EPA-600/2-76-144
May 1976
Environmental Protection Technology Series
                      ELECTROSTATIC  PRECIPITATORS:
                               RELATIONSHIP  BETWEEN
     RESISTIVITY,  PARTICLE  SIZE,  AND SPARKOYER
                                  Industrial Environmental Research Laboratory
                                       Office of Research and Development
                                      U.S. Environmental Protection Agency
                                 Research Triangle Park, North Carolina 27711

-------
                RESEARCH REPORTING SERIES

 Research reports of the Office of Research and Development, U.S. Environmental
 Protection Agency, have been grouped into five  series. These five broad
 categories were established to facilitate further development and application of
 environmental technology. Elimination of traditional grouping was consciously
 planned to foster technology transfer and a maximum interface in related fields.
 The five series are:

     1.    Environmental Health Effects Research
     2.    Environmental Protection Technology
     3.    Ecological Research
     4.    Environmental Monitoring
     5.    Socioeconomic Environmental Studies

 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 REVIEW NOTICE

This report has been reviewed by  the U.S.  Environmental
Protection Agency, and approved for publication.  Approval
does not signify that the contents necessarily reflect the
views and policy of the Agency, nor does mention of trade
names or commercial products constitute endorsement or
recommendation for use.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.

-------
                                          EPA-600/2-76-144

                                          May 1976



         ELECTROSTATIC  PRE CIPITATORS:

              RELATIONSHIP  BETWEEN

RESISTIVITY, PARTICLE SIZE, AND SPARKOVER
                            by

                   HerbertW. Spencer,  HI

                 Southern Research Institute
                   2000 Ninth Avenue South
                Birmingham, Alabama  35205


                   Contract No.  68-02-1303
                   ROAP No. 21ADL-027
                Program Element No. 1AB012


            EPA Project Officer:  Leslie E.  Sparks

         Industrial Environmental Research Laboratory
           Office of Energy, Minerals, and Industry
             Research Triangle  Park, NC  27711


                       P repared for

        U.S. ENVIRONMENTAL PROTECTION AGENCY
             Office of Research  and Development
                   Washington,  DC  20.460

-------
                             ABSTRACT

The report gives results of a study of the relationships of the
electrical resistivity of fly ash, its particle size, the occur-
rence of back corona and sparkover, and the electrical character-
istics of electrostatic precipitators (ESP's).   The study included
laboratory measurement of the dielectric strengths and resistivity
of five particle-size fractions of a fly ash sample and measure-
ment of the current densities and voltages at which back corona
and sparkover occurred for a 3-mm dust layer covering the plate of
a wire-plate negative-corona discharge device.   Results showed
that the peak current density for the formation of back corona
                                                      v •  t "•v'i
depended on the resistivity of the dust covering the positive
                                                      . »
electrode.  Operating current densities for full-scale ESP's are
discussed in relation to fly ash resistivity.
                               11

-------
                          CONTENTS
                                                     Page

Abstract                                              ii

List of Figures                                       iv

List of Tables                                        vi

Acknowledgments                                       vii

Sections

I      Conclusions                                     1

II     Introduction                                    6

III    Historical Review of the Effects of
       Dust Layers on Electrical Characteristics
       of Corona Discharges                            8

IV     Apparatus and Procedures                       15

V      Results of Laboratory Measurements             24

VI     Measurements on Full-Scale Precipitators       52
                             111

-------
                               FIGURES
No.                                                           Page

 1   Allowable current density as a function of resistivity
     according to Hall13                                       13

 2   Precipitation rate parameter vs. resistivity,  dashed
     curve White's field data11*                                14

 3   Wire plate corona discharge device                        16

 4   Photographs of the wire plate corona discharge device     17

 5   Wire plate corona discharge device circuit schematic      18

 6   Electrode configuration for testing dielectric strength
     of ash layer                                              20

 7   Resistivity vs. temperature for different particle
     size fractions; 1) 74.5% P(Porosity), 0-3 ym;
     2) 67.8% P, 3-7 ym; 3)  64.7% P, 7-15 urn; 4)  58.0% P,
     15-25 ym; 5) 54.3% P,  >25 ym; 9.4% water vapor by
     volume                                                    28

 8   Photograph of dust surface on the wire plate corona
     discharge device                                          30

 9   Additional photographs  of dust surfaces on the wire
     plate corona discharge  device                             31

10   Clean plate voltage - current characteristics  at
     60°C and 120°C                                            33

11   Voltage - current characteristics for the wire plate
     discharge device with 3mm dust layers                     34

12   Oscilloscope traces of  back corona pulses                 36

13   Wire plate corona discharge device voltages for spark-
     over and for formation  of back corona as a function
     of resistivity; wire to plate spacing 6 cm;  solid curve
     estimate of the average voltage at formation of back
     corona; dashed curve potential between dust surface and
     wire at formation of back corona                          41

14   Current densities for formation of back corona in the
     wire plate corona discharge device as a function of
     resistivity                                               42

15   Current densities at sparkover and for formation of
     back corona as a function of wire to plate spacing        46
                                IV

-------
                               FIGURES
                               (Cont'd)


No.                                                           Page

16   Sparkover voltages as a function of wire to plate
     spacing                                                   47

17   Current density distributions:   Wire potential 21 KV,
     clean plate spacings of 1)  3 cm and 2)  5 cm               49

18   Current density distributions with and without back
     corona                                                    50

19   Current density distributions for clean plate at
     several applied potentials and for dust covered plate
     with back corona                                          51

20   Operating current densities as  a function of resistivity
     for various plants tested by SRI; Numbers refer to data
     in Table V; Circles inlet sections, triangles outlet
     sections, squares unknown sections, solid symbols either
     NH3 or SO3 injection                                      53

-------
                              TABLES


No.                                                       Page

  I    Dielectric strenghths of ash layers                 25

 II    Physical properties of dust layers formed
       from Gaston Power Station fly ash particle
       size fractions                                      27

III    Sparkover and back corona voltages for five
       different particle size fractions                   38

 IV    Chemical analyses of size fractionated fly ash
       samples                                             39

  V    Precipitator electrical data                        54
                              VI

-------
                           SECTION I
                          CONCLUSIONS

The efficiency of an electrostatic precipitator is a direct
function of the electrical conditions that are obtained in the
precipitator.  By increasing the electric field and particulate
charge, the migration velocity of the particles to the collection
plate can be increased, resulting in increased efficiency or
reducing the size of the precipitator needed to meet a given
emission standard.  Unfortunately, precipitators are forced to
operate at voltages and current densities significantly below
what can be obtained with a corona discharge system containing
no particulate.  The resistivity of the particulate collected on
the plates of an electrostatic precipitator inversely affects
the current density at which the precipitator can operate.

Data are presented in this report for full scale precipitators
and for a laboratory corona discharge device that show the depen-
dence of precipitator electrical operation on particulate
properties.  The field data indicate that precipitators collecting
particulate with resistivities below ^1x10:° ft-cm can operate
at current densities on the order of 80 nA/cm2.  At this current
density the electric field in the collected dust layer is less
than 0.8 kV/cm2 for particulate with resistivities below 1x1010 fl-cm.
This electric field is less than the dielectric strength of the
collected particulate, which is usually on the order of 20 kV/cm.
Current densities for precipitators collecting particulate with
resistivities above 1x1010 fi-cm were observed to decrease with
resistivity.   However, considerable scatter was obtained in
the data.   Some units operated at current densities producing
electric fields in the collected particulate layer near the di-
electric strength of the layer, while others operated at current
densities a factor of 10 to 20 times below the limit set by the
dielectric strength and resistivity of the dust layer.  It is
estimated that spatial and temporal variations in current density

-------
can account for a suppression in the operating point a factor
of 2 or more below the limit set by the particulate resistivity
and dielectric strength.

The peak current densities obtained in the laboratory  for the
formation of back corona were within approximately 20% of the
point at which the dielectric strength of the layer would have
been exceeded.  Field operating points were expected to be set
by the point at which back corona occurs; a comparison of our
field and laboratory measurements indicates a discrepancy between
field operation and small scale corona discharge operation.  For
low resistivities, the data indicate that a considerable increase in
operating current densities might be obtained with proper design.
For high resistivity dust the scatter in the field data was such
that a clear interpretation is not possible.  However, it
appears as mentioned before that the field units operate a
factor of 10 to 20 times below the expected operating point.

The comparison of the laboratory and field measurements in this
report shows that in the design of precipitators the minimum
and maximum clean plate current density limits must be considered.
For very high resistivity dust, the minimum clean plate stable
operating density must be below the current density for formation
of back corona.  For low resistivity ashes  (<109 ft-cm), pre-
cipitator electrical design should allow for operating at
current densities in the range of 100 nA/cm2 if maximum
performance is to be obtained.

The objective of the laboratory study was to determine if there
were other particulate properties besides dust resistivity
that affect the operation of a corona discharge when a dust
layer is deposited on, the collection electrode.

-------
It was determined that the particulate resistivity was the main
factor.  Other factors such as particle size, bulk density,
porosity, cross sectional area of voids, and specific surface
area appeared to affect operating points for precipitators only
by their effect on the resistivity of the collected dust layer.
A measurable change in the dielectric strength of the collected
dust layer as a function of the above factors was not observed
for the ash samples studied during this work.

It was observed that ash resistivity varied as a function of
particle size and that if the ash is sufficiently fractionated
by particle size in the precipitator, a decrease in ash
resistivity from the inlet to the outlet can occur.  The decrease
can exceed a factor of 2 and result  in a corresponding increase
in operating current density from the inlet to the outlet of
the precipitator.  The operating electrical data tabulated in this
report for full scale units show an  increase in current
density from the inlet to the outlet.  The data also showed
that lower operating voltages are normally obtained in the
outlet sections of the precipitator.  Precipitator electrical
behavior cannot be entirely explained by dust resistivity alone;
the decrease in space charge effects due to suspended dust
from the inlet to the outlet are probably responsible for the
observed increase in current density for a given voltage.l  Space
charge effects due to particulate should be included in any
future laboratory studies.  Theoretical calculations of the
voltage-current characteristic for a wire-duct configuration
that include space charge effects should be developed for inter-
preting variations in the voltage-current characteristics of
full scale units.

-------
A misaligned corona wire was shown to increase the chance for
formation of back corona and sparkover.  It was observed that
by decreasing the wire to plate spacing the ratio of the peak
current density to average current density was increased.  It
was also observed that the peak current density at the formation
of back corona is independent of the wire to plate spacing.
The result of these two observations is that if a wire is
misaligned, back corona will form at an average current density
less than the expected value.  Misalignment may account for some
of the discrepancy between the operating electrical characteris-
tics of field units and laboratory devices.  The magnitude of
the effect for precipitators would need to be determined with
a multi-wire system.

Other factors such as the variations in dust layer thickness and
errors in in-situ resistivity measurements may also play a part
in the discrepancy between field units and laboratory measure-
ments.  An investigation of the in-situ resistivity measurement
procedure is needed to determine if the procedure is affecting
the results.  The variations in dust layer thickness produced
by nonuniform rapping probably produce significant variations in
current density, increasing the probability of back corona for
a given average current density, and should be investigated.
The effects on current density distribution could be determined
by measuring the current densities to various points on a
precipitator plate covered with a non-uniform dust layer.  The
non-uniform dust layer would be formed by rapping the plate
to remove some of the dust, and then precipitating a new dust
layer on the surface.

Measurements of the current density distributions for the following
conditions: no dust layer, dust layer without back corona, and
dust layer with back corona showed a drastic change in current
density distribution with the formation of back corona.  This
had been previously observed by Kercher.2  Back corona had the

-------
effect of increasing the average current density while leaving
the current densities in some places unchanged.
FUTURE RESEARCH
Further investigation will be needed to clearly delineate the
difference between full scale units and a small scale corona
discharge when a particulate is present.  A determination of
the exact current density at which a precipitator can operate
for a given set of dust properties will depend on further
investigations.  Such studies are needed if accurate performance
characteristics are to be theoretically determined for
electrostatic precipitators.

Specific areas for future research are as follows:
1.  Gathering of additional data correlating operating points
    of field units and dust resistivity.
2.  Further development of the correlation between laboratory
    resistivity measurements and in-situ measurements.
3.  Test in a dry pilot scale precipitator to study the effect
    the following factors have on the electrical behavior of
    electrostatic precipitators:
    a.  Design of electrodes (wires and plates)
    b.  Dust properties
    c.  Uniformity of current density distribution with
        and without dust layers
    d.  Thickness and uniformity of dust layers
    e.  Space charge
    f.  Gas composition
4.  Theoretical calculations of V-I characteristics for a
    parallel plate precipitator including space charge effects.

-------
                         SECTION II
                        INTRODUCTION

The objective of this research is to provide information for
interpreting the electrical behavior of electrostatic pre-
cipitators when that behavior is governed by the characteristics
of the collected dust layer.

Dust layers affect precipitator electrical behavior in several
ways.  For one, they introduce a resistance element into the
electrical circuit, which, except for the nonlinear character-
istics of the resistivity of dust layers, behaves in the same
manner as the incorporation of an ohmic resistor in the circuit.
Second, the electrical breakdown of the dust layer and the
resulting formation of point corona in the dust layer, called
by the descriptive terms "back corona", "back ionization",
"back sprays", and "back discharge", drastically affect the
electrical behavior of a precipitator.  Back corona occurs when
a highly conductive point is formed in the dust layer, usually
by electrical breakdown of the dust at a point where the
current density is such that- the ohmic buildup of voltage exceeds
the dielectric strength of the dust.  These visible discharges
in negative corona  affect precipitator operation by decreasing
sparkover potential, increasing the average current density, and
producing positive ions that neutralize the negative space
charge in the corona gap.

The sparking and back corona conditions for fly ash were investi-
gated using a negative corona wire-plate discharge device in
the laboratory.  In particular, the variation of corona
characteristics as a function of the properties of dust layers
formed from different particle size fractions of fly ash was
studied.   The resistivities, dielectric strengths, porosities,
mass median diameters, and chemical compositions of 5 particle
size fractions of a fly ash were determined.
                               6

-------
Different particle size fractions were studied because precipi-
tators fractionate the inlet dust, the particles collected in
the inlet sections being larger than those in the outlet sections,
Previous attempts to relate sparkover to particle size have not
been successful.3

-------
                         SECTION III
 HISTORICAL  REVIEW OF STUDIES ON THE EFFECTS OF DUST LAYERS ON
       ELECTRICAL CHARACTERISTICS OF CORONA DISCHARGES
 In  1918, Wolcott4 found that a dielectric sheet of discontinuous
 surface placed over the plate electrode of a point-plane electrode
 system lowered the sparkover voltage by about 50% for negative
 corona, although it affected positive corona only slightly.  In
 1933, Franck5 observed that dust layers also influence the charac-
 teristics of discharges, especially in an asymmetric electrode
 arrangement.  For a negative corona discharge, he found that the
 sparking voltage decreased to about 1/3 of that without a dust
 layer.  The sparkover voltages increased with thicker dust layers
 from a minimum at approximately 1.5 mm, presumably because of the
 increased potential drop across the dust layer needed to reach
 breakdown.

 Franck interpreted the phenomenon of breakdown  in terms of
 the relative dielectric constants and conductivities of dust
 and air.  The dust layer and air at the dust layer boundary were
 assumed to be homogeneous dielectrics with homogeneous fields.
 The ratio of the electric field in the gas  (Ei) to the electric
 field in the dust layer (E2) is given by
      Ei _ £2
      E2   £i

when there is no current and

      E_j_ _ 0_2_
      E2   (Ji

when there is a current, where ei,e2 are the dielectric constants
and ai,a2 are the conductivities of  the  air  and  the  dust  layer,
respectively.  Because  the  dielectric  constant of the  dust is  greater
                              8

-------
than the dielectric constant of air, the electric field in
the dust layer is smaller than the electric field in the
air above the dust layer when the potential is first applied.
At steady state, the electric field in the dust layer is
larger than the field in the air above the layer if the
conductivity of the layer is low.

When the break-down voltage of the dust layer is reached, thin
conductive channels appear  as spots in the dust layer.  These
spots redistribute the field and behave like corona points  (back
corona).  For negative corona, this produces a strong anodic
field, lowering the potential for sparkover.  For positive corona,
a strong anodic field exists prior to the breakdown of the dust,
and thus the effect on sparkover is not as great.

In 1948, White6 ascribed the detrimental effects of back corona
in reducing the collection efficiency of an electrostatic pre-
cipitator to a lowering of the sparkover potential and to the
production of positive ions which decrease the efficiency of
charging of suspended dust particles in a negative corona.  He
noted that the electrical field  (E) in the dust is determined
by the product of the current density and the dust resistivity,
according to Ohm's Law.  Therefore, in an electrostatic precipi-
tator collecting a typical dust having a resistivity of 101° fi-cm
and a dielectric strength of 10 kV/cm, back corona would set in
at an operating current density of 1 yA/cm2.

Using voltage pulses of 1-2 microsecond duration, White obtained
peak current densities which were far greater than the direct
current densities for the formation of back corona.  This can
be explained in part by the estimated charging time of 30 micro-
seconds for a layer of dust with 10:1 ^-cm resistivity.  The
voltage across the dust layer for a pulsed current is less than
the voltage for a direct current when the pulses are shorter
than the charging time and when the time between pulses is longer
than the discharge time.

-------
In a series of papers, Penney7"10, reported on the
effects of resistivity layers on both the electrodes
of corona discharges.  The term "flare" was used to denote a
discharge consisting of repetitive current pulses or streamers
which originate from a fixed location on the anode when the
anode is partially covered by a high resistivity material.
Penney's experiments showed a reduction in sparkover potential
for negative corona with increased resistivity and with de-
creased wire size.  Although not reported, the reduction as a
function of wire size was probably related to increased current
densities obtained with the smaller wires, for a given potential.
By using a negatively charged ball, Penney was able to show that
the flares were a source of positive ions.  The flares consisted
of current pulses of 0.05 y-sec rise time and 0.5 y-sec decay
times with a period of approximately 0.33 y-sec between peak
current pulses of 30-80 mA.  Penney's experiments indicated that
with an imperfect insulating coating on the anode and a source
of ionization,current pulses followed by sparkover can occur
if the average gradient is of the order of 5 kV/cm.
This gradient is significantly below the gradient of 30 kV/cm
for breakdown of gases between clean metallic electrodes.
Penney suggested that the space charges resulting from the flare
produce a gradient which is favorable to the development of a
spark or at least alter the field, so that the initial voltage
distribution is rather unimportant.

An investigation of back corona in high-resistivity dust layers
          o
by Kercher  showed that there was a slight variation in the current
density at which back corona formed for a layer thickness < 3mm
but that Ohm's Law held for a thickness > 3mm.  He also showed
that when back corona formed the discharge current was concen-
trated at the breakdown sites.
                            10

-------
The effect of resistivity on the formation of back corona has
also been studied by Simm.ll  He substantiated previous work,
but observed that the sharp increase in current with the for-
mation of back corona was at a somewhat higher current density
than Ohm's Law predicted.  This indicates that back corona may
occur before a distinct change in the voltage-current charac-
teristics is visible.

Apparently the effect of particle size of the dust on back corona
has not been adequately studied.  In experiments with different
particle-size fractions of fly ash, Penney3found that the larger
particle-size fraction he was using had a lower sparkover poten-
tial than a smaller particle-size fraction,  even though the re-
sistivity of the large particle-size fraction was lower.  However,
the larger particle-size fraction contained highly conductive
carbon particles.

Electrical operating conditions in industrial electrostatic pre-
cipitators have been simulated by Herceg"12 using a point-to-plane
corona device and layers of porous insulating materials to simu-
late precipitated dust layers.  In this work he attempted to physi-
cally model the corona discharge device by a lumped-element circuit
based on a systematization of corona phenomena into four regions
described as the primary ionization, accumulation, transport, and
secondary-ionization regions.  A solid-state analog was designed
on the basis of the lumped-element circuit model.  However, no
data were presented with the design to show that it did model
a point-to-plane corona discharge.

Hall13 has indicated that the practical limits of operating
current density are calculated from Ohm's Law, in which the
breakdown voltage (E) has a value of 1-2 kV/cm, approximately
an order of magnitude smaller than the measured dielectric

                             11

-------
strengths of dust layers.  White1"* has indicated that the
precipitation rate parameter of field units (an empirical
parameter that characterizes the performance of a given
precipitator under a specified set of operating conditions)
varies inversely with the resistivity of the dust being
collected by the unit.  The practical effects of dust layers
on precipitation are summed up in graphs by Hall13 (Figure 1)
and by Whitelk (Figure 2).
                             12

-------
500
                       RESISTIVITY , ohm-cm

 Figure 1.   Allowable current density  as  a function of
            resistivity according  to Hall.13

                        13

-------
   16
   14
o
0)
in
\

E
u


cc.
UJ
I-
Ul

S
   12
£  10
UJ
o

|5

E
cc
Q.
   8
             T	1—I  I  I  I II
       *
         »
            \
               \
    Figure 2.
                  \
                    »


                    \
                       \
                          \
                             \
                                \
                                   \
                                      \
                                         ^
                                            »
                                                 \
                                                    \
             I      I    I  I   I  I I  I
                                           I     I   I
                                  10"                           I012

                          RESISTIVITY, Ohm-cm



               Precipitation rate  parameter vs resistivity;

               Dashed curve White's  field data.1"
                            14

-------
                         SECTION IV
                  APPARATUS AND PROCEDURES
This  section describes the equipment and the procedures used
to  study the influence of dust layers on the electrical be-
havior of a wire-plate corona discharge.

CORONA DISCHARGE DEVICE
The wire-plate corona discharge device used in the laboratory
experiments is shown schematically in Figures 3 and 4.  It
consists of a 0.089 cm (0.035 in.)-diameter stainless steel
corona wire supported 6.0 cm above a 10-cm square stainless
steel plate.  The plate electrode is recessed 3 mm in a Teflon
block.  Five insulated segments are incorporated for measuring
the current density at five separate points on the plate.

A Hipotronics DC power supply capable of providing 100 kilovolts
maintains high negative potentials on the corona wire of the
corona discharge device (ripple is less than 0.5% of output
voltage, when operating the device).  The corona discharge
device is installed in an environmental chamber so that tempera-
ture and humidity can be controlled during the measurements.
A Teflon-insulated copper tube extending to the device at the
bottom of the chamber is connected at the top of the chamber to
a voltage divider (a Hipotronics Model 100 high voltage meter).
The corona current to any one of the five insulated plate seg-
ments or the current to the outer plate is measured by selective-
ly connecting the desired element to the current-measuring circuit
displayed in Figure 5 while shorting the other elements of the
flat-plate electrode to ground.  The current measuring circuit
consists of a spark protector, a 100 n resistor, and a Keithley
414A picoammeter.  The voltage across the 100 fi resistor in series
with the electrometer is amplified five times by a differential
amplifier and displayed on a 10 MHz oscilloscope for recognition
of back corona onset.
                             15

-------
                                        TEFLON
                                        INSULATION
                             1/4 IN.
                             BRASS ROD
        CORONA BALL
        2 IN. OIA
BRASS ROD
8-32 THREADS
                                                   5/8 IN.DIA.
                                                    PLATE I , 1-1/2 IN. x 3/4 IN.
                                                    4 PLATES 1-1/2 IN. x 3/8 IN.
Figure  3.  Wire plate  corona  discharge device

-------
Figure 4.   Photographs of the wire plate corona
           discharge device.
                    17

-------
                                                      OSCILLOSCOPE
                                                            DIFFERENTIAL
                                                            AMPLIFIER
r<
Li
ft
i i
f 1
i SWITCH
1 1 	
•
1 — ^ X-N
« i '•' "* v"
; 5 inner plates • 1
1
1
1
"1 	
1
bKAKK
PROTECTOR
1
| 20
I A A A
1 vVV
1 0.02

_L
— i
i
f i
4,3V 1
i
1
1
1 I
                                                                       500 K
oo
        Figure 5.  Wire plate corona  discharge device circuit  schematic

-------
By  rounding  the  edges of the  flat collecting plate  and by mount-
ing corona balls on the ends  of the corona wire support, the
maximum clean-plate voltages  were increased to above  70 kV  and
the maximum  clean-plate current densities were increased to
~3  pA/cm2  (8 mA/ft2) .  This current density limit is  about  a
factor of 80 larger than peak current densities in  full-scale
precipitators.

RESISTIVITY AND ELECTRICAL BREAKDOWN APPARATUS
The electrical resistivity of the fly ash was measured in a
conductivity cell built to the specifications of the American
Society of Mechanical Engineers Power Test Code 28.   The cell
consists of an electrode cup  for holding the ash sample and a
flat electrode with a guard ring that is placed lightly on the
dust, layer.  Voltage is supplied to the electrode by a Keithley
240 A high-voltage  supply.  The current through the cell is
measured with  an electrometer assembly similar to that used
with the corona discharge device.

Dielectric strength measurements on the fly ash samples were
made in a specially designed cell (Figure 6).   The cell in-
corporates a Power Test Code 28 conductivity cell cup for holding
the sample and an upper electrode with rounded edges machined
from stainless steel.   A Peschel Instruments 24-'cV high-voltage
supply furnishes a negative potential.  The overload protection
indicator on the power supply is used for detecting breakdown.
Dielectric breakdown of air was measured with the same equipment.
These pieces of equipment were installed in an environmental
chamber so that temperature and humidity could be controlled
during the measurements.

ENVIRONMENTAL CHAMBER
Temperature and humidity are controlled and monitored in an in-
sulated chamber housing the wire-plate corona discharge device,
                              19

-------
                                      3mm DUST LAYER
Figure 6.  Electrode configuration for testing
           dielectric strength of ash layer
                       20

-------
 resistivity cells, and ash and air breakdown cells.  System air
 is bubbled through a temperature-controlled water bath for humidity
 regulation.  This air is heated by two thermostatically-controlled
 heating elements mounted directly in front of a fan which circu-
 lates air through the chamber.  Two other heaters, rheostatically
 controlled, heat the air before it passes through the fan and as
 it is blown into the chamber.  This combination of heaters allows
 continuous control of temperature to 180°C.  The humidity in the
 chamber is controllable from 1% to 15% water vapor by volume with
 an approximate accuracy of 1%.

Temperature inside the chamber is monitored by chromel-alumel
thermocouples at various locations in the chamber.  A 150°C
mercury thermometer is visible through a glass window in the
chamber.

Moisture content of the chamber atmosphere is determined by
weighing the water absorbed by a cylinder of Drierite (CaSCK)
after a known volume of air is drawn through the cylinder from
the chamber.   An Atkins dew point indicator continuously moni-
tors the relative moisture content.

PROCEDURES
Ash from the Gaston Power Station, Wilsonville, Alabama, was
separated into five particle-size fractions by the Donaldson
Company of Minneapolis, Minnesota.  From these size fractions,
samples were quartered out and screened through an 80-mesh screen,
to break up large agglomerates and to remove any trash present
in the sample.  Smaller samples were quartered out from the
above sample for chemical analysis, resistivity measurements,
dielectric strength measurements, and back corona measurements.

The sample for the back corona measurement is placed on the
plate of the corona discharge device and is smoothed to a depth
of 3 mm while shaking and vibrating the device to compact the
                             21

-------
dust layer.  The same procedure is used to fill the two resistivity
cells and the dielectric breakdown cell.

These devices are then placed in the environmental chamber.  The
electrodes are put in place on the dust layers and electrical
connections are made to each device.  A check of electrical connec-
tions is made before closing the chamber.  After the chamber is
closed the temperature and humidity controls are adjusted to give
the desired conditions by increasing temperature and moisture from
ambient conditions.

Once the environmental conditions are stabilized, electrical
readings are made.

The plate current distribution in the corona discharge device is
measured as a function of voltage for determining the voltage-
current characteristics, back corona, and sparkover data.  The
voltage is adjusted and read from a Hipotronics Model 100 high
voltage meter.  At constant voltage, inputs from the plate segments
are consecutively switched to a picoammeter and these currents re-
corded.  As current readings are seldom constant, only as many
significant figures as indicated by the current range are recorded,
or the range of current fluctuation is recorded with the most
consistent value of current.  Current readings are taken starting
at applied voltages of 10 kV and then at applied voltages up to
60-70 kV in increments of 2 to 4 kV.  The size of the increment
depends on the stability of the voltage-current characteristics.

Oscilloscope waveforms of the current are noted concurrently
with the current measurements.   Visual observations of the
device are also recorded.

The resistivity-cell current measurements are taken one minute
after voltage is applied.  Current values from both cells are
consecutively read at one voltage by switching the picoammeter
input from one to the other.  The data are taken so that each

                             22

-------
cell is read at two applied voltages, 200 V and 1000 V.

Determination of breakdown potentials in dust and air follows
an identical procedure.   Voltage to the cell is slowly and
steadily increased at about 500 V/sec until the supply over-
load indicator shows sparkover.  The potential at that point
is recorded.
CALCULATIONS
Current densities for the wire-plate and resistivity cells are
calculated as the current divided by the appropriate electrode
area.  Resistivity, p, from the resistivity cell data is
calculated by Ohm's Law and the definition of resistivity:
                    RA
                P = D-
                    V
                R ~ I

so that         p =  l" "  D
where
       I = current from cell
       V = voltage applied (200 or 1000 V)
       R = resistance of ash layer
       D = depth of ash layer
       A = area of electrode
                            23

-------
                          SECTION V
             RESULTS OF LABORATORY MEASUREMENTS
The current-voltage characteristics of the test device were
measured with a clean and with a dust-covered plate at various
temperatures from 60°C to 180°C and at humidities in the range
of one percent to fifteen percent water vapor by volume.  Simul-
taneously, the resistivity and dielectric strengths of the ash
layers were measured.  Layers of fly ash having mass median
diameters of 2.3 jam, 3.6 ym, 8.2 ym, 14.5 ym, and 40 ym were used
for the study.  Particle size fractions having these mass median
diameters were obtained by mechanically separating fly ash from
the Gaston Power Station into the following size fractions:
0-3 ym, 3-5 ym, 5-7 ym, 7-15 ym, and >25 ym.  The mass median
diameter of these particle size fractions were determined by
Bahco analysis of each size fraction.

The current density at which back corona forms theoretically
depends on the ratio of the dielectric strength and the resis-
tivity of the ash.  The dielectric strengths of the different
particle size fractions of the Gaston ash are tabulated in Table
I.  Corresponding values for air are given in this table for a
3 mm gap between electrodes similar to the ones used for the ash
electrical breakdown measurements.

The average dielectric strength of the 3 mm dust layers is
22.7 kV/cm, with a standard deviation of 2.3 kV/cm.  The average
dielectric strength of the 3 mm air gap was 36.9 kV/cm* with a
standard deviation of 0.5 kV/cm.  The dielectric strengths of
the different particle-size fractions bear no definite relation
to the particle size.  The data also indicate no correlation of
the dust dielectric strength with either temperature or resistivity,
*According to Paschen's Law for a 0.3 mm gap with a uniform
field the dielectric strength of air at p = 720 Torr and T = 293°C
is 35.23 kV/cm.1S
                             24

-------
                                                 TABLE  I

                                    DIELECTRIC STRENGTHS OF ASH LAYERS
NJ
Ash
Particle
Size

15-25 ym







0-3 ym

3-7 ym

7-15 ym

>25 ym


T(°C)
-
60
70
80
100
120
140
160
180
100
160
100
160
100
160
100
160

H20(%)
1.0
2.0
1.6
2.4
3.0
3.4
3.9
4.7
3.0
4.0
2.7
4.3
3.0
4.0
2.4
3.6

Ash
Breakdown
(kV)
6.8
6.1
6.6
6.8
7.2
6.5
8.0
7.9
7.3
7.5
6.8
7.3
5.4
6.5
6.8
6.1
Dielectric
Strength
Ash Resistivity
kV/cm p(n-cm)
22.7
20.3
22.0
22.8
24.0
21.7
26.8
26.3
24.3
25.0
22.8
24.2
18.0
21.7
22.8
20.2
4.4xl010
2. 9x10 12
1.3xl012
5. 6x10 12
2. 0x10 1 3
5. 6x10 13
1.3xl013
6.4xlO)2
2.9xl010
8.3X101 i
1.2xl012
2. 2x10 12
8. 1x10 1 1
2.0xl01£
2.4xl012
2.6xl012
Dielectric Strength
Air
kV/cm
•TV v / Will
36.3
36.7
37.3
37.7
37.3
36.5
36.7
37.0








                               Average ash breakdown 6.8 kV
              D.  C. voltage increased at the rate of 0.5 kV/sec.
              Dust layer thickness  was 3 mm.

-------
Some investigators studying thin glass plates as well as 50 jjm
hollow glass spheres have observed that breakdown strengths
change from approximately 30 kV/cm to approximately 4 kV/cm with
a temperature change from 20°C to 200°C, while other investigators
found no temperature dependence.16  Results from this investigation
with Gaston Power  Station ash  showed  no  temperature  dependence.

The physical properties of the dust layers studied in this in-
vestigation are tabulated in Table II.  The resistivities of the
different particle size fractions are shown in Figure 7.  The re-
sistivities in the surface conduction region shown in this figure
increased with increased particle size'.  Bickelhaupt18
discusses the relationship of resistivity and particle size and
surface area.   For the different particle size fractions, the
estimated diameter of voids in the dust layer ranged from 3 to
29 mm.   The surface area per unit volume likewise changed an order
of magnitude.

These results indicate that changes in the physical properties
of the collected ash layer in a precipitator will influence pre-
cipitator operation by changing the ash  resistivity and not by
significantly changing the dielectric strength of the ash layer.

EROSION AND DUST LAYER SURFACES
The dust layer surfaces of large particles at low resistivities
are eroded when current producing voltages are applied.  A com-
bination of electrostatic forces and mechanical forces due to
"electric wind" scatters the dust, thins the layer and creates
surface irregularities.  This phenomenon is especially detrimental
to close study of dust layers.

Gaston Power Station dust resistivities at ambient temperature
and humidity (23°C and 1-2% H20 by vol) are two or three orders

                             26

-------
                                                 TABLE  II

                               PHYSICAL PROPERTIES OF DUST LAYERS FORMED
                               FROM THE GASTON ASH PARTICLE SIZE FRACTIONS
Size Range Medium Average
of Fraction, Diameter, Measured
ym ym Porosity,*
0-3
3-7
7-15
15-25
+ 25
2.3
3.6
8.2
14.5
40
76.7
69.9
62.4
57.3
54.7
Cross sectional Diameter, of Surface of the
average area circle particle with
of void formed by with equivalent No. of median
contact of spheres,** pore area, Particles/ diameter,++
ym2 ym cm" ym2
6
10
39
98
676
.15
.6
.2
.9
.5
2.
3.
7.
11.
29.
8
7
1
2
3
3
1
1
2
1
. 66x10 10
.232xl010
.30xl09
.675x10"
-35xl07
1
4
2
6
5
.66x10'
.072x10*
.llxlO2
.60xl02
.03xl03
Surface per
cm3 of dust, +
m2
.608
.502
.274
.177
.068
  Determined by Bahco  analysis

 *Porosity = P = 100%  x (1  - Apparent^densitv^f material,


**Cross sectional area of

  void formed by contact of spheres = A =  .36d2 [—'-—„— ]
                                                1-
                                                   Too
  +number of particles per unit vol.  = N =  (1 —— }/ov d3

 ++surface area of particle = Ag =  ird2

+++surface are per cubic centimeter of dust  layer = NAg = ir(l -
d = diameter based on assuming all the
    particles are of the same diameter.
                                                               av = 7 for spheres1 7
        for spheres = 6(1 -
                                                                                                /d

-------
            10
         o
         2
         O
        CO
        CO
        UJ
        tr
            10
                                                      IOOO/T(°K)
                                                      °C
                              TEMPERATURE
Figure 7.  Resistivity vs. temperature for different particle
           size fractions; 1)  74.5%  P(Porosity), 0-3 ym;
           2) 67.8% P, 3-7 ym;  3)  64.7% P, 7-15 ym;
           4) .58.0% P, 15-25  ym;  5)  54.3% P,  >25 ym;
           9.4% water vapor by  volume.
                               28

-------
of magnitude lower than those at higher temperatures  (180°C).
Lower resistivities cause correspondingly smaller voltage drops
across the dust layer for a given current density.  This
decreases the tensile strength of the layer.  The total cohesive
effects of van der Waals' forces are more pronounced  for
smaller particles, making them more difficult to free from a
surface than larger ones.

The largest particle fraction studied, with particle  diameters
greater than 25 microns, was blown badly under all moisture
conditions studied.  A 3 millimeter layer was cleared from the
plate in about three minutes at room conditions with  30 kV
applied.  Higher temperatures and moisture contents reduced this
effect; 0-3 ym and 3-7 ym size fractions were not significantly
affected by corona wind under any of the conditions investigated.

Dust was first disturbed at the outer edge of the layer,
particularly at corners.  Higher voltages caused blowing at
all points on the surface (Figure 8 upper photograph).  Craters
were formed at :the sites of back corona and sparkover (Figure
8 lower photograph).  Large particle sizes showed peculiar
patterns in the dust layer at sites of back corona and spark-
over.   Particles appeared to be redeposited and realigned on
the surface (Figure 9 upper photograph).

With a dust layer, sparkover normally occurred:on the inner plates,
forming craters.  It often occurred at several points on the
surface.  In high temperature and low moisture content conditions
cracks in the surface formed between plates.  Sparkover and back
corona tended to occur in these cracks (Figure 9 lower photograph).
                                29

-------
            Blown dust  layer, large particles

              25 )_im  (Temp 86°C, Moisture 12V/O)
            Craters  formed by sparkover to un-
            disturbed  dust layer
Figure 8.  Photographs of dust surface on the wire plate
           corona discharge device.
                           ii

-------
             Patterns  formed  on dust  layer composed
            of large particles by back-corona and
            sparkover

             Cracks  formed  in dust layer of 15-25
            particles at 180°C
Figure 9.   Additional photographs of dust surfaces on the wire
           plate corona discharge device.
                              31

-------
THE WIRE PLATE CORONA DISCHARGE CHARACTERISTICS
The voltage-current  (V-I) characteristics  for  the  wire-plate
negative corona discharge device* discussed  previously  are
plotted in Figure 10 for temperatures of 12,0'°G1,and 60°C.   The
shift in the V-I characteristic with increasing  temperature re-
sults from the decrease in gas density with  increasing  tempera-
ture.  The series of graphs in Figure 11 show  the  changes  in the
voltage-current characteristics* that occur.with .an increase in
ash resistivity"'from 3x109 fl-cm to 1x1012  fl-cm.-- The solid line
in each of the'se plots is the characteristic obtained without
an ash layer..;'; The voltage-current characteristicvwith  an  ash
layer is indicated by the circles.  The dashed., line is  the clean
plate characteristic shifted according to  the  theoretical  voltage
                                               \ " '
drop across the ash layer.  The voltage drop is  calculated using
Ohm's Law, the measured resistivity, and the measured thickness
of the dust layer.
      AV =  jpl
      AV =  potential across  the  dust  layer
        j =  current ..density                   ,•  .  ...
             f- ";/.  .                            •. . '  '•£
      p  =  resistivity                        "..:
        1 =  depth of  the ash  layer.           ;;
The shift in the V-I characteristic with a 109 ^ft-cm resistivity
ash is shown in graph 1 of Figure 11.  The difference between
these two curves is masked in practice by variations in moisture,
temperature, and corona wire characteristics between measurements
with and without ash layers.

For a 1010  ft-cm,resistivity ash, the shift is  several kilovolts
and is easily observed as shown in graph 2 of  Figure 11.   The
data points are near the shifted curve until back  corona occurs.
*Center segment of the wire plate discharge device.  Note  that  the
V-I characteristics are plotted on semi-log scales.  The curves
have a normal appearance when plotted on  linear  scales.

                             32

-------
z
UJ
cr
ct

o

UJ
o
tr
<

o
   10
    -4
   10
     -5
   IO'6
   10
     -7
                                         I20°C
      10       20      30      40      50


                             POTENTIAL , kV
60
70
80
   Figure  10.   Clean plate voltage-current characteristics
                            33

-------
I0-4i—I GRAPH I  h
    ic
      5
oc
UJ
I-

UJ
o
I  I0~4
o:

o
UJ
          P~6.6xl09 Q -cm
                           H20
                  	 CLEAN PLATE .

                  — - SHIFTED CLEAN

                      PLATE

                   O  ASH PLATE
                                           (GRAPH 2 }-
                                              , IOO°C,3v/o
o

-------
With the formation of  back  corona,  there is a sharp increase in
current, which  is clearly observed  on a semi-log plot.  This break
in the characteristic  determines  the start of back corona.
For higher} current'S^r glpfaswere(observed at various locations on
the dust layer;  in  addition,  bursts 6;f current occurred at random
intervals/of approximately  0^:1 to'l^sec.  The burst lasted for
        ,>'\ •'.  -~ ,  , '•.,,   . { -... •   .   - •• •  . i  jjj'^'xi-,'
approximately 2  visec'with peak curr4|ijfes of 2mA.  Typical pulses
        '.•f :'V<-  -.-•'    "'   -••   -1  • '•••'"^•:;s-£,h^*
are displayed in Figure 12.   The\in'di%Ldual pulses occurred less
frequently than  the flares  observed by; Penney-1 1 and lasted four
times longer'*- The.,peak currents-were a factor of 10 smaller than
the 30-80 *lnAc peak: current that Penney?'reported..  Two separate
types of'pulses were observed both of'which occurred only when back
corona spots appeared on the  dust layer surfaces.  The positive
pulses had a high frequency component superimposed on the pulse
and appear to be related to changes in the characteristics of the
brushes on the corona  wire.   Some of the brushes appeared to change
from blue to,yellow when back corona formed, in agreement with an
        •-•*..  ' ^'l~-,- ~~ > f -  "^"'         _ -       ^
earlier observation by White.    The pulses indicated that the
back corona sp'bt was pulsing. These purrent pulses are much larger
        *>•"-" -  .-. .  •' '.  .  .[ •*
than the'^Trichel pulses19 that are observed when negative corona
is firstiMnitiated.                    *
         '             '
The calculated voltage  drop across  th^ dust with a resistivity of
3x1011 J|reji|is "indicated by the  dasjled! curve in graph 3 of Figure 11.
        $ •',t!-?f?-'--:'•: •_•';'.: • •" '•  '-'"'  •--         ""  i ?^y
The measured V-i did not follow  this :curve.  The V-I characteristics
with a 10:1 fi-cm dust layer initially  had higher currents than the
clean plate characteristic and paralleled the shifted clean plate
characteristic until back corona developed.  Starting currents
higher than clean plate starting currents were observed several times
for dust layers with 1011 J^-cm resistivities.  A satisfactory ex-
planation  for this behavior does not exist at this time.
                                  35

-------
 to
 LU
 to
 LL)
 (T
 CC
 73
 O
 £
 o
LJ
Q:
a:

o
                                      OSCILLOSCOPE

                                      OUTPUT
ELECTROMETER

OUTPUT
                TIME, lOms/cm
                                       OSCILLOSCOPE

                                       OUTPUT
                TIME , O.Susec/cnn

Figure  12.  Oscilloscope  traces  of back  corona pulses
                     36

-------
 In  graph 4, the V-I characteristic for a 1012 fi-cm ash  layer  is
 plotted.  Back corona formed near the corona onset.  The current
 increased sharply.

 A comparison of the V-I characteristics of the different particle
 size fractions indicates that ash resistivity is the important
 parameter determining where back corona and where sparkover occur.
 The results of the measurements are shown in Table III.  The  fines
 have the highest sparkover potential and the lowest resistivity.
 Penney2 also found higher sparkover voltages with his fine particle
 size fraction, which had higher resistivities than his  larger par-
 ticle sizes.  However, Penney's large particle size fraction  con-
 tained large carbon particles, which reduced the resistivity  of
 the ash.

 The chemical compositions of the ash fractions used in  this study
 are tabulated in Table IV.  The data indicate that the  carbon
 content-loss on ignition (LOI) of these size fractions  was largest
 in the small particle size fraction and least in the large par-
 ticle size fraction.  Measurements by Bickelhaupt2° show that re-
 sistivity of fly ash is unaffected by a carbon content  of 10% or
 less.  Except for the carbon content, the compositions  of the
 size fractions are similar.  However, as shown by Bickelhaupt18
 the resistivity of the ash decreased with decreasing particle size
 in the surface conduction region.

As mentioned previously, the surface characteristics of ash layers
 formed from the different particle size fractions varied signifi-
 cantly because of the decrease in the cohesiveness of the dust
with increasing particle size.  The small particle size fraction
produced dust layers that cracked when the Teflon base  of the
wire plate discharge base expanded.  The large particle size
 fractions produced dust layers that were easily blown by the corona
                             37

-------
                                          TABLE  III

       SPARKOVER AND  BACK  CORONA VOLTAGES FOR FIVE DIFFERENT PARTICLE SIZE FRACTIONS
co
CD
                                                                 Back
        Particle                                        Back    Corona
          Size                  Moisture*  Sparkover   Corona   Current
        Fraction,  Temperature*   Volume,    Voltage,   Voltage.  Density.
                      _      S                 _                '      .  o'
ym
0-3
3-7
7-15
15-25
>25
°C
90
90
90
90
90
%
13.
11.
14
12
12

5
8



Resistivity
kv
52
48
46
42
42
kV
52
44
42
42
34
yA/cm
>1.7
.99
.76
.71
.25


1
2
3
1
ft- cm
IxlO9
.9xl010
.4xl010
.6xl010
.SxlO1 '
         * Chosen to obtain reasonable resistivities.

-------
                            TABLE  IV




    CHEMICAL ANALYSES OF  SIZE  FRACTIONATED FLY ASH SAMPLES






                        Particle  Size  Range
Chemical
Constituent
Li20
Na20
K20
MgO
CaO
Fe203
A1203
Si02
Ti02
P205
SO 3
LOI


0-3y
.08
1.16
2.86
1.13
1.59
6.62
28.136
45.04
3.19
0.705
0.81
8.47
98.981
+ 0.81
3-7y
.08
1.07
2.63
1.08
1.84
5.93
29.52
45.35
2.74
0.482
0.54
8.71
99.432
0.54
7-15y
.08
.544
2.63
1.17
1.54
6.71
30.09
47.39
2.80
0.321
0.29
6.16
99.435
0.29
15-25y
.08
1.34
2.67
1.21
1.41
7.72
27.23
49.74
2.87
0.356
0.22
3.67
98.30
0.22
+25y
.08
1.23
2.67
1.28
1.50
12.84
24.66
51.14
2.74
0.496
0.09
0.85
99.48
0.09
Avg.
.08
1.07
2.69
1.17
1.58
7.96
27.92
47.73
2.87
0.47
0.39
5.57
99.50

SUM       99.79    99.97    99.725     98.52      99.57
                             39

-------
wind when using conditions which produced low resistivities.
Penney2 also experienced this problem.  These conditions affected
our data.  For example, sparking normally occurred at locations on
the anode plate where a crack was present in the dust layer.
However, the effects appeared to be slight compared to changes in
resistivity.

The influence of ash resistivity on the formation of back corona
is displayed in Figures 13 and 14.  In Figure 13 the voltages at
which back corona occurred and at which sparkover occurred as
determined from the measured V-I characteristics are plotted as
function of resistivity.  The horizontal lines at the top of the
graph indicate the voltages at which sparkover occurred for
various temperatures and show a decrease in sparkover potential
with increasing temperature.  The dashed curve indicates the
potential between the outer surface of the dust layer and the
corona wire at the formation of back corona or sparkover for a
3 mm dust layer with a dielectric strength of 24 kV/cm.  The
solid curve is an estimate of the average voltage at the formation
of back corona.  The data points for the voltages at which back
corona occurred are denoted by circles and the voltages at which
sparkover occurred are denoted by triangles.

For ash with a resistivity greater than 2x1010 ft-cm, the current
density at sparkover is large enough that the ohmic voltage drop
in the dust layer exceeds the dielectric strength of the ash layer
and back corona is observed at voltages significantly below the
sparkover voltage.

The dashed curve in Figure 13 was obtained by subtracting the
theoretical voltage drops in the ash layer at sparkover (or at
the formation of back corona if back corona existed before spark-
over)  from the average voltage curve for the formation of back
                              40

-------
   68


   64


   60


   56


   52


   48
UJ
o
CLEAN PLATE SPARKOVER VOLTAGE AT IOO°C


CLEAN PLATE SPARKOVER VOLTAGE AT  160°C



     O VOLTAGE AT FORMATION OF BACK CORNA

     A VOLTAGE AT SPARKOVER
                                                                                 10
                                                                                  15
                                     RESISTIVITY ,ohm-cm

     Figure 13.   Wire plate corona  discharge  device voltages  for  sparkover
                  and for formation  of back corona as a function of
                  resistivity; wire  to plate spacing 6 cm.  Solid  curve
                  estimate of the average  voltage at formation of  back
                  corona, dashed curve potential between dust  surface and
                  wire at formation  of back corona.

-------
  10
    r5
CJ

 o
 \
 <
V)


LJ
O
LU
or
oc

o
  IO'6
   10
    -7
  10"
                                                   D MEASURED CURRENT DENSITY
                                                     AT START OF BACK CORONA
\       X       MAXIMUM CLEAN PLATE CURRENT DENSITY  WITHOUT SPARKOVER

  \       >y ^^- CALCULATED MAXIMUM

     V      ^^"^  PEAK CURRENT DENSITY
      \       \s   WITHOUT BACK CORONA

        \  o  o


—  o      \   cPn
             N   o o

 •               ^



  CALCULATED          v

  MAXIMUM AVERAGE      \

  CURRENT DENSITY

 _WITHOUT BACK CORONA




                                 \  o



  MINIMUM STABLE STARTING CURRENT DENSlV^

  FOR CLEAN PLATE
                     10
                      10
                                     10'
                                              10
                                                      12
10
                                                                13
                                    ASH RESISTIVITY, ohm-cm
     Figure  14.
                  Current densities for  formation of back corona in
                  the wire plate corona  discharge device as  a function
                  of  resistivity
10
                 14

-------
corona.  The voltage drop across the ash layer after the for-
mation of back corona is assumed to remain fixed at a voltage
equal to the product of the layer thickness and the dielectric
strength of the ash.

For ash with a resistivity below 2x1010 fl-cm, the current density
at sparkover produced voltage drops in the dust layer that were
less than the measured breakdown voltage of the dust layer.
However, as shown in Figure 13, the voltages at sparkover were
substantially below the clean plate sparkover voltages and back
corona was observed just before sparkover.  A full explanation
is not available.  Cracks and distortion? in the dust layers
can produce high electric field regions that probably account
for some of the reduction in sparkover voltages.  The reduction
in the sparkover voltage is also partially the result of the
smaller spacing between the wire and dust layer when compared
to the separation between the wire and plate.

In a precipitator, the voltage drop across the dust layer will
vary in time as the thickness of the dust varies and, depending
on the dust resistivity and operating current densities, the
changing voltage drop can account for changes in precipitator
performance with time, especially when a precipitator is first
turned on.

Although there is considerable scatter, the data in Figure 13
show that the potential difference between sparkover and the
formation of back corona increases with increasing resistivity.

Theoretically, back corona occurs when the electric field pro-
duced in the dust layer is greater than the dielectric strength
of the ash.  The average electric field in the dust layer is
given by the product of the current density and the resistivity
                             43

-------
of the layer.  Thus, current density at the formation of back
corona is given by

         . _  breakdown
            p at breakdown
where E^^^^^ is the dielectric strength of the dust layer and
pbreakdown is the resistivity of the ash layer at breakdown.
 The  solid  diagonal  line  in Figure  14  is  a  graph  of  the  above
 expression for E,   ,  ,    =24 kV/cm.  The two dashed horizontal
                 breakdown
 lines  represent  the respective maximum and the minimum  current
 densities  achievable with the wire plate corona  device  used in
 this series of experiments.  These horizontal lines set the
 operating  limits for the device.

 The  squares in Figure 14 represent either  the minimum current
 density obtained if back corona occurred as soon as the corona
 discharge  was initiated  or the current density at which the
 back corona occurred as  determined from  the V-I  curves  and back
 corona current pulses.

 A  comparison of  the solid diagonal line  and the  squares indi-
 cates  reasonable agreement between the theory of the formation
 of back corona as function of resistivity  and the actual processes
 The  two data points near 10 *1 fi-cm that  lie above the diagonal
 line correspond  to  data  from V-I curves  similar  to  the  one shown
 in graph 3 of Figure 11.

 The  average current density  for the total  plate  area of the wire
 plate  discharge  device was 0.3 to 0.5 times  the  peak current
 density.
                             44

-------
Hence, the dashed diagonal line in Figure 14 represents the maxi-
mum average current density at which the device could be operated
without the formation of back corona.  Likewise, the average
operating current density of precipitators is below the peak
theoretical predicted values.

VARIATION IN SPARKOVER VOLTAGE AS A FUNCTION OF THE WIRE TO
PLATE SPACING
The current densities at which back corona developed as a
function of the wire to plate spacing for the corona discharge de-
vice are displayed in Figure 15.   The current densities at which
sparkover occurred are indicated by squares for the clean plate.
The sparkover voltages as a function of wire to plate spacing for
both the clean plate and the ash covered plate are plotted in
Figure 16.

The current density plot suggests that the current density for
the formation of back corona is independent of the wire to
plate spacing as predicted by theory.  Using E = jp to calculate
the electric field in the dust layer when back corona forms, we
obtain a value of 16.5 kV/cm*, which is approximately 6 kV/cm
less than the measured dielectric strength of the dust layer.
The estimated error in the resistivity measurement can account
for the difference between this calculated dielectric strength
and the measured dielectric strength.

Although changing the wire to plate spacing does not affect the
current density at which back corona forms, it does decrease the
voltage at which back corona and sparkover occurs as illustrated
by Figure 16.  The decrease in the sparkover voltage when a dust
layer is present results from the increase in peak current density
that is obtained for a given voltage when the wire to plate spacing
is decreased.  The significant increase .in current density with
* j = 1.1 yA/cm2
  p - l.SxlO10 ft-cm
                              45

-------
      o
      (O
      Z
      UJ
      o

      I-

      UJ
      a:
      a:

      o
         102
         10'
10
 -I
        10
         -2
Q
                           O CURRENT DENSITY AT FORMATION

                             OF BACKCORONA, p/vl.5xlOlon-cm


                           D CURRENT DENSITY AT SPARKOVER

                             FOR CLEAN PLATE
                            2345


                             WIRE TO PLATE SPACING, cm
Figure  15.   Current densities for sparkover and for  formation

             of back corona as a function of wire to  plate spacing,
                                   46

-------
     UJ
     cr
     ui
     o
        80
        70
        60
        50
        40
        30
        20
        10 —
                                          ACLEAN PLATE
                                          O3mm ASH LAYER
                                           P=lxlOl°fi-cm
          0        I        2       3        4        56
                            WIRE TO PLATE SPACING.cm
Figure  16.   Sparkover voltage as a function of wire to  plate spacing,
                                  47

-------
decreasing wire plate spacing is illustrated in Figure 17 by
the current density distributions obtained at a wire potential
of 21 kV with wire to clean plate spacings of 3 cm and 5 cm.
The ratio between the peak current density and the average
current density is also increased by decreasing the wire to
plate spacing.

The effects discussed above can be expected to occur in full
scale units where there is poor alignment of the wire and
plate.  Back corona will form at a lower average current density
than the resistivity of the ash would indicate and the operating
voltage throughout the section with poor alignment will be greatly
reduced.

CURRENT DENSITY VARIATIONS
The current densities to small rectangles on the plate parallel
to the corona wire are plotted in Figure 18 and 19.  The solid
lines in Figure 18 represent the clean plate current density
distributions perpendicular to the corona wire for voltages of
22 kV, 28 kV, 40 kV,. and 52 kV.

The data points represent the current density distributions for
a dust layer with and without back corona.  This data shows that
the effect of back corona is localized in the region in which
the electrical breakdown occurs.  In Figure 19, current density
distributions for a 6x1012 fi-cm resistivity dust layer are plotted,
The area covered by back corona is much larger than in the pre-
vious case.  The outer two segments appear to be free of back
corona.
                               48

-------
      10
        -5
      10
        -6
   CVJ
    E
    o
   >-
   K
   lf>

   LJ
   O
   UJ
   CCL
   a:
      10"
      10
        -8
Figure 17,
      01        2345

             DISTANCE FROM CENTER,cm


Current density distributions:   Wire potential  of
21 kV, clean  plate spacings  of 1)  3 cm and 2) 5 cm
                              49

-------
 2x10
     -5
    10
     ,-5
CVJ
 
-------
 2x10'
   10
    -5
CM
 E
 o
                  - CLEAN PLATE

                    DUST COVERED PLATES

                    WITH BACK, CORONA

                    P=5.6xlOl2Q-cm

                    WIRE TO PLATE

                  O 28 kV

                  O24kV

                  A 22 kV
 UJ
 Q
 UJ
 CC
 cr
 3
 o
   10-6
               O

               A
                                                              52 kV
                                                              40 kV
                                                               28 kV  _
   10
    -7
    -8
                                                               22 kV
   10
  Figure  19
               0
I
                   2345

             DISTANCE FROM CENTER LINE,cm

Current  density distributions for  clean plate  at

several  applied potentials and  for  dust covered

plate with back corona
                                   51

-------
                         SECTION VI
          MEASUREMENTS ON FULL-SCALE PRECIPITATORS
The operating current densities versus resistivity for several
full-scale precipitators are plotted in Figure 20.  The resistivi-
ties were measured using a point-to-plane resistivity probe.21
The operating current densities are the typical spatial and time
average current densities that existed during efficiency tests.
A large portion of the data in this plot came from an ash con-
ditioning study by Dismukes.1

An analysis of the data plotted in Figure 20 is difficult for
several reasons.  There is no well defined point for the most
efficient current density setting when sparking and back corona
occur.  Thus, the operating current densities are set to some
degree by the operator's opinion as to what the best setting is.
A second problem is that in several cases the operating current
densities are set by the power or current capability of the power
supply and not by the electrical characteristics of the precipi-
tator.  A third problem is that the average current density is
plotted; and back corona and sparking depend on peak current
densities, which vary from one unit to another.  In one case,
plant personnel stated that the plates had buckled and that the
wire to plate spacing varied throughout the unit (data point 6).
The operating average current densities were substantially below
those that would normally be expected for the resistivity measured
at that installation.  But, when consideration is given to the
effect that misalignment has on peak current densities, the ob-
served current densities are not surprising.  Another difficulty
is in the variation in resistivity across the width of the pre-
cipitator when a Lundstrom preheater precedes the unit.*  Laboratory
measurements also suggest that variations in resistivity along the
length of the precipitator may occur for temperatures at which
*Documentation showing the location of the resistivity measure-
ments with respect to the precipitator sections was not avail-
able for much of the data plotted in Figure 20.
                               52

-------
U)
        10'
                                                                    10'
                                        RESISTIVITY, ohm-cm
         Figure 20.
Operating current densities as a function of resistivity
for various plants tested by Southern Research Institute.
Numbers refer to data in Table V.  Circles, inlet  sections;
triangles, outlet sections; squares, unknown sections;
solid symbols, either NH3 or SO3 injection.

-------
          TABLE V
PRECIPITATOR ELECTRICAL DATA


Plant
No.
1

2

3
4



5



6


7

8



9

10



11


12

13



Resistivity,
ft- cm
2x10"

5.3x10"

2x10"
7.3xl010

7.3xl010

2.8xl010

1.3xl07

1.2xlO:o
3.0x10'

2.0xl011
4.0xl011
3. 0x10 M

2.2x10"

3.0x10"
3.8xl010
2.8x10°

2.0xl08

2. 0x10 e

8. 0x10 7
1.6x10"

3.0xl010



Electrical
Section
Inlet
Outlet
Inlet
Outlet
Inlet
Outlet
Inlet
Outlet
Inlet
Outlet
Inlet
Outlet
Inlet
Outlet
inlet
Outlet
Inlet
Outlet
Inlet
Inlet
Inlet
Outlet
Inlet
Outlet
Inlet
Inlet
Inlet
Outlet
Inlet
Outlet
Inlet
Outlet
Inlet
Outlet
Inlet
Inlet
Inlet
Outlet


Operating
Voltage, kV
40
36
48
44
50
22
40
46
41
44
39
35
38
36
24
30
30
36
38
41
39
_
42
40
_
35
45
40
47
41
38
29
46
30
37
41
31
29
Operating
Current
Density,
nA/cnr
33
33
49
33
2
25
25
95
5
95
20
33
35
45
8
14
11
13
25
22
40
70
22
60
31
29
30
62
15
60
45
45
20
45
4
38
83
95


Injection







NH3
NH3


32 ppm SOs
32 ppm SO 3

14 ppm SO 3
14 ppm SO 3

14 ppm NH3
14 ppm NH3

13 ppm NHs
13 ppm NH3

5 ppm NHj


20 ppm NH3
20 ppm NH3


10 ppm NH3






Comment
All plants coal
fired boilers
except as noted

Cement plant








Electrode
misalignment
Electrode
misalignment



















          54

-------
surface conduction is important  (< 230°C).

For resistivities less than 109 ft-cm, direct current densities
larger than 1 yA/cm2 are feasible without exceeding the di-
electric strength of the collected dust layer.  However, current
densities in field units did not exceed 0.1 yA/cm2 even for these
low resistivities.

There are several reasons for this, one of which is that several
of the units are equipped with power supplies that are inadequate.
A second reason is the large size of field units.  It is not un-
usual for a power supply to handle a plate area of 300 m2 (3000 ft2)
which is much larger than the 100 cm2 that our laboratory
corona discharge device had.  This increased area and the imprac-
ticality of machining rounded edges on all parts of a field unit
greatly increase the probability for sparkover.   Our laboratory
measurements also indicated that clean plate sparkover voltage
and current density are reduced to some degree even by low re-
sistivity dust layers.

In some installations the current appeared to be limited by space
charge, especially when ammonia was injected as  an agent.
This point is discussed by Dismukes in a recent report.1  He
suggests that NHs combines with S03 to form a fine particulate
which results in a greatly increased space charge.  This in turn
results in increased operating voltages, decreased current den-
sities, and increased sparkover rates.  These occur without a
significant increase in resistivity and show that resistivity is
not the only factor that can increase sparkover rates.  Calcu-
lations of the effects  of particle space charge for cylindrical
corona indicate  that the space  charge results in a higher electric
field at the passive electrode.22  For negative corona, the field
strengths at the anode  would be increased and this could lead to
                            55

-------
 lower sparkover potentials than those obtained without a dust
 loading.

 A comparison of the inlet and outlet data shown in Figure 20 in-
 dicates that in most cases the outlet current density exceeds
 the inlet current density, while the inlet voltage exceeds the
 outlet voltage.  There are two possible explanations for the
 variations in the voltage-current characteristics from the inlet
 to outlet.  One attributes the variations to changes in dust load-
 ing and the other to changes in the thickness and resistivity
 of the collected dust layer.

 The effect of a variation in resistivity from the inlet to the
 outlet is illustrated by the following considerations.  If a
 resistivity of 10J1 ft-cm is obtained in the inlet and 5x1010 fi-cm
 in the outlet, the inlet peak current density could be as high as
 200 nA/cm2 without back corona and the outlet could be as high
 as 400 nA/cm2  for  a breakdown  strength  of  20  kV/cm.

A half order of magnitude change in the resistivity of the
collected dust layer from the inlet section to the outlet section
of a precipitator is reasonable.  Our laboratory measurements
show a decrease in ash resistivity of nearly one order of magni-
tude for particle size fractions with mass median diameters of
40 to 2.3 ym.  The resistivity decreases with decreased particle
size for temperatures  at which surface conduction is important.
Thus,  a decrease in resistivity from the inlet section to the
outlet section results from the fractionation of the ash into a
larger particle size fraction in the inlet section and smaller
particle size fractions throughout the rest of the unit.  In
addition,  a change in  resistivity can result from a temperature
variation from the inlet to the outlet.   For hot precipitators,
the resistivity would  be expected to increase from inlet to outlet
and for cold precipitators, the resistivity would decrease from
the inlet to outlet.

                             56

-------
Average current densities depend on the spatial variation and
time variation of the current.  For full wave rectification
and no filtering, the time average current density is 70.7% of
the peak current.  Typical spatial variations result in spatial
average currents equal to 60% of the peak currents.

Thus, the time and spatial average current equals approximately
50% of the peak current.  For our hypothetical problem, the
maximum possible average currents without back corona are
estimated to be 100 nA/cm2 for the inlet and 200 nA/cm2 for the
outlet.  If the depth of the collected dust layer is the same
in the inlet and outlet, for the above conditions, the voltage
drop across the dust layers is the same in both sections.  For
the same configuration in both sections, the operating voltage
across the discharge should be higher in the outlet since the
currents are higher.  However, this does not agree with most of
the data tabulated in Table V.  Theoretically, the thickness of
the dust layer decreases exponentially throughout the precipi-
tator if rapping variations are neglected.  This results in
larger voltage drops across the inlet dust layers when both
units are operated with current densities near the critical
value for the electrical breakdown of the dust layer.  This can
lead to higher voltages in the inlet sections than in the
outlet sections.

The above explanation for the variations in electrical charac^
teristics from the inlet to the outlet is unsatisfactory for
low resistivities (<5xl09 fi-cm), where voltage drops less than
250 V occur for 0.5 cm thick dust layers and a current density
of 0.1 yA/cm2.  The decrease in the space charge produced by
the charged suspended particulate from the inlet sections to the
outlet sections appears to be the dominant reason for larger
currents and lower voltages in the outlet sections.
                              57

-------
The time required for changes in the voltage-current character-
istics gives some indication of the importance of the above
mechanism.  Dismukes1 observed that when ammonia was injected
the time lag between injection and the effect on precipitator
operation was very short.  In contrast, resistivity effects take
longer to occur, since they depend on the accumulation of dust
on the collecting plates.
                            58

-------
                         REFERENCES

1.  Dismukes, E. B.  Conditioning of Fly Ash with Ammonia.
    In:  Proceedings, Symposium on Electrostatic Precipitators
    for the Control of Fine Particles.  Pensacola Beach.
    Environmental Protection Agency, Washington, D. C.  Publi-
    cation Number EPA-650/2-75-016.  January, 1975.  NTIS PB
    240440/8WP.  p. 257-287.
2.  Kercher, H.  Electric Wind, Back Discharge and Dust Re-
    sistance as Parameters in Electrostatic Precipitators.
    Staub-Reinhalt.  Luft  (in English) .  2_9_(8) : 14-20, 1969.
3.  Penney, G. W., and T. E. Alverson.  Influence of Mechanical
    Collection on Electrostatic Precipitator Sparkover Voltage—
    A Laboratory Simulation.  IEEE Trans. Ind. Gen. Appl. 7^(3):
    433-438, May-June, 1971.
4.  Wolcott, E. R.  Effects of Dielectrics on the Sparking
    Voltage.  Phys. Rev. 12:284-292, 1918.
5.  Franck, S.  Funkenentladungen in Luft-Staubgemischen  [Spark
    Discharges in Air Dust Mixtures].  Z. Phys.  (Berlin).
    8^7:323-229, 1933.
6.  White, H. J.  Characteristics and Fundamentals of the
    'Back Corona1 Discharge.  (Presented at Gas Discharge
    Conference, Brookhaven National Laboratory,  Upton, N. Y.,
    1948.)
7.  Penney, G. W.  Electrostatic Precipitation of High-Resis-
    tivity Dust.  Trans. Am. Inst. Electr. Eng., Part 2.
    713:1192-1196, 1951.
8.  Penney, G. W., and J. G. Hewitt.  Some Measurements of
    Abnormal Corona.  Trans. Am. Inst. Electr. Eng., Part 1.
    77:319-327, July, 1958.
                            59

-------
 9.  Penney, G. W., and S. E. Craig.  Sparkover as Influenced
     by Surface Conditions in D-C Corona.  Trans, Am. Inst.
     Electr. Eng., Part 1.  21:112~118' MaY- I960.
10.  Penney, G. W., and S. Craig.  Pulse Discharges Preceding
     Sparkover at Low Voltage  Gradients.  Trans. Am. Inst.
     Electr. Eng., Part 1.  80^:156-162, May. 1961.
11.  Simm, W.  Untersuchungen xiber des Riickspruhen bei der
     elektrischen Staubabscheidung.[Studies of the Back Spray
     in Electrical Dust Removal] Chem. Ing. Tech. (Weinheim).
     2:43-49, 1959.
12.  Herceg, Z., and R. M. Huey.  Model for Corona Modes in
     Point-to-Plane Device with Coated Electrodes.  Proc. Inst.
     Electr. Eng. (London).  12_0:394-399, 1973.
13.  Hall, H. J.  Trends in Electrical Precipitation of Elec-
     trostatic Precipitators.  In:  Proceedings of Electrostatic
     Precipitator Symposium, Birmingham, Alabama.  February
     23-25, 1971.  p.  75-116.
14.  White, H. J.  Industrial Electrostatic Precipitation.
     Reading, Massachusetts, Addison-Wesley, 1963.  p. 327.
15.  Alston, L. L.  High Voltage Technology, Oxford University
     Press, London, 1968.  p. 48.
16.  Morey, George W.   The Properties of Glass.  2nd Edition.
     Reinhold Publishing Corp., New York, 1954.  p.  532.
17.  Dallavalle, J. M.  Micromeritics.  2nd Edition.  Pitman
     Publishing Corp., New York, 1948. p. 131.
18.  Bickelhaupt, R. E.  Surface Resistivity and the Chemical
     Composition of Fly Ash.  J. Air Pollut. Contr.  Assoc.
     i25_:148-152, February, 1975.
19.  Loeb, L. B.  Fundamental Processes of Electrical Discharges
     in Gases.  John Wiley and Sons,  Inc., New York, 1939.  p. 517
                             60

-------
20.  Bickelhaupt, R. E.   Personal Communication.
21.  Nichols, G.  B.  and H.  W.  Spencer.   Test Methods and
     Apparatus for Conducting Resistivity Measurements.
     Southern Research Institute, Birmingham, Alabama,
     Contract No. 68-02-1083.   Environmental Protection
     Agency, Research Triangle Park,  N.C., 1975.
22.  Lowe, H. J.  and D.  H.  Lucas.  The  Physics of Electro-
     static Precipitation.   Brit. J.  Appl. Physics,  (London),
     Suppl. _2:540-47, 1953.
                             61

-------
                                 TECHNICAL REPORT DATA
                           (Please read Innmctions on the reverse before completing)
 1. REPORT NO.
  EPA-600/2-76-144
                            2.
                                                        3. RECIPIENT'S ACCESSION-NO.
 4. TITLE AND SUBTITLE
 Electrostatic Precipitators:  Relationship Between
 Resistivity,  Particle Size, and Sparkover
                                   5. REPORT DATE
                                    May 1976
                                   6. PERFORMING ORGANIZATION CODE
 7. AUTHOR(S)

 HerbertW. Spencer, HI
                                   8. PERFORMING ORGANIZATION REPORT NO.
                                   SORI-EAS-75-629
                                   3134-XVI
 9. PERFORMING ORGANIZATION NAME AND ADDRESS
 Southern Research Institute
 2000 Ninth Avenue South
 Birmingham, Alabama 35205
                                   10. PROGRAM ELEMENT NO.
                                   1AB012; ROAP 21ADL-027
                                   11. CONTRACT/GRANT NO.

                                   68-02-1303
 12. SPONSORING AGENCY NAME AND ADDRESS
 EPA, Office of Research and Development
 Industrial Environmental Research Laboratory
 Research  Triangle Park, NC 27711
                                                        13. TYPE OF REPORT AND I
                                                        Final; 4/74-12/75
                                                                         PERIOD COVERED
                                   14. SPONSORING AGENCY CODE
                                    EPA-ORD
 ^.SUPPLEMENTARY NOTES JERL-RTP Project Officer for this report is L.E. Sparks, Mail
 Drop 61, Ext 2925.
  . ABSTRACT
               rep0rt gives results of a study of the relationships of the electrical
 resistivity of fly ash, its particle size, the occurrence of back corona and sparkover,
 and the electrical characteristics of electrostatic precipitators (ESP's).  The study
 included laboratory measurement of the dielectric strengths and resistivity of five
 particle-size fractions of a fly ash sample and measurement of the current densities
 and voltages at which back corona and sparkover occurred for a 3 -mm dust layer
 covering the plate of a wire-plate negative -corona discharge device.  Results showed
 that the peak current density for the formation of back corona depended on the resis-
 tivity of the dust covering the positive electrode.  Operating current densities for
 full-scale ESP's are discussed in relation to fly ash resistivity.
                              KEY WORDS AND DOCUMENT ANALYSIS
                 DESCRIPTORS
                                           b.lDENTIFIERS/OPEN ENDED TERMS
                                               c. COSATI Field/Group
 Air Pollution
 Electrostatic
   Precipitators
 Fly Ash
Electric Corona
Dust
Electric Sparks
Flashover
Air Pollution Control
Stationary Sources
Back Corona
Particulate
 Electrical Resistivity
 Measurement
13B
2 IB
20C
14B
         11G
 B. DISTRIBUTION STATEMENT

 Unlimited
                      19. SECURITY CLASS (ThisReport)
                      Unclassified
                         21. NO. OF PAGES

                            68
                                           20. SECURITY CLASS (Thispage)
                                           Unclassified
                                                                    22. PRICE
EPA Form 2220-1 (9-73)
                                          62

-------