flue gas stream.
The system is simple and durable in operation. Control of the process could
be accomplished similar to the one described earlier using a logic proven
with traditional external S03 generation systems.
The major benefits of this modular design are: installation is simple and
easy and does not require any special tools and instrumentation; it provides
for standardization and 'off-the-shelf design; it allows for variable S03
distribution to match existing gas flow non-uniformity; only relatively minor
duct modifications would be required; virtually no moving parts; maintenance is very easy, and periodic catalyst cleaning as well as change-out is
simple to accomplish.
16B.5

Catalyst selection

The catalyst is the heart of any IGC system. The proper catalyst will
improve system efficiency, longevity and economics. The major requirements
for SOz to S03 oxidation catalysts are:

High conversion efficiency


Low 'light-off' temperature
Activity in specified range of space velocity (SV)
High stability and low sensitivity to poisoning
Resistance to erosion/abrasion
Clean ability

It is interesting that from a position of catalyst chemistry, the same types


of catalyst developed for selective catalytic reduction (SCR) of NO x are
suitable for SOz oxidation. The only problem is that the catalyst manufacturers spent years sorcering with catalyst formulations to reduce SOz to S03
conversion down to only a few percentage points for a 'hot' catalyst and to
a fraction of that for a 'normal' catalyst. For IGC applications, the goal is
quite the opposite - to maximize SOz oxidation while still maintaining the
mechanical integrity and chemical stability of the catalyst.
For the typical temperature range of 288-399 C (550-750 OF), usually
316-321 C (600-700 OF), two types of catalyst are applicable: based on base
metal oxides and on precious metals. The first type usually has a maximum

500

FLUE GAS CONDITIONING TODA Y AND TOMORROW

efficiency closer to the higher part of this temperature range. The noble
metal catalyst is more active at lower temperatures.
The expected conversion efficiency is different for different types of IGC
system:
for variable exposed area IGC system 3-8%, with SV of 1 x 10 5 -3 X
105 h- 1
for variable catalyst temperature IGC system 1-3%, with SV of 3 x 10 5 _
6 X 105 h - 1 and
for variable flow IGC design 40-60%, with SV of 1 x 104 -3 X 104 h- 1
Preliminary tests conducted in the simulated environment have shown that
these target values are conservative, at least for a fresh catalyst.
Different types of substrates have been considered for this application:
extruded monolith, metal foil, and plate type. The desired range of cell
densities is 10-50 cpsi, with variable flow IGC at the higher side of this
range. The catalysts, based on these substrates have specific features that
may suit different IGC systems.
One of the important issues is the capability to withstand fly ash erosion.
This may become critical for variable exposed area IGC systems, where the
catalyst sees full particulate loading and the flue gas velocities are even
higher than in non-obstructed ducts. The rate of erosion is usually proportional to the velocity to the power of 4 to 6 and at some point (approximately 10-15 mjs (35-50 ftjs it starts to increase dramatically. From this
point of view the variable flow systems (like modular IGC) have definite
advantages. Deflecting coarse particles from direct contact with the catalyst
significantly reduces the danger of erosion. The smallest particles usually
result in catalyst masking that may be effectively addressed by sootblowing.
Some formulations, based on noble metals allow even washing the catalyst
with chemical solutions and significantly extend its effective life by a
properly planned maintenance program.
References (16B)
1. Krigmont, H.Y. and Coe, E.L. (1992) Flue gas conditioning: key advances in recent years.

Power, August, 30-6.


2. Ferrigen, J.J. et al. (1993) A naval approach to flue gas conditioning. EPRIjDOE International Conference on Managing Hazardous and Particulate Air Pollutants. Poster Session.
August 16-18th, Toronto, Canada, in print.
3. Kanowski, S. and Coughlin, R.W. (1977) Catalytic conditioning of fly ash without addition
of S03 from external sources. Environ. Sci. Technol., 11(1), 67 - 70.
4. Bibbo, P.P. (1994) EPRICON: agentless flue gas conditioning for electrostatic precipitators.
Proc. POWERGEN America, Orlando, Renwell. Inst. Houston, USA.
5. Woracek, D. et al. (1993) Catalytic conversion of native sulphur trioxide in flue gas for
resistivity conditioning., 10th EPRI Particulate Control Symposium and 5th International
Conference on Electrostatic Precipitators, October, Washington, pp. 13.1-16, EPRI TR103048 2, Palo Alto, CA, USA.

16C

High temperature/high pressure precipitators for


advanced power generation systems

Worldwide environmental concern over 'green house' gases, particularly


those arising from the power generation industry, has led to new designs for
advanced power generation having far greater efficiencies than those obtainable from the conventional steam cycle of around 38%. Efficiencies of some
42-44% can be realised with pf units utilising supercritical steam conditions,
mid 40s%, using pressurised fluid bed combustors' combined cycles and
much higher using integrated gasification and combined cycle and hybrid
approaches.
A schematic arrangement of an Air Blown Gasificaton Cycle is illustrated
in Figure 16C.1 [1]. This employs an air blown gasifier stage, the gases from
which drive a turbine, while the resultant char passes to a circulating fluid
bed combustor in the steam cycle stage. Alternate approaches using oxygen,
rather than air, for the gasifier result in improved efficiency, dispensing with
the 'char' stage to produce a simpler steam-raising block.
In addition to the normal clean-up of the gases finally exiting the equipment,
the gasifier gases need to be supercleaned to avoid erosion and high
temperature corrosion difficulties arising with the turbine blades. Following
the gasifier cyclone, indicated in Figure 16C.1, the emission is high in condensed alkali material, which needs to be removed to minimise adhesion and
other problems. Cleaning as indicated is achieved by a ceramic filter operating at a temperature of some 400C and a pressure of 20 bar. An alternate
approach would be to use high efficiency cyclones operating at a higher
temperature, plus a 'ruggedised' turbine to accept a higher particulate loading.
In many high temperature but atmospheric pressure type applications,
the increase in kinetic energy of the gas molecules leads to very conductive
gases, this with the difficulties of suitable insulator materials, which have low
resistance surface properties at high temperature, as indicated in Figure
4.9, has meant that electrostatic precipitation is not feasible for these
processes. There are exceptions and precipitators have been successfully
used at temperatures above 550C, in gases containing high levels of SOz
and also at 800C, where the gases contain high levels of HF plus water
vapour but basically no air.
With the new advanced power generation systems operating at high
pressure and high temperature, the problems of the kinetic energy of the
molecules disappear, and the difficulties of providing suitable insulator
materials have been overcome by moving the suspension insulators outside
the hot gas stream, so that they can operate, without breakdown, in a lower

Coal and
sorbent

Gasifier

Pressure ~
let down

,ASh

Fluidising
air

Gas
filter

Air from heater

Pulse gas

Figure 16C.l Air blown gasification cycle [1].

--

I i

Steam turbine
and generator

Air to CFBC

Condenser

c::::,,:

Waste heat

. . Gas to
. . stack

CirCUit

.. Tosteam

FUNDAMENTALS

503

temperature environment. While there will be engineering difficulties associated with the mechanical design of large precipitator units because of
differential temperatures, these are probably no more onerous than those
experienced by the ceramic filter and its cleaning system.
Over the next decade it is anticipated that there will be a significant
increase in the demand for power worldwide, particularly with China, India
and Russia becoming fully industrialised; the need for high efficiency power
generation becomes of paramount importance to limit the discharge of
carbon dioxide gases. To meet these opportunities a great deal of preemptive work has been carried out by the precipitator companies to clean
up the high temperature/high pressure gases associated with these advanced
power generation systems.
The theory and practical aspects of high temperature/high pressure
precipitation are presented in the following resume prepared by Dr Claus
Riehle. This gives the case for the electrostatic precipitator together with
some of the engineering difficulties which have to be resolved for its
successful application in the advanced generation systems now being considered for implementation, both now and in the next century.

16C.l

Fundamentals

Changing gas temperature and/or changing gas pressure usually leads to a


change in gas density. This can be easily seen when considering the equation
of state for ideal gases (equation 16Cl).

P
-=RT
P

(16Cl)

For handling different states, it is convenient to introduce the so-called


relative gas density b, defined by equation (16C2). This relates the gas
density of a state 2 to the gas density of a known state 1. Obviously from
Figure 16C2 the relative gas density is a function of the pressure ratio and
the reciprocal temperature ratio.

b = pz
PI

Pz. TI
PI T2

(16C2)

Usually temperature and pressure are related to normal conditions, i.e.


TI = To = 273K and PI = Po = 1 bar. Figure 16C2 illustrates the value
range of the relative gas density for temperatures from 300K up to 1500K
and pressures from 1 bar up to 30 bar, represented by corresponding
isolines.
A decreasing relative gas density corresponds to an increasing average
distance between the gas molecules (which can be deduced from kinetic gas
theory) or, in other words, comparing two different gas states, the mean free

504

HTjHP PRECIPITATORS FOR ADVANCED POWER GENERATION

Po = 1 bar
To = 273 K

....>-

'iii
&::

41

!fj

III

c.?
41

>

'';::
III

iii

a:

500

700

900

1100

1300

1500

Temperature T 2 in K

Figure 16C.2 Relative gas density as a function of temperature; isolines represent constant
pressures.

path of the gas molecules behaves like the reciprocal of the relative gas
density (equation (16C.3)). At normal conditions the mean free path
;'(Po, To) is about 0.065 Jlm. Therefore, under normal conditions for particles
smaller than 1 Jlm the mean free path of the gas molecules is of a similar
order and has to be taken into account.
'(

I.

p,

T) = A(Po, To)

(16C.3)

Closely connected with the mean free path of the gas molecules is the
mobility of the gas ions, The mobility of the charge carriers, b, is defined as
the ratio of the mean drift velocity v of the charge carriers to the electrical
field strength E generating the drift and a typical value for gas ions at
normal conditions is given in equation (16C.4), The mean drift velocity of
the charge carriers is determined by the collision frequency with other
(neutral) gas molecules; obviously the collision frequency decreases with
increasing intermolecular distances; thus the mobility increases in the same
way as the mean free path of the gas molecules (equation (16C.5)). The
mobility of gas ions influences the current -voltage relationship as will be
discussed in the following sections.
b(To,Po) =2.10- 4

m2
Vs

(16C.4)
(16C.5)

VOLTAGE AND CURRENT

505

Finally we have to consider how the viscosity of the gas flow is affected
by high temperatures and/or high pressures. It can be easily demonstrated
that viscosity is not influenced by pressure but only by temperature, since
the p, T-dependence of gas density and mean free path compensate each
other (equation 16C.6). The well known .JkT-dependence of the mean
thermal velocity of a gas molecule with mass m remains valid (equation
16C.7). Therefore, viscosity is not modified in terms of relative gas density,
but in terms of .J(T2 /T1 ) (equation 16C.8).
1]- p),<v)

fkT

(16C.7)

<v) - ~-;;
1]( T)

if

(16C.6)

1]( To)

(16C.8)

Gas viscosity is decisive in the migration velocity of the particles which


results as a balance between electrical and drag forces. Therefore, viscosity
determines particle migration while relative gas density and mobility mainly
govern the operating conditions. Finally, some particle properties important
for the ESP process are also a function of temperature, e.g. electrical
resistivity, particle-particle and particle-collector adhesive forces, as discussed elsewhere.
16C.2

Voltage and current

The corona initiation field strength Eo was introduced in chapter 3.


According to equation (3.1), Figure 16C.3 shows Eo as function of wire
radius and isolines for different relative gas densities. In general, an increase
of wire radius leads to a decrease in electrical field strength; this trend
softens for radii > 1 mm. Relative gas densities > 1 need higher electrical
field strength values to initiate corona in the same geometry. This seems
reasonable because in a denser gas the mean free path of molecules reduces
and with the time available for acceleration; this can be compensated by
higher electrical field values. Obviously, with relative gas densities < 1 the
opposite behaviour occurs.
The corona onset voltage for a tube-type precipitator can be estimated
according to equation (3.3) and its dependence on relative gas density is
plotted in Figure 16C.4. Generally for fixed geometries, an increasing
relative gas density needs higher voltage levels to initiate corona. The thick
line in Figure 16C.4 illustrates the onset condition for a wire radius of
1.0 mm and a tube radius of 150 mm. This figure also shows that a variation
of tube radius has only small effects on voltage levels compared with

506

HT/HP PRECIPITATORS FOR ADVANCED POWER GENERATION

300

;::Ol_E
e: >
~o

250
5.0

00"-

~=----~.
--_____

"C -., 200

'ij W

u:::_
c:

CI)

150

III

100

o (,)
~~
ca ::l

E
c:

:~-=CC-.-

~ 'j

eo~-

O.~

50

o
o

0.2

0.6

0.4

0.8

1.2

Radius of Discharge Wire rSE

1.4

1.6

mm

Figure 16C.3 Corona initiation field strength at wire surface as a function of wire radius;
isolines represent constant relative gas densities.

1S0

rNE'mm
100
1S0
200

>

...

'0

~ 100

j
I
0

SO

rSE = O.S mm
rNE = 1S0mm
0

4
3
Relative Density

Figure 16C.4 Corona onset voltage as a function of relative gas density. For different tube and
wire radii.

507

VOLTAGE AND CURRENT

--......

b(po,To) = 2"104 m'Ns

'"E

<t

1.6

"~

1.2

'NE

= 150 mm

rSE

;;:

1 mm

ca E
>- L1J

";;;
C

Q)

c 3:
c

Q)

::l

iii
C)

0.8

;;
CJ

U ~

'0

0.4

3.0

4.0
0

20

40

60

Applied Voltage U
Figure 16C.5 Electrical current density at collecting wall as a function of applied voltage for
different relative gas densities.

changes of discharge wire radius. It is interesting to note that the smaller


the discharge wire, the weaker the dependence of onset voltage on gas
density becomes.
In order to discuss the dependence of the current - voltage relationship on
the relative gas density 6, one has to keep in mind that corona onset voltage
V 0' and the mobility of gas ions b, are functions of 6. Figure 16C.5 shows
calculations according to equation (3.13) for different relative gas densities
for a tube type design of rSE = 1 mm and r NE = 150 mm. The higher the
relative gas density the flatter the curves become, or in other words, at
higher relative gas densities, a constant applied voltage can force less gas
ions through the gas phase.
The flow of charge carriers is limited by the electrical breakdown
condition. This sparkover voltage plays an important role in the operation
of electrical precipitators, because stable operation is only possible if an
essential difference between corona onset and spark over (V eril - V o) exists.
As the experimental results in Figure 16C.6 show [2], the region of stable
operation V eril - Vo in a tube-type ESP continuously decreases for increasing temperatures when the pressure is kept constant at 1 bar. However,
when the pressure is also increased, then the region V eril - Vo widens and
extends to higher temperatures. For example, at room temperature and
1 bar, V eril - Vo is about 25 kV (Figure 16C.6), whilst approximately the
same voltage difference is observed at about 800C but at a pressure level

508

HT/HP PRECIPITATORS FOR ADVANCED POWER GENERATION

Temperature Tin C
0

200

400

600

800 1000

200

400

600

800 1000

200

>
.:.:
c

::

(II

CI

as

~
"C

1 bar
100
50

..

0
200

---.--- .
..

150

.!!!
0..

100

c:r:

50

Co

5 bar

150

21 bar

.......

0
273

473

673

873

1073 1273 273

473

673

873 1073 1273

Temperature T in K
Figure 16C.6 Corona onset voltage (0) together with sparkover voltage (e) as a function of
temperature for four different pressures. The measurements were done by Weber [2] in a
tube-type ESP with RNE/ RSE = 62.5.

of 21 bar. A stable operation of ESPs at high temperatures must, therefore,


be accompanied by an appropriate high pressure level, i.e. high relative gas
densities.
Furthermore, the high gas densities enable higher voltages to be applied,
i.e. higher electrical fields can be sustained between the electrodes (Figure
16C.6), producing corresponding improvements in the particle collection, as
will be shown. Since no reasonable forecast of the sparkover voltage can be
made, no reliable estimates for the maximum electrical field strengths are
possible and experiments are necessary. For example, Figure 16C.7 shows
jSE/U -curves experimentally determined by Bush et al. [3] in a laboratory-scale tube-type ESP covering temperatures in the range 533-1366K
and pressures 3.4-35.5 bar (R NE = 3.63 cm and RSE = 0.8-1.6 mm). As other
authors, they did not find a general limitation for corona discharge at high
pressures and high temperatures in the range investigated. In most cases
negative corona was found to be superior to positive corona. In Figure
16C.7 relative gas densities of about b ~ 6 enable voltage levels of about
100 kV; for a tube radius of about 3 cm, this corresponds to electric field
strengths in the order of 30 k V/cm! For normal operating conditions the
field strength is about 3 k V/cm, so this increase represents, obviously, an
outstanding potential for successful particle collection at high temperatures
and high pressures.

509

PARTICLE CHARGING

T = 811 K

T=1089K

10-r------------------~ -.------------------~
0,:14
0,5

...oo

2,83

3,81

:;::

"en

<C
E
w

"Je

C'CI

D...

20-.------------------~ ,---------------------,

4'!7

"en

-...
C

18

Q)

16

14

o::s

12

10

"i:

C'CI

:;::
C'CI

en
Q)

jjj

6
q,34

4
2

40

80

120

1600

40

80

120

160

Applied Voltage U in kV
Figure 16C.7 Current-voltage characteristics iSE(U) for dry air measured by Bush et al. [13].
Diagrams on the left for T = 811 K; diagrams on the right for T = I089K. Top diagrams
represent positive corona, bottom diagrams hold for negative corona. The top end of the curves
corresponds to electrical breakdown. The value of the relative gas density is written along each
curve. The tube radius was RNE = 3.63 cm and that of the discharge wire RSE = 1.17 mm.

16C.3

Particle charging

The consequences of temperature and pressure on the particle saturation


charge can be calculated according to Cochet's equation given in section 3.3
(equation (3.56)). Figure 16C.8 shows the calculated influence of the relative

510

HT/HP PRECIPITATORS FOR ADVANCED POWER GENERATION

10 4

r--~~~~~-'-'---~-~~~ ,---,--~~~~

10 3

~~ IE
I ~

Q)
~

....
....<'?

= 3.0*10 s V/m
epsr = 10

Q.

delta

Q)

El
IU

..c:

10'

1.0
1.5

2.0
3.0
4.0
5.0
6.0

c:

:;:

IU
~

::J
IU

en

10'
0.01

0.1

10.

Particle Size dp 11m


Figure 16C.S Particle saturation charge according to Cochet [4] plotted against particle
diameter. An electrical field strength of 3 kV/cm and a relative electrical permittivity of 10 are
assumed for the calculations. Different lines correspond to different relative gas densities.

gas density on particle charge for an electrical field strength of 3 k Vjcm and
an electrical permittivity of the particle of er = 10. With decreasing particle
size, the achievable particle charge is generally reduced because of decreasing surface. This is why a straight line emerges for large particles on the
log-log scale. At low relative gas densities, small sized particles can carry
much more charge than at atmospheric conditions.
According to Figure 16C.2, gas densities < 1 correspond to operation at
atmospheric pressure and high temperature. To generate the charge carriers,
however, additionally high pressure conditions are necessary, as previously
discussed. Both factors do not correlate; thus, one could suppose at this
stage, that ESPs will work less effectively in the fine particle region at high
temperature and high pressure; on the other hand, the efficiency might be
compensated by potentially higher electrical field strengths.
16C.4

Particle migration

Figure 16C.9 illustrates the theoretical migration velocity as a function of


particle size under atmospheric pressure and raised temperatures, which
corresponds to relative gas densities < 1. The absolute value of the characteristic minimum in the migration velocity function is weakly lifted and its
location is shifted with decreasing relative gas density to larger particle sizes.

511

PARTICLE MIGRATION

---D-

= 1375 K (or delta = 0.2)


= 546 K (or delta = 0.5)

= 273 K (or delta = 1.0)

= Po = 1 bar

E
3.0*105V/m
epsr = 10

10.2

L-~~~~~_ _~~~~~_ _~~~~L-~~~~~

0.01

10

0.1

100

Particle Size dp 11m


Figure 16C.9 Theoretical migration velocity calculated by equation (3.70) as a fum:tion of
particle size for atmospheric pressure and three temperatures (or three relative gas densities,
respectively). An electrical field strength of 3.0 kV/cm and a relative electrical permittivity of 10
are assumed for the calculations.

0.1

= 0.5
= 1.0
= 2.0
--m- delta = 4.0

d p = 1.01lm
epsr = 10
E = 3.0*1 ()5V/m

-0-

~
~!

delta

delta
__ delta

~ ~ ~ t---

0.02

300

500

700

900

1100

1300

1500

Temperature T in K
Figure 16C.l0 Theoretical migration velocity of a 1.0/lm particle (E = 3.0 kV /cm; e, = 10) as
a function of temperature; isolines correspond to constant relative gas densities.

512

HT/HP PRECIPITATORS FOR ADVANCED POWER GENERATION

For particles> 1 /lm the influence of the dynamic viscosity '1 dominates the
b-dependence, i.e. at higher temperatures or relative gas densities < 1,
respectively, the gas phase becomes more viscous and, hence, the particles
migrate more slowly. For particles < 1 /lm this behaviour reverses: here the
b-dependence of particle charge and Cunningham correction dominate over
the viscosity, i.e. at higher temperatures or relative gas densities < 1,
respectively, particles can penetrate the gas phase easier resulting in higher
migration velocities.
The situation is more complex in an ESP, because for a corona discharge
to develop the pressure level has to be raised (section 4.7.2). Therefore, it is
necessary to look at migration velocities of particles with a fixed size as a
function of temperature, keeping the relative gas density constant. This is
illustrated in Figure 16C.10 for 1.0/lm particles. Here it can be seen that, for
constant relative gas density, an increase in temperature will weakly decrease the migration velocity. For constant temperatures, doubling the
relative gas density shows less effect on the theoretical migration velocity
than reducing it by half.
The necessity of applying much higher electrical field strengths at high
relative gas densities to increase the corona currents still has to be considered. An increase in the electrical field E results in an enormous increase
of the migration velocity, because W th quadratically depends on E (equation
3.70). This is illustrated in Figure 16C.ll for a 1.0/lm particle at a relative
gas density of b = 4.0 for electrical field values of E = 3/6/9/12 kV/cm.

.:

0.6
0.5

dp

~ E = 12.0*1()5V/m

0.4
0.3

ii

0.2
0.1

-..............

...........

~m

----

= 1.0pm

epsr = 10
delta = 4.0

r--

6.0*105V/~
~ ,.....
E = 3.0j1()5V/m

300

500

700

900

1100

1300

1500

Temperature T In K
Figure 16C.ll Theoretical migration velocity of a 1.0 Jlm particle (6, = 10) for a relative gas
density of 4.0 as a function of temperature; different lines correspond to different electrical field
strengths.

513

GRADE EFFICIENCY

16C.5 Grade efficiency

When looking for the Tip-dependence of the grade efficiency T(d p ), the ESP
design must first be specified. Here a tube-type precipitator was assumed
with rNE = 150 mm and LNE = 5.0 m and operation conditions of E =
3.0 k VIcm and Vo = 1.0 m/s. Grade efficiency was calculated according to the
Deutsch equation (equation 3.87) for raised temperatures and atmospheric
pressure which corresponds to relative gas densities < 1.
The results are illustrated in Figure 16C.12. The efficiency curves show,
of course, the same tendencies as those of the migration velocities in Figure
16C.9; the absolute value of the typical minimum value is significantly raised
and its location is shifted as before to larger particle sizes for decreasing
relative gas densities. It can be seen that reducing the relative gas densities
leads to a dramatic improvement in efficiency; however, this neglects the
high pressure levels needed for corona generation.
Figure 16C.13 shows efficiency as a function of temperature with the
relative gas density as parameter for 1.0 Jim particles. Obviously, particles
will be less efficiently collected at high temperatures when the relative gas
density is kept constant and assuming a constant electric field strength.
However, when the potential higher electric field strengths at raised
relative gas densities are considered, things look different. Figure 16C.14
shows the efficiency values of a 1.0 Jim particle, calculated and plotted as a
function of temperature as before, but for a relative gas density of c5 = 4.0

--

~ 0.95
"'Q.

iii

"

t------'r-r\---jff-----;======::::L-,
P

= Po= 1 bar

E = 3.0*105V/m
0.9 t - - - - - - - 1 r l - - - \ - - - - f f - - t - - - j Vo = 1.0 m/s

LNE = 5.0 m
rNE = 150 mm
0.851--------\+-----(---+--4 epsr = 10
-0- T = 1375 K (or delta = 0.2)
0.8 1 - - - - - - - - - + - 1 . - - - , 1 - - 1 ___ T = 546 K (or delta = 0.5)

-m-- T = 273 K (or delta = 1.0)

0.1

10

Particle Size dp~


Figure 16C.12 Grade efficiencies at atmospheric pressure calculated according to Deutsch
(equation 3.87) for three different temperatures; the ESP design was assumed to be LNE = 5.0 m,
rNE = 150mm and its operation E = 3.0kV/cm and Vo = l.Om/s.

514

HT/HP PRECIPITATORS FOR ADVANCED POWER GENERATION

0.1

'i:
.
C!
.,..

0.9

0.8

j:'

>- 0.7
.1
.S:!

-0-

!!l

ffi

delta = 0.5 f--!----!---....!j!::==-...;:::::::t


delta = 1.0

- - - delta 2.0
_
delta = 4.0

0.6

=30*10
. 5 V/m; LNE =5.0 m

f--!--:--,-+---':=--I---~

0.5

300

700

900

1100

1300

1500

Temperature T in K
Figure 16C.13 Efficiencies according to Deutsch (equation 3.87) for a 1.0)lm particle as a
function of temperature; different lines correspond to different relative gas densities. ESP design
and operation conditions as before.

= 6.0*105V/m
E = 5.0*10 5 V/m

E
0.95

~
E
::1.
C!
.,..

dp

>~

'u
==
w

0.85

j:'
u
c:

0.9

0.8
0.75

0.15

= 4.0'105 V/m

= 3.5*1 05V/m

0.1

-o-...
"'C

CD
:::I
CD

III

:::I

0.15 "'C

= 1.011m

(:)

= 4.0
LNE = 5.0 m
rNE = 150 mm
Vo = 1.0 m/s
delta

0.2
E = 3.0*1 05VI

-Z.
:.l!

0.25

epsr = 10
0.7
300

500

700

900

1100

1300

0.3
1500

Temperature T in K

Figure 16C.14 Efficiency of a 1.0)lm particle (B, = 10) for a relative gas density of 4.0 as a
function of temperature; different lines correspond to different electrical field strengths. ESP
design and operation conditions as before.

OPEN QUESTIONS

515

with the applied electric field as parameter. Obviously, particle collection is


extremely sensitive to electrical field strength; e.g. at a temperature level of
1300K, doubling the electrical field strength from 3 to 6 k V/cm increases the
migration velocity by a factor 4 (compare Figure 16C.ll), which reduces
penetration (1 - efficiency) by more than a factor of 25!
Experimental results for total mass efficiencies (integrating over all
particle sizes) have been published, e.g. by Feldman and Bush [5] and
Rinard et al. [6]. As expected, they found a strong influence of the electrical
field strength on particle collection. Feldman and Bush refer to a wire-pipe
ESP of Union Carbide Olefins Co. operating at temperatures of 870-1000K,
at pressures of 3-8 bar and gas velocities of 0.2-1.2 m/s. The ESP consisted
of 19 pipes of 15.2 cm in diameter, 1.8 m in length and discharge wires of
2.1/3.4 mm in diameter. Rinard et al. refer to a test facility located at the
Denver Research Institute with one tube, 30.5 cm in diameter and 2.1 m in
length, operating at temperatures up to 1200K, at pressures up to 10 bar and
a flow rate of 0.078 m 3 /s at these conditions.

16C.6 Open questions


For a successful high temperature, high pressure ESP to be designed and
developed the following factors need careful consideration.

16C6.1

Electrical resistivity

The combination of low ash resistivity at high operating temperatures,


together with the correspondingly high current densities, will not necessarily
lead to back-corona since the high gas densities should suppress it. On the
other hand electrical re-entrainment of particles might result from the low
resistivity values.

16C6.2

Mechanical stability of material

The suitability and integrity of the materials used at temperatures ~ 1000C


has to be considered carefully. For long-term applications at elevated
temperatures, the creep behaviour of the material has to be taken into
account. Furthermore, the flue gas and particularly the ash particles at high
temperature are much more aggressive; therefore, corrosion could be a
severe problem.

16C6.3 Rapping
For some materials, at temperatures above 700C the region of forgeability
starts. This can become a problem if rapping is done by conventional

516

HTjHP PRECIPITATORS FOR ADVANCED POWER GENERATION

hammer systems. Cleaning by pulse jets, analogous to those used in bag


houses, or by ultrasonic horns are under discussion.
16C6.4 Electrical insulation
Insulator arrangements commonly used on the roofs of ESP housings
operate at temperatures about 50C less than the gas and are stressed
thermally, mechanically and electrically. Unfortunately the electrical resistivity of most insulator materials drastically decreases at temperatures above
about 300C. An application of the so-called advanced ceramics for insulating material for the extreme requirements of both temperature and pressure, might be promising and should be investigated.
16C6.5 Emptying of hoppers
To bring the collected dust out of the hoppers, it is necessary to overcome
a pressure barrier of some 10 bar. In order to guarantee secure dust handling
a carefully designed pressure sluice is absolutely essential.
16 C 6. 6 Electrical power consumption
The electric power consumption, as a fraction of the total electric output
from a PCFB electric generating plant, has been observed to be between 1.5
and 2% [7]. If the ESP efficiency has to be increased, the collecting area has
to be enlarged, which could lead to probably unacceptable high power
consumption. A solution might be to use power-conserving means on the
transformer rectifier sets, such as intermittent electrical energisation.
For successful ESP operation at high temperatures, correspondingly high
pressure levels are necessary. Thus, the gas can withstand the electrical
breakdown much better than at normal pressure conditions, leading to
extremely high electrical field strengths. For the same reason back-corona is
not expected to cause severe problems. The high electrical field strengths
therefore suggest rather smaller specific collecting areas for efficiencies
~ 99% than for 'normal' operation.
Tassicker [7J concludes that 'the data available would be sufficient for
the commercial-scale of an ESP for conditions of 5-15 bar and 400-700 dc.
The available data are less definite for a firm design at 850-900 dc. More
pilot-plant work is needed before a commercial-scale plant in this range
could be confidently sized'.

16C.7

Symbols

mobility of gas ions

REFERENCES

Eo
jNE

~E

Q';

r NE
r SE

T
T(d p )

Uo
Ucrit

(v)

1'/

517

electric field strength


corona initiation field strength
electrical current density at collecting wall
length of collecting tube
particle saturation charge
pressure
radius of collecting tube
radius of discharge wire
temperature
grade efficiency
applied voltage
corona onset voltage
sparkover voltage
mean thermal velocity of gas molecules
fluid velocity
theoretical migration velocity of an individual particle
relative gas density
(epsr) electrical permittivity of particle material
viscosity of fluid
mean free path of gas molecules
density of fluid

References (16C)
1. Perrin, A.1. (1994) Clean Coal Technology - An Industry Perspective. Presented at
Prospects for Clean Coal-A Contractors' Meeting, I-2nd November, Nottingham, U.K.
Sponsored by the DTI/DOE. RTSU Harwell, U.K.
2. Weber, E. (1984) Electrostatic precipitation under extreme conditions of temperature and
pressure. Proc. 2nd Int. Con! Electrostatic Precipitation, Kyoto, Japan, pp. 85-95, EPA,
Pittsburgh, USA.
3. Bush, R.1., Feldman, P.L. and Robinson, M. (1979) High pressure, high temperature
electrostatic precipitation. 1. Air Pollut. Control Assoc., 29, 365-71.
4. Cachet, R. (1961) Lois charge des fines particules (submicroniques) etudes theoriquescon troles recents spectre de particules. ColI. Int. la physique des forces electrostatiques et
leurs application. Centre National de la Research Scientifique, 102, 231-8, Paris, France.
5. Feldman, P.L. and Bush, J.R. (1980) Performance of electrostatic precipitators at high
temperatures and high pressures. VDI-Berichte,363, 87-92.
6. Rinard, G., Rugg, D.E. and Yamamoto, T. (1987) High-temperature high-pressure electrostatic precipitator electrical characterization and collection efficiency. IEEE Trans. Ind.
App/., IA-23, 114-19.
7. Tassicker, 0.1. (1986) High temperature-pressure electrostatic precipitator for electric
power generation technologies: an overview of the status. [ChernE Syrnp. Ser., 99,331-9.

16D

Computer sizing of precipitators

Although at present, precipitators are normally sized by the supplier, using


his databank knowledge and experience from similar processes, the advent
of very high speed computers for solving Poisson and Laplace equations
may eventually mean that sizing will be computer generated from first
principles. A number of companies and research investigators are already
working on this approach, as indicated in chapter 9, but for computer based
sizing to be generally accepted by industry, a great deal of research and
confirmatory investigations will have to be completed in the following areas
to determine how they impact on performance from a theoretical standpoint
for inclusion in the programming.
Some of these have been identified in previous chapters from various
operational standpoints, but there is a need for them to be incorporated in
the overall sizing programme. Some in fact will have their own subroutine
to the major format on sizing if the approach is to prove acceptable.
1. Rapping and re-entrainment.
2. Effect of ionic wind on particle transport phenomena and on gas
distribution under operational conditions.
3. Initiation conditions leading to reverse ionisation/back-ionisation from
basic input information.
4. Influence of small variations in carrier gas analysis and temperature.
5. Improved usage of electrical power.
6. Effect of corona and collector plate current distribution.

Some of the above items must also be considered in their own right, as
indicated in the introduction, since they can have considerable impact on
precipitator performance and hence size and cost. In summary, although the
electrostatic precipitator has been part of our industrial heritage for almost
a century, there is a growing need for higher efficiency, improved reliability
and availability of equipment, to ensure that future generations enjoy a
clean environment, particularly free from man-made pollution in the form
of particulate matter.

Index
Page numbers appearing in bold refer to figures and page numbers appearing in italic refer to
tables.

Aerodynamic factors affecting performance


113-50
Aerosols 154
Agglomeration 157, 159, 162
Air inleakage hoppers 109
Alkali chlorides 177, 360, 363, 368
Appli~ations of ESPs, see Contents v, ix, x,
Xl

Aspect ratio 188, 189, 355


Automatic voltage control 22, 217-29
analogue type 23
microprocessor 23, 91, 217-29
supervisory type 241-6
Back corona 169, 195,226-8, 241, 265, 364
onset of back ionisation 22
severe back ionisation 216
Bag filters, see Fabric filters
Brownian motion 3, 154, 162, 184
Bus section/field 25
Carbon absorption 7, 301
Carbon particles 160, 353, 371, 379
Casings 97-100, 306-11, 324, 325, 387-90
pressure testing of casings 318, 324- 5
purging of casings 340-1, 395,415
Cenospheres 357
Cochet charging model 52-4, 84, 114,
509-10
Cohesion/cohesivity 163-6, 178, 262, 357,
452
Collector elements
catch space 39, 95, 313
concentric ring 405
FRP 97, 391, 393, 403, 408
rolled channel 94
tubular 15, 97, 403
Commissioning of precipitators 305-48
electrical 326-339
mechanical 306-326
process 339-348
Composition of dust, effect of 17-28, 183,
272-4
Computational fluid dynamics 74, 80,
142-8
Conditioning of gases and dusts 429-81
ammonia 357, 445-52

dual S03 + NH3 352, 452-4


injection rate optimisation 471-80
moisture 16, 169, 184, 362, 365, 372,425
other reagents 426
sulphur trioxide 352, 429, 436-8
S03 injection rate equation 440-5
S03 injection rate prediction 438-40
Conditioning towers 362, 363, 369, 373
Corona discharge
breakdown voltage 29, 30, 31, 508
corona power 193-5, 228-9
corona suppression 91, 195, 358, 368,
376
coaxial system 38-44
current density 38, 40, 42, 45, 241
initiation/onset voltage 33-6, 34, 262,
505, 508
parallel plate 34-44, 246
space charge 134-6, 195,451
streamers 30, 170
threshold voltage 31-44, 196-8
see also Particle charging mechanisms
Cunningham correction factor 56-9
Cyclones/inertial separators 2, 3, 4, 9
Design considerations 180-90
Detarring 17, 342, 406, 411-15
Deutsch 20, 62-5, 84,113-17,182,185,255,
281-5,514
Deutsch number 64, 116, 117, 122
Diffusers 129,317
Diffusion 114-17, 122
Diffusivity model 82-5
Discharge electrode spacing 34-6
Discharge electrode support 92-4, 187
Dry precipitators 349-81
applications 349-81
Dry scrubbing 17, 369, 379
Duct spacing 34-8, 40-4, 48-50, 185,424
Earthing and grounding 335
Effective migration velocity 65-70, 182-6,
194,434
Electron attachment 30
Electrical clearances
external 110-11
internal 110, 200, 309, 320

520

INDEX

Electrical energization 192~46


full wave 192, 201~1O
high frequency power conversion 192,
487~90

intermittent energization 192, 210~ 17,


423
pulse energization 121, 192, 230~41, 423
transformer power ratings 206~ 8
Electric fields 49, 50
distribution 44~52, 237, 238
plate precipitators 48, 49, 50
tube precipitators 46~8, 47
Electrodes
controlled emission type 90, 122
discharge elements 38, 89~92, 90, 187,
315,386
high emission types 90
mounting of electrodes 92~4, 187
pubescent type 13
Electropositive/negative gases 30
Extended Deutsch equation 256~ 8
Fabric filters 3, 5, 9, 35,421
Flares (gas distribution) 317, 318
Flue gas conditioning systems 429~81,
492~500

Fly ash composition, importance of


183, 272~3
Fume 13,370

172~8,

Gas distribution 129~34, 138, 317, 355


CFD modelling 80, 81, 142~8, 485
field testing 148~9, 323
flow bypass 109, 137
gas distribution methods 129~ 31
scale modelling 139~42
visualization 122, 148
see also Aerodynamic factors affecting
performance
Heavy metals 298~303, 358, 369
High resistivity particles, see Particle
resistivity
High temperature/high pressure
precipitation 501 ~ 16
High voltage supplies, see Electrical
energization
History of precipitation II ~23
early designs and applications 11 ~ 20
energization systems 20~ 3
Hoppers 108~ 110, 311
air in leakage 109, 325, 377
heating requirements 334, 351
overfilling effects of 109
pyramid type 316
trough type 389
Hydrofiners 16

Inertial sampling 296~ 7


Insulators
H.T. lead through 100~2, 322, 387, 409,
516
heating 334, 351, 387
purging 387, 388
[on mobility 37, 504
Ionic space charge 46, 50, 51
Isokinetic sampling 293~4
see also Testing of precipitators
Lambert's law 161
Linear inductors 208~ 10
Mass flux 70~6
Materials of construction
high temperature applications 515
wet precipitators 391, 392, 393
Maxwell equation 126, 246
Mechanical collectors 2, 4, 9
Mist precipitators
applications 407 ~ 16
design considerations 402~ 7
Modelling of precipitators 250~91
Caiiadas 286~ 7
computer 274, 285
CSIRO 272
full scale 275, 277
Pad ova University 288, 289
pilot scale 276
Southern Research Institute 285
Modified Deutsch Formulae 20, 68, 250,
255, 256, 284, 420
CSIRO 250, 256~ 74
Matts Ohnfeldt 68, 250, 255, 284
Petersen(FLS) 250, 255
Numerical flow model 126
Parallel plate nomenclatures 25, 34, 283
Particle charging mechanisms
Cochet's model 52~4, 84, 114, 509
collision charging 52
ion diffusion 52
saturation charge 53, 54, 55, 510
Particle migration 54~85, 51O~ 13
exponential law 20
practical consideration 59, 69
theoretical 57, 81
Particle re-entrainment 69, 128, 136~ 7, 356,
485
Particle resistivity 166~ 72, 263, 264, 426,
430~6

critical value 176


effect on corona 170, 262

INDEX
effect on performance 263-4, 362, 433,
434,435
effect of temperature 167, 168, 263, 271,
427
measurement of 167
prediction of resistivity 432-6
Particle size
distributions 59, 60, 153-4, 258, 259, 296
effect on performance 58, 65,
76-85, 184-5,511,513
grade efficiency relations 61, 63, 64, 66
optical properties 161
shape 153, 160
Particle tracking model 76-82
Particle transport 28, 76-82, 114,254, 510
see also Effective migration velocity
Peclet number 116, 121-6
Performance line 267-74
Performance testing
impactors inertial sampling 296
isokinetic sampling 293-4
methods and standards 292, 300
Pilot precipitators 81, 117, 263, 266, 276,
364, 508
Poisson's equation 45, 46
Rapping 102-8, 346
collectors 103, 104, 314,316, 321
discharge elements 103, 104, 315, 321
Rapping optimization 107,241,
245, 246, 346, 485
Re-entrainment effects 69, 103, 136, 425
Rectifier control methods
transductor/magnetic saturable reactors
22
thyristors 22, 201-46
see also Automatic voltage control
Residence time 131-4
Reynolds number 56, 126
Safety interlock systems 331
Sampling of gases 292-303
standards 292, 295, 300
Scouring of dust from internals 137, 189
Scrubbers 2, 3, 4, 9, 396

521

Secondary flow in precips 113, 117, 122-6


Sectionalization 189, 199-201
Selective dust separation 15, 19, 382
Sliding bearings 97, 98, 310
Smog 1
Sneakage and sweepage 137
Sodium depletion 176, 464
Space charge effects 91, 134-6, 188, 386
Spray irrigation 322
Stokes-Cunningham correction 56, 162
Stokes' diameter 3, 159
Stokes' law 56, 258
Sulphate reducing bacteria 413
Supervisory AVCs 241-6
Temperature effect on performance 167,
183,184,261-6,271-2,353,427,514
Testing of precipitators 292-5
Theory of precipitation 25-87
Thermal expansion 344
Thermal insulation 317
Top housings 101,308
Treatment time 410, 420, 424
Triche! pulses 30
Tube type precipitators 15, 26, 33, 38, 46,
403,406,413
Turbulence 27-8, 76-85, 113-27
Two stage precipitation 25, 27
Upgrading of precipitator performance
418-424
Velocity of gas 127-9, 138, 189,355
Viscosity 28, 56, 113, 261, 505, 512
Volatiles 298, 299, 367, 370, 501
Voltage/current relationships 29, 42, 91,
197,337,338,507,509
Water treatment 390, 391
Wet precipitators 7, 9, 8, 17, 25, 382-400
applications 394-401
collector film flow 19, 383, 384
design considerations 383-94
spray irrigation 384, 386
washdown spray system 322, 337, 385

You might also like