EPA-650/2-74-132
JULY 1972
Environmental Protection Technology Series
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S. Environ-
mental Protection Agency, have been grouped into series. These broad
categories were established to facilitate further development and applica-
tion of environmental technology. Elimination of traditional grouping was
consciously planned to foster technology transfer and maximum interface
in related fields. These series are:
1. ENVIRONMENTAL HEALTH EFFECTS RESEARCH
2. ENVIRONMENTAL PROTECTION TECHNOLOGY
3. ECOLOGICAL RESEARCH
4. ENVIRONMENTAL MONITORING
5. SOCIOECONOMIC ENVIRONMENTAL STUDIES
6. SCIENTIFIC AND TECHNICAL ASSESSMENT REPORTS
9. MISCELLANEOUS
This report has been assigned to the ENVIRONMENTAL PROTECTION
TECHNOLOGY series. This series describes research performed to
develop and demonstrate instrumentation, equipment and methodology
to repair or prevent environmental degradation from point and non-
point sources of pollution. This work provides the new or improved
technology required for the control and treatment of pollution sources
to meet environmental quality standards.
-------
EPA-650/2-74-132
AN ELECTROSTATIC PRECIPITATOR
PERFORMANCE MODEL
by
Grady B. Nichols and John P. Gooch
Southern Research Institute
2000 Ninth Avenue South
Birmingham, Alabama 35205
Contract No. CPA 70-166
ROAP No. 21ADJ-026
Program Element No. 1AB012
EPA Project Officer: R. C.Lorentz
Control Systems Laboratory
National Environmental Research Center
Research Triangle Park, North Carolina 27711
Prepared for
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON, D.C. 20460
July 1972
-------
EPA REVIEW NOTICE
This report has been reviewed by the National Environmental Research
Center - Research Triangle Park, Office of Research and Development,
EPA, and approved for publication. Approval does not signify that the
contents necessarily reflect the views and policies of the Environmental
Protection Agency, nor does mention of trade names or commercial
products constitute endorsement or recommendation for use.
This document is available to the public for sale through the National
Technical Information Service, Springfield, Virginia 22161.
11
-------
TABLE OF CONTENTS
Section Page
INTRODUCTION ................................... 1
1. DEVELOPMENT OF A PILOT-SCALE PRECIPITATOR ...... 3
A. Purpose for Building ....................... 3
B. Philosophy of Design ....................... 3
C. Design Requirements and Details ............ 5
D. Design Parameters .......................... 12
2 . FUNDAMENTAL STUDIES ............................ 13
A. Particle Concentration Distribution ........ 17
B. Particle Size Effects ............... : ...... 23
C. Resistivity ................................ 27
D . Sparkover and Back Corona .................. 53
E. Optimization of Precipitator Design for
High Resistivity Dusts ................. 67
F. Bibliography .............................. 73
3 . REENTRAINMENT STUDIES .......................... 74
A. Scouring ................................... 76
B. Rapping Reentrainment ...................... 78
C. Electrical Forces .......................... 80
D. Sparking ................................... 80
E. Saltation .................................. 81
F. Hopper Losses .............................. 81
G. Full-Scale Precipitator Tests .............. 81
H . Conclusions ................................ 84
I . Bibliography ............................... 84
4. REFINEMENT OF PRECIPITATOR MATHEMATICAL MODEL.. 85
A. Collection Efficiency ...................... 85
B. Migration Velocity ......................... 86
C. Particle Size Considerations ............... 87
D. Particle Charging Time ..................... 88
E. Particle Reentrainment ..................... 88
F. Verification of the Precipitator
Mathematical Model ..................... 90
G. Conclusions ................................ 94
-i-
SOUTHERN RESEARCH INSTITUTE
-------
TABLE OF CONTENTS (Continued)
Section
Page No,
5. ECONOMIC COMPARISONS FOR COLLECTION OF HIGH
RESISTIVITY DUST 97
A. Basis for Comparison 97
B. Enlarged Precipitator at Normal Temperature 98
C. Fly Ash Conditioning with Sulfuric Acid
Vapor 99
D. High Temperature Operation 100
E. Low Temperature Operation 103
F. Overall Comparison 108
G. Bibliography Ill
6. LOW TEMPERATURE CORROSION AND FOULING 112
A. Introduction 112
B. Sulfuric Acid Occurrence in Flue Gas 112
C. Factors Influencing Corrosion Rates 126
D. Fouling of Low Temperature Surfaces 139
E. Laboratory Corrosion Studies 140
F. Summary of Field Experience 146
G. Methods of Assessing Corrosion Tendencies
of Flue Gases 153
H. Summary and Conclusions 154
I. Bibliography 157
Appendix
1. LIST OF SYMBOLS 162
2. COMPUTER PRINTOUT FOR ELECTROSTATIC PRECIPITATOR
MODEL 164
-ii-
-------
LIST OF FIGURES
Figure Ho. Page No.
1.1 Electrostatic Precipitator System Model 4
1.2 Pilot Electrostatic Precipitator 6
1.3 Mechanical Configuration of the Pilot
Precipitator 7
1.4 Arrangement of Flow Control Plates and Kicker
Plates 9
1.5 Velocity Contours at 4.94 ft/sec Average
Velocity 10
1.6 Details of the Collection System 11
2.1 Variation in Length Required for Collection
for Two Gas Velocities with a Constant
Migration Velocity (Laminar Flow Case).... 15
2.2 Obscuration as a Function of Distance from
the Collecting Electrode for Various Gas
Flow Ratios 18
2.3 Laser Extinction Probe Assembly 20
2.4 Obscuration of White Light by a Varying Dust
Load for Two Particle Size Distributions.. 22
2.5 Effect of Current Density Changes on
Obscuration at a Flow Velocity of 3.5
ft/sec 24
2.6 Graph of Computed Precipitation Rate
Parameter for Example Showing Increase in
Wp with Increasing Gas Velocity 28
2.7 Precipitator Rate Parameter vs. Gas Velocity
- Precipitator Model 29
2.8 Cyclone Collector - Cylindrical Electrode Cell
for Collection External to the Duct 31
2.9 Schematic of Point-Plane Resistivity Probe.... 33
2.10 Kevatron Australian Resistivity Probe 35
2.11 Comparison of Particle Size Distributions from
Precipitator Hopper and Two Resistivity
Probes 36
2.12 Resistivity vs. Temperature for Different Size
Fractions of Beulah Standard Electrostatic
Precipitator Ash 38
2.13 Variation of Fly Ash Resistivity with Time
Using the In-Sit^u Cyclone Probe at a
Temperature of =280°F 39
2.14 Comparison of Kevatron and Cyclone Resistivities
with Point'-Plane Resistivities at an Electric
Field of 2.5 kV/cm. Settled Values for
Cyclone, Peak Values for Kevatron 46
-iii-
SOUTHERN RESEARCH INSTITUTE
-------
LIST OF FIGURES (Continued)
Figure No. Page No.
2.15 Comparison of Kevatron and Cyclone Resis-
tivities with Point-Plane Resistivities
at an Electric Field of 2.5 kV/cm. Peak
Current Values Used for Cyclone and Kevatron 47
2.16 Comparison of Cyclone Resistivities with
Point-Plane Resistivities. Maximum
Electric Field on Point Plane. Settled
Values for Cyclone 48
2.17 Laboratory Resistivity Measurements for Two
Fly Ash Samples from Western Coal 52
2.18 Variation in Resistivity for Various Oxides as
a Function of the Parameter 1/kT 54
2.19 Composite Plot of Resistivity Data for Several
Fly Ash Samples 55
2.20 Behavior of a Point-to-Plane Electrostatic
Precipitator Based on Theoretical Consid-
erations of Sparking and Back Corona.
Clean Plate Curve is Measured Curve,
Remaining Curves Computed 58
2.21 Dimensions of Point-Plane Precipitator Used in
Laboratory Studies of Sparking and Back
Corona Conditions 61
2.22 Experimental Volt-Current Curves for Point-
Plane Device with a Variety of Dust Resis-
tivities 63
2.23 Voltage-Current Curves for a Precipitator
Section 65
2.24 Voltage-Current Curve for a Precipitator
Operating on Fluorspar Dust with Very High
Resistivity 66
2.25 Schematic of Standard and Reduced Current
Density Test Conditions for Southern
Research Institute Model Precharging Tests 69
2.26 Comparison Between the Collection Efficiency
for Standard Conditons and Reduced Current
Density Conditions for Constant Electric
Field 70
2.27 Comparison Between the Percentage of Material
Removed within Each Increment of Length for
Two Values of Current Density 72
-iv-
-------
LIST OF FIGURES (Continued)
Figure No. Page No.
3.1 Effect of Reentrainment on the Efficiency of
a Four-Section Precipitator Designed for
a No Reentrainment Efficiency as
Indicated 75
3.2 Relationship Between Scouring for Various
Applied Voltages Defined as Excess Loss
Over No Overrun Collection Efficiency 77
3.3 Precipitator Losses as a Function of Gas
Velocity for Two Full-size Precipitators.. 83
4.1 Approximation of Inlet Particle Size Distri-
bution 89
4.2 Computer System Flow Diagram 91
4.3 Comparison of Computed and Measured Collection
Efficiency for Various Test Conditions
Utilizing the Model 93
4.4 Comparison of Computed and Measured Collection
Efficiency Under Field Test Conditions.... 95
5.1 Resistivity as a Function of Temperature at
Plant 6 104
5.2 Schematic of Flue Gas Flow Plan 105
5.3 Cost of Boiler Efficiency Loss 107
6.1 Equilibrium Conversion of S02 to SO3 113
6.2 Equilibrium Conversion of S03 to H2SO,, at 8.0
vol % H20 in Flue Gas 115
6.3 Dew Point and Condensate Composition for Vapor
Mixtures of H20 and H2SOW at 760 mm Hg
Total Pressure 118
6.4 H2SO,, Dew Points for Typical Flue Gas Mois-
ture Concentrations 119
6.5 H2SOi, Dew Point Obtained by Various Investi-
tigators 122
6.6 Percent HjSO^ Available for Condensation for
Flue Gas of 100 ppm H2SOi, and 10% H20
Vapor 124
6.7 Variation in Condensation Rate with Surface
Temperature 125
6.8 Equilibrium Sulfuric Acid Condensate Compo-
sition 128
6.9 Corrosion of Steel in Flue Gas as a Function
of Calculated H2SOn Condensate Strength.. 129
6.10 Corrosion of Steel as a Function of H2SOi,
Concentration at 75°F 130
-v-
SOUTHERN RESEARCH INSTITUTE
-------
LIST OF FIGURES (Continued)
Figure No. Page
6.11 Variation of Condensation and Corrosion with
Surface Temperature .....................
6.12 Variation in Rate of Acid Buildup (RBU) and
Excess Cation Content of Fly Ash as a
Function of Surface Temperature ......... 134
6.13 Consumption of the Available Base on Fly Ash
as a Function of the Concentration of
Neutralizing Acid in Flue Gas with
5 gr/scf Fly Ash ........................ 135
6.14 The Effect of Chlorine Addition on Corrosion
of Mild Steel in a Synthetic Flue Gas... 138
6.15 Schematic Diagram of Apparatus Used in
Corrosion Experiments ................... 143
-vi-
-------
LIST OF TABLES
Table No. Page No.
1.1 Design Parameters 12
2.1 Calculated Collection Efficiency for a Dust
Composite, Based on a Charging Field of
5 kV/cm and a Collecting Field of
3 kV/cm 26
2.2 Summary of Ash Analysis - Western Coal 51
2.3 Comparison of Resistivity Determined by the
Parallel Disc Method and the V-I Curve
Method 62
2.4 Model Precipitator Operating Conditions for
Comparative Efficiency Tests 68
4.1 Operating Data for the Pilot-Scale Precipita-
tor 92
4.2 Operating Data for Selected Field Installa-
tions 96
5.1 Cost of Hot Precipitators Reported to Southern
Research Institute 102
5.2 Estimated Total Capital Investment - 99% Effi-
cient Precipitator, 250 MW Unit 109
5.3 Estimated Incremental Annual Cost in Dollars -
High Resistivity Dust 110
6.1 Composition, Percent by Weight, Spectrographic
Analysis of Specimens Tested 127
6.2 Sulfur and Chlorine Concentrations in Flue Gas 136
6.3 Fly Ash Properties 141
6.4 Corrosion Rate Experiments 144
6.5 Properties of Flue Gas and Fly Ash for Various
Coal-Fired Boilers 148
-vii-
SOUTHERN RESEARCH INSTITUTE
-------
ENVIRONMENTAL PROTECTION AGENCY REVIEW NOTICE
This report has been reviewed by EPA and approved for publi
cation. Approval does not signify that the contents
necessarily reflect the views and policies of EPA, nor
does the mention of trade names or commercial products
constitute endorsement or recommendation for use.
-viii-
-------
ACKNOWLEDGEMENTS
The information contained in this document is the result
of a team effort at the Southern Research Institute. As such,
many individual contributions are included. Specifically, the
overall project was under the direction of Sabert Oglesby, Jr.,
Vice President and Director of the Engineering and Applied
Sciences Department. He supervised the entire program and made
significant contributions to the writing of this final report.
Dr. E. B. Dismukes was responsible for the gas conditioning
and resistivity portion of the work and made significant
contributions to that section. Dr. R. E. Bickelhaupt contributed
in the area of resistivity and conduction mechanisms portions
of the work. N. L. Francis contributed to the upgrading of the
precipitator performance model in the area of gas flow and
reentrainment, as well as in the design and checkout of the
pilot precipitator.
-ix-
SOUTHERN RESEARCH INSTITUTE
-------
ABSTRACT
The objectives of the study covered by this report were:
1) extend the precipitator computer model to include factors
not included in the first model influencing its accuracy,
2) design and build a pilot precipitator for further studies
of the factors influencing precipitation processes, 3) to
review limitations to precipitator performance due to back
corona and sparking, 4) to investigate particle concentration
profile in the interelectrode space, 5) obtain data from both
field and pilot plant tests to attempt to verify the computer
model, and 6) analyze potential for optimizing precipitator
performance by design or operating modifications.
This report reviews the design details of the pilot
precipitator and the results of particle concentration profile
studies. Discussions of measurement of resistivity and
correlations between resistivity and precipitator operation
are also reviewed.
-x-
-------
AN ELECTROSTATIC PRECIPITATOR PERFORMANCE MODEL
INTRODUCTION
This is the final report under Contract CPA 70-166, cover-
ing a study of an electrostatic precipitator performance model.
The purposes of this study were to analyze some of the critical
factors limiting precipitator performance, to establish quanti-
tative relationships between precipitator performance and these
factors, and to review operating experience and costs to deter-
mine the influence of gas temperature on overall system econo-
mics and problems related to the tendency toward corrosion and
fouling of the air heater and precipitator.
In carrying out this program, a small-scale precipitator
was designed and constructed to permit studies to be made of pre-
cipitator fundamentals, such as the effect of charging time,
relation of particle distribution in the interelectrode space to
collection efficiency, reentrainment factors, etc. This report
outlines the results of these fundamental studies utilizing the
pilot precipitator. These studies provide a more thorough under-
standing of the basic precipitation process in addition to pro-
viding a basis for estimating precipitator performance based
upon dust characteristics and precipitator operating parameters.
In addition to the small-scale precipitator studies,
operation of several full-scale plants was reviewed, particularly
the plants where unusual problems were encountered. Many of these
problems involved dust resistivity, both high and low values. In
attempting to correlate resistivities with precipitator perfor-
mance, it was found that resistivities determined by various
methods differed by such a wide margin that no analysis was pos-
sible without a better understanding of resistivity measurement
techniques. Measurements were made at several power stations
burning different types of coal in an effort to compare resis-
tivities measured by several types of apparatus and to provide
data for analysis of precipitator performance.
Finally, the mathematical precipitator model developed
under Contract CPA 22-69-73 was upgraded by providing refinements
that were not included in the initial simplified version. This
model includes some input from the small- and full-scale precipi-
tators to determine how closely the predicted and experimental
values matched.
-1-
SOUTHERN RESEARCH INSTITUTE
-------
-2-
The various studies carried out under this contract can
be grouped into categories which include the development of a
pilot-scale precipitator model, fundamental studies of the pre-
cipitation process, reentrainment, refinement of the precipita-
tor mathematical model, economics of collecting high resistivity
dusts, and studies of problems of low temperature corrosion and
fouling.
-------
-3-
SECTION 1. DEVELOPMENT OF A
PILOT-SCALE PRECIPITATOR
A. Purpose for Building
An electrostatic precipitator can be described in terms
of the system input variables and subfunctions as shown in Figure
1.1. The complexity of the system precludes a detailed evalua-
tion of each subfunction in full-scale precipitators operating
on industrial processes because of both the costs and difficulty
in controlling variables in industrial operations. Thus, it was
desirable to construct a general purpose precipitator with suffi-
cient flexibility to provide a means for experimentally investiga-
ting the effects of modifications to individual system subfunc-
tions.
In addition to providing a tool for evaluating the effects
of variations of the system inputs and variables, a pilot-scale
precipitator provides the means for evaluating a computer systems
analysis model for describing precipitator behavior. A prelim-
inary computer model of an electrostatic precipitator system
which was developed under a previous contract was modified under
this program. The purpose of this modification, as described in
Section 4,was to upgrade the model to more closely predict the
behavior of physical precipitator systems.
Thus, the purpose for constructing a pilot-scale precipita-
tor is twofold:
it provides a means for evaluating the effects
of modifying subsystem inputs and variables on
precipitator performance, and
it provides an easily controlled physical model
to provide data for evaluating the computer
systems model of an electrostatic precipitator.
B. Philosophy of Design
The pilot-scale precipitator constructed for this project is
fundamentally a research tool. Therefore, the design philosophy
utilized is based on the three considerations of flexibility,
accuracy, and utility. The flexibility requirement is necessary
because of the wide range of possible operating configurations
available. The accuracy requirement is important because of the
significance of the conclusions that may be drawn from the
SOUTHERN RESEARCH INSTITUTE
-------
WIRE RADIUS
COLLECTOR RADIUS
WIRE ROUGHNESS
SECONDARY EMISSION
AVALANCHE COEFFICIENT
IONIZING RADIATION
RESISTIVITY
VAN DER WAALS, MOLECULAR.
AND MECHANICAL
ELECTRONEGATIVE GAS
GAS VELOCITY
VELOCITY DISTRIBUTION
COLLECTION AREA
VOLUME FLOW
APPLIED VOLTAGE
GAS DENSITY
ION MOBILITY
DUST THICKNESS
SECTIONALIZATION
COLLECTION AREA
WIRE RADIUS
COLLECTOR RADIUS
PARTICLE SIZE
DIELECTRIC CONSTANT
TIME
TEMPERATURE
DUST LOAD
GAS AND DUST
ION VELOCITY
, GAS AND UNCOLLECTED DUST
COLLECTED DUST
DUST LOAD
PLATE DESIGN
HOPPER DESIGN
GAS VELOCITY
•GAS DISTRIBUTION
RAPPING FORCE
RAPPING INTERVAL
PARTICLE SIZE
DUST PROPERTIES
TEMPERATURE
Figure 1.1. Electrostatic Precipitator System Model
-------
-5-
experiments conducted with the apparatus, and the utility
requirement is dictated by the need to minimize the manpower
required to conduct tests and change conditions between tests.
In addition to the above requirements, it is important
for the general configuration of the pilot unit to be similar
to existing full-scale collectors. This factor is important
since the pilot unit was used to evaluate a computer systems
model that ultimately is planned to be used for evaluating full-
scale precipitator systems.
C. Design Requirements and Details
A precipitator system can be described in terms of the
fixed and variable parameters that are associated with the
mechanical and electrical components of the installation. Since
this device is fundamentally a research tool, the fixed param-
eters were intentionally kept at a minimum, while the variable
parameters were allowed as much range as was thought to be prac-
tical.
The general layout of the pilot precipitator is shown in
the photograph in Figure 1.2. The electrical power supply con-
trols are located above and to the rear of the precipitator
proper. The air inlet, dust feed, and temperature controls are
located at the right of the photograph; and the fan, velocity
control, and measurement systems are located at the left side.
The collection hoppers are shown between the precipitator support
structure in the lower center.
1. Mechanical design
The mechanical configuration of the pilot model precipita-
tor is shown in Figure 1.3. The air is introduced into the elec-
trical heater section (11-E) at the inlet. The gas flows into a
rotary mixing section (10-C) where the dust from the dispenser
system (21-B) is fed into the gas stream through a paddle wheel
fan (21-D) and then into a venturi mixing chamber (13-A). .
The gas stream then flows through Sections 11-C and 10-A
where devices for generating a more nearly uniform gas stream
are located. The gas flows from these sections through the
4-section precipitator, the rectangular-to-circular transition
(10-B), the flow control orifice (1007), the velocity control
unit (12-A), out through the induced draft fan (2045), and is sub-
sequently emitted to the atmosphere.
-------
Figure 1.2. Pilot Electrostatic Precipitator
-------
t=
\
Jl
p
_j
1
4-~
\
\ 1 —
/
J
r~
«e»
R
.
/
^eee ^ ~
"i?1 . . .1.
•
«H
©^(^~
; © /»-''• (3p"! ®"'
\ \ / r i / © f •
r5 ~ \ ' — n r, i"
S-
1
a
3,
5
n
>
5
5
2
— j
i
j
i
\
'!\
i
1
1
- -^j
w
1 ;-
*i
CO
1— il 1
v
r--L
,5
r^J
^^J
® f-rft.
O
o i
O
1" -J U 1
k
liiirs©
uj
Figure 1.3
til
^ /n 11- iy
/ /
' § "
a
w
n HI
'1 1!!! ' ' II It '
Q
- f-
0
rik A ^ A
r\ V^
| Li J ; U J
a -. & -J
1 -'. 2> ^ '
,"^-^_ \ '• • -^ .-* j«
0 -. ! i i^:.- :-C
-. J' - ! , '-M_f«i ---
•^_^J ij % ' .-5
n • ,n r~i n n j' ! P«»i-
l| ' ' •"'•
8^p=j a , !: - :
3 • , <-;?1 ,1
- - — ' • • 1 1 , ._!, _ , . . .
\ .^ses
@
Mechanical Configuration of the Pilot Precipitator
-------
-8-
The gas flow characteristics of the pilot unit were deter-
mined from measurements made with a pitot tube. The necessary
flow control devices were installed, to provide a nearly
uniform flow; see Figure 1.4. A velocity contour map of gas
flow conditions is shown in Figure 1.5. The final flow system
provides a flow with a generally smooth shape and a velocity
standard deviation of about 20%.
The dust is collected on the collection electrodes, rapped
free from the plates, and retained in the eight collector jars
below the precipitator. The collection hopper is divided into
eight zones so that the characteristics of the collected dust can
be determined as a function of position through the collector.
2. Electrical design
The electrode system shown in Figure 1.6 consists of corona
wires suspended between collection plates spaced 10 in. apart.
The corona wires are suspended from Teflon insulators with drill
chucks (1504) utilized as the wire restraining device. This pro-
vides flexibility in the type of corona wire to be used. Plexiglas
guard frames cover the corona lead wires for personnel protection
(901).
The collection electrode system is divided into four mechan-
ically independent plates with dimensions of 3 ft. These elec-
trodes are suspended from spring-loaded mounting pins (1409) so
that plate rapping can be easily accomplished.
Electrical power is supplied to the electrode system from
three independent high voltage power supplies. These supplies are
capable of providing half-wave filtered or unfiltered and full-
wave filtered or unfiltered, positive or negative polarity elec-
trical power to the electrode system. The peak voltage levels are
continuously variable from 0 to 100 kV. This provides a suffi-
ciently flexible power supply for most conditions.
The current and voltage waveforms can be displayed on an
oscilloscope for routine monitoring. Meters are provided on each
power supply system.
-------
-9-
11
I
36
1
2-1/16-+
3-1/8
4-1/4.
5-1/2-
6-7/8.
8-3/8
10
12
10
13
18
19
20
21
22
Flow Control
Plates
3 6
Kicker Plates
Figure 1.4
Arrangement of Flow Control Plates and
Kicker Plates
SOUTHERN RESEARCH INSTITUTE
-------
-10-
Figure 1.5. Velocity Contours at 4.94 ft/sec
Average Velocity
-------
!<«•=•. •=•=_•=. =«©=.-== = .=•<
) " - •• -i'
7] : * *|. x ,
T
r~ '
?
" — -.*»- -* • - -;}
• • .
e
8
m
I
n
H
Electrode
Figure 1.6. Details of the Collection System
-------
-12-
D. Design Parameters
The design parameters that are pertinent to the operation of
the pilot precipitator are given in Table 1.1.
Table 1.1. Design Parameters
Collection electrode area, ft2 72.0
Inlet cross-sectional area, ft2 2.5
Collection electrode spacing, in. 10.0
Wire spacing, in. 9.0
Corona wire length, ft 48.0
Corona wire size, in. 0.10
Gas velocity range, ft/sec 2-20
Volume flow rate range, ft'/sec 5-50
-------
-13-
SECTION 2. FUNDAMENTAL STUDIES
A review of electrostatic precipitator technology was
undertaken to resolve some of the questions regarding the
factors influencing collection and reentrainment. Further clar-
ification of these factors is desired to permit a more thorough
understanding of the principles and to form the basis of a
better precipitatcr model. These theoretical studies include:
an analysis of the change in precipitation
rate parameter with increasing gas
velocity,
a study of the particle concentration distri-
bution in the interelectrode space,
an investigation of the influence of resis-
tivity of the dust layer on sparkover and back
corona conditions, and
verification of the theoretical concept that
migration velocity was dependent upon particle
charge and field.
These studies were conducted with the small-scale precipitator
described in the previous section.
The collection of a monodisperse dust in a precipitator is
described by the Deutsch-Anderson equation
1 - exp -
2.1
The derivation of the equation is based or the assumption that
there is no reentrainment, that the particle concentration at any
cross section is uniform, that the particles are spherical
and monodisperse, and that particle charging occurred instantan-
eously. White1 derived a similar expression for collection
efficiency based upon the probability that a particle was
*See List of Symbols for definitions, Appendix 1.
1 Refer to the Bibliography at the end of this section.
SOUTHERN RESEARCH INSTITUTE
-------
-14-
captured. The derivation by White assumes that particles
entering the boundary zone defined by the gas flow conditions
will be captured. If turbulent mixing and diffusion are
sufficient to provide uniform mixing of the particles, the
equation defining collection efficiency is the conventional
Deutsch-Anderson equation. If, however, there is an uneven
distribution of particles within a given cross section, the
probability of a particle entering the boundary zone would be
modified by the ratio of the particle concentration in the
boundary zone to the average concentration in the cross
section.
The collection concept can be illustrated by considering
the boundary zone to be defined by the laminar flow region in
the vicinity of the collection plate. For a given electrical
migration velocity, w, the length of plate required for collec-
tion will be that defined by the vector sum of the migration
velocity and the particle velocity due to gas flow. As gas velocity
is increased, the distance required for collection of all of the
particles in the boundary zone will increase as illustrated in
Figure 2.1. Conversely, if the migration velocity is increased by
increasing the particle charge or the field, the component of
velocity toward the plate would increase and collection of the
particles within the zone would occur in a shorter distance or a
shorter time.
The thickness of the boundary layer is dependent upon gas
velocity. Increasing the gas velocity in the turbulent region
decreases the thickness of the laminar flow region, and hence
reduces the volume of material that would be enclosed by the
boundary layer as well as increasing the time required for collec-
tion. Consequently, by the Deutsch-Anderson equation alone, one
would predict a decrease in collection efficiency due to an
increased gas velocity. Written in another form,
Thus as the gas velocity increases, there would be a corresponding
decrease in efficiency. (AC = precipitator cross-sectional area;
v = gas velocity.)
-------
-15-
.Collection
Plate
'91
'92
Figure 2.1.
Variation in Length Required for
Collection for Two Gas Velocities
with a Constant Migration Velocity
(Laminar Flow Case)
SOUTHERN RESEARCH INSTITUTE
-------
-16-
The Deutsch-Anderson equation as originally derived
describes the behavior of a monodisperse aerosol in an electro-
static precipitator. For this condition, the parameter w is
the terminal velocity of a charged dust particle under the
influence of an electric field where this motion is opposed by
the viscous drag force of the gas stream. In this report the
parameter w is written with a subscript i denoting that this
is the terminal velocity of the ith particle when charged to
saturation as defined by equation 2.3 below. For this condition
the term migration velocity applies.
In reality, the particles are not instantaneously charged
to the saturation value of charge because of the non-zero
charging time that exists in a real precipitator system. For
this case, the collection efficiency for a monodisperse particle
will be somewhat less than would occur for instantaneously
charged particles. Thus, if one computes the parameter w from
the Deutsch-Anderson equation from the measured performance (or
integrated average of the computed performance) of an operating
installation, an "effective" migration will be determined that is
less than the migration velocity (wi). This value of w is denoted
as we, the effective migration velocity of the particle.
A third definition has evolved in the use of the Deutsch-
Anderson equation in the precipitator technology community. It
is used as a measure of performance of an installation collecting
a polydisperse aerosol. For this special (and widely used) case,
a value of the parameter w is computed from the measured
performance of a real installation. If the efficiency and gas
volume flow rate of an electrostatic precipitator is measured and
the collection electrode area is known, the parameter w may be
computed from the Deutsch-Anderson equation. The parameter w is
now referred to as the precipitation rate parameter and is denoted
wp. when used in this manner, the value of WE is strictly a
measure of performance, with the dimensions or velocity, that
only describes the behavior of a polydisperse dust in this
particular electrostatic precipitator.
Several investigators have observed that for a precipitator
collecting a polydisperse aerosol the change in efficiency with
gas velocity was less than would be predicted from the Deutsch-
Anderson equation. Mathematically, this would appear as though
the precipitation rate parameter, wp, increased with increasing gas
velocity. Since w is defined by Detusch in terms of particle
-------
-17-
charge and electric field, and
2.3
it would be more appropriate to investigate other causes for the
phenomenon.
Several authors have addressed themselves to this apparent
anomaly in the behavior of precipitators where the theoretical
considerations would predict a constant value of the parameter
w independent of velocity, whereas in practice, it increased.
Cooperman2 attributed this phenomenon to a diffusional transport
mechanism in the direction of the gas flow. The particulate con-
centration gradient in the direction of gas flow was thought to
give rise to diffusional transport for the particulate that caused
the mass transport to exceed the gas flow velocity. As the gas
velocity increased, the concentration gradient in the direction
of the flow was theorized to decrease and become less significant
for higher gas velocities.
Robinson3 relates this velocity dependent precipitation
rate parameter (wp) to an assumption that the inlet dust contains a
nonprecipitable reentraining fraction. Heinrichk suggests that
the increased turbulence transports a larger percentage of the
dust particles near the corona wire in a higher field region where
a larger charge is applied. Williams and Jackson5 consider dif-
fusion transport augmented by electrostatic particle convection
as the mechaism for this behavior.
A. Particle Concentration Distribution
A set of experiments was conducted to determine the varia-
tion in the particle concentration distribution in the cross section
at the outlet of the pilot precipitator. The purpose was to deter-
mine if there was a depletion of particles near the collection
electrode to account for the apparent increase in the precipitation
rate parameter (wp) with an increase in gas velocity. The particle
concentration distribution was determined by the use of a laser
light source with a photodetector to serve as a sensor. This
device is referred to as an obscurometer and is described later.
The precipitator was operated with ga>s velocities that covered the
range from 2.5 to 11 ft/sec. Dust was fed into the precipitator at a
fixed rate so that a uniform dust load as a function of time was
presented to the precipitator inlet. The results of this series
of tests are shown in Figure 2.2. The data are expressed as a
percentage of light lost in traversing a 36 in. path.
SOUTHERN RESEARCH INSTITUTE
-------
-18-
°"11 ft/sec
- 7 ft/sec
A-4.4 ft/sec
D-2.5 ft/sec
3 sections operating
at 300 yA/section
(pilot precipitator)
4 -
Figure 2.2,
2345
Distance from Hall, inches
Obscuration as a Function of Distance from
the Collecting Electrode for Various Gas
Flow Ratios
-------
-19-
It is to be noted that the light obscuration is not
linearly related to the mass concentration of the particulate.
The He-Ne laser provides a light source with a wavelength of
about 0.6 y which scatters better from the small particles, thus
the light output is more nearly related to the fine particle
fraction concentration than to the total mass. However, since
the larger particles are preferentially collected, this should
not cause too great an error in the interpretation of results
for the particular experiment just described.
1. Experimental procedure and apparatus
A sketch of the obscurometer, showing two views, is illus-
trated in Figure 2.3. In order to achieve good spatial resolution,
particularly at very short distances from the collecting plates,
a laser was used as a light source rather than a conventional col-
limated "white light" source. With the laser utilized in these
experiments, the effective beam diameter at the point of emergence
from the precipitator was about 0.08 in. The use of a lead screw,
driving the entire obscurometer assembly, allowed the beam to be
scanned across the precipitator in steps of slightly more than one
beam diameter with good positional repeatability.
The extinction of light produced by small particles is a
function of the number and size distribution of the particles with-
in the light beam. As a consequence, if the opacity of an aerosol
varies from point to point in space, either the mass concentration
or size distribution, or both, must be nonuniform. If the size
distribution of the particles is constant, a mass of aerosol, M,
within the beam causes a fractional intensity loss, f, given by
the expression
f = 1 - e-*M = 1 - e-kdm 2.4
where
d = the path length, and
m = the mass concentration.
The value of k, the extinction coefficient per unit mass, is deter-
mined by the wavelength distribution of the light and the size
distribution of the aerosol.
The obscuration measurements were made at 1/13 in. intervals
within 1 in. of the precipitator plates and at 1 in. intervals
over the remaining 8 in. of the duct width (10 in. total span from
plate to plate). The dust feed rate was maintained as nearly con-
stant as possible; however, it did show significant temporal
SOUTHERN RESEARCH INSTITUTE
-------
Beam Splitter &
Detector Assembly
Detector
Assembly
rh
Laser
S i
I
I
i
I
|r v«'
to
Laser Beam
Figure 2.3. Laser Extinction Probe Assembly
-------
-21-
variations. In order to minimize the effect of these variations
and to average out temporal concentration variations resulting
from turbulence, from three to five scans across the duct were
made and averaged for each test condition. The sampling time at
each point on a scan was about 15 seconds.
Because the size distribution of the aerosol at the outlet
of the precipitator was unknown/ and in fact probably was variable/
depending on the exact conditions for a given test (flow velocity,
current density/ temperature, humidity, source of inlet dust
sample, etc.), the results are reported only in terms of the obscu-
ration produced by the aerosol.
The size distribution of the dust at the outlet was not
determined and variations due to gas velocity, humidity, etc.,
probably account for some scatter and uncertainty in the data.
However, since conditions were maintained reasonably constant,
obscuration changes are thought to be due primarily to mass
concentration changes for the fine particle fraction of the dust.
The inlet dust was similar to the MMD of 10 in Figure 2.4.
In all cases of measurements made with the precipitator
operating with the dust feed on, the obscuration was found to
increase by about a factor of 2 within the last one-half inch-to-
inch of the wall of the duct, and to increase slightly from the
center to within one inch of the wall. The shape of the opacity
curve is fairly well represented by an equation of the form
f = ae-b/x 2.4
where a and b depend on operating conditions and x is the distance
from the precipitator plate to the beam center. If the size dis-
tribution were constant over the width of the duct, an equation of
this form would produce a very good fit for the resulting mass
concentration profile across the duct. It is not impossible,
however, that the mass concentration profiles would actually
increase more rapidly from the center of the duct than would be
inferred from the obscuration curves assuming a constant size dis-
tribution. This would result from the action of the precipitator
which probably tends to preferentially move larger particles from
the center of the duct toward the collection plates. As can be
seen from the curves in Figure 2.4, a much higher mass concentra-
tion of large particles is required to produce a given amount of
obscuration than is required for small particles.
SOUTHERN RESEARCH INSTITUTE
-------
100
5
•S
§ 10
§ i
-22-
NMD - 10
a - 5
I L
I
I
.001 2 4 6 8.01 2 4 6 8 .1 2 4681.02 4 6 8 10
Diut Load, gr/ft*
Figure 2.4 Computed Obscuration of White Light by a Varying Dust
Load for Two Particle Size Distributions.
-------
-23-
Increasing the flow velocity results in an opacity increase
that varies approximately as (v)+3' . It is not unlikely that the
mass concentration might vary more nearly as a linear function of
v because of a probable shift toward a larger mean particle
diameter as the precipitator efficiency drops. If the foregoing
conjecture is the case, the overall inefficiency would vary as v2
after accounting for the inlet dust loading variations induced by
the velocity changes (inlet concentration is proportional to 1/v
for a constant dust feed rate).
At velocities of 3.5 and 8 ft/sec, measurements were obtained
at several corona currents ranging from 50 to 300 viA per section.
Typical obscuration profiles for these tests are shown in Figure
2.5. Increasing collection currents leads to a flattening of the
central parts of the profile, a narrowing of the high opacity
region near the boundary, and a steepening of the profile near the
boundary.
Following most of the foregoing tests, scans were made with
the dust feed shut off, but with the collection plates unrapped.
In all of the latter scans, no measurable obscuration was detected
at any point across the duct, which would tend to indicate the
absence of any direct erosion or scouring effects resulting solely
from the air stream.
2. Conclusions
Results of the particle concentration measurements failed to
explain the reasons for the increase in precipitation rate parameter
in terms of increased turbulent mixing. The increase in particle
concentration near the collection electrode would result in a modi-
fication of the precipitation rate because of the increased concen-
tration of particles near the collection surface. However, the
ratio of particle concentration near the collection surface to
that in the interelectrode space is in the opposite direction to
account for the increase in precipitation rate parameter with gas
velocity.
B. Particle Size Effects
Further exploration of the cause of the increased precipitation
rate parameter with gas velocity was made using the SRI model precipi-
tator and the mathematical model. It is apparent from theoretical
considerations that precipitation rate parameter varies over a wide
range for various particle sizes. For large sizes, the migration
velocity is so large that collection is virtually 100% for particles
greater than around 10 y. Thus, an increase in gas velocity
would not materially reduce the collection of the large particles.
Consequently, the increase in precipitation rate parameter with gas
velocity is predictable from theory when considering a polydisperse
aerosol.
SOUTHERN RESEARCH INSTITUTE
-------
-24-
O - 3 sections at 100 yA/Beation
O - 3 sections at 200 vA/Beotion
O - 3 sections at 300 yA/saotion
A - 3 sections at 450 yA/seotion
Figure 2.5,
2 3
Oiatanoe from Hall, inches
Effect of Current Density Changes on
Obscuration at a Flow Velocity of 3.5 ft/sec
-------
-25-
This apparent anomaly is caused by the fact that the
Deutsch-Anderson equation was derived for a single particle
size with a constant migration velocity, while the precipita-
tion rate parameter is obtained by measuring the collection
efficiency of an installation and back-calculating.
If we are dealing with a monodisperse particle distribution
where all particles have the same migration velocity, the
Deutsch-Anderson equation effectively describes the behavior of
the precipitator for various gas velocities. It is to be noted
that in all the discussions of efficiency, the basis is for a
weight or mass collecting efficiency. The behavior of a preci-
pitator operating on a material with a range of particle sizes
becomes clear if one considers the behavior of that device
while collecting a material consisting of a number of mixed mono-
disperse particle groups. In this case, each particle size will
be characterized by a mass concentration and its associated migra-
tion velocity.
The behavior of the precipitator for the dust composite can
be determined by applying the Deutsch-Anderson equation to each
individual particle size to determine the percentage of the weight
of material removed within each size increment. The overall effect
is then determined by relating the total weight of material col-
lected to the total inlet material. Only now are we able to com-
pute a value for the precipitation rate parameter.
The precipitation rate parameter for this same dust but for
an increased gas velocity can be determined by completing the
entire computation cycle for each particle size, again summing the
results, and finally computing a new precipitation rate parameter.
This yields a value different from that obtained by modifying only
the volume flow rate for the composite system equation. This fact
is illustrated by the data shown in Table 2.1.
SOUTHERN RESEARCH INSTITUTE
-------
Table 2.1.
Calculated Collection Efficiency for
a Dust Composite, Based on a Charging
Field of 5 kv/cm and a Collecting
Field of 3 kV/cm
Particle Material Migration
size weight velocity
ym % cm/sec
>32.5
32.5
29.0
21.8
12.4
7.9
4.2
2.1
1.3
15
4
8
22
11
21
13
3
3
365
238
212
159
90.5
57.6
30.7
15.3
9.5
Overall efficiency:
Area-to-volume ratio, ft2/kcfm:
Precipitation rate parameter:
Fractional efficiency each size range for
indicated velocity
ft/sec
4.4
100
100
100
100
100
99.76
96
79.8
63
97.76
53
36
5.4
100
100
100
100
99.95
99.3
92.8
72.7
55.6
96.9
43
40.8
6.7
100
100
100
100
99.8
98.1
88.0
65.0
48.0
95.5
35
45.7
8
100
100
100
99.99
99.46
96.4
83.0
58.4
42.3
94.0
29.3
48.8
10
100
100
100
99.93
98.45
93.0
75.8
50.6
35.6
91.81
23.5
54.3
12
100
100
100
99.8
96.9
89.0
69.4
44.3
30.6
89.45
19.5
58.6
a\
-------
-27-
A plot of this variation in precipitation rate parameter
as a function of gas velocity is shown in Figure 2.6. Thus we
see a variation in precipitation rate parameter with gas velocity
based on purely theoretical considerations for polydisperse
dusts, while the collection efficiency for each size interval
is based on the Deutsch-Anderson equation with a fixed migration
velocity and an area-to-volume ratio depending upon the gas
velocity.
The computer performance model described in Section 4 also
predicts this behavior. The precipitator operating parameters for
the SRI pilot plant were used as inputs to the computer mathematical
model which was used to compute the charge, effective migration
velocity, and collection efficiency for each size range at each
position, and finally, an overall efficiency for each of a number
of test conditions. These computed efficiencies were used to
determine a precipitation rate parameter for each gas velocity.
These data are shown in Figure 2.7.
White1 reports similar results when collecting fine oil
smokes on two types of collection electrodes. Thus, the increase
in migration velocity with gas velocity is not an anomalous
behavior based on diffusion phenomena or electric wind, but
rather is predictable from purely theoretical considerations
due to the large variation in migration velocity for a wide
range of particle sizes. The primary misunderstanding comes
from a misuse of the Deutsch-Anderson equation in a manner that
violates the initial assumptions.
C. Resistivity
The influence of resistivity on the performance of electro-
static precipitatiors has been rather clearly established. White,1
Penney, and others have shown that resistivity of the collected
dust determines the voltage and current at which a precipitator
will operate due to sparking and back corona limitations.
A review of the methods of measuring resistivity and com-
parison of the resistivity data measured by the various methods
indicated that there were fundamental differences in the measure-
ment methods and probably wide variation in the resistivity values
obtained by the different techniques.
To resolve some of the questions regarding the measurement
of resistivity, a review was made of the methods used or proposed
for resistivity measurement and comparisons were made utilizing
several types of resistivity probes.
1. Types of apparatus
Dust resistivity can be measured in the flue-gas environment
(in situ) or the dust sample can be measured in an environment
SOUTHERN RESEARCH INSTITUTE
-------
-28-
0)
4J
M
Id
n
I8
43
Id
4*
•H
04
•H
I)
8
PI
60
50
40
30
20
10
I
2 4 6 8 10 12 14
Gas Velocity, ft/sec
Figure 2.6. Graph of Computed Precipitation Rate
Parameter for Example Showing Increase
in Wp with Increasing Gas Velocity.
-------
-29-
-------
-30-
simulating flue-gas conditions (laboratory). Because of the
difficulty in reproducing flue-gas conditions in the laboratory,
resistivities measured other than in the flue gas are of ques-
tionable value in the temperature region where surface conduction
predominates. At elevated temperatures (above around 450°F),
laboratory resistivity data are in substantial agreement with
those determined in situ.
Measurement of resistivity involves first a collection of
a sample of the dust followed by determination of the current and
voltage relationships within the dust layer itself. The various
methods of measuring resistivity differ primarily in the method
of dust collection, the method of applying voltage to the dust
layer, and the configuration of the measuring cell.
During the course of this investigation, four types of
resistivity measuring apparatus were studied. The first system
consisted of a mechanical cyclone collector with a cylindrical cell
electrode structure, as shown in Figure 2.8. This apparatus is of
the type described by Cohen and Dickinson7 and consists of a probe
that is inserted into the duct to withdraw a sample of the dust-
laden flue gas. The cyclone collector removes the dust which falls
into the hopper beneath the cyclone. The hopper consists of
metallic central and concentric electrodes supported by a Teflon
cup which constitute the measuring cell. When the cell is filled
with dust, a megohmeter is used to determine the dust resistivity.
Temperature of the cyclone and measuring cell is maintained
at flue-gas conditions by means of heaters located inside the
enclosure containing the cyclone. A mechanical vibrator is used
to compact the dust, thereby giving reproducible packing conditions.
Use of this probe was limited somewhat by the difficulty in
maintaining isothermal conditions throughout the system. It was
necessary to wrap the sampling probe with heater tape to insure
that the gas was not cooled in passing to the cyclone and to
measure the temperature of the cyclone and measuring cell with
thermocouples rather than depend upon thermometers in the enclo-
sure. With these modifications, the apparatus performs satis-
factorily.
To minimize the difficulties associated with maintaining
isothermal conditions, a second type of cyclone probe was con-
structed to be inserted directly into the duct. A thermocouple
on the cyclone was used to determine when the cyclone came to the
flue gas temperature. When stable temperature was reached, a
sample was taken by the cyclone collector. The collector was
-------
-31-
To Pump
I1
Th«
Insulated Sample
_• \ line
If:-:. ",-''..-.i;.':-v.;::.-.',:-..-;vl- •••»••
^^:••:'i•l^>v•-.•'•.:l,•.:^•.'*^i•>;^:•^
Heater''
jrmc
• '
:• »
- • f
. . •
* *
i
>meter
X
I-1. •;
f
.
<>
2 i
j
-^
I
< • ' * - *
i
\
(
/
t
i=
^
.../.... \ \.
— i
i
^
/ ^
I
|L 1
h «
1 •. .' • • ,
,*
. ' • *
• „ *
• •/:
•» .
i
••"
•t (
";V
*, .^
•. •'
• >• .
•-'
..—•**Power
— — 10 Mett
\
Insulated Box
Figure 2.8. Cyclone Collector - Cylindrical Electrode Cell
for Collection External to the Duct
SOUTHERN RESEARCH INSTITUTE
-------
-32-
vibrated or rapped to compact rhe dust sample in the measuring
cell. Resistivity was then measured by applying a voltage to
the electrodes and measuring the current by means of a picoam-
meter and current recorder.
Recording of the current through the resistivity cell
showed that the current decreased with time (increasing resis-
tance) for a period of around 20 min for fly ash samples. There-
fore, the question arose as to which value of resistivity was
most representative the initial value or that obtained when
steady-state conditions are reached. This phenomenon of current
decay will be discussed later.
The third type of probe used was the point-plane type,
shown in Figure 2.9. The apparatus consists of a disc and guard
rings as one electrode, and a point electrode located approxi-
mately 3.2 cm above the disc. When high voltage is applied across
the electrodes, a corona is generated at the point and dust is
electrostatically deposited on the disc. When a sufficient layer
has been deposited, a second disc, concentric with the point
electrode, is lowered onto the dust surface and the resistivity
determined from measurements of voltage across the discs and
current through the sample. Resistivity was then computed from
the sample resistance, cell dimensions, and dust thickness. The
apparatus shown in Figure 2.9 was designed so that constant
spring pressure was applied to the dust layer and the thickness
was measured by means of a dial gauge.
In determining resistivity by the point-plane apparatus, it
was observed that the current-voltage relationships were nonlinear.
Apparent resistivity decreased as the voltage or electric field
increased. The significance of this variation will be discussed
later.
The point-plane resistivity apparatus permits an alternative
method of measuring resistivity by determining the voltage-current
curves with and without a deposited dust layer. This alternative
method can serve as a check on the measured resistivity and lends
considerable confidence to the resistivity values determined by
the point-plane apparatus. However, for very low values of
resistivity, the voltage drop across the dust layer is small and
the accuracy of this method is not as great as it is for higher
resistivities. For high resistivities (1011 to 1012 ft-cm) , back
corona begins at very low values of current. The resistivity must
be determined from current-voltage relationships below the onset
of back corona. For intermediate values of resistivity, both tech-
niques provide comparable values for resistivity.
-------
-33-
~rr
Figure 2.9. Schematic of Point-
Plane Resisvitity Probe
SOUTHERN RESEARCH INSTITUTE
-------
-34-
A fourth type of resistivity apparatus utilized was the
Electrostatic Precipitator Analyzer (Kevatron) which utilizes a
cylindrical wire and tube precipitator for collection of the
dust and a cylindrical cell electrode for determining resistivity.
Following collection, the dust is rapped from the walls of the
collector and falls into the measuring cell where the resistivity
is determined from the resistance of a known geometry of the
material. Figure 2.10 is a schematic of the system.
A comparison of the various sytems shows that they differ
principally in the manner in which the dust is collected and
deposited in the cell, in the size and geometry of the measuring
cell, and in the manner in which the apparatus is operated.
Since the collection mechanisms are fundamentally different,
one question that arises is the particle size distribution of the
dust. Cyclone collectors of the type used in the Cohen-Dickinson
apparatus, or variations of this apparatus, would not be expected
to collect very small particles. This, coupled with the extreme
nonisokinetic sampling, would tend to introduce considerable un-
certainty as to how representative the collected sample is apt to
be.
The point-plane apparatus would tend to suffer from the
same difficulties, as would the Kevatron apparatus to some extent.
Electrostatic precipitators are in themselves size selective so
that larger particles would tend to be preferentially collected,
thus preventing the taking of a representative sample.
On the other hand, the dust layer on the precipitator is
not of the same particle size distribution as the dust in the gas
stream. In terms of particle size distribution, therefore, there
is considerable question as to how to extract a sample that would
be representative of the dust layer on the precipitator plate,
especially since this varies along the precipitator length.
An alternative approach, therefore, is to attempt to get a
reproducible sample either by maintaining constant collection
conditions on a nonisokinetic system or by attempting to maintain
isokinetic conditions and providing a means for collecting all the
dust above the minimum size fraction of interest.
Figure 2.11 shows the size distribution of the dusts col-
lected by the cyclone collector, the wire and tube precipitator,
and the dust from the precipitator hopper.
-------
Collecting
Electrode
Gas Outlet
Measuring
Electrode
-35-
Spring
dc Power
Supply
Insulator
/
Tension
Gas Inlet
Corona
Wire
Insulator
Central
Electrode
1
J
r
Recorder
do Power
Supply
Figure 2.10. Kevatron Australian Resistivity Probe
SOUTHERN RESEARCH INSTITUTE
-------
-36-
100
80
60
40
20
o
N
•H
10
o
•H
4*
10
8
6
4
3
I I
1 2
I I I
ii i i r
Precipitator
Hopper Dust
I I I I
''/I
/ SRI Cyclone
I I I
I I
10 20 30 40 50 60 70 80 90
Percent Less Than Size Indicated
95 98 99
Figure 2.11. Comparison of Particle Size Distributions
from Precipitator Hopper and Two Resis-
tivity Probes
-------
-37-
The quantitative influence of particle size of the dust on
its resistivity was not determined. The effect may vary with
the nature of the dust and type of conditioning. In general, we
have observed a larger percentage of sulfate on the smaller frac-
tion of fly ash than on the larger. However, size of the dust does
not appear to be a major factor. Figure 2.12 shows the variation
in resistivity of various size fractions of a fly ash as deter-
mined by Bahco as reported by Bucher.8
A second difference in the various types of resistivity
probes is the degree of compaction or density of the dust in the
cell. Cohen and Dickinson7 report variations in resistivity by an
order of magnitude with changes in dust density from around 0.5
to 1.0 gm/cm . With fly ash, our experience indicates that varia-
tions of the order of 2 may be more realistic. The degree of com-
paction does vary considerably between the point-plane type appa-
ratus and both the Kevatron and cyclone systems. The dust layer is
compacted by the measuring disc in the point-plane probe, whereas
the dust is rapped or vibrated into the cells in the Kevatron and
cyclone type collectors.
The method of deposition of the dust layer is also different
between the point-plane, cyclone, and Kevatron apparatus. In the point-
plane apparatus, dust is electrostatically deposited on the sur-
face of the disc and, perhaps, some alignment of the dust particles
does occur. Such alignment does not take place in the Kevatron and
cyclone type apparatus other than from localized dust electric fields.
All of the effects mentioned above are somewhat secondary to
the more pronounced influence of electric field and the variations
in resistivity with time as mentioned earlier. Figure 2.13 shows
a typical change in resistivity with time for fly ash as deter-
mined by the cyclone collector and cylindrical-cell apparatus.
Such time dependence of resistivity has been noted by McLean9 on
borosilicate glass and on fly ash at temperatures up to 225°C.
As discussed by White1 and others, resistivity of most
granular materials in the presence of moisture and/or other con-
ditioning materials varies with temperature in a manner indicating
two modes of conduction. At temperatures below around 350°F, con-
duction takes place primarily through an adsorbed layer on the sur-
face of the particles. Above around 400°F, conduction takes place
through the material itself. In the intermediate region between
350 and 400°F, both surface and bulk conduction are significant.
The equivalent electrical circuit would be a parallel resistance,
the value of each being temperature dependent. The mechanisms of
current conduction in each of the two modes of conductivity are
significant in understanding the observed behavior of resistivity
measuring apparatus.
SOUTHERN RESEARCH INSTITUTE
-------
-38-
10
10
10
12
1 1
1 0
c:
4*
•H
10:
10'
10s
I
I
I
I
100 200 300 400 500
Temperature, °F
600
Figure 2.12. Resistivity vs. Temperature for Different
Size Fractions of Beulah Standard Electro-
static Precipitator Ash
-------
-39-
10
15 20
Time, min
25
30
35
Figure 2.13. Variation of Fly Ash Resistivity with Time Using the
In-Situ Cyclone Probe at a Temperature of =280°F
SOUTHERN RESEARCH INSTITUTE
-------
-40-
2. Surface conduction
The results of studies under EPA Contract CPA 70-149 show
that conduction on the surface of a fly ash particle is due to
an adsorbed layer of water and sulfuric acid. Conduction through
this adsorbed layer roust be primarily electrolytic, the charge
carriers being mainly H+, OH~, HSO^~,and SO,,""2 ions.
In accordance with Faraday's laws, conduction through the
layer must be accompanied by charge-transfer processes at inter-
faces where the fly ash adjoins other materials containing charge
carriers of a different type. Such interfaces exist in cells
used for resistivity measurements, where the metal electrodes
conduct electricity through the migration of electrons. They
also exist in deposited layers of fly ash on the electrodes of
an electrostatic precipitator, where the electrodes again conduct
electricity through the migration of electrons, and the adjacent
gas stream conducts primarily through the migration of charged
molecules, such as 02~. Moreover, in accordance with observa-
tions that different ions in solution vary in electric mobility
and thus in transference number (the fraction of the current
transported), variations in the electric mobilities of different
ions in the adsorbed layer on fly ash must be expected. Such
variations are to be expected especially as the result of dif-
ferences in attractive forces between different ions and the sur-
face components of the fly ash.
In an adsorbed layer of only H20 molecules, the charge
carriers in the ash must consist of only H+ and OH~ ions. If
electrodes of electrochemically inert metals are used to determine
the resistivity of the ash, the charge-transfer processes at the
electrode-fly ash interfaces will lead to the formation of H2 gas
at the negative electrode as the result of the reduction of H+
ions, and the formation of 02 gas at the positive electrode as
the result of the oxidation of OH" ions. Alternatively, if the
electrodes are not electrochemically inert, another process may
occur at either electrode; for example, a metal-oxide coating
may be reduced to the metal at the negative electrode, or the
metal may be oxidized to produce either metal ions or the metal
oxide at the positive electrode. For purposes of illustration,
however, it may be hypothesized that only H2 and 02 are produced
by the electrode reactions, and consideration may then be given to
the changes in the composition of the adsorbed layer that occur
concurrently as a result of the electric mobilities of the H+ and
OH~ ions.
-------
-41-
If, as in liquid water, the electric mobilities of both H+
and OH~ ions in an adsorbed layer of H20 molecules have signi-
ficant values, the quantity of water on the fly ash near each
electrode will gradually decrease, as a result of the fact that
neither icn can be supplied at the electrode at the rate that
reduction or oxidation occurs. However, if the fly ash has
acidic surface components such as Si02 that react specifically
with the OH" ion to form an immobile ion such as Si02(OH)~, the
mobility ratio of H+ to OH~ ion will be increased and the rate
of depletion of the amount of water at the positive electrode
will exceed that at the negative electrode. Alternatively, if
the fly ash has basic surface components such as A12O3 that have
a specific attractive force for H* ion, the mobility ratio of H+
to OH~ ion will be lowered, perhaps to the degree that a deple-
tion of the amount of water will occur only at the negative elec-
trode rather than the positive electrode. In summary, the com-
bination of two phenomena described gas formation at the elec-
trodes and water depletion on the fly ash at one electrode or
both electrodes will increase the resistance through the ash
and lead gradually to an apparent increase in the resistivity of
the ash as the time of measurement and the amount of electricity
conducted through the ash increase.
If the adsorbed layer contains H2SO,,, the formation of Hz
and 02 gas molecules must again be expected on inert metal elec-
trodes. Furthermore, despite the near certainty that H+ ion will
have a much higher mobility than any ion of opposite charge that
may be present (HSO,," or SO,,'2) , the combination of electrode
reactions and the transference of the ions must lead to an increase
in the concentration of H2S04 at the positive electrode and, more
importantly, a decrease in concentration of H2SOU at the negative
electrode. The net result of the phenomena described is likely
to be a gradual increase in the apparent resistivity of the fly
ash during the time of measurement. Gas formation at both elec-
trodes and H2SOif depletion at the positive electrode should over-
come the opposing effect of H2SOt| enrichment at the negative
electrode.
The foregoing discussion represents an effort to explain
why the observed resistivity of fly ash may increase substantially
with the time of measurement. Now a relatively brief discussion
can be devoted to another experimentally observed phenomenon,
the decrease in fly ash resistivity with increased electric field.
A similar phenomenon, referred to as the Wien effect, occurs in
certain liquid electrolytes aqueous solutions of ionic
compounds that are analogous in composition to the adsorbed
layers on fly ash. A decrease in resistivity of aqueous solutions
SOUTHERN RESEARCH INSTITUTE
-------
-42-
occurs as increasing electric fields are produced with an ac
power supply (which is typically used in studies of aqueous
solutions, in contrast to the dc power supplies normally used
with fly ash). This decrease is attributed to the reduction
in the effects of inter-ionic attractions that limit the rates
at which individual ions of opposite charge migrate in different
directions. The corresponding decrease in the resistivity of
fly ash can be attributed to a similar reduction in ionic inter-
actions within the adsorbed layer, and perhaps, in addition, a
reduction in ionic interactions with immobile surface components
of the fly ash. (This phenomenon is not associated with a
temperature rise as insufficient electrical energy was supplied
for this to be significant).
3. Volume conduction
Variations of resistivity with time and electric field
have also been observed in the absence of surface conduction,
where the temperature is above 450°F. Typical current-time
curves show an initial high value of current decreasing with time
at a rate dependent upon many factors, as described by McLean.
The initial current is called absorption current and can be of
short duration in the case of solid borosilicate glass to minutes
in the case of particles of the same material.
Current conduction in siliceous-oxide glass is generally
accepted to be electrolytic. Although the possibility of some
electronic conduction cannot be eliminated, the nature of the
current-time and current-electric field data can be explained on
the basis of rather conventional electrolytic conduction.
In fly ash, conduction by an ionic mechanism is possible
due to the openness of the glass structure in contrast to that of
crystalline oxides possessing long-range order within the crystal
structure. This openness provides a "path" for the migrating charge
carrier and eliminates the need of energy to create structural
imperfections. Conduction is low in this type of system due to
the low ion mobility.
The charge carrier is usually thought to be a monovalent
cation such as Na+, K+, or Li+. However, the carrier may be some
other electrostatically unbalanced species resulting from, for
example, gases or water vapor dissolved in the glass. The charge
carriers, in general, are loosely attracted to the glass structure
and contribute to the openness. When the glass is subjected to an
electric field, mass migration of the carriers occurs toward
their respective electrodes of opposite polarity and a current is
detected.
-------
-43-
Below about 300°C, using dc measurements, it is noted that
the initial current will decrease with time and eventually reach
some "equilibrium" value. The higher the temperature, the
shorter the time to equilibrium. This phenomenon is called
absorption current. It is believed related to the initial random
distribution of the charge carriers and random network glass
structure. The glass structure does not provide a uniform energy
barrier to the migrating species. Therefore, certain avenues of
migration for the carrier are more energetically favorable than
others. Ultimately the migration attains an "equilibrium" value
when the diffusing species are finally limited in progress by
high energy barriers.
If one is concerned with a finite specimen and an electro-
lytic conduction mechanism, the direct current will gradually
decrease as the concentration of carriers builds up at the elec-
trodes, and will ultimately cease when the migrating ions are
depleted. The effect is called electrode polarization.
The volume electrolytic conductivity increases with
temperature. For conditions under which the current field strength
relationship is linear/ it can be expressed as an Arrhenius equa-
tion, log o ~ 1/T. The conductivity may also be affected by field
strength. In this case, the experimental activation energy is
altered due to distortion of the energy barriers inhibiting the
migration of charge carriers. The net effect is that conduction
is enhanced disproportionately at high field strengths.
From the above brief description of ionic conduction, the
observed current decay with time and the enhancement of conduc-
tivity with increased field strength can be rationalized. It would
appear entirely feasible that charge carrier depletion at the
points of contact in a particulate or granular material, together
with the absorption current phenomena would be adequate to explain
the observed behavior. However, in view of the complex nature of
the fly ash structure and the presence of many impurities, the
possibility of semi-conductor behavior or purely electronic con-
duction cannot be competely discounted.
In a precipitator, current is conducted to the dust sur-
face by ions either carried on the particulate or unattached.
Because of the higher mobility of the gas ions as compared with
particulates, approximately 99% of the current is carried by these
unattached ions. The method by which the ions arriving at the
dust surface are neutralized has not been completely determined
and may vary, depending on whether conduction through the dust
layer is primarily through the adsorbed layer or through dust par-
SOUTHERN RESEARCH INSTITUTE
-------
-44-
ticles themselves. Regardless of the exact nature of the
charge-transfer process, the result is that the arriving ions
are neutralized at the gas-dust interface and the charge trans-
fer through the dust layer takes place in the same manner that
it would in the resistivity cells previously described.
One significant difference, however, is that fresh dust
is being precipitated continuously, thus, the dust layer-gas
interface is constantly changing. The interface between the dust
layer and grounded electrode, however, does not change, and var-
iations in resistivity with time could occur at that interface in
much the same manner as in a resistivity probe. However, deple-
tion of the charge carriers occurs primarily at the cathode in
the case of a sulfuric acid conditioning system, which would mini-
mize the effect.
The electric field in the dust layer of a precipitator
depends upon the current density and the resistivity of the dust
(E = jp). The limiting field is the breakdown conditions of the
interstitial gases, usually around 10 to 20 kV/cm. If dust
resistivity is high (around 1010 to 10ll fl-cm) , the precipitator
would normally be operated near the breakdown field of the dust
layer. However, if the dust resistivity is low (around 107 to
10 * fl-cm), the limiting operating condition of the precipitator
is not breakdown of the dust layer, but rather sparking across
the interelectrode space. Thus, the resistivity seen by the pre-
cipitator would be that corresponding to the lower electric field.
The time dependence of resistivity is perhaps a more sig-
nificant variable. Rapping frequency varies depending upon dust
loading and is in the range of 5 to 10 min. If all of the dust
were removed from the plate during each rap, the average time that
the dust is on the plate would be 2.5 to 5 min. However, in nor-
mal rapping, dust moves vertically down the plate during a rap
and is either recollected several times prior to finally falling
into the hopper or some dust may fall the entire distance in a
single rap before being recollected. Thus, the dust may remain in
the precipitator during several rapping cycles with current flowing
through the dust layer for a period of, say, 10 to 20 min. Thus,
the resistivity in a precipitator could vary with time and perhaps
be a factor. Again, however, fresh dust being deposited would
tend to minimize the effect.
4. Comparison of resistivity measurements by various probes
Since the factors discussed in the previous sections affect
the resistivity as measured by the different cells, a comparison
was made of resistivities at the same plant utilizing several
-------
-45-
resistivity probes. Figures 2.14 and 2.15 are comparison plots
of the resistivity measured by the point-plane probe with that
measured by the cyclone collector, cylindrical cell probe, and
the EPA apparatus which utilizes a wire and pipe precipitator
for collection and a cylindrical cell electrode for resistivity
measurements. Figure 2.16 shows the comparative resistivities
between the steady-state values of the cylone probe and those
measured by the point-plane apparatus at maximum field strength.
In addition to the scatter, the steady-state values from the
cyclone probe are at least one order of magnitude higher than
those measured by the point-plane probe.
Figure 2.14 shows the settled out cyclone data plotted
against the point-plane data using the point-plane data at 2.5
kV/cm, which corresponds to the field in the cyclone apparatus.
A somewhat better agreement between the two methods is apparent
with compensation for the influence of electric field.
Figure 2.15 shows the peak values of resistivity from the
Kevatron and cyclone probes plotted against point-plane data for
the same (2.5 kV/cm) field. In this case, much better agreement is
obtained between the cyclone and point-plane data. The EPA data
are still higher than the average of the cyclone or point-plane
data, although there is statistically insufficient data to draw
firm conclusions regarding the Kevatron values.
The logic of comparing the peak current values of resistivity
from the cyclone with the point-plane data can be rationalized
to some extent by the fact that fresh dust is being deposited on
the surface during the precipitation process. Thus, even though
corona current is flowing during the collection interval, the dust
surface in contact with the upper disc has not been subjected to
a current flow for an appreciable time. The interface between
the dust and lower electrode has had current flowing during the
collection period. However, polarization effects are not likely
at the anode surface.
The agreement between resistivities as measured by the
methods compared is not extremely bad when one considers the
scatter of the data when using a single apparatus. However, some
basis on which to compare the data in a rational manner needs to
be resolved.
In further consideration of the factors that might cause
differences in resistivity values as measured by the various
methods, several observations were made in the laboratory and in
the field regarding the differences in time variation between the
SOUTHERN RESEARCH INSTITUTE
-------
-46-
10"
a
«
3
n
H 10»fl
1
I
g. 10*
10'
10'
Figure 2.14,
oo
D
O
O
o
o
Perfect Comparison
O - Cyclone
O - Kevatron
10" 10* 10*• 10**
Point-Plane Resistivity, Q-cm
10"
10
11
Comparison of Kevatron and Cyclone Resistivities
with Point-Plane Resistivities at an Electric
Field of 2.5 kV/cm. Settled Values for Cyclone,
Peak Values for Kevatron.
-------
-47-
10"
n
•H
n
" 10"
1
10*
10
p o
o oo
Perfect Comparison
O Line
O " Cyclone
O • Kevatron
10§
Point-Plane Resistivity, Q-om
Figure 2.15
Comparison of Kevatron and Cyclone Resistivities
with Point-Plane Resistivities at an Electric
Field of 2.5 kV/cm. Peak Current Values Used
for Cyclone and Kevatron.
SOUTHERN RESEARCH INSTITUTE
-------
-48-
10"
?
a
0
10"
* 10i»
I
fr 10"
n
•H
a
10 •
00
o
o
Perfect Comparison
Line
10
10* 10* 10" 1011 10"
Point-Plane Resistivity, fi-cm
Figure 2.16. Comparison of Cyclone Resistivities with Point-
Plane Resistivites. Maximum Electric Field on
Point Plane. Settled Values for Cyclone.
-------
-49-
two types of cells; parallel disc and cylindrical. In general,
a greater variation in resistivity with time was observed with
the cylindrical cell than with the point-plane apparatus.
A difference in the electrochemical properties of the
electrode materials is one possible explanation for the differ-
ence in the effects of the two electrolytic phenomena in the
cells with cylindrical and disc electrodes. In the cylindrical
cell constructed in this laboratory the electrodes are made from
stainless steel, whereas in the parallel disc cell, the elec-
trodes are made from copper. Gas formation may occur less exten-
sively at copper electrodes than at stainless steel; if so, the
overall effect from electrolytic processes in the parallel disc
cell will be less than that in the cylindrical cell.
A difference in current densities at the electrodes is still
another possible explanation for the difference in the effects of
the electrolytic phenomena in the two cells. At least superfi-
cially, the difference in current densities may appear to be small.
For fly ash with a resistivity of 1 x 1010 n-cm in the cylindrical
cell fabricated in this laboratory with electrode diameters of
0.47 and 1.27 cm and a height of 4.44 cm current densities of
0.54 and 0.20 yA/cm2 will be produced at the two electrodes with a
typical value of 2.5 kv/cm as the average electric field in the
sample. For the same ash in the parallel disc cell with an area
of 5 cm2 on each of the disc electrodes and a typical sample thick-
ness of 0.05 cm a current density of 0.25 yA/cm2 will be pro-
duced at the two electrodes with a typical value of 2.5 kV/cm as
the average electric field in the sample. In view of differences
in sampling packing densities, however, the actual current densi-
ties at the cylindrical electrodes may be substantially different
from those at the disc electrodes. Sample packing between the
cylindrical electrodes occurs only as the result of rapping or
vibration; sample packing between the disc electrodes, on the
other hand, occurs under an applied pressure. The number of con-
tact points per unit of surface area between the fly ash particles
and the cylindrical electrodes is, therefore, likely to be much
smaller than the number of contact points between the particles
and the disc electrodes. Consequently, even with approximately
equal current densities computed for the two types of electrodes
on a superficial basis, the actual current densities and the
intensities of electrolytic effects may be much greater at the
cylindrical electrodes than at the disc electrodes.
SOUTHERN RESEARCH INSTITUTE
-------
-50-
A difference in the degree of isolation of fly ash from
the flue gases following collection in the two types of
resistivity cells may also lead to a difference in the effects
of the electrolytic phenomena in the two types of cells.
Samples collected with cyclone probes and deposited between
cylindrical electrodes must be more completely isolated from
the flue gas than the samples collected with the point-plane
apparatus and located between parallel disc electrodes. If the
ash is collected with a reactive material such as H2SOU on its
surface and then isolated from the flue gases/ changes in surface
composition should occur as a result of chemical reactions,
accentuating the changes associated with electrolytic processes.
We have encountered evidence that fly ash collected in the cyclone
apparatus will contain more S0i»~z than ash collected simulta-
neously in the point-plane apparatus. It is reasonable to expect
that any H2SOi» reacting with fly ash after collection can be
less easily replenished by further adsorption of H2SO4 vapor
from the gas stream in the cyclone apparatus than in the point-
plane apparatus.
These factors need to be resolved in attempting to arrive
at comparable methods of measuring resistivity.
5. Factors influencing fly ash resistivity
The factors previously discussed point out the considera-
tions in making meaningful fly ash resistivity measurements and
comparing the values determined by the various types of appara-
tus. In addition to this, resistivity can change drastically
between fly ash from different types of coals. The factors
responsible for these variations are those that influence both
surface and volume conduction.
In the temperature region where surface conduction pre-
dominates, resistivity changes with temperature and the amount of
conditioning material present in the gas. In the case of fly ash,
the conditioning agent is H2SO,, formed by the water and S03 in
the flue gas. The sulfur content of the coal determines the sul-
fur dioxide content of the flue gas. Under most conditions,
around 1% of the S02 in the flue gas is converted to SO3, and
hence for normal circumstances, the sulfur content is a reasonably
good estimate of the SO3 present in the flue gas and therefore
is a good measure of the resistivity that can be expected at
any given temperature within the range where surface conduc-
tion predominates.
-------
-51-
However, several factors can alter these normal relations.
First/ the percentage of SO2 converted to S03 can change depend-
ing upon the temperature-time conditions, amount of excess air,
and perhaps upon the surface composition of the ash. Since a
large quantity of SO2 is present in the flue gas, even a small
increase in the percentage converted can alter resistivity of the
ash considerably.
The second factor that can alter the normal resistivity-*
sulfur relationship is the basicity of the ash. A highly basic
ash surface will react with the available SOa to form a sulfate
that in itself has high resistivity. Once the sulfate is formed,
a layer of H2SOk can be adsorbed on the surface and this is re-
sponsible for the change in resistivity.
In addition to uncertainties in fly ash resistivity in the
temperature region where surface conduction predominates, varia-
tions also occur in the high temperature range where volume
conduction is primarily responsible for the resistivity. Compo-
sition of fly ash can vary widely between various types of coal.
These variations can influence the resistivity of the ash as a
result of day-to-day variations in composition from the same coal
seam.
Table 2.2 lists average, maximum, and minimum values of the
various compounds in coal ash samples taken from the Colstrip
mine. Out of 21 samples, the CaO content varied from a minimum
of 13% to a maximum of around 30%. Figure 2.17 shows the
variation in resistivity with temperature for two samples of this
ash over a range of temperatures. It is apparent that resistivity
of the ash was above 2 x 1010 ft-cm, even at 600°F for this
particular sample.
Table 2.2 Variation in Chemical Analysis
of Fly Ash from Western Coal
Compound Average High Low
SiO2 36.3 40.6 32.2
A1203 19.6 22.2 15.6
Fe203 4.74 8.5 3.5
Ti02 0.69 0.9 0.3
P,0S 0.36 0.8 0.20
CaO 19.1 29.7 13.3
MgO 4.31 6.6 1.40
Na2O 0.27 0.4 0.20
K20 0.11 0.2 0.10
S03 12.85 14.6 11.0
98.33 12TT5" 77.&
SOUTHERN RESEARCH INSTITUTE
-------
-52-
10:
10 >
10"
3
CO
10
10
10'
\
A = CaO 22%
B = CaO 10%
200 300 400 500 600 700 800 900 1000
Temperature, °F
Figure 2.17. Laboratory Resistivity Measurements
for Two Fly Ash Samples from Western
Coal (Pre-Oried Ash Samples).
-------
-53-
Figure 2.18 shows the variation in resistivity for various
oxides as a function of the parameter 1/kT as reported by White.1
It is apparent from these data that rather wide changes in
resistivity can occur between different oxides, and hence it is
not surprising that variations in bulk resistivity of fly ash
occur. However, for the majority of fly ash, resistivity values
at 600°F are sufficiently low so as not to consitute a problem.
Figure 2.19 is a composite plot of the resistivity data
taken on several fly ash samples utilizing a point-plane probe.
D. Sparkover and Back Corona
The phenomena of sparkover and back corona constitute prac-
tical limits on the operation of an electrostatic precipitator
since they determine the maximum current and voltage at which the
precipitator will operate. From the standpoint of modeling or
predicting precipitator performance, it is necessary to establish
the maximum current and voltage conditions as related to measur-
able properties of the dust and precipitator electrode geometry.
Studies of sparking and back corona under this contract were
directed toward review of the theory of sparkover and experimental
verification of the influence of dust resistivity on sparkover
and back corona conditions.
Electrical properties of a precipitator are defined by the
curves relating secondary voltage and current. Under clean col-
lection electrode conditions, the secondary voltage can be
increased with a corresponding increase in secondary current
until the electric field in the interelectrode space breaks down
and a spark propagates from the anode to the cathode. In the
case of negative corona, the anode is the collection electrode,
which is at ground potential.
The shape of the voltage-current curve as well as the
sparkover voltage is determined by the electrode geometry and
spacing. For a wire and plate or cylinder precipitator, the
primary variables are the diameter of the wire and the separa-
tion between wire and plate or cylinder. Except for very small
wires, smaller diameter wire results in higher current for a given
voltage if the pipe diameter is held constant. Both the wire
diameter and spacing alter the electric field and hence determine
the conditions for spark initiation and propagation.
SOUTHERN RESEARCH INSTITUTE
-------
-54-
Temperature, °p
10"
10
18
I 1012
J?
" 10"
1011
A120,
Si02
Ply
Ash
1*0 1.2 1.4 1.6 1.8 2.0
1/kT x 10'20 MRS Unite
Figure 2.18. Variation in Resistivity for Various
Oxides as a Function of the Parameter
1/kT
-------
-55-
A. Bayowater. Auatralla
New Vali, Australia
C. U.S. Bureau of Mine« -
Anthracite
0. U.S. Bureau of Mines -
Ditumi nouB
Numbers re£er to American tnstallationo
by code. Bulk realativltlea are labora-
tory neMurementa. RealHtlvltlea on th*
left of the pe
with a point-plane disc probe at an
electric field (trength of 10 kV/cm.
loo 900 IflOO
10
100
400
SOO
(00
TEMPERATURE, *P
Figure 2.19. Composite Plot of Resistivity Data for
Several Fly Ash Samples
SOUTHERN RESEARCH INSTITUTE
-------
-56-
Fundamental studies of the mechanism of spark initiation
and propagation have been made by Penney10'11, Loeb12, and
others. These studies indicate that sparking originates from
streamers or flares at the anode and propagates across the inter-
electrode space provided there is sufficient voltage across the
electrodes.
In the case of clean electrodes, the conditions for sparking
are simply that localized breakdown occurs, usually at a small
region of high field strength near the collection electrode. The
conditions necessary to cause the localized breakdown are
generally sufficient to propagate the spark across the interelec-
trode space.
If dust is present on the collection electrode surface, the
voltage-current curves will be altered by the voltage drop in the
dust layer, which is given by
V = jpt . 2.5
a
If the dust layer resistivity is low (around 107 fl-cm) , the
voltage drop across it will be low and the voltage-current curve
will be shifted as though the electrode spacing were decreased.
High dust resistivity, on the other hand, can result in large
voltage drops and high electric fields within the dust layer itself.
The field in the deposit can be computed as E = jp. For reason-
ably high dust resistivities, electric fields in the deposit can
be rather large for currents normally encountered in precipitator
practice. As the field is increased, a point is reached where it
exceeds the breakdown strength of the interstitial gases. The
breakdown strength can vary depending on the size and composition
of the dust, as well as the gases; however, for the majority of
industrial dusts, breakdown generally occurs in the range of from
10 to 20 kV/cm.
Breakdown of the dust layer can cause one of two events to
take place. If the dust resistivity is in the intermediate range,
breakdown of the dust layer can cause a sudden increase in the
interelectrode voltage since the voltage drop that was previously
across the dust layer is now added to the voltage that previously
was established between the discharge electrode and dust layer
surface. If this voltage is sufficiently high, a spark will prop-
agate across the interelectrode space. Localized breakdown of the
dust can act in much the same fashion as a point, thus reducing
the voltage required for spark propagation.
-------
-57-
If the resistivity of the dust layer is increased further,
the voltage drop across the dust layer can be exceptionally high
even at low current densities, and breakdown can occur at volt-
ages lower than those required to propagate a spark across the
interelectrode space. If this occurs, a condition of back corona
develops and the voltage-current curves of the precipitator depart
from the normal curves characteristic of a pure resistive effect.
The condition of back corona can be observed in the labora-
tory as a diffuse glow over the dust layer surface. A continued
increase in voltage and current will result in sparking, provided
the power supply has sufficient capacity.
The significance of a back corona condition is that the
breakdown releases positive ions (for the negative corona preci-
pitator) , and these ions are propelled by the electric field into
the interelectrode region. If they impinge on negatively charged
particles, they serve to reduce the normal charge or charge the
particles with the opposite polarity.
When a back corona condition develops, the character of
the current-voltage curve changes. Increases in voltage beyond
the onset of back corona result in an abnormally rapid increase
in current as compared with the normal case.
From theoretical considerations, the conditions limiting
precipitator operation are summarized graphically in Figure 2.20.
Curve 1 in this figure is a clean plate, voltage-current curve
for a point-plane precipitator with a point-to-plane spacing of
1.12 cm. Curve 2 is calculated on the basis of a dust layer
thickness of 0.20 cm and a resistivity of 108 fl-cm. Curve 3 is
for a resistivity of 109 fi-cm, and Curve 4 corresponds to a dust
resistivity of 1010 ft-cm. The dotted line 5 represent the volt-
age drop in the dust layer corresponding to the breakdown field
strength for dust layers of 0.2 cm.
In the case of Curve 2, it is apparent that breakdown in
the dust layer would occur only at very high current densities.
Before this point is reached, the voltage across the interelec-
trode space is sufficient to cause a spark to propagate. This
would theoretically occur when the voltage from the dust layer
to the discharge electrode equals the clean plate sparking con-
dition. Because of the voltage drop across the dust layer,
sparking would theoretically occur at a voltage corresponding
SOUTHERN RESEARCH INSTITUTE
-------
-58-
E
O
(0
41
2
n
3
O
60
50
30
20
10
1. Clean Plate - Point-to-Plane Spacing, 1.12 cm
2. 2 mm layer dust - 10e fl-cm resistivity
3. 2 mm layer dust - 109 ft-cm resistivity
4. 2 mm layer dust - 1010 fi-cm resistivity 1
5. Line designating electric field in deposit
reaches 20 kV/cm for 2 mm layer
NOTE: Arrow designates sparking
..•3
12 15
Applied Voltage, kV
18
21
Figure 2.20. Behavior of a Point-to-Plane Electrostatic Preci-
fitator Based on Theoretical Considerations of
parking and Back Corona. Clean Plate Curve is
Measured Curve, Remaining Curves Computed.
-------
-59-
to Point A. This condition, therefore, represents one cri-
terion for sparking in which there is no breakdown in the
dust layer.
In the case of the dust with a resistivity of 109 fi-cm,
the voltage drop across the dust layer would exceed the break-
down field for a 0.2 cm dust layer at a current corresponding to
Point B. At this point, localized breakdown would result in a
voltage developing across the interelectrode space (dust layer
surface to discharge electrode) equal to the voltage required for
spark propagation on a clean plate electrode. When this condition
is reached, sparking will occur. This consitutes the second
criterion for sparking; that is, when breakdown of the inter-
stitial gases occurs at a voltage sufficient for the combined
electrode-to-dust-surface voltage and the voltage drop across the
dust layer to equal the clean plate sparkover voltage.
Point C represents the third condition which corresponds to
a high dust resistivity. Here breakdown occurs at a voltage
sufficiently low so that the combined electrode-to-dust-surface
voltage and the voltage drop across the dust layer is less than
that required for propagation of a spark. Under these conditions,
a back corona develops without sparking. However, if the voltage
is increased, the current will increase somewhat along the curve
for constant breakdown voltage until a voltage sufficient to cause
a spark to propagate is reached. This consitutes the third cri-
terion for sparking; that is, a back corona condition precedes
the sparkover voltage.
Several factors can modify sparkover conditions. If the
dust layer thickness is increased, the voltage drop across the
dust layer will increase proportionately. The breakdown voltage is
thus greater so that this voltage drop added to the electrode-to-
dust-surface voltage is sufficient for a spark to propagate. Thus,
the increased thickness can cause a change from a back corona to
a sparking condition for the same dust.
The waveform of the applied voltage is also thought to
influence the sparkover characteristics. The sharp rise rates
on the leading edges of the waveform results in a higher peak to
average voltage and consequently field on the device.
Changes in the character of the dust can also alter spark-
over and back corona conditions. The presence of conductive
particles in the dust (such as unburned carbon) can alter the
localized field strength and cause breakdown to occur at lower
voltages than would be predicted.
SOUTHERN RESEARCH INSTITUTE
-------
-60-
Also, very fine dust of high resistivity material can act
in somewhat the same fashion as a solid insulator to increase
breakdown strength. These factors can alter the specific con-
ditions; however, the generalized theory of sparkover and back
corona can be used to predict behavior if the specific condi-
tions are known.
1. Experimental verification
Studies of sparking and back corona conditions were made in
the laboratory utilizing dust of varying resistivities to deter-
mine how closely the limiting conditions of sparkover and back
corona could be predicted. The apparatus utilized in these studies
was a point-plane precipitator with dimensions as shown in Figure
2.21. The procedure followed was to obtain a voltage-current
curve for a given spacing from the point to the grounded plate.
A dust layer was then placed on the grounded electrode and screeded
to give the desired thickness. The plate spacing was then ad-
justed so that the distance from the point to the dust surface was
equal to the previous point-to-plate spacing. The V-I curve with
the dust layer was then obtained. In each case, the voltage was
increased until sparkover occurred. Dusts used in the experi-
mental work were fly ash from two sources, aluminum oxide (activa-
ted alumina, 80 to 325 mesh chromatographic grade) and sulfur.
All data were taken in air at ambient temperature (70°F) utilizing
an X-Y recorder to obtain the voltage-current curves. In the case
of the high resistivity dust, the current scale was expanded in
the low range to detect the initial rapid current increase indi-
cative of the onset of back corona. Resistivity of the dust was
measured with two parallel discs following each test.
Table 2.3 shows a comparison of the resistivity as deter-
mined from the V-I curve and that measured by the parallel disc
method. The values agree reasonably well when compared at equiva-
lent field strengths and current densities.
Figure 2.22 shows the V-I curves for the dusts with various
resistivities. The data are for a dust layer thickness of 0.23
cm.
The data shown illustrate the conditions for spark propa-
gation reasonably well. Point A of Curve 2 indicates sparking at
a voltage reasonably close to the clean plate sparking voltage plus
the voltage drop in the dust layer. Points B and C show sparking
occurring at near the predicted breakdown field strength; however,
the voltage at which sparking occurs is less than that required to
propagate a spark under clean plate conditions.
-------
-61-
A - High Voltage Supply
B - Current Measurement
C - Ground
r
T
» /i
-}
Point Electrode
Metal Electrode
1.5 en Area = 5 cm2
^
UB
rc
3 Guard
p Ring
Figure 2.21. Dimensions of Point-Plane Precipitator Used in
Laboratory Studies of Sparking and Back Corona
Conditions
SOUTHERN RESEARCH INSTITUTE
-------
-62-
Table 2.3. Comparison of Resistivity Determined by
the Parallel Disc Method and the V-I Curve Method
(Direct Current Tests)
Parallel disc
6 x 107
7 x 108
1 x 1010
8 x 1012
V-I curve
4 x 107
3.1 x 108
5.7 x 109
1.4 x 1011*
•Back corona noted.
In reviewing the probable causes for spark propagation
occurring at lower than clean plate voltage, it was concluded
that two factors could contribute to this condition. First,
breakdown in the dust layer is accompanied by visible light
emission. Acker11 states that one mechanism of spark propa-
gation is by photoionization since the current rise is more rapid
than would be possible by electron flow. If so, the emissions
accompanying breakdown could account for the lower spark propaga-
tion voltage. A second factor in the lower breakdown voltage is
the probability that the breakdown in the dust layer acts as a point.
Since point-to-point breakdown occurs at much lower voltage, this
could account for the behavior noted.
2.
Relation to precipitator operation
The theory of sparkover and back corona applies to full-size
precipitators. Since a precipitator is normally operated with a
time-varying voltage, the peak current and waveform would determine
sparkover conditions. Also, variations in spacings from wire to
plate, as well as areas of localized high fields, can alter the
sparkover conditions.
-------
-63-
E
u
-p
•H
U)
c
M
M
3
U
Point Plane Spacing, 1.12 cm/ / . ..7^
r ^ / /D = 6xl07n-cm
' Reference Line
' E = 20 kV/on
- ' cjl." 10>-
r.. — 1 r> ^-* *
Applied Voltage, kV
Figure 2.22. Experimental Volt~Current Curves for Point-Plane
Device with a Variety of Dust Resistivities.
Dust layer thickness is 2.3 mm.
SOUTHERN RESEARCH INSTITUTE
-------
-64-
Figure 2.23 shows the voltage-current curves for a pre-
cipitator section. Curve 1 was taken when the precipitator was
operating at 260°F with a 3% sulfur coal. Dust resistivity under
these conditions is low and the curve should approximate a clean
plate curve very closely.
Curve 2 corresponds to conditions of around 1% sulfur coal
with a correspondingly higher resistivity. Also plotted on the
same figure are the theoretical voltage-current curves corres-
ponding to resistivities of 1 x 1012 to 1 x 1013 fl-cm. From the
appearance of the curves, a back corona condition would be pre-
dicted around Point A. The V-I curves would indicate the dust
resistivity to be high (1 x 1013 ft-cm). Once back corona is
established, the voltage drop across the dust layer would tend to
be constant and the V-I curve would depart from the typical pure-
ly resistive behavior. The gaseous breakdown would cause the
curve to act more like a gaseous discharge tube with a constant
voltage drop which is equivalent to the breakdown field strength.
Figure 2.24 is a similar curve for a second precipitator
operating on fluorspar dust with very high resistivity. Again,
the behavior suggests that breakdown of the 'dust layer occurs at
very low current and that the shape of the curve parallel to the
clean plate curve suggests that the back corfona condition causes
a constant voltage drop across the dust layer. In the case of
very high resistivity dusts, the shape of the V-I curve can be
further altered as a result of more complete breakdown of the
dust layer.
Results of the studies so far have not correlated measured
resistivities with the apparent resistivity as indicated by the
V-I curves, the latter suggesting higher resistivities in most
cases than is indicated by resistivity measurements on the parti-
cular dust. Additional studies to correlate measured resistivi-
ties with V-I curves are indicated.
The optimum operating point for a precipitator in a back
corona condition also needs study. Many precipitators with high
resistivity dusts are set to operate under spark-limiting condi-
tions. Since, for high resistivity dust, back corona precedes
sparkover, studies of optimum operating conditions need to be
made.
-------
-65-
20
N
E
U
15
•H
0)
c
s
c
0)
3
U
10
10
Clean
Electrode
20 30 40
Applied Voltage, kV
50
Figure 2.23. Voltage-Current Curves for a Precipitator
Section
SOUTHERN RESEARCH INSTITUTE
-------
-66-
ISO
140
130
120
110
100
90
S 80
70
60
50
40
30
20
10
0
M
«
•O
(1) Atmospheric Air Load O
(2) Hot Air Load
/ /,(3) Calculated
*; V-I for 1 cm
Layer of 1010 f)-ca
p =
4x10 ll n-cm
(4) 320-P.
I
10 15 20
Secondary Voltage, kV
25
30
35
Figure 2.24. Voltage-Current Curve for a Precipitator Operating
on Fluorspar Dust with Very High Resistivity
-------
-67-
E. Optimization of Precipitator Design
for High Resistivity Dusts
The theory of precipitator operation states that the
migration velocity of a dust particle is proportional to both
the charge on the dust particle and to the value of the electric
field in the vicinity of the collection electrode as shown by
w = qE /6Tran 2.6
Thus, the two electrical quantities of interest are the charge and
the field.
The operating point for electrostatic precipitators used to
collect high resistivity materials is limited by the sparking rate
of the power supply or by the current limit in the absence of
sparking. The spark rate of the power supply is determined by
the electrical conditions at the surface of the dust layer as dis-
cussed in Section D. With a reasonably high resistivity dust
layer on the plates, sparking occurs when the electric field in
the dust layer reaches that value that causes an electrical break-
down of the interstitial gas.
The electric field in the dust layer is related to the cur-
rent density and resistivity as shown by
E = jp 2.7
When E approaches the breakdown strength of air, about 20 kV/cm, a
sparkover in the dust layer leads to a spark between the elec-
trodes of the electrostatic precipitator. Equation 2.7 suggests
that either the resistivity or the current density may be modified
to improve the operating characteristics of a precipitator collect-
ing high resistivity dusts.
In such an installation, two requirements must be met. In
one case, it is necessary to experimentally verify Equation 2.6
to be sure that the migration velocity, and hence collection effi-
ciency, is really dependent on the collection electric field rather
than on power density or current density, as is sometimes assumed.
In the second case, the significance of particle charging time
must be known. These two pieces of evidence will be sufficient
SOUTHERN RESEARCH INSTITUTE
-------
-68-
to determine the desirability of pursuing a low current density
precipitator with a precharging section.
Equation 2.6 was evaluated by operating the model precipi-
tator at conditions which would provide an electric field at the
collection plate that was essentially constant for the values of
current density that differed by a factor of 10. The two sets
of electrical conditions are shown in Table 2.4.
Table 2.4. Model Precipitator Operating Conditions
for Comparative Efficiency Tests
Item Standard Reduced
conditions current
Voltage, kV 37 45
Current, yA 300 30
Electric field, kV/cm 1.85 1.95
Current density, A/cm 17.0 x 10~9 1.7 x 10~9
Electrical power, watts 11.1 1.35
Corona wire diameter, in. 0.109 0.250
The precipitator model was operated with the inlet section function-
ing as a charging section and with the collection plates replaced
by a rod curtain arrangement as shown in Figure 2.25. This
charging section was installed to assure that the particles were
uniformly charged for both test conditions.
The evaluation of Equation 2.6 was made by determining the
percentage collection efficiency as a function of position for
the two conditions as shown in Figure 2.26. This figure indicates
that for the low current density, the collection efficiency is
actually improved over that for the standard current density. The
actual electrical power expended in the standard condition is
about a factor of 8 greater than that for the condition with re-
duced current density. These tests show that if the particulate
material is charged, then the collection efficiency is determined
by the collection electric field at the surface of the dust.
-------
-69-
Charging Section Collection Section
0.109 in. wire
OOOO OOOOOOOOOO
Standard Configuration
0.250 in, wire
oooo OOOOOOOOOOOO
Low Current Density Configuration
Figure 2.25. Schematic of Standard and Reduced Current Density
Test Conditions for Southern Research Institute
Model Precharging Tests
SOUTHERN RESEARCH INSTITUTE
-------
-70-
n
o
H
M
01
04
0)
•H
U
•H
H
g
•rl
4J
U
0)
8
c
0)
U
M
0)
40
30
20
10
Collection Section
E -1.9 kV/cra - both cases
Hire dia • 0.205"
j - 1.7 x 10-'
A/cm
Hire dia =
0.109"
j - 17.0 x 10-9 A/cm
23 45 67 8
Collection Position in Pilot Precipitator
Figure 2.26.
Comparison Between the Collection Efficiency
for Standard Conditions and Reduced Current
Density Conditions for Constant Electric
Field
-------
-71-
The significance of particle charging considerations was
also verified experimentally. (If contemplated to develop a precip-
itator with a reduced current density in the collection zone, it
is necessary to determine if a high current density charging
section ahead of this collection zone is also required.)
If the precipitator could be operated at reduced currents
throughout the entire precipitation zone, then the solution to
the high resistivity problem would be simplified. However, since
theory states that the particle charging time constant is also
increased for reduced currents, a reduced collection efficiency
would result.
The significance of particle charging was evaluated by
operating the pilot precipitator at normal current densities and
with current densities reduced by a factor of 5. The percentage
of material collected as a function of position through the pre-
cipitator model is shown in Figure 2.27. If the material were
instantaneously charged to saturation, the collection as a func-
tion of position as predicted by the Deutsch-Anderson equation
would be a decreasing exponential function. As the charging
time increases, the collection vs position curve should be shifted
to the right. The high current density curve in Figure 2.27 shows
the dust to be charged to near saturation at Position 2, while for
the reduced current density case, the near-saturation charge occurs
at Position 3. Thus, particle charging time is seen to be signi-
ficant.
The requirements for a precipitator that is optimized for
collecting high resistivity dust particles can be summarized.
The current density in the collection zone must be reduced below
a value that will cause the electric field in the deposit to
exceed the breakdown strength of the gas in the interstitial
regions of the deposit; and some means must be provided to
apply a near-saturation charge to the dust prior to intro-
duction into the collection zone. The ability to apply this
charge to the dust is also limited by the electrical resistivity
of the dust. Some preliminary work has been done utilizing a
small corona wire in conjunction with a rod curtain type of
grounded electrode system to provide a high current density
charging section. The rods were heated to reduce the resistivity
of the thin deposit that collected on the rods. It seems that
this may be one technique for providing a high current charging
section for dusts with high resistivity, but insufficient
experimental evidence was available to prove this point con-
clusively. The evidence does point to the possibility of col-
lecting high resistivity materials by tailoring the current
density and electric field combinations to provide an optimized
design.
SOUTHERN RESEARCH INSTITUTE
-------
-72-
O j = 5 x 10~8 A/cm2
j = 1 x KT8 A/c2
o
u
1.5 3.0 4.5 6.0 7.5 9.0 10.5 12.0
Position Along Precipitator, ft
Figure 2.27.
Comparison Between the Percentage of
Material Removed within Each Increment
of Length for Two Values of Current
Density
-------
-73-
F. Bibliography - Section 2
1. White, H. J., Industrial Electrostatic Precipitation,
Addison-Wesley Publishing Co., Inc., Reading, Mass. ("1963).
2. Cooperman, P., "A New Theory of Precipitator Efficiency,"
Paper No. 69-4, APCA (1969).
3. Robinson, M., Atmos. Environ. 1, 193 (1967).
4. Heinrich, D. 0., Staub, 2_3, 83 (1963).
5. Williams, J. C. and Jackson, R., Proc. Symp. Interaction
Fluids Particles, Inst. Chem. Engrs. (London) pp 282-288,
discussion pp 291-293, 297-298 (1962).
6. Penney, G. and Hewitt, J. G., "Some Measurements of Abnormal
Corona," Communication and Electronics, published by AIEE
(July 1958T
7. Cohen, L. and Dickinson, R. W., "The Measurement of the
Resistivity of Power Station Flue Dust," J. Sci. Instrum.
4£ (1963).
8. Bucher, W. E., "A Study of the Bulk Electrical Resistivity
Characteristics of Fly Ash from Lignite and Other Western
Coals," a Thesis submitted to the faculty of the University
of North Dakota, Grand Forks, North Dakota (December 1970).
9. McLean, K. J., "Electrical Conduction in High Resistivity
Particulate Solids," a Thesis submitted to Wollongong
University College, University of New South Wales (December
1969).
10. Penney, G. and Craig, S., "Pulse Discharges Preceding
Sparkover at Low Voltage Gradients," Amer. Inst. Elec. Engrs.
Meeting, New York City, Paper No. 61-91 (1961).
11. Acker, F. E. and Penney, G., "Influence of Previous Positive
Streamers on Streamer Propagation and Breakdown in a
Positive Point-to-Plane Gap," J of Appl. Physics 39, No. 5,
pp 2363-2369 (April 1968).
12. Loeb, Leonard B., Electrical Coronas, Their Basic Physical
Mechanisms. Uni. of Cal. Press, Berkeley and Los Angeles,
1965.
SOUTHERN RESEARCH INSTITUTE
-------
-74-
SECTION 3. REENTRAINMEHT STUDIES
In dry-type precipitators, efficiency is determined by the
loss of dust due to reentrainment in addition to those factors
that determine collectability. Reentrainment losses can be quite
serious for some types of dust, causing substantial degradation
of precipitator performance.
The study of reentrainment under the current contract was
undertaken to review the present technology relative to reentrain-
ment losses and to provide a basis for determining those losses
quantitatively.
The effect of reentrainment on overall precipitator
efficiency depends upon the length of the precipitator, the
plate area rapped at one time, and the percentage of material
lost from each section and the number of sections. Figure 3.1
shows the influence on precipitator efficiency of loss of various
percentages of the collected dust from a precipitator composed
of four sections. Assuming that a constant percentage of the
dust collected is subsequently reentrained by any mechanism,
the degradation in efficiency can be computed on the basis of the
Deutsch-Anderson equation. That is, the material reentrained
from the first section will add to the dust burden entering the
second section, etc. The amount of dust removed by the second
section will be greater than it would if no reentrainment occurred,
thus, the degradation of efficiency is not as great as would
perhaps be expected.
Reentrainment of dust in a precipitator can result from:
direct scouring of the dust by aerodynamic drag
of high velocity gases,
carry through of some dust during rapping or
falling of its own weight,
release of collected dust when sparking occurs,
saltation due to impaction of a large particle
on the dust layer,
repulsion of dust due to electrical forces
hopper losses caused by aerodynamic penetration.
air inleakage, and full hoppers.
-------
-75-
100
90
I - (i - mar
ibT
% Reentrainment Per Section (1-r)
80 70 60 50 40 30 20 10
0 10 20 30 40 50 60 70 80 90 100
% of Collected Dust Reaching Hopper
Vertical Transport Efficiency (r)
Figure 3.1. Effect of Reentrainment on the Efficiency of
a Four-Section Precipitator Designed for a
No Reentrainment Efficiency as Indicated
SOUTHERN RESEARCH INSTITUTE
-------
-76-
Qualitative and quantitative studies were made during this
program to determine the magnitude of the reentrainment due to
these various effects.
A. Scouring
One potential mechanism of reentrainment loss is direct
scouring of the dust by a high velocity gas stream. In terms of
precipitator performance, scouring could be significant since loss
of collected dust would take place more or less continuously and
seriously reduce precipitation efficiency.
Scouring in'a precipitator can take place if the aerodynamic
drag force exceeds the adhesive and cohesive forces acting on an
individual particle or on an agglomeration of particles.
In a practical precipitator, these forces are dependent upon
the type of dust being collected, the electrical properties of the
dust, and the gas velocity, turbulence, and uniformity.
Studies of reentrainment due to scouring were made with the
small-scale precipitator by precipitating a dust layer of about
2.0 mm thickness, cutting off the dust feed, and running the pre-
cipitator at various gas velocities for a period of 30 minutes.
If scouring were present, a greater percentage of the dust would
be collected in the downstream sections and less in the upstream
sections.
Since scouring is related to the electrical holding force,
the test series was repeated at various voltages ranging from 0
to 30 kV. Figure 3.2 shows the percentage of the zero overrun
collection efficiency for the range of voltages used. This shows
that some scouring takes place at voltages below around 15 kV,
which is around the corona onset voltage.
The air temperature for these tests was around 125°F so
that the fly ash resistivity was in the vicinity of 1 x 109 fl-cm,
which is moderately low for fly ash precipitators.
The lack of scouring under these conditions was somewhat
unexpected, especially since there is evidence that some scouring
might take place in a full-size precipitator at lower velocities
than the maximum used in these tests. It was suspected that the
uniformity of gas flow in the small-scale precipitator with the
smooth plate might not give the same conditions for scouring as
in a full-size precipitator with baffle plates. To explore this
effect, 1% in. baffles spaced 9 in. apart were added to the
-------
-77-
15
OP
C
o
•H
in
O
M
w
10
10 20
Applied Voltage, kV
30
Figure 3.2.
Relationship Between Scouring
for Various Applied Voltages
Defined as Excess Loss Over No
Overrun Collection Efficiency
SOUTHERN RESEARCH INSTITUTE
-------
-78-
plates and the tests repeated. Again, no evidence of direct
scouring was found at gas velocities up to 12 ft/sec.
The conclusion to be drawn from these tests is that scour-
ing of the collected dust does not appear to take place at gas
velocities up to 12 ft/sec with a rather typical fly ash (10 ym
mmd). Larger particle size or higher gas velocities would perhaps
show scouring effects.
In full-size precipitators, the same general conclusion may
be reached, that is, that no scouring as defined here is likely
to take place at gas velocities below 12 ft/sec. When extremely
poor gas flow or a dust with rather poor cohesive properties is
encountered, some scouring may occur. However, the influence
appears to be considerably smaller than might be expected.
B. Rapping Reentrainment
Loss of dust during rapping constitutes one of the major
sources of reentrainment losses in a precipitator. Under some
conditions, these losses can be visually observed as puffs of
dust that escape during and immediately following a rapping cycle.
These rapping losses can be minimized by proper adjustment of the
rapping intensity and frequency.
Studies were made with the small-scale precipitator of the
conditions required to give minimum rapping losses. Some problems
peculiar to the plate height as compared with a full-scale precipi-
tator were encountered, particularly with regard to the influence
of the dust layer falling into the hopper and causing breakup and
reentrainment. However, by installing a plexiglas window on one
side of a precipitator section and observing the nature of the
dust breakup and fall, qualitative observations were made regard-
ing optimum conditions for rapping.
The condition required for optimum dust removal is that
the dust fall as a sheet layer into the hopper. This requires
that a sufficient thickness be built up so that a rap will
dislodge it as a unit. Otherwise, if the dust layer is thin, a
more severe rap will be required to remove it and, in addition,
the dust will powder and be propelled into the gas stream where
it will be carried out of the precipitator. At the other ex-
treme, the dust layer can be allowed to build up until the weight
of the dust is sufficient to overcome the adhesive or cohesive
forces and falls of its own accord. When this occurs, the dust
layer falls free with a velocity sufficient to cause self-reentrain-
ment and extensive loss of dust. On free-fall from tall plates,
-------
-79-
velocities in the vicinity of 50 ft/sec can be reached causing
excessive localized scouring. Thus, rapping frequencies should
be adjusted between these two extreme conditions.
Rapping intensity can also be adjusted to give minimum loss
of dust. If rapping is too light dust will not be released from
the plate at the time of rap, and on the succeeding rap the
thickness will be sufficient to cause an avalanche effect. On
the other hand, too severe a rap will cause the dust to be pro-
pelled from the plate a sufficient distance to fall freely. In
addition, too severe a rap causes powdering of the dust with sub-
sequent reentrainment loss.
Under ideal rapping conditions, the dust layer is dislodged
from the plate and slides down the plate until it is recaptured
by the precipitation process.
The factors that influence dust loss due to reentrainment
are the gas velocity, resistivity, particle size, and adhesive
and cohesive properties of the dust.
The influence of resistivity of the dust on reentrainment
has not been sufficiently recognized. In the case of a high
resistivity dust, the electrical forces holding the dust to the
plate are large and rapping intensities required to dislodge the
dust are high. Under these conditions, considerable powdering
of the dust can occur with accompanying high losses. In some
instances, dust resistivity is so high that it is impractical to
remove it by conventional rapping. In such cases, power-off
rapping is required and rapping puffs are visible during the rap
and the losses are excessive as compared with more conventional
rapping.
At the other extreme, low resistivity can also result in
excessive reentrainment losses. In the case of fly ash precipi-
tators operated at low temperature (260°F) with high sulfur coal,
reentrainment losses can be as high as 20 to 30%.
The effect of low dust resistivity on performance is mani-
fested by excessive rapping puffs when viewed visually or with
optical obscuration instruments.
Qualitatively, the effect of low resistivity would appear
to be that the dislodged dust is not re-precipitated following
rapping and falls freely into the hopper. It would, therefore,
act in much the same manner as permitting too thick a dust layer
to form. Adjustments to the rapping intensity and rapping frequen-
cy can minimize the effect of these losses. Use of impact absorbing
SOUTHERN RESEARCH INSTITUTE
-------
-80-
materials as well as minimizing the rapping impact have been
attempted in an effort to control rapping intensity and minimize
losses. However, when very low dust resistivity is encountered,
such as results from high sulfur coal at low gas temperatures,
it has not been practical to control rapping sufficiently to
overcome the higher rapping losses.
C. Electrical Forces
The electrical field in the dust layer is determined by the
current density and dust resistivity. If the resistivity of the
dust is lower than around 1010 ft-cm, the field in the dust layer
will be lower than in the space immediately adjacent to it. This
abrupt change in electric field can occur only if there is an ac-
companying positive surface charge. The magnitude of this charge
is dependent upon the difference between the electric field in
the dust layer and in the adjacent gas stream. For dust resis-
tivities in the range of 106 to 107 fl-cm, this positive surface
charge may propel the collected dust back into the gas stream and
contribute to reentrainment.
Tests made on the small-scale precipitator demonstrated this
effect when a layer of low resistivity dust (108 ft-cm) was pre-
cipitated onto the plates. The voltage was then increased, which
increased the electric field in the gas more than in the dust
deposit. The result was that with an increase in voltage, dust
particles were removed from the plates, dropping into the hopper.
No measure of this type of reentrainment was attempted.
D. Sparking
Sparking can influence reentrainment in two ways. The
spark itself can cause disruption of the current in a localized
section of the precipitator. This interruption of the corona
current causes loss of electrical holding force in the affected
area and dust can be reentrained during this period. A second
influence of sparking is that energy dissipated in the spark
causes the dust in the vicinity of the spark to be expelled into
the gas stream.
The conditions contributing to reentrainment due to spark-
ing can vary depending upon the precipitator design. If the
spark is rapidly quenched, the effect is minimized. However,
with poorly designed electrical energization equipment a spark
or power arc can persist for a reasonably long period of time,
and the effect can become quite serious in terms of reentrainment
as well as other factors influencing performance.
-------
-81-
E. Saltation
The impact of a large particle on a precipitated dust layer
can dislodge previously collected dust and constitutes another
potential source of reentrainment.
The saltation effect has been studied by Penney,1 Robinson,2
and others. However, it is difficult to distinguish this type of
reentrainment loss from others in a precipitator since it occurs
during the precipitation process. It may be regarded as a type
of scouring loss since the dislodged particles will be picked up
by the gas stream and probably is dependent upon gas velocity.
No attempts were made during this program to identify this type
of reentrainment loss independently.
The effects of saltation, sparking, and electric field
differences are difficult to separate quantitatively. Sparking
and electric field effects are perhaps aggravated during rapping
and are manifested by increased rapping losses.
F. Hopper Losses
In addition to losses of the dust from the collection
electrodes, losses also occur in the vicinity of the hopper.
Vertical dust concentration profiles can show dust levels at the
bottom of the precipitator considerably higher than at the top.
This can be due to gas sweeping into the hoppers and reentrain-
ing dust or to rebounding of the dust as it falls into the hop-
pers. These effects can be minimized by proper baffling.
In the small-scale precipitator, studies of reentrainment
were clouded by the magnitude of the losses associated with dust
falling into the hopper. Visual observation with the plexiglas
window showed that dust falling into the hopper following a rap
was powdered by the impact and rebounded approximately one-third
of the height of the collection plate. Similar effects occur in
full-size precipitators, resulting in a smaller percentage of the
total reentrainment for the tall plates.
G. Full-Scale Precipitator Tests
Detection of reentrainment losses in a precipitator can be
made by several methods as discussed by White.3 These include:
(1) plotting precipitator losses as a function of gas velocity
1 Refer to the Bibliography at the end of this section.
SOUTHERN RESEARCH INSTITUTE
-------
-82-
on semi-log paper and observing the point of departure from
linearity, (2) measuring the particle size distribution of the
inlet and outlet dust and observing the presence of larger parti-
cles in the outlet, and (3) measuring the electrical charge on
particles and observing the increase in the number of positively
charged particles.
In view of the previous discussions of reentrainment
factors, it is reasonable to assume that each of the methods of
detecting reentrainment might be more sensitive to particular
reentrainment mechanisms. If reentrainment is due to powdering
of the dust as a result of intense rapping, large particles may
not be present. If dust resistivity is very high, reentrained
particles may still have a negative charge. On the other hand,
if reentrainment is due to erosion occurring either during rapping
or as direct scouring, one would expect to find a preponderance
of large particles. Low dust resistivity would also tend to
produce positively charged particles regardless of the mechanism
by which reentrainment occurred.
In general, each of these methods is primarily useful in
detecting the point at which reentrainment becomes severe.
Reentrainment losses obviously occur over the entire range of
conditions. Detection by the above techniques, therefore,
primarily indicates an abnormal condition of operation.
A number of field tests were made and data examined to
indicate conditions of excessive reentrainment losses. One of
the primary objectives of these studies was to identify the cause
of poor performance associated with the use of low gas tempera-
tures on precipitators used with boilers burning high sulfur coal.
Figure 3.3 shows the precipitator penetration as a function of
gas velocity for two full-size precipitators. These curves show
the excess penetration over that predicted from the Deutsch equation.
Curves 1A and IB are for a precipitator used on a boiler burning
3 to 4% sulfur coal at temperatures of 275 to 300°F and hence
widely varying resistivities. In the case of the low resistivity
ash (275°F), the departure from the Deutsch equation penetration
occurred at around 5.5 ft/sec, whereas the precipitator would
follow the Deutsch efficiency prediction to around 7.5 ft/sec when
collecting the higher resistivity ash corresponding to the 300°F
flue gas temperature. Curves 2 and 3 show similar effects for
two different plants.
The significance of these data is that excessive losses
can occur within the range of gas velocities normally encountered
-------
-83-
Jl
41
a
: 20
§
U
n
3
LR = Low Resistivity
HR = High Resistivity
678
Gas Velocity, ft/seo
10
Figure 3.3. Precipitator Losses as a Function of Gas Velocity
for Two Full-Size Precipitators
SOUTHERN RESEARCH INSTITUTE
-------
-84-
in fly ash precipitators and that these losses are dependent on
dust resistivity. Thus in many instances where high gas veloci-
ties are encountered along with high dust resistivities, attempts
to improve performance by reducing dust resistivity can result in
little or no improvement due to excessive reentrainment. When
changes are made in dust resistivity, for example, by addition of
conditioning agents or reduction of gas temperature, rapping
should be optimized. Even so, when gas velocities are high and
perhaps nonuniform, reentrainment losses can prevent substantial
improvement in performance by reducing dust resistivity.
H. Conclusions
The individual factors related to particulate reentrainment
have been discussed. However, at this point, the theoretical
relationships between the variation in these factors and the
reentrainment conditions in a precipitator cannot be definitive-
ly given. The measurements made with the pilot-scale precipita-
tor, while yielding insight into the phenomenon of reentrainment,
do not provide data that can be applied in an engineering manner
to full-scale units. At this time, it is necessary to approach
the problem of reducing reentrainment losses in any particular
installation on an individual basis. The descriptions given above
will provide guidance, but the particular solution to a given
reentrainment problem must be determined individually.
I. Bibliography - Section 3
1. Penney, G. W., "Adhesion and Cohesion in Dust Collection,"
progress report, Carnegie Institute of Technology,
Pittsburgh, Pennsylvania, Grant #RG-6402(C2) (January 1961).
2. Robinson, M., "Collection and Erosion Mechanisms in
Electrostatic Precipitation," precipitator evaluation in
terms of dimensionless parameters, Research-Cottrell, Inc.,
Bound Brook, New Jersey.
3. White, H. J., Industrial Electrostatic Precipitation,
Addison-Wesley Publishing Co., Inc., Reading, Mass. (1963) .
-------
-85-
SECTION 4. REFINEMENT OF PRECIPITATOR
MATHEMATICAL MODEL
A simplified computerized model of an electrostatic pre-
cipitator model was developed under a previous contract. This
model was developed using several simplifying assumptions for the
purpose of illustrating the utility of a theoretically based
mathematical approach to the problem of predicting precipitator
performance.
One purpose of the current contract was to further develop
and refine this simplified model to more closely predict the
behavior of actual precipitator installations. The first step
in this development process was to upgrade the model by incor-
poration of factors covered by present theory. The next step was
to modify the system of equations from the electrostatic system
of units to the more popular MKS units. And finally, factors
were added to include those functions that were omitted because
of the simplifying assumption in the initial model.
A. Collection Efficiency
The Deutsch-Anderson equation is a mathematical expression
that is used to relate the collection efficiency of electrostatic
precipitators to the physical parameters of collection electrode
area and volume flow rate of a precipitator system, together with
a term that is associated with the collectability of the dust
particles, designated precipitation rate parameter. This
expression is repeated from a previous section below
4.1
and is used both for empirical and theoretical considerations and
forms the basic equation for the precipitator mathematical model.
The Deutsch-Anderson equation has been used as an empirical
equation to develop proprietary data that serves as a measure
of the electrostatic collectability of various material. When
used in this sense, for an installation with a known volume flow
rate, collection plate area, and collection efficiency, a value
for w can be determined for that installation. In reality this
w is a measure of the performance for that installation utilizing
that particular dust, and is termed a precipitation rate
parameter.
1 - exp - - w,
SOUTHERN RESEARCH INSTITUTE
-------
-86-
B. Migration Velocity
The Deutsch-Anderson equation is also useful for theoret-
ical analyses. In this case, the term, w, applies specifically
to the electrical migration velocity that results from the
charged dust driven by the force resulting from the applied
electric field, which yields a velocity where the viscous drag
force from the gas balances the electrical force. The rela-
tionship that is used for describing this electrical migration
velocity given previously as equation 2.3 is
which becomes for field charging
c e0E0Epa 1
w = 2 — - -- 4.2
e+2 n
Thus, the theoretical migration velocity for field charging is
seen to be a function of the dielectric constants, e,e0; collec-
tion electric field, E0Ep; the gas viscosity, n; the particle
radius, a; and the charging time constant expression. These
factors are utilized in the computer model to predict a particle
migration velocity and a collection efficiency for each particle
size.
1. Particle charging
There are two charging mechanisms active in electrostatic
precipitation; field and diffusion charging. Even though both
mechanisms are active in charging, field charging is the domi-
nant charge mechanism for large particles while diffusion charging
is dominant for small ones. The computer model can be programed
to compute the charge for each mechanism and select the one that
yields the greater value. This technique will lead to some error
in the value of charge for those particle sizes where both mechaisms
can provide significant charge. The equations relating charge to
the various system parameters are given in Equations 4.3 and 4.4
for field and diffusion charging, respectively.
-------
-87-
q(t) = 12
e+2
,
ira2E0
4.3
2.
4irE0 akT
q(t) = - In
Electric fields
1 +
avN0ezt
4E0kT
4.4
The two electric fields used in Equation 4.2b refer to the
charging and collection electric fields in the precipitator. The
charging electric field used is the space and time averaged field
in the interelectrode space. Thus, a single value is determined
for the charging field that results from the applied voltage and
resultant current that is present in the precipitators. This
electric field is averaged over the interelectrode space and is
utilized for computing the charge for each particle size. The
expression for the electric field as a function of position is
given in Equation 4.5.
E(r) =
4.5
The collection electric field used in the computer model is
the time averaged value of the electric field adjacent to the
collection electrode surface. This field is described by the
same equation as the charging field. However, the numerical
value is different because the collection field is determined by
evaluating Equation 4.5 at a radius that corresponds to the col-
lection electrode spacing, while the charging field results from
a spatial average of Equation 4.5.
C.
Particle Size Consideration
The Deutsch-Anderson equation specifically applies to the
collection efficiency for particles with a given migration veloci-
ty. Since the migration velocity is a function of particle size,
a single value of migration applies to a single particle size.
SOUTHERN RESEARCH INSTITUTE
-------
-88-
Thus, for a particle size distribution that occurs in practice,
a range of migration velocities will be required. To approximate
this continuum of migration velocities, the inlet particle size
distribution is approximated by a first-order approximation such
as is shown in Figure 4.1. Within each increment, the total
material is considered to be effectively represented by a number
of particles with the same particle size that represents the
total mass of particles within that size increment. This solution
for discrete particles represents an approximation to the solution
for a continuous size distribution.
D. Particle Charging Time
The particle charging rate is related to the free ion
density in the vicinity of the particle being charged. The free
ion density is related to current flow at the point of charge.
As the particles acquire an electrical charge, this rather im-
mobile charge carrier causes a reduction in current, with the
resultant decrease in the free ion density. These factors are
included in the particle charge time expression in the computer
program.
E. Particle Reentrainment
At the beginning of this contract period, particle reentrain-
ment was considered to be associated with several variables from
the precipitator. Specifically, reentrainment was thought to occur
by:
scouring of the dust from the plates,
scouring of the dust from the hopper,
erosion from the dust layer falling
after rapping,
• erosion from dust clumps falling
voluntarily, and
erosion by impaction from particles being
collected (saltation).
The above effects were included in the computer equations as
terms that reintroduced collected material into the subsequent
sections of the precipitator. Thus, the concentration of materi-
al in the jth increment of the precipitator is expressed as the
-------
-89-
Material Within Interval, %
M M KJ to
o tn o in o m
1
1 1
1 1 1
1
MB
1 1 1 1 1
1.0 2.0 4.0 6.0 10 20
Particle Diameter, pm
40 60 80 100
Figure 4.1. Approximation of Inlet Particle Size
Distribution
SOUTHERN RESEARCH INSTITUTE
-------
-90-
concentration in the j-1 increment, less that collected in the
j-1 increment plus that reentrained in the j-1 increment due to
the various causes expressed in the equation shown below:
XNO(j) = XNO(j-l)-DXNO + K(V - V ) DXO + R(j"l) DXNO 4.6
where
XNO(j) = the number density of particles of size
increment j in the inlet of a section,
DXNO(j-1) = the number of particles in size range j
collected,
K(V - Vc) = the percentage of particles collected in
size range j that were reentrained by
velocity erosion, and
R(j-l) = the percentage of particles collected in
size range j that were reentrained due
to rapping.
These factors were included in the computer program prior
to the experimental investigations conducted with the pilot model.
The model work with reentrainment studies suggests that these
terms will require modification (see Section 3 - Reentrainment).
A system flow diagram with the pertinent equations is shown
in Figure 4.2. A computer program is included in Appendix 2.
F. Verification of the Precipitator
Mathematical Model
The mathematical model of an electrostatic precipitator
system, described in Section 2, was developed fundamentally for
use as a design and analysis tool. In order for this systems
model to be useful, the behavior of physical models must be
reasonably predictable. The operating data from a variety of
tests with the SRI pilot-scale precipitator were used as input
data for the model. A list of these data is shown in Table 4.1.
The computer model predicted a collection efficiency for each
test condition. The results of these computations are compared
with the measured efficiencies for each of these tests in Figure
4.3. The data taken from the pilot precipitator were without
rapping. Since erosion only occurs for gas velocities in excess
of about 30 ft/sec, the reentrainment factors were set to zero.
Thus, tne model was set to neglect reentrainment.
-------
E{r)
(0
0
c
X
PI
X
z
PI
XI
n
(I)
H
H
PI
Average
Value
Reentrain-
ment
Factor,R
Lw
i
y
s
*!
E
c
ri
vt
wi
V
n
A
z
R
Note:
n ^efficiency)
Figure 4.2 Computer System Flow Diagram
Hire Radius
Collector Radius
Roughness Factor
Relative Air Density
Breakdown Field Strength
Breakdown Voltage
Total Current
Corona Hire Length
Current/Length
Ion Nobility
Surface Area of Dust per cm* of Gas
Area of Plate
Current Density of Plate
Average Electric Field
Saturation and Present Charge
Dielectric Strength Relative
Radius of Particulate
Volume Flow Rate
Time Lapsed in Collector
Migration Velocity
Number of Particles
Precipitator Length
Gas Velocity
Efficiency, viscosity
Increment
Summation
Reentrainment Factor
Expressions for diffusion charging have
been added to the program, but since we
do not have sufficiently valid size
distribution data for particles smaller
than 1.3 ym, diffusion charging as yet
contributes no significant charge.
vo
H
I
-------
Table 4.1. Operating Data for the Pilot-Scale Precipitator
Test
No.
1
2
3
4
5
6
7
8
9
10
11
Collection
area
ft2
72
72
72
72
72
72
72
72
72
72
72
Gas flow
rate
ft'/min
1500
1800
654
804
654
1800
1800
1500
1200
1000
804
Applied
voltage
volts
36,000
36,000
37,000
36,000
36,000
36,000
35,000
34,000
35,000
36,000
36,000
Total
current
amps
0.0012
0.0012
0.0012
0.0012
0.0012
0.0012
0.0012
0.0012
0.0012
0.0012
0.0012
Dust load
gr/ft»
0.50
0.54
1.2
1.03
1.05
0.99
0.35
0.56
0.71
0.83
0.90
Wire radius
in.
0.049
0.049
0.049
0.049
0.049
0.049
0.049
0.049
0.049
0.049
0.049
Wire-to-Plate
spacing
in.
5
5
5
5
5
5
5
5
5
5
5
Efficiency
. %
Measured Computed
77.1
76.4
84.0
87.3
83.5
68.7
74.5
75.6
83.0
86.0
87.3
80.2
76.9
92.1
89.8
92.0
77.5
76.5
80.3
84.2
86.9
89.7
(O
-------
-93-
100
95
90
<*> 85
c 80
0)
•rl
U
•H __
»w 75
u-i
W
TJ
-------
-94-
The laboratory data for the model precipitator show good
correlation with the mathematically predicted values. Thus, the
model is concluded to be fairly representative of the pilot model
precipitator. As a means of determining how closely the computer
model approximates the behavior of full-scale field installations,
data from selected field tests where reentrainment was considered
to be minimal were used for comparison. The results of these
computations are shown in Figure 4.4, with the input data for
operating characteristics shown in Table 4.2.
G. Conclusions
The precipitator performance model requires additional
refinement before it can be used to reliably predict the behavior
of field installations. The areas considered to be important are
gas velocity, temperature and resistivity variations across the
face of the precipitator, as well as a better definition of reen-
trainment losses.
-------
-95-
10
10 20 30 40 50 60 70
Measured Efficiency, %
80
90 100
Figure 4.4.
Comparison of Computed and Measured
Collection Efficiency Under Field
Test Conditions
SOUTHERN RESEARCH INSTITUTE
-------
Table 4.2. Operating Data for Selected Field Installations
Test
No.
1
2
3
4
5
6
7
8
9
Collection
area
M
2300
2300
3310
3310
3374
3374
8623
3374
3310
Gas flow
rate
M'/sec
119.7
123.5
162.2
156.7
191.5
185.8
295.6
184.3
155.3
Applied
voltage
volts
44,300
40,600
33,700
37,800
32,300
34,000
34,200
33,900
32,500
Total
current
amps
0.83
0.76
0.37
0.92
0.14
0.14
1.605
0.14
0.29
Oust load
kg/M
0.0014
0.0018
0.0014
0.0012
0.005
0.006
0.007
0.005
0.0012
Hire radius
H
0.0014
0.0014
0.0014
0.0014
0.0014
0.0014
0.0014
0.0014
0.0014
Hire-to-Plate
spacing
M
0.114
0.114
0.114
0.114
0.114
0.114
0.114
0.114
0.114
Efficiency
, %
Measured Computed
95.6
96.7
92.2
95.6
89.9
89.2
98.3
88.8
78.4
95.2
93.6
90.4
95.7
96.9
96.4
99.8
97.2
87.0
vo
at
1
-------
-97-
SECTION 5. ECONOMIC COMPARISONS FOR
COLLECTION OF HIGH RESISTIVITY DUST
It is well known that a dust resistivity above a value of
about 2 x 1010 ft-cm is detrimental to precipitator performance.
The electric field in a dust layer of excessive resistivity
will, at normal current densities, exceed the breakdown strength
in localized regions, resulting in excessive sparking and back
corona. When high efficiency collection of high resistivity
dust is desired, it is necessary to effect changes in design or
operating conditions of precipitators collecting such material.
The options available are as follows:
enlargement at normal operating temperature,
conditioning at normal operating temperature,
high temperature operation, and
low temperature operation.
Each of these options requires an appreciable expenditure
above that required for electrostatic precipitation of a dust
with acceptable resistivity values at normal operating temperature.
This section presents a comparison of the economic and engineering
considerations pertaining to each method of overcoming the problem
of high dust resistivity. Factors not associated with resistivity,
such as design of mechanical and electrical components, are not
considered in this discussion.
A. Basis for Comparison
In order to establish a basis for comparison of the various
solutions to a high resistivity problem, a precipitator designed
for a 250 MW power plant with an acceptable resistivity at 270°F
will be used as a base point. It will be assumed that the pre-
cipitation rate parameter used for design of the base unit is
13 cm/sec, or 25.8 ft/min. This particular value was selected
because it was reported by Oglesby and Nichols1 to represent an
average design precipitation rate parameter for the electrostatic
precipitation of fly ash produced from coal-fired utility boilers.
1. Refer to Bibliography at the end of this section.
SOUTHERN RESEARCH INSTITUTE
-------
-98-
From the Deutsch-Anderson equation, the plate area required
for a given collection efficiency and gas flow is obtained by
The gas flow from a 250 MW plant at 270°F will be approximately
800,000 acfm. Thus, for a 99% collection efficiency
,a?fm In 100 = 143,000 ft2 or 179 ft2/1000 acfm.
,
25.8 ft/rain
The cost of such a precipitator can be estimated from a
recent compilation of precipitator costs.1 The average cost per
acfm of gas treated, for precipitators treating 500,000 to one mil-
lion acfm at a collection efficiency of 99% or greater, was repor-
ted as about $0.827 per acfm in the period 1965 through 1969. This
cost includes foundation, erection, and electrical work. If the
mid-point of this period, 1967, is used as a basis, the cost of the
precipitator, corrected to the first quarter of 1971 by means of
the Marshall and Stevens Equipment Cost Index, may be estimated by
Cost = 8.0 x 10s acfm x $0.827/acfm x
= $793,000 or $3.20/kW.
B. Enlarged Precipitator at Normal Temperature
It is theoretically possible to obtain desirable collection
efficiencies in precipitators whose efficiencies are limited by
high resistivity dusts simply by increasing the plate area. This
approach has been employed in Australia, with specific collecting
areas as high as 550 ft /1000 acfm in use.2 For a dust with very
high resistivity, however, the increase in plate area required
can be excessive to the point of impractical! ty.
Consider as an example a fly ash at 270°F with a resistivity
in excess of 10 12 fi-cm, which results in a precipitation rate
parameter of 3 cm/sec. The increase in plate area required over
the previously described base unit is proportional to the decrease
in precipitation rate parameter. Thus, to obtain 99% collection
efficiency
Area = i x 143,000 = 625,000 ft2 or 780 ft2/1000 acfm.
-------
-99-
The cost of such a unit may be estimated fronv data given by Oglesby
and Nichols1 in which erected costs of precipitators with col-
lection efficiencies greater than 99% are given as a function of
the volume of gas treated. Since the plate area required at a
given effective migration velocity and efficiency is directly
proportional to the volume of gas treated, it may be assumed
that the functional relationship between plate area and cost is
the same as that between gas flow and cost if the precipitation
rate parameter and efficiency have no bias with respect to volume
flow. Such an assumption should be sufficiently accurate for
cost estimation purposes.
The slope of erected precipitator cost on log-log-coordinates
as a function of volume flow was found to be 0.75. Thus, using
the reasoning indicated above, the cost of the enlarged unit is
given by
Cost = $793,000 (?o' S^ = $793,000 (4.38)
$2,400,000 or $9.60/kW.
This cost estimate does not include a penalty for the additional
space requirements resulting from the installation of a precipi-
tator with a plate area 4.38 times larger than would be required
under normal circumstances. These space requirements would, in
many cases, make such an enlargement at an existing installation
completely impractical.
C. Fly Ash Conditioning with Sulfuric Acid Vapor
Data obtained by Southern Research Institute under EPA Contract
CPA 70-1493 have established that a high dust resistivity problem
can be solved by injecting sulfuric acid vapor into the flue gas
ahead of the precipitator. Successful conditioning was obtained
with fly ashes of widely different compositions, and with systems
employing both anhydrous SO3 and 66° Be H2SO,t as sulfuric acid vapor
sources. The details and method of operation of different types
of conditioning systems are described in the final report on the
above contract .
In order to compare fly ash conditioning with other approaches
to solving a high resistivity problem, it is necessary to consider
both capital and operating costs of the conditioning system. A
cost comparison of existing conditioning facilities has indicated
that an anhydrous S03 system requires less capital cost than does
a system utilizing sulfuric acid. Therefore, conditioning cost
estimates presented here are based on such an installation.
SOUTHERN RESEARCH INSTITUTE
-------
-100-
For a 250 MW plant, the capital expenditure required for
installing an anhydrous SOs conditioning system is estimated as
$169,000. Total estimated operating costs, including capital
charges at 14.5%/year, labor and maintenance, and energy and raw
materials range from $57,100/year at an injection "level of 10 ppm
sulfuric acid vapor to $79,200/year at 20 ppm. For these two
injection levels, energy and raw material costs comprise $24,100/
year and $46,200/year of the respective total costs. The calcu-
lations were based on 7000 operating hr/year at full load.
An improvement in the precipitation rate parameter of 467%
has been reported as a result of conditioning for a unit which was
operating with a very high resistivity dust. It is therefore
reasonable to assume that the installation of a conditioning unit
would increase the effective w value from 3-5 cm/sec to 13 cm/sec,
which is the w assumed for the "normal" precipitator used as a
basis for comparison. Thus, the total capital expenditure would
be $793,000 for the precioitator plus $169,000 for the conditioning
plant, or $962,000.
D. High Temperature Operation
There is a growing trend in recent fly ash precipitator
installations to locate the precipitator ahead of the air heater
at temperatures in the neighborhood of 700°F. At this temperature,
the controlling conduction mechanism in the precipitated dust layer
is intrinsic or volume conduction, instead of the surface conduction
mechanism which predominates downstream from the air heater at lower
temperatures. Thus, the fly ash resistivity at high temperature is
not sensitive to the S03 or moisture content of the flue gas. Most
published resistivity data indicate that resistivities below the
critical 2 x 1010 fi-cm will occur above 600°F; therefore, high
temperature operation offers an alternative solution to high dust
resistivities resulting from combustion of low sulfur coals.
Another advantage of high temperature operation is that fouling
of the air heater by fly ash is reduced. However, in installations
burning high sulfur coal with a basic fly ash, it is probable that
removal of this ash ahead of the air heater would result in increased
corrosion rates of air heater cold end elements. For installations
in which coal and oil firing are employed, high temperature operation
minimizes oil ash handling problems.
The principal motivation for installing a hot precipitator is
to overcome the excessive size requirements brought about by high
dust resistivity at lower temperatures. There are, however, two
-------
-101-
factors which partially offset the reduced size as a result of
lowered resistivity: (1) a higher gas volume must be treated due
to the higher temperature, and (2) the decreased gas density results
in lower corona voltages and electric fields.
Design precipitation rate parameters, w, for hot precipitators
reported to Southern Research Institute range from 8.8 to 10.5
cm/sec. For the 250 MW unit discussed previously, the gas flow at
700°F is given by
800,000 x (460 + 700^ = 1,270,000 acfm.
460 + 270
To obtain a 99% collection efficiency, the plate area required using
8.8 cm/sec (17.3 ft/min) as w
A = X In 100 = If270,000 ln 10Q = 338,000 ft2 or
w 17.3
266 ft2/1000 Cfm at 700°F, or 423 ft2/1000 cfm
at 270°F.
It can be seen that, in spite of the partially offsetting factors
of increased gas flow and lowered electric fields, high temperature
operation results in design migration velocities sufficiently high
to allow a reasonable plate area. Recalling the earlier discussion
of precipitator enlargement at 270°F for a high resistivity dust
with a w of 3 cm/sec, a plate area of 625,000 ft2 was required for
the same efficiency.
Cost data relating to hot precipitators are rather limited
because of the small number of installations in operation. Table 5.1
summarizes the costs of hot precipitators reported by various
utilities to Southern Research Institute. The high total cost for
Plant F may be due to the fact that this installation was the first
of its type. As would be expected, the precipitators installed on
existing units are more expensive than those installed on new
installations. Plant D provides a basis for comparing the use of
a hot precipitator with the other options. It is a new installation
in the same size range as the 250 MW unit employed as a basis for
comparison, and the total installed costs are given. The costs of
ductwork should be included in comparing hot precipitators with other
options because of the additional ducting cost associated with such
an installation.
SOUTHERN RESEARCH INSTITUTE
-------
-102-
Table 5.1. Cost of Hot Precipitators Reported
to Southern Research Institute
No. of units
Plant and capacity
A 2 units,
285 MW total
B 2 units,
360 MW total
C 1 unit,
920 MW
1 unit,
350 MW
1 unit,
110 MW
1 unit,
1000 MW
Basis for cost
Precipitator, installation
& ductwork-retrofit, 99% eff.
Precipitator, installation
& ductwork-retrofit, 99% eff.
Precipitator, delivered &
erected, no ductwork-new unit,
99% eff.
Total installed cost, includ-
ing ductwork, new unit, 99%
eff.
FOB cost only
Precipitator (followed by
mechanical collector), and
erection
Complete installation, includ-
ing ductwork, ash handling and
foundations.
7.2-8.6
6.80
4.70
10-12
Plant D has a plate area of 564,000 ft2 and a median cost of
$7.90/kW, or $2,770,000. The cost of a 250 MW unit can be estimated
by again assuming that precipitator costs show a power rule vari-
ation with area, using an exponent of 0.75. Thus, for a 250 MW unit
with 8.8 cm/sec migration velocity at 700°F,
$2,770,000 (§f£$£>
0.75
$1,890,000, or about $7.55/kW.
Similarly, if a migration velocity of 10.5 cm/sec is used as a
design basis, the cost estimate would be
$2,770,000
0.75
= $1,660,000, or about $6.65/kW.
-------
-103-
E. Low Temperature Operation
The shape of the resistivity-temperature relationship in
stack gases indicates that low temperature operation can be used
as a solution to the problem of high dust resistivity. The dis-
cussion of low temperature corrosion and fouling in Section 6
has shown that most plants with high fly ash resistivity will not
be prevented from operating in the 220-240°F range as a result of
corrosion and fouling in the cold end elements of the air preheater.
Thus, the principal economic considerations are the limits of the
temperature range required for proper resistivity and the feasi-
bility of obtaining and controlling such temperatures at the pre-
cipitator inlet.
Figure 5.1 shows resistivity as a function of temperature
at Plant 6.1* The in-situ resistivity data indicate that accept-
able resistivity values are obtained in the 220-240°F range.
This plant was originally designed for a precipitator inlet
temperature of 300°F, and changes in the boiler were made to
lower stack temperature. The economizer surface in the boiler
was increased and provisions were made, using excess fan capacity,
to dump excess air blown through the air heater. The use of a
steam coil heater, which preheats combustion air prior to the
air heater, was also reduced. Another modification was the in-
stallation of a mixing baffle to overcome gas velocity and
temperature distribution problems at the precipitator inlet.
Figure 5.2 gives the flue gas flow plan.1*
It should be noted that in-situ resistivity measurements
are needed to determine the required range of precipitator inlet
temperatures. For a new installation, such data could be obtained
from combustion of the projected fuel in a similar full-size unit,
or from measurements at the plant under consideration in a
temporary duct prior to installation of the precipitator. Selec-
tion of a design temperature and precipitation rate parameter in
the absence of such data would be hazardous.
The effect of low temperature operation on boiler economics
varies considerably with local conditions. The selection of an
optimum stack gas temperature in the absence of secondary con-
straints, suph as space limitations and corrosion hazards, is a
trade-off between the savings resulting from high heat recovery
and the expense involved in installing and maintaining additional
heat exchanger surface. Some utilities operating stations with
low-cost, low sulfur lignite and sub-bituminous coals have de-
signed units with stack temperatures in the neighborhood of 300°F,
apparently concluding that the addition of more heat recovery
surface is not justified. Thus, the procedure of lowering stack
SOUTHERN RESEARCH INSTITUTE
-------
1x10l"
IxlO11
V IxlO11
c
•* 1x10 M
1x10*
100
200
O In-situ Tests
0.7% S in Coal (oven dried)
10.7% BSO in Flue Gas
Laboratory Tests
o
£fc
I
300
400 500
Temperature, *F
600
700
800
Figure 5.1. Resistivity as a Function of Temperature at Plant 6
-------
r-
Hot Air to
Boiler
c
x
m
a
m
V)
n
a
n
i
z
10
H
M
C
m
Boi
Ljungstrom
Air Beater
Precipitator
_ . _ . .
Dump to Atmosphere
s . I
J Alternate Return
^^ .*-_ «4^-^_l_ ^_1_A_ ^B
o
tn
Figure 5.2. Schematic of Flue Gas Flow Plan
-------
-106-
temperature by dumping excess combustion air is not an unreason-
able concept, particularly with low fuel cost. The fuel cost at
Plant 6, for example, is about $0.18/106 BTU. At higher fuel
costs, the theoretical economic optimum stack temperature
decreases. If an optimum design temperature is assumed for
higher fuel costs, a smaller relative amount of excess com-
bustion air needs to be dumped to lower stack gas temperature
to the degree dictated by resistivity considerations.
Although it is recognized that the procedure illustrated
in Figure 5.2 is not practical with some boiler designs, it can
be used to illustrate the comparative costs of the low temper-
ature solution to high resistivity. A precipitation rate para-
meter of 10.1 cm/sec, which is based on the experience of Plant 6,
was used as a basis of calculation.
For a 250 MW plant, the cost of a precipitator designed for
99% collection efficiency with a w of 10.1 cm/sec at 220 to 240°F
is estimated as $920,000. It is assumed that sufficient heat
exchanger surface is included in the boiler to lower stack gas
temperature from 300°F to 270°F, and that a mixing baffle and
provision for dumping excess air are added to allow a uniform
temperature of 220°F to be obtained. The cost of these additions
is estimated at about $164,000. If a permanent by-pass is added
to the system to reheat the stack gas, the total cost of the
alterations is estimated as $332,000. Thus, the total estimated
cost associated with the low temperature precipitator is
$1,250,000 with stack gas reheating provisions, and $1,080,000
without these provisions. Stack gas reheating using dumped
excess air may be desirable because of plume bouyancy problems.
No provisions have been made in the estimate for the
incremental costs associated with purchasing excess fan capacity.
However, the amount of excess air dumped relative to net com-
bustion air at Plant 6 was about 13%, with 80°F ambient air, to
achieve a 50°F reduction in stack temperature. The incremental
costs associated with the purchase of this amount of additional
capacity for a new installation are not likely to be significant
when compared with the other items.
The fuel loss and fan energy consumption cost estimates for
low temperature operation are given in Table 5.3. A 0.8%
degradation in boiler efficiency with 13% excess air dumped was
assumed for 7000 operating hours/year in the calculations.
Electric power costs were assumed at $0.005/kWh. These estimates
are rather conservative in that it is unlikely to be necessary
to dump 13% excess air for 7000 hours. Figure 5.3 gives the cost
of losses in boiler efficiency for various fuel costs.
-------
-107-
1.0
0.8
o
•H
-P
<0
•O
0.6
(U
Q
U
c
0)
•H
•H 0.
U
M
Q)
•H
•H
n o.
r
t
<0
•^
/ /
//
I
L
V,
s
/
I
^
^
JP
/
/
/
N
-fa ^
^
250 !•
Base
Load
Furne
/
iX^
1W Pic
Effic
Factc
tee H(
/
/
tnt
:ienc}
>r, 8C
;at Ir
/
/
X
r, 36?
)%
iput,
X
/
^
k
9450
x
x
BTU/J
/
.Wh
X
10 20 30 40 50
Annual Fuel Penalty, 103 dollars/yr
60
Figure 5.3. Cost of Boiler Efficiency Loss
SOUTHERN RESEARCH INSTITUTE
-------
-108-
F. Overall Comparison
The estimated total capital investment required for the
various precipitator configurations is given in Table 5.2.
Annual incremental capital charges given in Table 5.3 were
computed by taking 14.5% of the difference between the total
capital costs of the various options and the base precipitator
cost of $793,000. Thus, the totals in Table 5=3 represent the
estimated additional annual expenditures required for collection
of high resistivity dusts. It is assumed that no appreciable
differences exist between the operating and maintenance costs
of the precipitators employed in the different configurations.
This comparison indicates that the least desirable option
for very high resistivity dusts is a large precipitator at
normal temperatures. Even though such an arrangement can be
less expensive than a hot precipitator, considerably more area
is required, and the probability of meeting design collection
efficiencies is less due to the fact that the enlarged precipi-
tator must operate with resistivities outside the desirable
range.
The use of a hot precipitator, based on presently available
cost data, is likely to be more expensive due to higher capital
cost than either conditioning or low temperature operation.
There are certain advantages associated with this extra cost,
such as operational simplicity and relative insensitivity of dust
resistivity to changes in coal sulfur content.
The relative desirability of low temperature operation as
compared to conditioning obviously depends on such factors as
boiler and air heater design, fuel costs, and concentration of
conditioning agent required. The estimates in Table 5.3 indicate
that low temperature operation, even when air dumping is required,
can be an economically competitive solution to high resistivity.
The air dumping procedure is costly, but it allows for flexibility
in that increasing cold end temperatures is possible by dumping
less excess air. This may be desirable if it becomes necessary
to burn a fuel of higher than planned sulfur content, which could
lower resistivity excessively and cause low temperature corrosion
and fouling. The principal drawback with low temperature opera-
tion, for a new installation, is that rather comprehensive
in-situ resistivity-temperature data are required to establish a
firm basis for design. Also, if a plant is designed for operation
at 220-240°F exit temperatures with no provision for temperature
variation via air dumping, changes in fuel composition could cause
problems with both air heater and precipitator operation.
-------
n
i
in
Table 5.2. Estimated Total Capital Investment
99% Efficient Precipitator
250 MW unit
c
x
m
a
z
Configuration
Temperature, °F
Precipitation rate
parameter, cm/sec
Capital investment, $
Base
270
13
793,000
Enlarged
270
3.0
2,400,000
4.5
1,770,000
Condi-
tioned
270
13
962,000
Hot
600-700
8.8
1,890,000
10.5
1,660,000
Cold via
air dump
220-240
10.1
With
stack
reheat
1,250,000
Without
stack
reheat
1,080,000
H
O
vo
I
n
-------
Table 5.3. Estimated Incremental Annual Coot in Dollars - High Resistivity Dust
99.0% Efficiency - 250 MW
Load Factor » 0.8
Configuration :
w, cm/sec:
Precipitator area, ft2:
Injected SO, concentration:
Fuel COSt/lO* BTD:
Incremental capital charge
§ 14.5%/yr:
SO, plant:
Maintenance
Operations
Enlargement Conditioning Hot
4.5 3.0 13 8.8 10.5
417,000 625,000 143,000 338,000 284,000
10 ppm 20 ppm
142,000 233,000 24,500 24,500 159,000 126,000
8,500 8,500
24,100 46,200
Cold via air dump
10.1
175,000
With Without
stack reheat stack reheat
15* 25* 15* 25*
66,000 66,000 42,000 42,000
Air dump:
Fuel loss
Fuel gain (economizer)
Incremental fan energy
19,900 33,200
-11.900 -20,000
7,600 7,600
19,900 33,200
-11,900 -20,000
7,600 7,600
TOTAL:
142,000 233,000 57,100 79,200 159,000 126,000 81,600 86,800 57,600 62,800
-------
-111-
For a case such as that given in Table 5.3, in which
conditioning is economically competitive with low temperature
operation via air dumping, some consideration must be given to
the relative operational advantage of the low temperature pro-
cedure. Clearly, less maintenance and reliability problems
would be expected from the operation of a fan-damper combination
than with an S03 conditioning plant. The experience at Plant 6
with low temperature operation using this procedure has revealed
no unexpected problems.
In summary, the use of a hot precipitator, S03 conditioning,
or low temperature operation offers potential solutions to the
problem of high dust resistivity. Each option has undergone some
degree of successful operation in full-size installations, and
the relative economics of the options are strongly influenced by
local conditions. Low temperature operation and conditioning
require less capital expenditure than a hot precipitator, but
the capital saving is partially offset by the operating costs
involved with these procedures. For installations in which low
temperature operation via air dumping is compatible with basic
boiler design requirements and where resistivity data are avail-
able, this method offers the simplest, and in some cases, the
most economical solution to the resistivity problem.
G. Bibliography - Section 5
1. Oglesby, S. and Nichols, G. B., A Manual of Electrostatic
Precipitator Technology - Part II, prepared for the
National Air Pollution Control Administration under Contract
CPA 22-69-73 by Southern Research Institute, Birmingham,
Alabama (August 1970).
2. McLean, K. J., "Some Effects of High Resistivity Fly Ash
on Electrostatic Precipitator Operation," Paper II-A,
Proceedings of the Electrostatic Precipitator Symposium,
Birmingham, Alabama (February 1971) sponsored by the Air
Pollution Control Office, Environmental Protection Agency.
3. Southern Research Institute, Final Report on Contract
CPA 70-149 (A Study of Resistivity and Conditioning of Fly
Ash) to Division of Control Systems, Office of Air Programs,
Environmental Protection Agency (in preparation).
4. Berube, D. T., "Low Gas Temperature Solution to High
Resistivity Ash Problems," Paper II-E, Proceedings of the
Electrostatic Precipitator Symposium, Birmingham, Alabama
(February 1971) sponsored by the Air Pollution Control
Office, Environmental Protection Agency.
SOUTHERN RESEARCH INSTITUTE
-------
-112-
SECTION 6. LOW TEMPERATURE CORROSION AND FOULING
A. Introduction
Reducing precipitator inlet gas temperatures to the 220 to
250°F range offers a promising method of lowering the high fly
ash resistivity which occurs with the combustion of most low
sulfur coals. However, operation at such low temperatures has
causcu corrosion and fouling of air heater elements in some in-
stallations, while others have experienced no difficulty with
air heater exit temperatures as low as 220°F. An understanding
of the factors which cause corrosion and fouling problems is
therefore required if low temperature operation is to be con-
sidered as a means of lowering resistivity and increasing precipi-
tator performance.
The purpose of this study, then, is to relate corrosion to
fly ash and flue gas composition, fly ash resistivity, and temper-
ature. If a basic understanding of this relationship could be
gained, a decision on the advisability of low operating tempera-
ture could be made on the basis of fly ash and flue gas analytical
data.
B. Sulfuric Acid Occurrence in Flue Gas
1. SOX, H20, and I^SOi, equilibria
A knowledge of the S03 concentration in the air heater and
precipitator region of power plant exhaust systems is important
from a standpoint of both corrosion and fly ash resistivity. The
principal cause of corrosion in air heaters, and the most im-
portant factor in determining fly ash resistivity, is sulfuric
acid, which results from the reaction of SO3 with water vapor.
Most of the sulfur in power plant flue gases appears as
S02, with typical S03 levels ranging from 1 to 2.5% of the S02.
However, as Figure 6.1 shows, the equilibrium constant for the
reaction
S02(g) + Js02(g) = S03(g)
strongly favors the formation of SO 3 at temperatures below 1000°F
with 3% oxygen. This graph was calculated from data cited by
Hedley.1 The kinetics of the reaction are, of course, unfavorable
in the absence of a catalyst, but it is thermodynamically feasible
'Refer to the Bibliography at the end of this section.
-------
-113-
o
c/i
o
4J
8
I
2
0)
*
M
a
H
w
100
90
80
70
60
50
30
20
10
600
3% 0.
j i
700
800 900 1000 1100 1200
Temperature, °F
Figure 6.1. Equilibrium Conversion of S02 to S03
SOUTHERN RESEARCH INSTITUTE
-------
-114-
for SO 3 concentrations to exist at levels much greater than those
normally encountered. Ratios of SO 3 to SO2 as high as 0.1 have
been reported.2 Thus, since the formation of S03 is controlled
by catalytic effects as well as the amount of excess air present,
the concentration of S03 resulting from the combustion of a
particular fuel can only be estimated in the absence of direct
measurements .
The reaction between water vapor and S03 is given by
H20(g) + S03(g) = H2SO,,(g) .
Figure 6.2 shows the equilibrium conversion of SO3 to HjSO^ as a
function of temperature for a typical flue gas water concentration
of 8%. At temperatures below 400 °F, essentially all of the S03
present is converted to H2S04 at equilibrium. In contrast to the
formation of SO3, the formation of H2SOU occurs rapidly in the
thermodynamically feasible temperature range.3 Thus, all SO3
below the air heater in a power plant will exist as H2SO,,, either
in the vapor or liquid state. Since corrosion problems are
associated with the presence of liquid phase sulfuric acid, the
determination of the condensation characteristics of sulfuric acid
from flue gas containing sulfuric acid and water vapor is a neces-
sary step in evaluating the corrosion potential of a particular
stack gas.
2. Determination of the sulfuric acid dew point
Fly ash particles can influence the apparent dew point, or
saturation temperature of H2SOU in flue gas, but experience has
shown that one commits practically no error by neglecting the
presence of other gases and considering only the system sulfuric
acid - water. ** A thermodynamic analysis of the sulfuric acid -
water - flue gas system, ignoring for the present the effect of
fly ash, provides a theoretical basis for predicting acid dew
points and condensate composition from vapor-liquid equilibria
data.
For the case of ideal or quasi-ideal binary solutions, dew
points of vapor mixtures composed of the binary solution vapor
and noncondensible gases can easily be calculated from a knowledge
of the pure component vapor pressures as a function of temperature.
The H2SOW-H20 system presents special problems because:
the H2S04 and water undergo chemical reaction to
form the various hydrates of sulfuric acid, and
therefore the equilibrium relationships are
-------
100
o
CO
90
80
s
8 70
I-
I 50
8 40
20
10
200
-115-
300 400 500
Temperature, •?
600
700
Figure 6.2. Equilibrium Conversion of S03 to
HjSO,, at 8.0 vol % H20 in Flue Gas
SOUTHERN RESEARCH INSTITUTE
-------
-116-
strongly composition-dependent, and
H2S04 has a very low pure component vapor pressure,
thus making direct measurements extremely difficult.
The total vapor pressure of H2S(\ at low temperature is essential-
ly the partial pressure of water above the acid solution/ and
this is available from the existing literature. In order to
determine the dew point, however, the H.SO^ partial pressure at
low temperature must be known, and the literature lacks such
data."
As a result of the experimental difficulties encountered
in low temperature vapor pressure measurements, efforts have been
made to calculate the partial pressures from liquid phase thermo-
dynamic data. Abel5 was the first to derive a relationship en-
abling the calculation of H2SO^, H2O, and S03 partial pressures
from standard state values of enthalpy, entropy, and heat capacity;
and partial molal values of enthalpy, entropy, free energy, and
heat capacity. Muller,1* using Abel's calculated data, computed
dew points of gases with low H2SOW concentrations. Gmitro and
Vermuelen6 utilized thermodynamic data, which are claimed to be
more recent and more complete, to calculate H2SOU, S03, and H20
partial pressures from -50 to 400°C with solutions ranging from
10 to 100 weight percent H2SOI|. Snowden and Ryan3 have used
Gmitro and Vermuelen*s partial pressure data to construct a
chart which gives the dew point temperature of a gas as a function
of HjSO^ and H2O partial pressures. The composition of the acid
condensate occurring at a given dew point is also provided.
The dew points predicted from Abel's data are about 30°F
higher than those arrived at with Gmitro and Vermuelen's data.
The difference in these two works lies mainly in the data availa-
ble for the calculation of the partial pressures. Gmitro and
Vermuelen had access to much more accurate data and should have
obtained the more accurate results. However, their results do not
agree with direct dew point measurements by the condensation
technique, whereas Abel's partial pressures have been verified
in part by use of this method.
A suspect assumption common to predictions of acid dew
points based on both the Abel and Gmitro calculations is that
the vapor state is an ideal gas, and that the vapor solution is
also ideal. A gas mixture may behave nearly ideally volumetrical-
ly, but a component present in small amounts may exhibit signifi-
cant departure from ideality if that component is associated in
the vapor state.
-------
-117-
Among the limitations of some presentations in the litera-
ture of the Miiller correlation with 10% water vapor is that they
do not indicate the effect of variations in the water vapor concen-
tration on sulfuric acid dew points. The concentration of the
condensate is also not provided. Figures 6.3 and 6.4 were
prepared to present this information.
Figure 6.3 is a sulfuric acid - water dew point chart
prepared from Abel's H2SOU partial pressures and Greenewalt's7
water partial pressures above sulfuric acid solutions. The
partial pressure data were calculated by computer from the follow-
ing equations:
P = exp [2.303(A + | + !
and
PH = exp [2.303(A' - £')] 6.2
H2O T
where T is in degrees Kelvin and partial pressures are in mm Hg.
The constants in these equations are given by Abel and Greenewalt
for various sulfuric acid concentrations. It should be noted that
the range of uncertainty indicated by Abel for the constant B in
Equation 6.1 results in a dew point uncertainty of 8°F at 10% water
vapor.
The information contained in Figure 6.3, if it were accurate,
would be of value in assessing the corrosion potential of a flue
gas. The dew point temperature can be predicted from an analysis
of HjSO,, and water vapor content, and if the gas is cooled to some
temperature below the dew point, the equilibrium concentration of
condensate and the amount condensed can be obtained. It should
be pointed out, however, that the amount of condensate predicted
from the use of a dew point chart such as Figure 6.3 is actually
a prediction of the amount available for condensation. The amount
of condensate depositing on a metal surface may differ from the
chart prediction because of mass transfer considerations.
As an example of the use of the chart, consider a flue gas
containing 10 ppm HjSO,, and 10% HZ0. Condensation would occur
at about 275°F, and the condensate composition at that point
would be about 79% HjSO,, by weight. If the gas were cooled to
250°F, 85% of the H2S04 should be removed from the gas phase, and
an insignificant amount of the water vapor would also be condensed.
The condensation, therefore, follows the 10% water line, result-
ing in a condensate which would be the equilibrium composition of
SOUTHERN RESEARCH INSTITUTE
-------
-118-
6 8 10 20
Water Vapor, Vol %
40
60 80 100
Figure 6.3. Dew Point and Condensate Composition for Vapor
Mixtures of H20 and H2SOU at 760 mm Hg Total
Pressure (Abel and Greenewalt)
-------
-119-
100
80
60
40
I
i
s
o
w
20
10
8
6
220
Z
240
260 280
Dew Point, °F
300
320
Figure 6.4. H2S04 Dew Points for Typical Flue Gas
Moisture Concentrations
SOUTHERN RESEARCH INSTITUTE
-------
-120-
the condensate at 250°F, assuming the vapor phase is in equilibri-
um with the total liquid condensed. The composition change of the
liquid is small over the temperature interval given as an example,
ranging from 79% at 275°F to 75% at 250°F.
It is apparent from Figure 6.3 that a knowledge of water
vapor concentration is of fundamental importance. Appreciable
changes in this variable can have a rather significant effect on
the predicted sulfuric acid dew point, and if a gas is saturated
with HjjSO^, the condensate composition is determined by the water
vapor content and temperature. Thus, if a surface is maintained
at a known temperature lower than the sulfuric acid dew point,
but higher than the water dew point, the concentration of acid
condensate which occurs can be predicted from Figure 6.3 if the
water vapor content of the gas is known.
In addition to the procedure based on calculated partial
pressures, a number of efforts have been made to determine
sulfuric acid dew points using instrumental and chemical pro-
cedures. Two methods will be discussed briefly: the conden-
sation method and an electrical conductivity method.
The problem of measuring S0a concentration and acid dew
point has been studied since Johnstone8 examined the problem in
1929. Many papers9'20 have been presented which employ the
electrical conductivity method which Johnstone originated. The
British Coal Utilization Research Association (BCURA) designed
an instrument which has found widespread usage employing Johnstone's
concept. This instrument, known as BCURA dew point meter, has been
described in detail by Flint.9 It is a portable instrument which
measures the conductivity of a condensing film. The detector
element is glass and contains two electrodes mounted flush with
the surface. A tube inside the glass probe transports compressed
air which is used to maintain the glass surface of the probe at
the desired temperature. A thermocouple provides a readout of
the glass surface temperature.
If an electrically conductive film forms on the detector
element, a current will flow that is proportional to the magnitude
of the externally impressed voltage and the conductivity of the
condensing film. The current flow is measured with a microammeter.
A dew point is determined by inserting the detector element into
a gas stream with the instrument temperature held at some value
above the dew point. The element temperature is then alternately
increased and decreased slowly to establish the exact temperature
at which the increase in conductivity, and thus the dew point,
occurs.
-------
-121-
The condensation method is widely used for determinations
of SO3 in stack gases. The basic procedure employed consists of
pumping the flue gas through a condenser coil maintained below
the dew point of sulfuric acid, but above the normal water dew
point. A heated sampling probe is used to obtain the flue gas
samples, and a filter is inserted at the probe entrance to
exclude particulate matter. A fritted glass filter follows the
condenser to serve as a spray trap. When the sampling period is
concluded, the H2SO,, is washed from the condenser, and the
washings are collected and titrated.21
The condensation of a binary vapor mixture from a noncon-
densible gas is normally path-dependent, and the composition of
the vapor leaving a condenser is not fixed merely by stating that
the gas is saturated at a particular temperature. This is,true
because the degree of fractionation occurring during condensation
depends on conditions which exist in the condenser. For the case
of HjSO^-HjO vapor mixtures in flue gas, however, the water vapor
is in large excess, and no appreciable change in its concentration
occurs until the water dew point is reached. The composition of
the gas is, therefore, not path-dependent, and the state of the
system is fixed if the gas is saturated with H2S04 at a certain
temperature and water vapor content. As a result, the conden-
sation method can be used to obtain dew points of HjSO^-flue gas
mixtures. Since the gas leaving the condenser is saturated with
HjjSO^ at the condenser exit temperature, the concentration of
the exit vapor represents the dew point, or saturation tempera-
ture, of the gas.
Figure 6.5 presents the results obtained for flue gas dew
points as a function of H2S(\ (g) content by various investigators.
To make an exact comparison, all of the curves should be for a
gas of the same volume percent water vapor. However, reference
to Figure 6.3 will indicate that a variation in water vapor concen-
trations from 7 to 10% can cause only about a 5 to 8°F change in
the dew point. Taylor's results were obtained with the BCURA
dew point meter in a mixture of air, water vapor, and sulfuric
acid.20 Lisle's data were obtained using the condensation method,
again with a mixture of air, water vapor, and sulfuric acid.21
The dew point curves of Gmitro, Miiller, and from Figure 6.3, are
based on the previously discussed calculated partial pressures.
It is obvious from Figure 6.5 that, except for Lisle and
Sensenbaugh's checks of the data based on Abel's sulfuric acid
partial pressures (Miiller's data and Figures 6.3 and 6.4), there
is1 little agreement between the results of the various investi-
gators . The data obtained from calculated partial pressures agree
SOUTHERN RESEARCH INSTITUTE
-------
-122-
60
50
40
30
20
o
W
D P. Muller Calculated Data Points
10% H.O
?• s. Lisle
.9-9.4% Ha
A. A. Taylor
J. I. Gmitro
Abel and Greenewalt
160 180 200 220 240
Dew Point, °F
260
280 300
Figure 6.5. H2SO^ Dew Point Obtained by Various Investi-
gators .
-------
-123-
in form, which is to be expected since the equations used to calcu-
late the partial pressures are also of the same form. The nature
of the disagreement between the calculated dew point and those
obtained with the dew point meter suggest there is a sensitivity
problem with the instrument at low sulfuric acid partial pressures.
In view of the difficulties with calculations based on
liquid phase thermodynamic properties and the probable inaccuracy
of dew point meters at low acid partial pressure, we have con-
cluded that the only reliable method of correlating sulfuric acid
dew points with water and H2S(\ vapor concentration is a carefully
planned experimental program based on the condensation method
employed by Lisle and Sensenbaugh. In the absence of such data,
the dew points based on Abel's partial pressure data will be used,
since they have been verified in part by experiment and by the
operational experience of several power plants. A resolution of
the difference between the two sets of thermodynamic data (Abel's
and Gmitro's) is needed, but is beyond the scope of the present
study.
3. Condensation characteristics
As stated previously, the amount of acid condensate predicted
from the use of a chart such as Figure 6.3 as a result of cooling
to a temperature below the sulfuric acid dew point is a prediction
of the amount available for condensation. Figure 6.6 shows that
the predicted percentage of H2S(\ condensed increases and asymp-
totically approaches 100% as the temperature is lowered below the
dew point. However, peak values of acid deposition rates at temper-
atures between the water and acid dew points have been observed by
numerous investigators.
The occurrence of such a peak in the condensation rate may
be caused by a change in the diffusivity of the HjSO^ in the
region close to the condensing surface. The rate of condensation
is dependent on the diffusion rate of H2S(\ and water vapor to the
surface. Small droplets of H2SO,, will form in the cooled gas ad-
jacent to the surface, and the size of these droplets is likely to
increase with decreasing temperature. The growth of the droplets
would slow their diffusion to the surface and increase the proba-
bility that they would be carried forward in the gas stream. Thus,
a temperature can be reached at which the slowed diffusion becomes
dominant over the increased amount of condensate available for
collision with the surface. This explanation is similar to one
offered by Flint and Rear.1" A typical condensate rate curve,
obtained in a spiral condenser with a vapor mixture consisting of
7.5 vol % H20, 69 ppm HaSO^, and the balance air, is shown in Figure
6.7.22
SOUTHERN RESEARCH INSTITUTE
-------
-124-
100
240
260 280 300
Temperature, °F
320
Figure 6.6. Percent H.SO^ Available for
Condensation for Flue Gas of
100 ppm H2S<\ and 10% H20
Vapor (Calculated from
Figure 6.3)
-------
-125-
48
o
o
o
J-
o
M
«M
K
16
50
75
100 125
Temperature, °C
150
175
Figure 6.7. Variation in Condensation Rate with
Surface Temperature (From H. D. Taylor)
SOUTHERN RESEARCH INSTITUTE
-------
-126-
C. Factors Influencing Corrosion Rates
1. Acid strength
If a flue gas is known to be saturated with H2SO,, vapor
at a temperature below the acid dew point, it is possible to pre-
dict the initial condenaate composition as a function of the water
vapor partial pressure and temperature. Since data are available
in the literature concerning the corrosion rates of various
materials as a function of acid concentrations, it is of interest
to determine whether there is any relationship between corrosion
rates measured in flue gas and the acid condensate strength
predicted from a gas analysis.
A study of flue gas corrosion of low alloy steels by Piper
and Van Vliet23 provides data which illustrate the difficulty
encountered in predicting corrosion rates of metals from acid
condensate strength alone. The compositions of the low alloy
steel specimens used in this study are given in Table 6.1. The
corrosion tests were conducted by inserting specimens maintained
at known temperatures into stack gas produced from a pulverized-
fuel-fired steam generator. The average H2SO,, content of the
stack gas was about 30 ppm. Figure 6.8 gives the predicted
sulfuric acid condensate compositions for the range of stack gas
water vapor concentrations experienced during the study.
Figure 6.9 shows the average corrosion rate of selected
steel specimens as a function of predicted H2S(\ condensate
strength. The condensate strengths shown in Figures 6.8 and 6.9
were obtained from the computer printout of partial pressure for
the H2S04-H20 system, using Greenewalt's equation (Equation 6.2)
for the partial pressure of water over sulfuric acid solutions.
The widths of the surface in Figure 6.9 indicate the possible
acid concentrations at each temperature over the range of water
vapor partial pressures encountered in the stack gas.
Figure 6.10 is a plot of corrosion rates of steel given
by M. G. Font ana21* at 75°F as a function of acid concentration.
The corrosion rates for steel specimens immersed in acid are
orders of magnitude higher than those observed by Piper. Since
corrosion increases with temperature, the differences between
the Fontana and Piper data are even greater than indicated
because the latter's data were obtained at high temperatures.
-------
-127-
Table 6.1. Composition, Percent by Weight, Spectrographic
Analysis of Specimens Tested
(from Piper and Van Vliet10)
Name Mn Si Cu Ni Cr Zr
Cor-ten
NAX-A 0.85 0.90 0.07 <0.1 0.59 Present
NAX-B 0.82 0.79 0.29 <0.1 0.60 Present
NAX-C 0.53 0.54 0.07 <0.1 <0.1 Present
The low alloy steels used in the Piper study would not be
expected to exhibit greatly different corrosion rates in sulfuric
acid solution than the ordinary carbon steel on which Fontana's
data are based. Therefore, the orders of magnitude differences
in corrosion rates indicated are largely a reflection of the
differences in environment between the two situations. Another
contributing factor is the parabolic nature of the corrosion-time
relationship usually found in corrosion work. Thus, because of
the effects of fly ash and condensate deposition rates, it is not
practical to predict or correlate corrosion rates of materials in
flue gas solely on the basis of equilibrium condensate compositions.
2. Acid deposition rate
The corrosion rate of metal surfaces in flue gas at tempera-
tures well above the water dew point is more strongly related to
the amount of condensate deposited than to the concentration of
the condensate. Consider, for example, a steel surface at 260°F
exposed to a flue gas with a bulk gas phase concentration of
10 ppm sulfuric acid vapor and 10% water vapor. A condensate
strength of 77% HjSO,, would be expected, and if fly ash neutral-
izing ability is ignored, some non-zero rate of corrosion would
be expected. If the same steel surface were exposed to a similar
flue gas with 80 ppm sulfuric acid vapor, the predicted conden-
sate strength would remain at 77% HjSO^, but the corrosion rate
would be greater because of the increased quantity of acid conden-
sate depositing on the metal. In both cases, decreasing the metal
surface temperatures to a value approaching the water dew point
(100 to 110°F) of the flue gas would result in increased corrosion
rates because of the highly corrosive dilute acid formed at these
temperatures.
SOUTHERN RESEARCH INSTITUTE
-------
-128-
310
290
270
250
tu
\ 230
I
2 210
M
f 190
170
150
130
110
20
30
7.
5 vol
% H
5.1
vol 1
//
B0
40 50 60 70
Weight % HSSO% Condensate
80
so
Figure 6.8.
Equilibrium Sulfuric Acid Condensate
Composition
-------
-129-
s,
§
•H
10
o
O
0
tt)
0»
Id
115°F
7.5 vol % H20
I
5.1 vol % H20
MAX - A, - B,
and Cor-Ten
- C,
S><£;
161°F
25
30
35 40 45 50
Condensate Strength, Weight Percent
55
60
65
Figure 6.9,
Corrosion of Steel in Flue Gas as a Function of
Calculated HjSO^ Condensate Strength (Corrosion
Data from Piper and Van Vliet; H2SO,, Data from
Greenewalt)
SOUTHERN RESEARCH INSTITUTE
-------
-130-
•
H
i
§
•rt
8
M
8
10 30
60 70
Weight Percent H2SO,.
80
90
100
Figure 6.10. Corrosion of Steel as a Function of
HSO, Concentration at 75°F
-------
-131-
The temperature at which the maximum condensation rate of
acid occurs has been correlated with the temperature of maximum
corrosion in flue gases. Figure 6.11 was taken from a study by
G. G. Thurlow, in which an air-cooled corrosion probe was exposed
to flue gas produced from burning a 0.8% sulfur coal.25 The
rate of sulfate deposition shows a peak at the same surface temper-
ature as the corrosion rate. This peak rate effect is often not
observed with coal firing, but Black18 and Clark19 have found
this phenomenon quite useful in correlating corrosion of air
preheaters in oil fired units. The sulfur content of the fuel
used in these studies ranged from 1.4 to 4.0%.
Black and Clark's work was done with the BCURA dew point
meter, and the peak rate of acid deposition was indicated by a
peak rate of increase in current, measured as microamps per
minute. The maximum corrosion rate is expected to occur in a
regenerative air preheater at the point where the average metal
temperature corresponds to the peak rate temperature indicated by
the BCURA meter. By superimposing a plot of the dew point meter
readings in the region of the peak over lines of average metal
temperature, it was possible to match the peak rate temperatures
with actual corrosion experience.
The above authors also found that the BCURA indication of
the acid dew point was a poor indicator of flue gas corrosion
potential, particularly when oil and gas mixtures are fired. This
observation is not surprising since, as Figure 6.3 indicates, the
dew point alone does not specify how much acid is available for
condensation. The accuracy of the dew point meter may also be an
important factor, because instructions for use of the meter state26
that changes in dew point readings of less than 20°F are not to be
regarded as significant. Referring again to Figure 6.3, a change
of dew point at 10% water vapor from 270 to 290°F indicates a 370%
increase in the HjSO^ vapor content of the flue gas.
Studies conducted by Lee, Freidrich and Mitchell,16 in
which the BCURA meter was employed with flue gas produced from
burning low sulfur lignite, showed that the meter was unable to
detect acid dew points with low sulfur coals. 'In one experiment,
no acid dew point was detected by the meter in the presence of
sulfuric acid vapor levels as high as 27 ppm. The author's
explanation for this is that the condensed acid was completely
neutralized by basic constituents in the fly ash.
Thus, since high fly ash resistivity is associated with low
sulfuric acid vapor concentrations, the BCURA meter is not likely
to be of value in assessing the low corrosion potential associated
with a flue gas containing high resistivity fly ash.
SOUTHERN RESEARCH INSTITUTE
-------
-132-
20
If
o
M
20
Uio
(0
160
I
I
200 240 280 320
Surface Temperature, °F
360
Figure 6.11. Variation of Condensation and Corrosion
with Surface Temperature (Data from
Thurlow)
-------
-133-
3. Fly ash alkalinity
Although fly ash can cause severe plugging problems in air-
heaters, it is well established that alkaline ashes can neutral-
ize a portion of the SO3 and H2S(\ occurring in stack gases,
thereby acting to reduce corrosion. Lee provides data which
illustrate the interaction of acid condensate with fly ash.
Figure 6.12 illustrates the effect of surface temperature on acid
condensation rate when burning a 7% sulfur coal with 3% excess
oxygen. The RBU plotted on the y axis in the upper graph is a
measure of the rate of acid condensation when the BCURA dew point
meter is maintained at the indicated temperatures. Data for the
lower graph were obtained by isokinetically sampling the flue gas
and collecting the fly ash and acid condensate in a Teflon vial
maintained at 180, 212, 245, and 275°F. The contents of the vial
were then extracted, and the extract was analyzed for acid or base
content. If the extract pH was less than 7, the solution was
titrated with sodium hydroxide, and the results were reported as
a negative cation content. If the extract was basic, the solution
was titrated with HCl, and the results were reported as an excess
cation content, indicating that the condensed sulfuric acid had
been completely neutralized.
The acid neutralizing ability of fly ash with various base
contents is illustrated in Figure 6.13 for a flue gas with a
typical dust loading of 5 gr/scf. The parallel lines each repre-
sent a base content of fly ash, expressed as milliequivalents
reactive base per gram fly ash. Data obtained on Contract CPA 70-
149 (A Study of Resistivity and Conditioning of Fly Ash) indicate
that fly ash produced from burning a high sulfur coal has as much
as 0.6 milliequivalents soluble base (1.7% CaO) per gram fly ash.27
This quantity of base is capable of neutralizing 80 ppm H2SC\ in
the gas phase, assuming that the flue gas has an ash concentration
of 5 gr/ft9. This is not to say that complete neutralization will
occur, since the degree of neutralization obtained in the flue gas
is a function of the rate of transfer of HjSO^ to the fly ash
particles and the rate of reaction occurring on the particle
surface.
4. Hydrochloric acid
Sulfur, chlorine, and alkali metal compounds are associated
with high temperature corrosion in coal-fired boilers, but low
temperature corrosion is usually thought of only in terms of
sulfuric acid. However, metals with surface temperature below the
moisture dew point would be subjected to HCl attack if the chlorine
SOUTHERN RESEARCH INSTITUTE
-------
-134-
600
500 -
4001P-
300 —
200 —
100 -
IdO
220 240 260
Surface Temperature,*F
280
300
Figure 6.12. Variation in Rate of Acid Buildup
(RBU) and Excess Cation Content of
Fly Ash as a Function of Surface
Temperature. Coal Contains 7% Sulfur
with 3% Excess 02 (Data from Lee)
-------
-135-
100
•o
U)
5
fl
CO
01
rH
J3
(0
B
0)
01
o.
10
1.0
0.1
0.1
10
I I I
100
ppm
Figure 6.13,
Consumption of the Available Base on Fly Ash as
a Function of the Concentration of Neutralizing
Acid in Flue Gas with 5 gr/scf Flv Ash
SOUTHERN RESEARCH INSTITUTE
-------
-136-
content of the coal is converted to HC1. Although not all of
the chlorine in coal appears as Nad, it is of interest to
examine the chemical reactions undergone by NaCl in the com-
bustion process. The following discussion is taken from a study
by Halstead26 in which chloride and sulfate deposit formations
were examined with probe tests and by thermodynamic calculations.
In pulverized coal firing, the NaCl .can be expected to
evaporate and undergo some degree of vapor phase hydrolysis.
NaCl(g) + H20(g) = NaOH(g) + HCl(g)
The reactions of the chloride and NaOH with S02 to form Na2S(\
are, however, of greater importance. They are
2NaCl(g) + H20(g) + S02(g) + %02(g) = Na2SOw(g) + 2HCl(g)
2NaOH(g) + S02 (g) + »s02 (g) = Na2SOlf (g) + H20(g)
Halstead calculated the equilibrium partial pressures of
NajSO^ and NaCl in flue gases produced from burning the coals
listed in Table 6.2 at 5% 02 excess, stoichiometric 02, and 2% 02
deficient. These calculations, together with deposition studies
conducted with a cooled probe, indicate that almost total con-
version of NaCl to Na2S(\ takes place with 3 to 5% excess oxygen
in large boilers with good mixing of fuel and air. With lower
oxygen levels, and when poor mixing and short residence times are
encountered, the conversion of NaCl to Na2SOlf may be incomplete.
Table 6.2. Sulfur and Chlorine Concentrations
in Flue Gas (from Halstead16)
Sulfur in Chlorine in Sulfur compoundsa Chlorine compounds3
coal coal in flue gas in flue gas
% % vol ppm vol ppm
0.8 0.8 750 680
1.2 0.4 1100 340
1.8 0.07 1700 60
a. Calculated by assuming complete volatization of all sulfur and
chlorine in coal and one atom of sulfur or chlorine present in
each gas molecule.
-------
-137-
Thus, it can be seen that significant concentrations of
HC1 are likely to result from the combustion of chlorine-contain-
ing coal. The subject of HCl corrosion in flue gases has received
comparatively little attention in the literature because it is
not likely to occur unless temperatures near the water dew point
are encountered. Air preheater elements, however, can drop below
the moisture dew point if excessive water vapor, such as would
occur from a steam leak, is present.
Figure 6.14, taken from a study by R. W. Rear,29 illustrates
the effect of HCl in a flue gas on corrosion of a test probe.
This experiment was conducted using an apparatus which produced
a synthetic flue gas by addition of S02 and C12 to the fuel supply
of a small laboratory burner. Analysis of the flue gas indicated
that all chlorine was converted to HCl, resulting in 400 ppm HCl
by volume. It should be noted, however, that corrosion could be
caused by the presence of chlorine gas. The assumption that
Figure 6.14 is an illustration of the effect of HCl gas is
therefore dependent upon Rear's conclusion that all chlorine is
converted to HCl in the burner flame. The S03 , or H2SOlj, content
of this gas was reported as 36 ppm. The temperature at which the
corrosion rate accelerates corresponds to the water dew point of
the synthetic flue gas, which is about 7% by volume water vapor.
When the metal surface temperature is above the water dew point,
the presence of HCl has no effect on corrosion, but it can be
seen from Figure 6.14 that drastic increases in corrosion occur
due to HCl as the metal surface falls below the water dew point.
The corrosion probe was exposed for a 30-minute period in each
experiment.
Data obtained by Piper and Van Vliet23 confirm Rear's
results. Piper's data were obtained by exposing metal condensers,
which could be cooled to selected temperatures, to flue gas
produced from burning a 0.066% chloride coal. Analysis of the
flue gas showed that HCl concentrations ranged from 16 to 82 ppm,
and the sulfuric acid vapor concentration averaged 30 ppm. The
relative rates of corrosion of low alloy steel specimens maintained
at 161, 141, 115, and 87°F for 2-month exposures were 1, 1, 3, and
66, respectively. The water dew point of the flue gas during the
exposure period ranged from 91 to 104°F. It is thus apparent that
the rate of attack greatly accelerated below the water dew point.
This corrosion is a result of both H2SO,, and HCl, but the im-
portance of the effect of HCl is indicated by the fact that at
the water dew point, the chemical equivalents of chloride exceeded
those of sulphate. Another important observation of the Piper
study was that a vitreous enamel coating on Cor-Ten, which was
used in a pilot-plant air preheater, was considerably attacked at
temperatures below the water dew point.
SOUTHERN RESEARCH INSTITUTE
-------
-138-
0.1% of SO, + 0.02% of Cl, in
flue gas a
0.1% of SO. in flue gas
30 40
60 70 80 90 100 110 120 130 140 150
Surface Temperature, °C
Figure 6.14. The Effect of Chlorine Addition on Corrosion
of Mild Steel in a Synthetic Flue Gas (from
R. W. Kear)
-------
-139-
Since high resistivity fly ash usually occurs in the absence
of sulfuric acid vapor, it is of interest to consider such a
situation in which appreciable concentrations of HC1 exist. Piper
analyzed the vapor-liquid equilibria data for the system HC1-H20,
and concluded that, with an HCl vapor concentration of 82 ppm,
the hydrochloric acid dew point would be 7°F above the water dew
point. A similar analysis of the water-S02 system indicated that
the sulfurous acid dew point, for a stack gas with about 1900 ppm
S02 and typical water vapor concentrations, would be the same as
the water dew point.
D. Fouling of Low Temperature Surfaces
Deposit formation, or fouling, in air heater elements is a
combination of chemical and physical processes. At 600 to 700°F,
which is the range of temperature normally encountered at the
hot end of regenerative air heaters, the saturation partial
pressure of the mineral components of fly ash is extremely low.
Thus deposit formation in this region is not a result of conden-
sation from the vapor phase, but is instead a mechanical process
in which slag and refractory material are carried by the flue gas
into the air heater elements. These particles can lodge within
the passages of hot end elements and thereby accumulate additional
deposits of finer dust particles.30 Procedures are reported in
the literature for removing such deposits.
If the flue gas contains appreciable amounts of H2SOU,
corrosion and deposit buildup will occur simultaneously in the
cooler regions of the air heater.31 The following reaction will
occur on steel surfaces which are below the H2SOi» dew point.
Fe + H2SO,, -" FeSO,, + H2
The ferrous sulfate can then oxidize to form ferric sulfate.
4FeSO,, + 21^50,, + 02 -»• 2Fe2(SOj3 + 2H20
An extensive study of regenerative air heater deposits by
the Bureau of Mines32 found that deposits built up in thickness
at the cold end of the air heater, and that this area was the
principal region of corrosion and destruction of the element.
All deposits found in this area exhibited the following charac-
teristics: partial solubility in water, presence of sulfates,
and acidity. The solubilities in water of these deposits varied
over a wide range 13 to 98%. Deposits with highest solubities
were found on preheater test plates which were most severely
SOUTHERN RESEARCH INSTITUTE
-------
-140-
attacked by acid. Some of the variations in deposit solubility
were attributed to variations in the ability of the deposits to
trap fly ash.
Reaction of the ferrous and ferric salts formed during
corrosion with alkaline compounds sometimes used in washing air
heaters can produce compounds that will result in additional
fouling. Ferric sulfate, for example, can undergo the following
reactions.3 *
Fe2(S(\)3 + 3Ca(OH)2 (lime) * 2Fe(OH)3 + SCaSO,,
Fe^SOjj + 6NaOH •* 2Fe(OH)3 + 3Na2SO^
Fe2(SOu)3 + 3Na2C03(soda ash) + 3H20 •*• 2Fe(OH)3 + 3Na2S01|
+ 3C02
The Fe(OH)3 (ferric hydroxide) is undesirable because it is a
sticky, gelatinous precipitate which can cause severe fouling.
The above reactions indicate that, in washing air heater elements
or tubes, removing the soluble sulfates with a neutral water wash
is desirable prior to a caustic wash.
It is important to note that deposit formation can occur in
air heater elements in the absence of significant amounts of H2SO,,,
Chemical analysis of deposits from air heaters installed in some
lignite-burning power stations has revealed no chemical evidence
for deposition." In one instance, moisture from steam cleaning
action was found to be responsible for trapping ash deposits.
Deposits formed in this manner are similar to cement and very
difficult to remove.
In the absence of moisture and acid condensate problems, the
nature of the fouling mechanisms discussed herein suggests that
lowered cold end temperature would not result in increased deposit
formation.
E. Laboratory Corrosion Studies
Samples of fly ash were obtained for corrosion studies from
the precipitator hoppers of two plants with high dust resistivity
problems. These ash samples have widely different soluble base
contents, as can be seen from Table 6.3. Sulfur contents of the
coal burned in the two plants range from 0.6 to 1.0%. Laboratory
experiments were conducted to determine whether deposited layers
of these ashes exhibit differing capabilities for neutralizing
-------
Table 6.3. Fly Ash Properties
Neutral (from Plant 1)
As received
Following experiment
(Experiment 4, Table 6.4)
pH of
suspension
6.70
1.69
Soluble
sulfate
wt %
0.31
23.4
Soluble base
as CaO
meq/g wt %
0
0
0
0
Mass median
particle
diameter, p
38
Basic (from Plant 6)
As received
Following experiment
(Experiment 3, Table 6.4)
12.25
8.72
1.2 2.7 7.6
23.1 Not deter-
mined
18
c
X
PI
a
z
a
m
m
m
a
O
H
H
PI
-------
-142-
acid and inhibiting corrosion.
A schematic diagram of the apparatus used for the experiments
is given in Figure 6.15, and data obtained are presented in Tables
6.3 and 6.4. The corrosion specimen was a 1 in. diameter mild
steel disc, and the amount of corrosion occurring as a result of
exposure to H2SC\ was determined by measuring the weight loss.
The experiments in Table 6.4 can be divided into two groups.
In Experiments 1 through 4, the acid condensation rate on the disc
was relatively low, but high condensation rates were achieved in
Experiments 5 through 10 by increasing the strength of oleum used
as an SO3 vapor source and by lowering the temperature of the
water bath. Water vapor concentrations of 2 - 2.5% by volume were
provided by the water spargers. Since the air streams bearing
H2O and SO3 vapor mix in the heated glass "T," a saturated mixture
of air and H2SO^ is formed, and the condensation rate will depend
on the temperature of the condensing surface and the concentration
of HjSO^ in the gas phase. For both sets of experimental con-
ditions, an examination of the corrosion rates (meg basis) and
acid deposition rates in Table 6.4 shows that an excess of acid
was present with respect to the amount of iron corroded in all
experiments.
For the experiments with fly ash, the ash was deposited in
the sample container in such a manner that the disc was covered
to a thickness of approximately 0.2 mm. Acid did not sufficiently
penetrate the ash to reach the underside of the disc in Experiments
3 and 4, and the penetration rates were calculated on the basis of
one side only. Corrosion was observed on both sides in all other
experiments; therefore, the total area of both disc surfaces was
used as a basis of calculation.
A comparison of data from Experiment 3 with those from
Experiment 4 indicates that the basic fly ash was more effective
in reducing corrosion than the neutral ash. The equilibrium pH
values of the ash samples prior to and following these experi-
ments are given in Table 6.3. As would be expected, the neutral
ash slurry is much more acidic than that of the basic ash after
both have experienced an equivalent sulfate gain due to H2SO,,
condensation. The fact that the basic ash produced a pH greater
than 7 following the experiment shows that it was capable of
neutralizing all of the condensed acid. Complete neutralization
did not occur until the acid-ash mixture was slurried in water,
however, as evidenced by the measurable degree of corrosion which
occurred in Experiment 3.
-------
Rotometer
X
Room
Air
5
s
o
I
I
•V1 •
Spargers
3 x 3/4 in. dia x 6 in. Drying
Tubes with 8-Mesh Drierite
Thermometer
Magnetic Stirrers at Low Speed
(=120 rpra)
Mist Eliminator
Thermocouple:1
Steel Disc
Charcoal Test
Meter
I
M
U>
Vacuum Pump
Figure 6 ,,15. Schematic Diagram of Apparatus Used in Corrosion Experiments
-------
Table 6.4. Corrosion Rate Experiments
H,SO,,
vapor
Experiment generator,
No. % Acid used
1 104
2 104
3 104
4 104
5 107
6 107
7 107
8 107
9 107
10 107
Condensate
Duration compositior
hr wt % HzSO,
2
1
2
2
1
1
1
1
1
1
.0 56
.9 52
.1 —
.0
.0 36
.0 40
.0
.0
.0
.0
) Temperature f °F
i
383
379
388
412
349
374
388
392
390
388
water
bath
78
84
79
79
37
39
37
37
36
39
disc
surface
—
—
—
90
77
95
86
86
81
Condensate
rate
roeq/hr
~
1.3
1.5
1.6
5.0
6.0
10.1
8.6
12.4
12.4
Apparent corrosion rate
mq/hr
1.05
0.90
0.20
0.40
32
32
18
17
53
41
meq/hrb
0.056
0.04B
0.011
0.022
1.7
1.7
0.97
0.91
2.B
2.2
mlls/yr
46
39
17a
34a
1400
1400
790
740
2300
1800
Reacting
Ash with disc
layer wt *
None
None
Basic
Neutral
None
None
Basic
Basic
Neutral
Neutral
~
3.7
0.7
1.4
34
28
10
11
23
18
a. Based on exposure of one side of disc to acid rather than both sides as in all other runs.
b. Assuming formation of Fet(SO,),.
-------
-145-
For the experiments with low acid condensation rates, both
the neutral and basic ash deposits reduced the weight loss rate of
the disc, but the penetration rate calculated for Experiment 4
(neutral ash) is not significantly different from those of
Experiments 1 and 2 (no ash). These results are to be expected,
since the neutral character of the material from Plane 1 indicates
that any corrosion inhibiting value which it exhibits is likely
to be the result of physical rather than chemical factors.
High corrosion rates were obtained in Experiments 5 through
10 due to increased acid condensation rates and decreased conden-
sate composition. The high percentage of HjSO,, reacting with the
disc in these experiments is an indication of the greater corrosive-
ness of acid in the 36 - 40 wt % range. Some difficulty was
encountered in maintaining constant experimental conditions, as
indicated by variations in the disc surface temperatures and the
acid condensation rates. Once again, the data suggest that the
neutral ash has little corrosion inhibiting value, but signifi-
cantly lower corrosion rates were obtained with the basic substance.
In contrast to the conditions of Experiment 3, an excess of acid
was present with respect to the base content of the ash layer for
Experiments 7 and 8. If it is assumed that the same amount of
base reacts per unit weight-of basic ash in both sets of experi-
ments, it can be shown that less than 30% of the condensing acid
could have been neutralized in Experiments 7 and 8. The princi-
pal mechanism by which corrosion rates were reduced in Experiments
7 and 8 appeared to be the formation of a cement-like deposit
which reduced the amount of acid reaching the metal surface. Such
deposits would be likely to cause plugging of air heater elements
in plant operation.
Generalizations concerning the direct effect of basic and
neutral fly ashes on corrosion rates from these experiments are
hazardous because of the complex nature of the corrosion process.
However, it is possible to draw some conclusions regarding the
interaction of the fly ash with condensing acid.
The reduced corrosion rate obtained in Experiment 3 indi-
cates that the basic fly ash from Plant 6 neutralized a major
portion of the acid as it condensed. This is an important obser-
vation because data obtained under Contract CPA 70-149 has
revealed the presence of unreacted acid on the surface of fly
ash containing amounts of water soluble base substantially in
excess of the apparent surface acidity. Thus, basic ash deposited
on metal surfaces could conceivably present an acidic, and hence
corrosive, environment to a metal surface and exhibit little or
no neutralizing capability. A layer of CaSOi,, formed by reaction
SOUTHERN RESEARCH INSTITUTE
-------
-146-
between H2SOU and CaO, apparently can prevent the underlying solu-
ble base from being utilized. The ash from Plant 6 contained
appreciable sulfate when received from the precipitator hoppers
(1.2%), but the experimental data presented here indicate that
the sulfate did not present an impermeable barrier to the liquid
condensate.
The neutral ash from Plant 1 would not be expected to
provide a significant degree of protection from condensing acid,
and the experimental data tend to confirm this. However, even a
neutral ash can reduce the amount of acid available for corrosion
in a flue gas by adsorbing S03. The small amount of sulfate
(0.31%) present on the ash from Plant 1 when received indicates
that some adsorption of S03 at high temperatures occurred. The
operating temperature of the precipitator at Plant 1 is about
320°F, which is well above the H2S(\ dew point.
In conclusion, then, the data from these experiments indi-
cate that a basic ash such as that from Plant 6 can be of signifi-
cant value in neutralizing condensed acid and reducing air heater
corrosion rates. However, in the presence of an excess of
condensing acid, serious deposit formation problems could be
expected. The neutral ash was of little or no apparent value in
reducing corrosion rates, but it exhibited a lesser tendency to
form cement-like deposits than did the basic material. The most
important benefit to be expected from the presence of a basic fly
ash from the standpoint of corrosion is the consumption of SO3
by the basic material in the high temperature region prior to the
air heater. Unfortunately, this also creates a high resistivity
problem for precipitators operating in the 300°F range.
F. Summary of Field Experience
1. Discussion with Air Preheater Company
A conference with Mr. Norman D. Clark, Manager of Technical
Services of the Air Preheater Company, was held concerning low
temperature corrosion and fouling problems. Mr. Clark has
authored a number of papers concerning corrosion of regenerative
air heaters in various types of service.
The Air Preheater Company is of the opinion that low temper-
ature air heater corrosion and fouling will not be a limiting
factor in low temperature operation with low sulfur coals. No
problems with corrosion have been reported from installations
burning low sulfur coals, but fouling, caused principally by
moisture from steam leaks or steam cleaning, can cause plugging
-------
-147-
problems. Usually, no effort is made to maintain air heater ele-
ment temperatures above the acid dew point due to the loss of
boiler efficiency which this would involve, but the water dew
point (~110°F) is to be avoided because of potential fouling and
corrosion from acids other than sulfuric acid.
Typical air heater element lifetimes were given as follows.
Gas firing - almost never replaced
• Coal firing - 3 to 5 years
Oil firing - 2 to 3 years under normal conditions,
but severe corrosion can cause de-
struction after only a few months of
service.
2. Plant data
Table 6.5 is a compilation of available data from a number
of power plants concerning fly ash, flue gas and coal composition,
and fly ash resistivity. The data reported in this table were
either obtained by SRI personnel under Contracts CPA 70-149 and
CPA 70-166 or made available to SRI by the utility companies.
Of all the plants listed in Table 6.5, only Plants 10 and
9 have experienced significant air heater corrosion problems. As
the following discussion will indicate, the factors that result
in high resistivity fly ash usually indicate that no corrosion
problems are to be expected.
The ash samples for which analyses are given in Table 6.5
were either collected from the precipitator hoppers or obtained
with a resistivity apparatus at the precipitator inlet. The
values of pH and free acid obtained in a 95% ethanol slurry, which
are given for selected samples, are an indication of acid present
on the surface of the ash. Samples which show an acidic pH in 95%
ethanol generally exhibit a minimum pH in water, followed by a
rise to a basic equilibrium value as the water soluble base is
dissolved. The presence of significant amounts of unreacted acid
on the ash surface is thought to be an indication that the fly ash
has been "conditioned" by sulfuric acid.
Data for S02 - S03 were obtained by SRI personnel using
procedures described elsewhere.27 Resistivity data were also
obtained by SRI using either a point-plane or cyclone resistivity
SOUTHERN RESEARCH INSTITUTE
-------
Table 6.5. Properties of Flue Gas and Fly Ash for Various Coal-Fired Boilers
Coal analysis
(dry basis)
Plant
designation
6
1
2"
11
8-3«
5
7»
4
9a,b
10«.b
sulfur
0.7-
1.0
0.6
0.5
0.5
0.5
0.95-
1.90
2.1
3.6
-3.5
3.2
ash
t
8.5
12
5.9
15-25
8.6
15.8-
16.0
21.9
16.4
-14
11.2
Fly ash j
water slurry
_pH_
12.2
8.2
11.1
11.2
9.4
9.4
5.1
11.0
9.8
6.4
sol base
as CaO
t
7.6
Negligi-
ble
2.10
1.50
0.35
0.19
0
1.65
0.35
0
sol SO,
1.2
0.23
1.50
0.17
0.77
0.41-
0.47
0.36
0.77
1.15
0.40
••""lysis
ethanol slurry
free acid as H,SOt
pB 1
>9.1 0
4.6 0.008
8.1 0
„
—
~
3.8 0.037
3.9 0.088
4.4 0.02
Flue gas analysis
precip. inlet
vol ppm
375
387
—
365
610-
1030
1650
26 BO
~
--
(wet basis)
H,SO, vapor
vol ppn
<1
<1
—
<1
0.8-
4.4
8.7
15
~
—
H,0
vol «
10.7
8.9
--
7.7
7.0
5.7
8.0
—
—
Typical fly
U-cm
1.9 x 10"
1.0 x 10"
3.8 x 10"
4.5 x 10"
1.0 x 10"
3 x 10"-1.5
1.0 x 10"
2.0 x 10"
1.0 x 10"
1.0 x 10*
~
ash resistivity
temp *F
302
220
320
275
230
309
x 10" 256-
319
319
300
287
290
—
1
t-1
00
1
a. Preoipitator preceded by mechanical collector.
b. Corrosion of air heater has occurred.
-------
-149-
apparatus, with the exceptions of Plants 6 and 11. For these two
plants, the data were given to SRI by the operating utilities.
Plant 6 has successfully overcome a high dust resistivity
problem by lowering the precipitator operating temperature tc
about 220°F at full load. An inspection of the low temperature
zone of this installation was conducted while the unit was off
the line for routine maintenance. This plant had nine months of
operation with low gas temperatures.
The areas examined for evidence of corrosion were the cold
and intermediate zones of the air heater elements, the plates and
wires in the precipitator, and the sides of the duct encompass-
ing the precipitator assembly. No evidence of corrosion was
found in the air heater elements. Thin deposits were noted in
some areas of the cold-end elements, but these were insufficient
to cause measurable draft losses. Minor corrosion was observed
on the perforated plate distributors at the precipitator inlet.
The rusted areas corresponded to regions of low gas velocity
caused by duct geometry. The only significant corrosion in the
entire assembly was found on the under side of the top plate of
the precipitator housing. The top side of this plate is exposed
to streams of low temperature bleed air from the plant exterior,
and it is probable that temperatures below the water dew point
were reached. The purpose of the bleed air is to maintain a
positive pressure for prevention of dust buildup on the rapper
bushings.
We have no direct measurement of S03 at Plant 6, but
measurements from Plant 2, which uses a similar fuel, show that
SO3 levels above and below the air heater are less than 1 ppm.
The soluble sulfate content of fly ash taken from the precipi-
tator hoppers of Plant 6, if a dust concentration of 1.5 gr/ft3
is assumed, is equivalent to an S03 concentration of 10 ppm.
It is possible, however, that a portion of the sulfate origi-
nated from oxidation of S02 on the ash surface rather than from
SO3 in the bulk gas phase. Figure 6.3 shows that the dew point
of a flue gas with 10 ppm S03 and 10.7% water vapor is estimated
as 275°F. The minimum cold end average temperature of the air
heater at Plant 6 is 140°F. It is therefore probable that some
acid condensation, and possibly corrosion, would have occurred
if the basic ash had not been present to combine with the S03
in the high temperature zone prior to the air heater, thus pre-
venting the formation of H2SO^ vapor in the air heater region.
Furthermore, the data in Tables 6.3''and 6.4-and the lack of
surface acidity indicated in Table 6.5 show that any H2SO^ which
may form in the air heater region is likely to be neutralized.
SOUTHERN RESEARCH INSTITUTE
-------
-150-
In view of the known dependence of fly ash resistivity on
temperature and the presence of H2SO^ on the fly ash surface,
the hypothesis of negligible H2SO,, in the low temperature zone
at Plant 6 may seem inconsistent with the decrease in resistivity
with temperature which occurs at this installation. This apparent
inconsistency can be qualitatively resolved by attributing the
resistivity behavior to increasing adsorption of water vapor on
the fly ash surface with decreasing temperature. It is also
possible that oxidation of S02 to S03 occurs on the ash surface,
and provides surface H2S04 for conditioning for a brief time
period, after which the acid is neutralized. The following
reaction sequence may be used to represent this hypothesis.
S02 (g) + Js02 (g) •* SO, (g)
S03 (g) + H20(g) •> H2S<\(g) «•-»• H2SO,, (1)
H2SO,,(g or 1) + CaO(s) -»• CaSO,, (s) + H20(g)
Thus, by adsorption of water and/or surface formation of S03, it
is possible to explain the lowering of resistivity with decreas-
ing temperature in the absence of appreciable S03 concentrations
in the bulk gas phase.
Plant 11 and Plant 10 are the other plants listed in
Table 6.5 with lowered cold-end temperatures. Plant 11 operates
with a low sulfur coal which produced a highly basic fly ash with
a high resistivity. No corrosion problems have been experienced
at this installation, as would be expected. Precipitator inlet
temperatures range from 230-253°F.
Plant 10 has operated with precipitator inlet temperatures
from 228-246°F. Excessive deterioration of air heater cold end
elements occurred when gas temperatures were lowered to 228°F,
and as a result, operating temperature has now been raised to 243-
246°F. The reason for lowering the exit temperature was said to
be a desire to increase boiler efficiency rather than a need to
lower fly ash resistivity. Fly ash and coal samples supplied to
SRI were analyzed and are reported in Table 6.5. However, the
sulfur content of the coal normally used was reported by the
utility to be 1.2-1.35%. Analysis of the fly ash indicates a
neutral ash similar to that from Plant 1, and little or no acid
neutralizing ability would be expected. The low sulfate content
indicates that, in spite of the high sulfur content of the coal
and the relatively low temperature at which the ash was collected,
a comparatively small amount of H2SO^ is collected by the ash.
From the ash content of the coal, the grain loading at Plant 10
is estimated, prior to the mechanical collector, as 3 gr/scf,3U
and the sulfate content of the fly ash is equivalent to only
6.4 ppm H2SOU. It is therefore probable that most of the
-------
-151-
formed from the combustion of this relatively high sulfur coal
remained in the gas phase and was available for condensation.
Although we have no resistivity measurements from Plant 10,
it is possible to infer from the coal and fly ash analysis that
a low resistivity fly ash (significantly less than 2 x 1010 fl-cm)
is probable at this installation at the precipitator operating
temperatures. It has been shown from studies of H.SO^ condition-
ing under Contract CPA 70-149 at Plant 1 that a sulfate gain of
only 0,1-0.2% due to adsorption or condensation of H2SO^ is
sufficient to lower resistivity by two orders of magnitude for a
neutral fly ash.
Plants 9 and 4 normally operate with a high sulfur coal,
and typical air heater exit temperatures for both units range
from 275-285°F. These plants have low fly ash resistivities at
normal operating temperatures, and at times the resistivity value
at Plant 4 has been too low for proper precipitator operation with
high gas velocity. The cold-end portion of the air heaters at
both of these installations operates below the acid dew point, but
the corrosion experience has been somewhat different. Plant 4 has
an average cold-end temperature of about 200°F, and Figure 6.3 shows
that most of the H,S0lf vapor is available for condensation at this
temperature. Furthermore, measurements of SO3 before and after
the air heater have indicated, on at least one occasion, a signifi-
cant drop in S03 concentration across the heater. It is therefore
probable that significant amounts of H2S(\ are condensed, either
on the ash in the cool boundary layer adjacent to the metal surface,
or on the metal surface itself. In spite of this fact, the cold-
end baskets (made of low-alloy steel) have been in service for
at least ten years at Plant 4 without requiring replacement.
Table 6.5 shows that the fly ash at this unit is highly basic,
and would be expected to have significant acid neutralizing ability.
However, the presence of surface acidity, as indicated by data
obtained in a 95% ethanol slurry, suggests that a sulfate layer
on the ash is preventing a portion of the water soluble base from
being utilized.
Plant 9 has required some replacement of cold-end, air heater
elements, but not at an excessive rate. The data in Table 6.5
indicates that the fly ash from Plant 9 is less basic than that
from Plant 4, but the presence of a mechanical collector at Plant 9
makes a direct comparison of the two fly ash analyses difficult
because of the difference in particle size distribution. It is,
however, reasonable to conclude that without the presence of the
basic fly ashes at both installations, corrosion would have been
more severer
SOUTHERN RESEARCH INSTITUTE
-------
-152-
Plant 7 operates at high air heater exit, temperatures with
an intermediate sulfur coal. The resistivity values indicated
in Table 6.5 for this plant would be classified as high, but the
near-neutral character of the ash, together with the presence of
appreciable concentrations of HZS0.4 vapor in the gas phase and
the slope of the resistivity temperature curve, suggest that
acceptable resistivity values would occur at about 280°P. With
an 80°F inlet air side temperature, this would give a cold-end
average of 180"F for the air heater. The Air Preheater Company's
cold-end temperature and material selection guide gives a
suggested minimum average cold-end temperature of about 160°F
for a coal of 2.1% sulfur content and corrosion-resistant, low-
alloy steel cold end elements.35 Some degree of corrosion may
occur because the cold-end metal temperatures fall appreciably
below the acid dew point, and because the neutral ash at Plant 7
could be expected to have no significant acid neutralizing ability.
However, the experience of the Air Preheater Company as represented
by their materials and temperature guide, and the lack of excessive
Ho SO,, vapor concentrations found at 300-320°F are indications that
a severe corrosion problem should not occur at Plant 7 with the
presently used fuel if air heater exit temperatures as low as
280°F were employed.
The corrosion experience of Plant 5 (Unit 1) is of interest
because the average air heater exit temperature is about 260°F.
Sulfur content of the coal normally burned at this unit is approxi-
mately 1%, and a typical dust load would be 3.7 gr/scf. Coal compo-
sition varied during the time period in which resistivity data
were taken, and possibly as a result, the resistivity data show
considerable scatter and no strong variation with temperature.
Nonetheless, the relatively high resistivity values are to be
expected on the basis of the coal sulfur content and the moderate-
ly basic character of the fly ash. No corrosion problems have
occurred at this unit, and none would be expected with the relative-
ly low H2SO,, vapor concentrations which were measured.
Plants 8-3 and 2 are typical of installations burning very
low sulfur coal; that is, no appreciable H2SOW vapor concentrations
are found in the bulk gas phase, the fly ash produces a basic water
slurry, and the resistivity is unfavorably high in the normal oper-
ating temperature range of 275-300eF.
If the design of these plants were such that operation in
the 220-240°F range were possible, no corrosion problems would
be expected because of the absence of H2SOi, vapor. Unfortunately,
there is not a sufficient quantitative knowledge of the relationship
between resistivity and temperature to predict with confidence that
-------
-153-
low temperature operation at these installations would produce
resistivity below the critical value of 2 x 1010 fi-cm. The fact
that the flue gas water concentrations at Plants 2 and 8-3 are
about 30% lower than that at Plant 6 is an unfavorable condition
for achieving lowered resistivity. However, the fly ashes from
Plants 2 and 8-3, and in particular, that from Plant 8-3, are
less basic than the ash produced at Plant 6. Data obtained
under Contract CPA 70-149 indicate that a highly basic ash
requires a greater gain of HjSO^, either by condensation or
adsorption, to lower resistivity than does a neutral ash. Thus,
if lowering of resistivity is due to the combined effects of
water adsorption and the formation of S03 on a fly ash surface
discussed earlier, it could be argued that the resistivity of
the extremely basic ash of Plant 6 would show less sensitivity to
decreasing temperature than the fly ash at Plants 2 and 8-3.
Since the variables of ash composition and flue gas water concen-
trations indicate opposing effects when comparing Plant 6 with
Plants 2 and 8-3, it would be hazardous to equate the resistivity-
temperature experience of Plant 6 with the other two installations.
G. Methods of Assessing Corrosion Tendencies
of Flue Gases
A comprehensive discussion of methods developed in England
for assessing the corrosion and fouling potential of flue gases
is given in a bulletin entitled, "Testing Techniques for Determin-
ing the Corrosive and Fouling Tendencies of Boiler Flue Gases"
published by the Boiler Availability Committee.26 The following
discussion is a brief summary of the purpose and method of oper-
ation of those procedures which relate to low temperature corrosion
and fouling.
1. Corrosion probes
The purpose of corrosion probes is to measure the amount
of corrosion produced by acid condensed on metal surfaces in a
flue gas environment. These probes provide a means of supporting
a prepared metal test specimen in flue gas streams at a selected
temperature. The BCURA probe is an air-cooled device in which
the surface temperature of the test specimen is monitored with a
thermocouple brazed to the body of the probe. Exposure periods
of 15-30 minutes are recommended, and the amount of corrosion is
determined by measuring weight loss of the specimen.
Probes designed for short term experiments are of value
for comparing relative effects of variations in operating param-
eters, such as temperature and fuel composition. However, for
SOUTHERN RESEARCH INSTITUTE
-------
-154-
prediction of actual corrosion rates over extended periods, long
term tests of 100 hours or more are desirable, A liquid-cooled
probe has been designed by the Shell Petroleum Company, Ltd., for
such extended experiments.36
2. Acid deposition probes
An indirect measurement of the rate of acid deposition on a
cooled surface is given by the BCURA dew point meter, which has
been described previously. Since the conductivity readings of
the dew point meter can be influenced by substances other than
sulfuric acid, it is of interest to consider a direct means of
measuring acid deposition rates.
Alexander2 7 has described an air-cooled deposition probe
which accomplishes this purpose. The probe consists of an air-
cooled, one-inch diameter stainless steel tube in which the cool-
ing air passes through the tube and discharges into the flue gas.
The amount of acid depositing on test areas of the probe, the
surface temperature of which is known, is determined by analysis
of deposits obtained from the test surfaces.
3. Gas and ash analysis
An analysis of flue gas for S03, SOz, H20, and dust loading,
along with analysis of the fly ash for soluble components, is
necessary for a qualitative assessment of the flue gas corrosion
potential. Procedures used by SRI for these analyses are
described in the final report from Contract CPA 70-149.27
H. Summary and Conclusions
It has been established that the principal cause of corrosion
in the low temperature zone of power plant exhaust systems is
condensation of sulfuric acid, either directly onto metal surfaces
or onto fly ash particles which subsequently come in contact with
the metal. Other acids, in particular hydrochloric acid, can be
responsible for corrosion at temperatures approaching the water
dew point of flue gas, but such temperatures are not normally
encountered.
Fouling in the low temperature zone of air heaters is
primarily caused by reaction of sulfuric acid with fly ash and
the metal surfaces of the heat exchanger. A basic fly ash can
neutralize appreciable quantities of S03 upstream from the air
heater region, but laboratory experiments suggest that reaction of
highly basic fly ashes with high concentrations of HjSO,, in the low
-------
-155-
temperature zone can result in problems with deposit formation.
This conclusion is supported by the experience of the Central
Electricity Generating Broad of England, in which medium sulfur
coals with alkaline ashes have produced fouling,*s but little
air heater wastage accompanied the deposit formation. It is also
possible to have deposit formation in the low temperature zone
in the absence of sulfuric acid if excessive moisture from steam
leaks or soot blowing is present.
Severe corrosion and fouling problems in regenerative air
heaters are associated with the temperature at which peak rates
in acid deposition occur. These peak rates often are not observed
with coal firing due to the presence of fly ash, but in any case,
the existence of such a peak is a manifestation of relatively
high concentrations of free H2SO,, vapor. Thus, the resistivity
of fly ash, due to the presence of excessive H2SC\, would be
expected to be lower than desirable for proper precipitator oper-
ation with high gas velocity under these conditions. Resistivity
data taken at plants burning high sulfur coals with alkaline fly
ashes have demonstrated that resistivity values below the critical
2 x 10"° ft-cm are obtained at temperatures above 280CF. Therefore,
lowering precipitator operating temperatures is neither necessary
nor desirable for the case of high sulfur coals, which produce
relatively high concentrations of HjSO^ vapor.
An analysis of the factors which cause corrosion, and the
operating experience of at least two power plants, has demonstrated
that low temperature operation of precipitators (220-250°F) will
not cause low temperature corrosion and fouling problems with a
flue gas containing a basic fly ash and no appreciable concen-
trations of sulfuric acid vapor. The occurrence of corrosion and
high fly ash resistivity thus tend to be mutually exclusive phe-
nomena. A possible exception to the rule would be a stack gas with
high (over 100 ppm) HC1 concentration.
For the case of a plant burning a low to medium sulfur
coal which produces a near-neutral, high resistivity ash at
approximately 300°F and low concentrations of I^SO^ vapor, the
occurrence of some degree of corrosion as a result of lowered
cold-end temperatures cannot be rigorously excluded. However,
data obtained under Contract CPA 70-149 have shown that amounts
of sulfuric acid sufficient to "condition" a neutral ash can be
adsorbed at temperatures well above the sulfuric acid dew point.
It is therefore probable that an acceptable fly ash resistivity
could be obtained at a temperature sufficiently high to avoid
appreciable condensation of sulfuric acid on the cold-end elements
of an air preheater. A quantitative evaluation of resistivity
SOUTHERN RESEARCH INSTITUTE
-------
-156-
and corrosion under such circumstances would require fly ash
resistivity data and relative corrosion rates (obtained with a
corrosion probe such as described in Section G) as a function
of temperature in the flue gas.
-------
-157-
I. Bibliography— Section 6
1. Hedley, A. B., in The Mechanism of Corrosion by Fuel
Impurities (H. R. Johnson & D. L. Littler, editors),
Butterworth, London, p 204 (1963).
2. Cuffe, S. T., Gerstle, R. W., Orning, A. A., and
Schwartz, C. H., J. Air Poll. Control Assoc. 14, p 353
(1964).
3. Snowden, P. N., and Ryan, M. H., "Sulfuric Acid Condensation
from Flue Gases Containing Sulfur Oxides," J. Inst. Fuel 42,
p 188 (1969) .
4. Mueller, Peter, "Study of the Influence of Sulfuric Acid
on the Dew Point Temperature of the Flue Gas," Chemie -
Ing.-Tech. 31, p 345 (1959).
5. Abel, Emil, "The Vapor Phase Above the System Sulfuric
Acid - Water," J. Phys. Chem. 50_, p 260 (1946).
6. Gmitro, J. I., and Vermuelen, T., "Vapor-Liquid Equilibria
for Aqueous Sulfuric Acid," Univ. of California Radiation
Laboratory Report 10866, Berkeley, California (June 24,
1963).
7. Greenewalt, C. H., "Partial Pressure of Water Out of
Aqueous Solutions of Sulfuric Acid," Ind. and Eng. Chem.
17_, pp 522-523 (May 1925) .
8. Johnstons, H. F., "An Electrical Method for the Determination
of the Dew Point of Flue Gases," Univ. of Illinois Eng. Exp.
Station, Circular 20 (1929).
9. Flint, D., "The Investigation of Dew Point and Related
Condensation Phenomena in Flue Gases," J. Inst. Fuel 21,
p 248 (1948).
10. Burnside, W., Marshall, W. G., and Miller, J. M., "The
Influence of Superheater Metal Temperature on the Acid Dew
Point of Flue Gases," J. Inst. Fuel 29, p 261 (1956).
11. Corbett, P. P., and Flint, D., "The Influence of Certain
Smokes and Dusts on the SOS Content of the Flue Gases in
Power Station Boilers," J. Inst. Fuel 25, p 410 (1953).
SOUTHERN RESEARCH INSTITUTE
-------
-158-
12. Dooley, A., and Whittingham, G., "The Oxidation of Sulfur
Dioxide in Gas Flames," Trans. Faraday Soc. 42, p 354
(1946) .
13. Whittingham, G., "The Influence of Carbon Smokes on the
Dew Point and Sulfur Trioxide Content of Flame Gases,"
J. Appl. Chem. 1^ p 382 (September 1951).
14. Flint, D., and Kear, R. W., "The Corrosion of a Steel
Surface by Condensed Films of Sulfuric Acid," J. Appl.
Chem. 1, p 388 (1951).
15. Lee. G. K., Friedrich, F. D., and Mitchell, E. R., "Effect
of Fuel Characteristics and Excess Combustion Air on
Sulfuric Acid Formation in a Pulverized-Coal-Fired Boiler,"
Department of Energy, Mines, and Resources, Mines Branch
(Canada), 9p (1967).
16. Friedrich, F. D., Lee, G. K., and Mitchell, E. R.,
"Combustion and Fouling Characteristics of Two Canadian
Lignites," Department of Energy, Mines, and Resources,
Mines Branch (Canada), Research Report R208, 31p (August
1969).
17. Kear, R. W., "The Influence of Carbon Smokes on the
Corrosion of Metal Surfaces Exposed to Flue Gases," J. Appl.
Chem. JL, p 393 (September 1951) .
18. Black, A. W., Stark, C. F., and Underwood, W. H., "Dew
Point Meter Measurements in Boiler Flue Gases," ASME Paper
No. 60-WA-285 (December 1960).
19. Clark, N. D., and Childs, G. D., "Boiler Flue Gas Measurements
Using a Dew Point Meter," Trans ASME 87(A-1), p 8 (1965).
20. Taylor, A. A., "Relation Between Dew Point and the
Concentration of Sulfuric Acid in Flue Gases," J. Inst. Fuel
16., p 25 (1942).
21. Lisle, E. S. and Sensenbaugh, J. D., "The Determination of
Sulfur Trioxide and Acid Dew Point in Flue Gases,"
Combustion 36 (1), p 12 (1965).
22, Taylor, H. D., "The Condensation of Sulfuric Acid on Cooled
Surfaces Exposed to Hot Gases Containing Sulfur Trioxide,"
Trans. Faraday Soc. 47, p 1114 (1951).
-------
-159-
23. Piper, John D. and Van Vliet, H., "The Effect of
Temperature Variation on Composition, Fouling Tendency,
and Corrosiveness of Combustion Gas from Pulverized-Fuel-
Fired Steam Generators," Trans. ASME 80, p 1251 (August
1958) .
24. Fontana, M. G., Corrosion; A Compilation, The Press of
Hollenback (195771
25. Thurlow, G. G., "An Air Cooled Metal Probe for the
Investigation of the Corrosive Nature of Boiler Flue Gases/1
J. Inst. Fuel 25_, pp 252-255 and 260 (1952) .
26. The Boiler Availability Committee (London), "Testing
Techniques for Determining the Corrosive and Fouling
Tendencies of Boiler Flue Gases," (Bulletin No. MC/316),
p 18 (March 1961) .
27. Southern Research Institute, Final Report on Contract
CPA 70-149 (A Study of Resistivity and Conditioning of Fly
Ash) to Division of Control Systems, Office of Air Programs,
Environmental Protection Agency (in preparation).
28. Halstead, W. D., "The Behavior of Sulfur and Chlorine
Compounds in Pulverized-Coal-Fired Boilers," J. Inst. Fuel
£2, p 344 (September 1969).
29. Kear, R. W., "The Effect of Hydrochloric Acid on the
Corrosive Nature of Combustion Gases Containing Sulfur
Trioxide," J. Appl. Chem. 5, p 237 (May 1955).
30. Canady, B. L., "High Pressure Jetting of Regenerative Air
Preheaters," Combustion, p 55 (February 1955).
31. Roddy, Charles P., "Sulfur and Air Heater Corrosion,"
Power Engineering, p 40 (January 1968).
32. Barkley, J. F., et al., "Corrosion and Deposits in
Regenerative Air Heaters," U. S. Bureau of Mines Report of
Investigations 4996, 23 pp (August 1933).
33. Brownell, Wayne E., "Analysis of Fly Ash Deposits from
Hoot Lake Station," Report to The Air Preheater Corp.,
Wellesville, New York, 12 pp (December 1961).
SOUTHERN RESEARCH INSTITUTE
-------
-160-
34. IGCI/ABMA Joint Technical Committee Survey, "Criteria for
the Application of Dust Collectors to Coal-Fired Boilers,"
(April 1965).
35. Clark, Norman D., "Higher Efficiency Through Lower Stack
Temperature," The Air Preheater Corp., Wellesville,
New York.
36. Kear, R. W., "A Constant Temperature Corrosion Probe,"
J. Inst. Fuel 32, p 267 (1959).
37. Alexander, P. A., Fielder, R. S., Jackson, P. J., and
Raask, E., "An Air-Cooled Probe for Measuring Acid Deposition
in Boiler Flue Gases," J. Inst. Fuel 33, p 31 (1960).
38. CERL (private communication).
Respectfully submitted:
Grady'B. Nichols, Head
Particulate Control Section
A
-------
-161-
APPENDIX 1: LIST OF SYMBOLS
APPENDIX 2: COMPUTER PRINTOUT
SOUTHERN RESEARCH INSTITUTE
-------
-162-
APPENDIX 1.
LIST OF SYMBOLS
A = Collection electrode area or specific area of an object
C = Constant or coefficient
Cp = Current to charge particulate
D - Dust concentration or diffusion coefficient
E0 ,E ,EP = Electric field, starting, charging, at collecting plate
° P = Force, Farad
K = Aspect ratio, constant
Length of precipitator or corona wire
N = Densities of various quantities - number per units volume
P = Perimeter, probability
Q = Quality factor or quantity
R = Radius - specific dimension
S = Surface area per gm of dust, circumference
V = Voltage, ionization potential
W = Weight of dust
X = Fraction of molecules ionized.
a = Radius corona wire or dust particle
b = Radius collection pipe, wire and pipe precipitator
d = Air density, relative
e = Electronic charge
f = Wire roughness factor, reentrainment factor
h = Height of plates, velocity head
i = Current, per unit length
j = Current, per unit area
k = Boltzmann's constant, coefficient
In = Natural logarithm
n = Number of items per unit time
p = Velocity pressure
q = Electrical charge
r = Radius or reduction factor
t = Time , thickness
v = Velocity
Vg = Volume flow rate
w = Migration velocity or precipitation rate parameter
-------
-163-
LIST OF SYMBOLS (Continued)
a = lonization coefficient, acceleration, erosion coefficient
6 = Constant erosion coefficient
Y = Secondary emission coefficient
6 = Boundary layer thickness
A = Increment
, e0 = Relative dielectric constant, permittivity of free space
n = Collection efficiency, viscosity
^ = Mobility
p = Resistivity, space charge density, gas density
o = Conductivity
I = Summation
T = Time constant
X = Concentration ratio
V2 = Laplacian operator
SOUTHERN RESEARCH INSTITUTE
-------
-164-
APPENDIX 2.
COMPUTER PRINTOUT FOR
ELECTROSTATIC PRECIPITATOR MODEL
-------
-165-
PAGE 1
// JOB
LOG DRIVE
0000
*ESP*
CART SPEC
0002
*ESP*
CART AVAIL
0002
0001
PHY DRIVE
0002
0001
V2 M10 ACTUAL 16K CONFIG 16K
// FOR
•LIST SOURCE PROGRAM
*ONE WORD INTEGERS
* IOCSKAUDi1132 PRINTERtPLOTTER)
C»
ELECTROSTATIC PRECIPITATOR MODEL.
C PRECIPITATOR PARAMETERS
C
C
C A = COLLECTION AREA FT2 M2
C VG = GAS FLOW RATE FT3/SEC MS/SEC
C VO = APPLIED VOLTAGE VOLTS VOLTS
C TC = TOTAL CURRENT AMPS AMPS
C DL = DUST LOAD GR/FT3 KG/M3
C WL = LENGTH OF CORONA WIRE FT M
C AC = CORONA WIRE RADIUS IN — M
C 8 = PLATE SPACING IN M
C PL = LENGTH OF PRECIPITATOR FT M
C V = GAS VELOCITY FT/SEC M/SEC
C ETAO = DFSIGN EFFICIENCY (STATED EFF)
C DO = DUST DENSITY KG/M3
C RHO = DUST RESISTIVITY OHM-M
C RF = ROUGHNESS FACTOR-CORONA WIRE
C DEL = RELATIVE AIR DENSITY
C W = WEIGHT OF DUST KG
C U = MOBILITY M2/IVOLT-SEC)
C EPS = DIELECTRIC CONSTANT RELATIVE)
C CD = CURRENT DENSITY AMPS/M2
C E = CHARGE ON AN ELECTRON COU
C ET = ELECTRIC FIELD IN DEPOSIT VOLT/M
C CL = CURRENT PER M. OF CORONA WIRE AMP/ M
C EO = STARTING ELECTRIC FIELD VOLTS/M
C NS = NUMBER OF PARTICLE SIZES
C FID = FREE ION DENSITY NUMBER/MS
C VIS = GAS VISCOSITY KG/CM-SEC)
C NF = NO OF .3 METER INCREMENTS IN PL METERS OF PRECIPITATOR
C DV = DUST VOLUME
C 2MFP = MEAN FREE PATH
C FPLT » ELECTRIC FIELD AT PLATE
C CCF = CUNNINGHAM CORRECTION FACTOR
C AFID = AVERAGE FREE ION DENSITY
C PSAT = FRACTION OF SATURATION CHARGE
C ZMD = INTERPOLATED MMD OF COLLECTED DUST
C ROVRI * RATIO OF TOTAL TO IONIC SPACE CHARGE
C ZMDL = INTERPOLATED MMD OF EFFLUENT DUST
C WT = WEIGHT OF DUST COLLECTED IN EACH INCREMENT
C FRRI = F(ROVRII
SOUTHERN RESEARCH INSTITUTE
-------
PAGE
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
-166-
2 *ESP»
DER = D • FMROVRH
DRRI = -FIROVRD/F* (ROVRI)
ITLU) = NAME OF PLANT - IDENTIFICATION
DW(I) = AMOUNT OF MATERIAL REMOVED EACH INCREMENT
DIAM(J) = DIAMETER OF PARTICLES M
PCNT(J) = PERCENTAGE OF PARTICLES OF D1AM(J)
VOL(J) = VOLUME PER PARTICLE M3
XNO(J) - NO. PER M3 EACH SIZE RANGE
OSAT(J) = SATURATION CHARGES
Q(J) = CHARGE ON EACH PARTICLE
ONO(J) = INITIAL NO. OF PARTICLES IN EACH SIZE RANGE
DXS(J) = TOTAL NO. OF PARTICLES REMOVED IN EACH SIZE RANGE
RAD(J) = RADIUS OF PARTICLES M
DIMENSION
$ ONO(ll)i
DIMENSION
DIMENSION
DIMENSION
CONSTANTS
DlAM(ll),VOLiXNO(ll),QSAT(ll).Om>.
DXS(ll). PCNT(ll). RAD(ll)
DW(27)i ITLUOI
WS(ll)
WSL(20)»CCF(20)
PI = 3.1415927
E = 1.6E-19
U = 2.2E-4
EPS = 100.
RF=0.90
EPSO = 8.85E-12
VIS = 1.8E-5
READ(2*5) NS*(DIAM(I)»I«1*NS)
5 FORMAT!I2/I10F8.0) )
DO 3 I « liNS
DIAM(I)=DIAM
-------
-167-
PAGE 3 »ESP»
1000 READI2.7) ITL
7 FORMAT UOA2)
READ<2i4)A.VGtVOiTC»DL»WL.ACiB»PL»ViETAOtDD»RHO»X»TEMP*P
4 FORMAT110F8.0)
PL = PL » 0*305
NF = PL / .3 + .5
NF ) 9999t9999»8
C
C CONVERSION
C
8 A = A » 9.3E-02
VG = VG « 4.73E-A
DL » DL • 2«29E-03
WL = WL * 0*305
AC = AC • 2*54E-2
B = B « 2.54E-2
RHO = RHO * 10.**(X-2. )
DV = DL / DO
V = V » 0«305
WRITE(3i3010) ITL
3010 FORMAT! «0' i40A2/)
T = !• / V
C
C
DO 1 I = 1 .NS
VOL( I) = PCNTU ) * DV
1 CONTINUE
C
C COMPUTE WEIGHT OF DUST
C
W = DL * VG
C
C COMPUTE CURRENT DENSITY
C
CD = TC / A
C
C COMPUTE ELECTRIC FIELD IN DEPOSIT
C
ET * CD * RHO
C
C COMPUTE CURRENT PER M. OF CORONA WIRE
C
CL = TC / WL
C
C COMPUTE STARTING ELECTRIC FIELD
C
P0=1.0
T0=293.*1.8
TEMP=TEMP+A59.
DEL=(TO/TEMP)»(P/PO)
SOUTHERN RESEARCH INSTITUTE
-------
-168-
PAGE 4 »ESP»
C CALCULATE MEAN FREE PATH
TDK=TEMP/1.8
ZMV=8.205E-05*TDK/P
ZMFP = ZMV/(1.4l<»»PI«ll.6E-19)«(6.02E+23))
EO = 3.E6 « RF « DEL • ( 1. + .03 MDEL/AC )««.5)
C
C COMPUTE VALUE OF EXPONENT IN DEUTSCH EQUATION FOR THE STATED EFF.
C
305 X»ALOG(100./(100.-ETAO) )
C
00 9 I • It NS
DXS(I) * 0.0
XNO(I) « VOL(I)/ I 4./3. • PI » RAD!I)*«3 )
ONO(I) = XNO(I)
9 CONTINUE
C
C
C COMPUTE EFFICIENCY PER 0.3 METER INCREMENT
C
ETAPF * l.-EXP!-0.3«X/PL)
C
C COMPUTE AMOUNT OF MATERIAL REMOVED PER INCR.ON A TOTAL WEIGHT BASIS
C
SW a 0.0
C
DO 700 J=1»NF
DWU) » (W - SW) « ETAPF
SW " SW + OW ( I )
700 CONTINUE
C
C PRINT INPUT PARAMETERS
IF(LK)llltllltl60
C
111 WRITEJ3.10JPL.NF.A
10 FORMAT I • PPR LENGTH «» t£ll«4 »U» 'METERS' »T41 • 'NO. OF INCREMENTS •'
$, 13, T81, 'COLLECTION AREA • ' .Ell .4, IX , «M2 ' )
C
WRITEJ3.il) VG, VOf TC
11 FORMAT (• GAS FLOW RATE «'.E11.4»' M3/SEC1 »T4l» * APPLIED VOLTAGE -'t
SE11.4.1 VOLTS'. T81. 'TOTAL CURRENT -'.Ell. 4.' AMPS' )
C
WRITE (3. 12) DL» WL t AC
12 FORMAT! ' DUST LOAD *'*E11.4t* KG/M3' »T4l, • CORONA WIRE LENGTH •'»
t Ell. 4. * M* »T81t 'CORONA WIRE RADIUS «'»E11.4»' M' )
C
WRITEI3.13) B* Vt ETAO
13 FORMAT! • PLATE SPACING o'»EH.*»« M'iT4l.'GAS VELOCITY »',Ell.<»,
$' M/SEC' ,T81, 'DESIGN EFFICIENCY -SF6.2.' PERCENT1)
C
WRITE(3»1<») DD. RHO. DV
14 FORMAT!' OUST DENSITY «'tE11.4i< KG/M3 ' .Ul . 'DUST RESISTIVTY •'»
$ E11.4*' OHM-M' .TSl.'DUST VOLUME -',E11.4»' M3/M3')
WRITEO.15) W. CD. ET
15 FORMAT!' DUST WEIGHT >(*E11.4*' KG/SEC' »T*lt 'CURRENT DENSITY «'»
AMP/M2' .T81, 'DEPOSIT E FIELD -'.E11.4*' VOLT/M' )
WRITEI3.16) CL.EO.DEL.RF
-------
-169-
PAGE 5 *ESP«
16 FORMAT!' CURRENT/M -'.Ell.4.' AMP/M*.T«l•'START ING E FIELD = 't
1 Ell.^t' VOLT/M'.TBl.'REL. AIR DENSITY='tFT.ftt
2/' ROUGHNESS FACTOR"'»F6.3)
C
C
160 X=ETAPF*100.
WRITE(3il61)X
161 FORMAT!/' INPUT EFF1CIENCY/INCREMENT-•tF6.2)
WRITEI3.4322)
4322 FORMAT/T5i'ROVRI'iT16»'ERAVG1tT31.«EPLT'»T43.'PSAT•iT53f'TOAVG1»
1T63.'TEFF'.T76»'MMD'iT89.'WE1GHT«iT101.'SECT ION NO.1/)
C«»«««****«*»**##«#*»*i^«»*»**««**#**«»***»»•«**«***»***»*»»•»*»••»•*•••
C
C START INCREMENTAL ANALYSIS OF PRECIPITATOR
C
LK=1
PSAT = 0.0
ZWT=0.
ZCA*U2.*PI«EPSO*EPS)/(EPS+2.)
ROVRI=10*
DO 3000 I=liNF
C CONVERGE ON CORRECT ROVRI USING NEWTON'S METHOD
ZC=200.#(DW (<«. * EPSO)/(AFID«E*U)
C
C COMPUTE CHARGING TIME CONSTANT IN THIS INCREMENT
C
C
TEFF=(TOAVG»PSAT1/(1.-PSAT)
C
ZL=0.305
P^AT=i./ll.+((TOAVG)/(TEFF+T*ZL)))
C
C
C*»#**###***#«**»*****#»*###««*###****#«*##*##»*«#»*»##*»»»**»#*«»*##*»*
C
C START PARTICLE SIZE LOOP
C
SOUTHERN RESEARCH INSTITUTE
-------
-170-
PAGE 6 »ESP»
WT » 0.
DO 2900 J - 1. NS
C
C COMPUTE CHARGE ON EACH PARTICLE AFTER ONE INCREMENT OF TRAVEL
C
0(J) = QSAT(J) • PSAT
C
C
C
C COMPUTE MIGRATION VELOCITY FOR EACH SIZE RANGE
C
A2E02=AC*AC»EO«EO
B2=B«B
AC2=AC*AC
EPLT=((1.-(AC2/B2M*ROVRI»CL/<2.»PI»EPSO*U)+A2E02/B2)*».5
EMV*(Q(J)»EPLT)/(6.*PI»RAD(J)»VIS>
CCF( J1=1«+O.B6«ZMFP/RADU)
EMV=CCF
-------
-171-
PAGE 1 *ESP*
IF(KJ)2910. 2910. 2911
2910 ZMD = OIAMU)
GO TO 2912
2911 ZMD = DIAMIKJ) + (TL2/TLimDIAMU>-DIAMCKjn
2912 WRITE«3.O23)ROVRI.ERAVG.EPLT.PSAT.TOAVG.TEFF.ZMDiWT.I
4323 FORMAT(T5.F6.3tT13.Ell.<»»T27»Ell.<»»m»F7.4»T51*F7.<».T60»F7i3*T71*
3000 CONTINUE
C
ETC= / ONO(l)
X = X * EFESR * PCNTII)
XY=PCNT( I)«100.
XEP=EFESR*100.
WR I TE( 3.2291 )DIAM( I ) iXY *XEP .CCF ( I )
2291 FORMAT (2X.E 17.7 «T29tF10.6»T45*F10«6»T61(F7t4)
SL=(1.0-EFESR)»ONO( I)
WSL( I )=SL*(1.33333*PI*RAD(I)«*3)*OD
2990 CONTINUE
X = X » 100.
WRITE! 3.2292) ETAOt X
2292 FORMAT! 'O1 »4X. 'EFFICIENCY - STATED • • .F5.2t5Xi 'COMPUTED «• iF6.2
S.5X. 'CONVERGENCE OBTAINED')
C
C
C CALCULATE HMD OF EFFLUENT
WTL=(1.-(X/100. I )*DL
ZTM=0.
DO 2995 1=1. NS
ZTM=ZTM+WSL( I )
ITtCZA-0.5 12995 i 2995* 2996
2995 CONTINUE
2996 CZB=(ZTM-WSL(I) I/WTL
TL1=CZA-CZB
TL2=0.50-CZB
KJ=I-1
IF(KJ)2980i2960.2981
2980 ZMDL=DIAM( I )
GO TO 2982
SOUTHERN RESEARCH INSTITUTE
-------
-172-
PAGE 8 »tSP»
2981 ZMDL-DIAM(KJ>«-)
2982 WRITE(3i2997)ZMDL
2997 FORMAT(SXi'MMD OF EFFLUENT" ' tEll • <»//)
C
GO TO 1000
9999 CALL EXIT
END
FEATURES SUPPORTED
ONE WORD INTEGERS
IOCS
CORE REQUIREMENTS FOR
COMMON 0 VARIABLES 566 PROGRAM 2150
END OF COMPILATION
// XEO
-------
-173-
PAGE
// JOB
LOG DR
0000
V2 M10
// OUP
1 *EVSR»
IVE CART SPEC
0002
ACTUAL 16*
CART AVAIL
0002
0001
CONFIG 16K
PHY DRIVE
0002
0001
»EVSR»
•DELETE
CART ID 0002
EV5R
DB ADDR 20ED
DB CNT 0011
// FOR
•ONE WORD INTEGERS
*LIST SOURCE PROGRAM
»*««*»*»*«*«*n*i
C» THIS SUBROUTINE CALCULATES ELECTRIC FIELD AND DE/D(ROVR1) AS
C« A FUNCTION OF RADIUS. OUTPUT IS A SPACE AVERAGE.
C«#»*#»**it»«*»«***»#*»*»»*#«»**«#«»»#«»*»**»»»*#»«»«»»»»##»*»»»«*««
SUBROUTINE LVSR(AC.EO.CLiU.8.EAVGiDER.ROVRIi£PSO»PI I
A2e02=AC*AC"EO*EO
EAVG=0.0
DER=0.0
R = AC
DR=(B-<».»AC)/50.
J = 0
AC?=AC»AC
ZKL=ROVRI*CL/(2.*PI»EPSO»U)
DO 100 1=1*200
R2=R«R
ZKM=(1.-IAC2/R2))«ZKL
E=(ZKM+A2E02/R2)«*.5
DET=0.5*1l./E)*(ZKM/ROVRI)
•
«
DER=DER+DET
100 CONTINUE
90 EAVG=EAVG/FLOAT( J»
DER=DER/FLOAT(J)
RETURN
END
FEATURES SUPPORTED
ONE WORD INTEGERS
CORE REQUIREMENTS FOR EVSR
COMMON 0 VARIABLES 22 PROGRAM
RELATIVE ENTRY POINT ADDRESS IS 0025 (HEX)
END OF COMPILATION
// DUP
218
SOUTHERN RESEARCH INSTITUTE
-------
-174-
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1 REPORT NO
EPA- 650/2-74-132
3. RECIPIENT'S ACCESSION-NO.
4 TITLE AND SUBTITLE
An Electrostatic Precipitator Performance Model
5 REPORT DATE
July 1972
6. PERFORMING ORGANIZATION CODE
7 AUTHOR(S)
Grady B. Nichols and John P. Gooch
B PERFORMING ORGANIZATION REPORT NO.
SORT-EAS-74-344
9 PERFORMING OR6ANIZATION NAME AND ADDRESS
Southern Research Institute
2000 Ninth Avenue South
Birmingham, Alabama 35205
10. PROGRAM ELEMENT NO.
1AB012; ROAP 21ADJ-026
11. CONTRACT/GRANT NO.
CPA 70-166
12. SPONSORING AGENCY NAME AND ADDRESS
EPA, Office of Research and Development
NERC-RTP, Control Systems Laboratory
Research Triangle Park, NC 27711
13. TYPE OF REPORT AND PERIOD COVERED
Final; Through 7/6/72
14. SPONSORING AGENCY CODE
15 SUPPLEMENTARY NOTES
16. ABSTRACT.
The report gives results of: a review of the design details of a pilot
precipitator; and particle concentration profile studies. It also reviews discussions
of resistivity measurement and correlations between resistivity and precipitator
operation. Objectives of the study covered by the report were: to extend the precip-
itator model to include factors not included in the first model influencing its
accuracy: to design and build a pilot precipitator for further studies of the factors
influencing precipitation processes; to review limitations to precipitator perfor-
mance due to back corona and sparking; to investigate the particle concentration
profile in the interelectrode space; to obtain data from both field and pilot plant
tests to attempt to verify the computer model; and to analyze potential for opti-
mizing precipitator performance by design or operating modifications.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS
c. COSATI Field/Group
Air Pollution
Mathematical Models
Electrostatic Precipitators
Resistance
Measurement
Air Pollution Control
Stationary Sources
13B
12A
14B
18. DISTRIBUTION STATEMENT
Unlimited
19. SECURITY CLASS (This Report)
Unclassified
21. NO OF PAGES
184
20. SECURITY CLASS (Thispage)
Unclassified
22 PRICE
EPA Form 2220-1 (9-73)
------- |