-------
         TABLE 1.  SOME CALCULATED ELECTRICAL CHARACTERISTICS OF
                   WIRE-PLATE PRECIPITATORS AT A CONSTANT AVERAGE
                   PLATE CURRENT DENSITY OF 0.4 mA/m2  (0.04 mA/ft2)

2a
in.
0.105
0.105
0.105
0.105
0.105
0.105
0.105
0.070
0.105
0.140
0.175
0.210
0.245
0.280
0.105
0.105
0.105
0.105
0.105
0.105
0.105
0.070
0.105
0.140
0.175
0.210
0.245
0.280
2b
in.
6.0
9.0
12.0
15.0
18.0
21.0
24.0
6.0
9.0
12.0
15.0
18.0
21.0
24.0
6.0
9.0
12.0
15.0
18.0
21.0
24.0
6.0
9.0
12.0
15.0
18.0
21.0
24.0
2c
in.
9.0
9.0
9.0
9.0
9.0
9.0
9.0
9.0
9.0
9.0
9.0
9.0
9.0
9.0
13.50
9.00
6.75
5.40
4.50
3.86
3.38
13.50
9.00
6.75
5.40
4.50
3.86
3.38
Wire
Voltage,
kV
24.
30.
36.
43.
51.
60.
69.
19.
30.
41.
53.
67.
82.
97.
24.
30.
37.
47.
60.
72.
88.
19.
30.
43.
60.
83.
114.
153.
6
1
4
5
5
2
6
7
1
4
8
4
0
8
0
1
6
2
0
7
2
7
1
2
6
9
5
8
Electric Field Average Space
Average, W/m Charge Dens-
*
E
X
111
128
142
155
167
179
189
98
128
154
178
200
221
241
90
128
160
189
217
242
266
81
128
176
233
298
372
456
E
y
50.1
40.5
33.3
27.9
24.3
21.3
19.2
40.8
40.5
38.4
36.3
34.5
33.0
31.8
44.4
40.5
30.6
24.0
19.5
16.8
15.3
38.1
40.5
36.6
33.6
31.8
31.2
31.5
# ity,
p 10~6 Coul./m3
141
169
193
213
232
249
265
131
169
201
229
255
280
303
121
169
206
238
267
290
300
116
169
218
273
336
399
459
8
7
7
6
6
6
5
9
7
6
5
5
4
4
8
7
6
5
4
4
3
9
7
5
4
3
2
2
.1
.6
.1
.7
.3
.0
.8
.2
.6
.5
.8
.2
.7
.3
.6
.6
.4
.5
.8
.3
.9
.4
.6
.8
.4
.5
.8
.3
Total Space
Charge Per
Section,
10 Coul./m
2.
4.
4.
5.
6.
7.
8.
3.
4.
4.
5.
5.
5.
6.
4.
4.
3.
2.
2.
2.
2.
4.
4.
3.
2.
1.
1.
1.
8
0
9
8
6
3
0
2
0
6
0
4
7
0
5
0
4
9
5
3
1
9
0
0
3
8
5
2
* Ex is average field in gap between plates in a direction perpendicular
  to the plates.

+ Ey is average field in gap between plates in a direction parallel to
  the plates.

# Ep is the average total field at the plate.


  All calculations are performed for a gas temperature of 177°C (350°F)
  corresponding to an average mobility of 3.87xlO~4 (m/s)/ (v/m) for nega-
  tive charge carriers.  The emitting electrode smoothness factor is unity.

                                  158

-------
                : 2.  AVERAGE DRIFT VELOCITY AS ft  FUNCTION OF PARTICLE
                     SIZE FOR VARIATION IN_b, WITH FIXED a,  c
                                                                      TABLE 3.   AVERAGE DRIFT VELOCITY AS A FUNCTION OF PARTICLE SIZE FOR
                                                                      	VARIATION IN a AMD b SUCH THAT b/a = 65.7,  WITH FIXED c
2a 2b
in. in.




2c AVERAGE MIGRATION VELOCITY (cm/sec) AS ^
in. METHOD* FUNCTION OF PARTICLE DIAMETER (micron)
2.0 4.0 8.0 12.0 16.0 24.0 32.0 MEAN

B 0.94 2.52 4.02 4.8S 5.81 9.11 12.82 4.81
C 0.58 1.17 2.33 3.50 4.66 6.99 9.32 1.56
9.0 A 0.89 1.53 3-92 5.91 7.60 9.93 11.21 3.75
                           0.84  1.83  5.50    7.70    9.28  15.24  16.47
                           0.69  1.39  2.77    4.16    5.55   6.32  11.09
                                              4.87
                                              2.26
                           0.86  2.11   5.56    8.41   10.95  12.63  16-75  5.52
                           1.37  2.16  4.15    7.03    9.26  13.59  18.56

                           0.95  2.38  6.07    9.15   12.11  16.59  21.87

                           0.86  1.73  3-45    5.18    6.91  10-36  13.81
                                              6.00

                                              3.01
                           1.09  2.52  6.28   10.43   13.66  19.64  23.94  6.58
1.23  2.61  6.70  11.67  15.51  22.45  2B.83

1.04  2.09  4.18   6.27   8.35  12.53  16.71
                                                                                                     2b   2c
                                                                                                     in.  in. METHOD*
                                                                                                AVERAGE MIGRATION VELOCITY (cm/sec)  AS A
                                                                                                  FUNCTIONOF PARTICLE DIAHgTER {micron)
                                                                                           2-0   4.0   8.0   12.0   16.0   24.0   32.0   I
                                                                                                                      0.44  0.89  1.78   2.67   3.56   5.34   7.12  1.32
                                                                                                                      0.52  1.25  2.50   3.74   4.99   7.49   9-99  1.90
0

c
0.175 15.0 9.0 A
C
B
C

	 -
9

1.34 -?.4Q 5.67 8.01 10.83 15.96 21.74 5.65
1.00 2.00 4.00 6.00 8.00 12.00 16.01 3,10
1.40 3.24 8.86 12.70 14.85 23.47 29.02 7,95

                                                                                                                      1.41  2.82  5.63   8.45  11.26  16.89   22.52   4.35
                                                                                           2.17   3.89   8.13   11.86   14.61   24.21  28.75
                                                                                           1.87   4.53  10.92   15.37   19.03   30.53  37.33
                                                                                                                                                                    8.75

                                                                                                                                                                   10.67

                                                                                                                                                                    5.00
 *  A -  trajectory method for  Poisuelle  flow field

   C   drift velocity method  for uniform flow field
       i  log  normal size <
                                                            and o = 2.0.
   Average velocity and average plate current density for  all  determinatii
   are  1.2 m/s  (3.94 ft/s) and 0.4 raA/m2  (0.04 mA/ft2)  respectively.
                                                                ' A - trajectory method for  Poisuelle  flow  field
                                                                  B - trajectory method for  uniform  flow field
                                                                  C - drift velocity method  for  uniform flow field
                                                                  Mean value of drift velocity is  evaluated on the basis of number density
                                                                  for a log normal  size distribution with 650 = 5,0 micron and a = 2.0.

                                                                  Average  velocity  and average plate current density for all determinations
                                                                  are 1.2  m/s (3.94 ft/s) and 0.4 mft/m2 (0.04 mA/ft2) respectively.
     TABLE 4.  AVERAGE DRIFT VELOCITY AS A FUNCTION OF PARTICLE SIZE FOR
               VARIATION IN b AND c  SUCH THAT be  =  20.25,  WITH FIXED c
                                                                 TABLE 5.  AVERAGE DRIFT VELOCITY AS A FUNCTION OF PARTICLE SIZE FOR
                                                                 __	VARIATION IN a, b AMD C SUCH THAT b/a - 85.7 AND be = 20.25

2a 2b








0.105 15.0
Q.105 18.0
0.105 21. G

0.105 24,0



2c
in. METHOE


B




B
C
5.40 A
B
C
4.50 A
B
C
3.86 A
B

3.38 A



AVERAGE MIGRATION VELOCITY (cm/sec) AS A
)* FUNCTION OF PARTICLE DIAMETER (micron)
2.0 4.0 3.0 12.0 16.0 24.0 32.0 MEAN

0.87 2.25 4.27 4.37 5.20 6.72 7.09 4.66

0.89 1.53 3.92 5.91 7.60 9.93 11.21 3.75


0.92 2.11 S.89 8.68 11.62 13.85 16.46 5.27
0.79 1.58 3.17 4.75 6.34 9.50 12.67 2.47
1.34 2.38 5.26 7.77 9.99 14.49 17.62 5.33
1.12 2.78 6.44 9.49 12-74 17.30 23.83 o.62
0.99 1.98 3.97 5.95 7.93 11.90 15.87 3.08
1.67 2.93 6.56 9.97 12.66 17.79 25.31 6.49
1.36 3.51 8.89 12.93 16.56 23.74 32.95 8.10
1,21 2.41 4.83 7.24 9.65 14.48 19-31 3.74
1.99 3., 32 7.98 11.6.9 15.65 23.13 30.81 7-60
1.61 4.22 10.72 15.65 20.66 31.58 41.14 9.62

2,18 3-71 9,41 13.91 18.50 26.63 33.92' 8.67


2a 2b 2c
in. in. in. METHO
0.070 6.0 13.50

0.140 12.0 6.75

0.210 18.0 4.50

0.280 24.0 3.38
A
B
C
B
C
A
B
C
B
C
A
B
C
B
C
A
B
C
D*
2.0
0.63
0.58

0.75
0.62
1.23
1.10
0-96

1,44
2.59
2.83
2.13
4.66
3.05
5.12
7.30
4.18
. AVERAGE MIGRATION VELOCITY (cm/sec
FUNCTION OF PARTICLE DIAMETER (mi.
4.0
1.33
1.98

2.31

2.19
3.36
1.91

2.88
5.46
7.40
4.25
12.04
6.10
12.20
14.38
8.36
8.
3.
3.

4.

5,
6.
3.

5,
.0
,33
96

.52

.58
,62
.82

.76
11-70
14.09
8.50
20
12
.22
.20
22.56
28.83
16.72
12.0
1.26
5.54

6-16

8-42
9,86
5.73

8.63
16.35
20.01
12.75
32.65
18.31
30.41
43.94
25.09
16.0
4.74
5.15

7.37

9.66
10.02
7.64

11-51
22.47
26.01
17.00
41.17
24.41
40.35
48.05
33.45
24.0
7.30
10.73

10.74

15.04
19.66
11.47

17,27
32.25
40.79
25.51
45.38
57.97
36.61
62.95
50.17
) AS A
32.
U
7.48
7.78

14.

19.
19.
15.

23.
37.
51.
34,
56.
68.
48,
75
66

19

39
37
29

03
,74
,63
.01
.38
.92
.82
.84
.89
MEAN
2.87
3.91

4.61
1.90
5.18
6.53
2.77
7.18
9.77
3.87
9.97
13.73
5.29
14.27
17.63
7.04
18.82
9.01
* A - trajectory method  for Poisuelle  flow  field
  B - trajectory method  for uniform  flow  field
  C - drift velocity method for  uniform flow field
  Mean value of drift velocity is evaluated on  the basis
  for a  log normal size  distribution with £59 = 5.0 micr.
                             of number density
                             n and O   2.0.
  Average velocity and average plate current density  for  all  d
  are 1.2 m/s  (3,94 ft/s} and 0.4 mA/m2  (0.04 mA/ft2)  respect!
                                      irrent density  for  all  determinatii
'  A -  trajectory method  for  Poisuelle  flow  field
  B -  trajectory method  for  uniform flow  field
  C -  drift velocity method  for  uniform flow  field
  Mean value of  drift velocity is  evaluated on  the basis of number density
  for  a log normal  size  distribution with d50 - 5.0 micron and a = 2.0.

  Average velocity  and average plate current  density for all determinations
  are  1.2 m/s (3.94 ft/s) and 0.4  mA/m2 (0.04 mA/ft2) respectively.
                                                                                159

-------
      SIMULTANEOUS MEASUREMENTS OF AERODYNAMIC SIZE AND ELECTRIC CHARGE
        OF AEROSOL PARTICLES IN REAL TIME ON  A  SINGLE  PARTICLE  BASIS
                                     By
                M. K. Mazumder, R. G. Renninger, T* H. Chang
            R. W. Raible, W. G. Hood, R. E. Ware, and R. A. Sims
                Department of Electronics  and Instrumentation
           University of Arkansas  Graduate Institute of Technology
                               P. 0. Box 3017
                         Little Rock, Arkansas 72203
                                  ABSTRACT
      An  instrument has been developed for measuring the aerodynamic
 relaxation  time  T  and electrical mobility Z of aerosol particles in real
 time  and on an individual particle basis in the range from 0.3 to 10.0 pm
 in diameter.  The instrument, an electrical single particle aerodynamic
 relaxation  time  (E-SPART) analyzer, employs a laser Doppler velocimeter
 (LDV)  for measuring particle motion in an applied electric field.  It
 operates by electrically charging the particles, subjecting them to an
 oscillatory electric field and then measuring the amplitude V  and relative
 phase  lag (j> of the particles with respect to the electric field.  The phase
 lag  and
 Vp, computes d^ and q for individual particles, and stores the data to
 generate the size and charge distributions of the aerosol particles.  The
 application of this instrument in an electrostatic precipitator is briefly
 discussed.
                                INTRODUCTION


     Measurement of the charge distribution acquired by the particles
inside an electrostatic precipitator may give an insight into the
effectiveness of charging as a function of the magnitude and polarity of
the electric field, the geometry of the discharge electrode, and the
resistivity of the flyash inside the precipitator.  Such measurements can
also be used as a diagnostic tool for determining the effect of back
corona, particle sneakage, and reentrainment.  Particle charge is usually
measured using the Faraday cage method originally developed by Masters (1)
and later modified by a number of researchers (2).  Recently McDonald,
Anderson, and Mosley (3) reported a method of measuring the charge on'
individual particles using a Millikan cell.  They applied this technique to


                                    160

-------
measuring the charge on flyash particles  in  the range 0.3 to 1.5 um at
ambient temperatures ranging from 38°C  to 343°C.  While producing useful
data, this method is inherently slow because  it requires the visual
observation of individual particles inside the Millikan cell.  The method
discussed here is capable of measuring  a  particle's charge and its
aerodynamic diameter on a single particle basis and at a rate of 100
particles per second.  Similar in principle  to a Millikan cell, the method
measures the electrical mobility of charged  particles in an electric field
at 40 kHz to determine the charge and also measures the phase lag to
determine the aerodynamic diameter.
                           PRINCIPLES  OF OPERATION
     The basic principle  for measuring  the  size as well as the charge on an
 individual particle basis  is similar  to that of the single particle
 aerodynamic relaxation time  (SPART) analyzer (4).  The SPART analyzer
 determines aerodynamic diameter by subjecting particles to an acoustic
 field of known frequency.  The oscillatory  motion induced on the particle
 is measured by using a frequency-biased differential laser Doppler
 velocimeter, LDV,  (5).  The particle  motion lags behind the acoustic
 excitation because of its  inertia, and,  consequently, the phase lag <)>
 between the particle motion and the acoustic excitation provides a
 measurement of the aerodynamic diameter of  the particle.  In the present
 technique, an electrical  excitation replaces the acoustic excitation.
 Since it is possible to measure both  the velocity amplitude and the phase
 lag  of charged particles  inside the analyzer, the electrical SPART analyzer
 (6)  is capable of measuring both  the  electrical mobility Z and the
 aerodynamic relaxation time T  of individual charged particles in real
 time.  From the measured value of T   and Z, the aerodynamic diameter d& and
 the  particle charge q are  determines!.
 Measurement of Electrical Mobility
     The equation of motion of  an  electrically charged particle suspended
 in  air within a uniform AC electric  field  can be written as
     T  (dV /dt) + V  = ZE,sinu)t                               (1)
      p   p        p     d


where T  is the aerodynamic  relaxation  time  of  the particle, given by



     TP = P0da2cc


where d  is the aerodynamic  diameter, pQ  equals 1 gm/cc, Cca represents the

                                     161

-------
Cunningham correction factor for molecular slip evaluated as a function  of
dg, and n is the viscosity of air.  In Equation 1, V  is the particle
velocity, E, is the amplitude of the driving field and to is the angular
frequency of the electric field.  The electrical mobility Z is a function
of both the charge q and the particle size.  For a particle of aerodynamic
diameter d ,
     Z - qC,,a/3TTTidQ .                                          (3)
           C 9     9
The steady-state solution of Equation 1 can be written as


     V  (t) =  [ZEd/(l + u)2xp2)1/2] sln(o)t - *) ,                (4)
where
      tan  = COT   .


The phase  lag $  is therefore independent of both the driving field
amplitude  E, and the particle charge q.  From Equation 2, the amplitude of
the particle velocity is given by


      V  =  ZEd/(l + u>2Tp2)1/2                                  (5)


which depends upon q, E,, and d  .  All other quantities are known.  Since 
is measured independently from v , it is possible to measure the electrical
mobility or the  particle charge  q by using Equation 5.

      Since both  the phase lag  and the velocity amplitude V  are available
for each particle, the electrical SPART analyzer can measure the
aerodynamic diameter and the electric charge of individual aerosol
particles  in real time.
                             EXPERIMENTAL SETUP
      Figure  1  shows  the  relaxation  chamber of  the electrical  SPART
 analyzer.  An  aerosol  charging  device,  similar to that described by Langer
 et  al.  (5) is  currently  used  for  charging the  particles so  that the size
 and charge distribution  measurements  can be performed on particles that
 initially are  electrically neutral.   The charging section will be
 disconnected when  the  electrical  SPART  analyzer is used to  measure the
 charge  and size  distribution  of particles that are sampled  from inside of

                                     162

-------
Figure  1.  Schematic  of the  relaxation cell  of the  E-SPART
      analyzer for measuring  charge  arid aerodynamic  diameter
      simultaneously.
                          BEAM
                         SPUTTER
ARGON ON LASER


   23 »W
               RELAXATION CELL-v/-
                            OEFLECTON PLATES
                                                     RFTO
                                                     SIGNAL PROCESSOR
                                                     EXCITATION SIGNAL

                                                         NCETO
                                                     SIGNAL PROCESSOR
                    SOEVEWOF

                   RELAXATION CELL
Figure 2.   Schematic of  the LDV  optics  and electrical
      high  voltage  drive  circuits of E-SPART analyzer.
                                 163

-------
an electrostatic precipitator.

     In the charging section,  the aerosol is first exposed to a DC corona
established at a pointed electrode as shown in Figure 1.   The aerosol
delivery tube to the relaxation cell is made of copper and serves as the
ground electrode.  The charger is mounted at the top of the relaxation
chamber to minimize particle deposition inside the delivery tube between
the charger and the sensing volume of the LDV-  The sensing volume is
positioned between electrodes across which a high voltage signal (6500
V _ ) at 40 kHz is applied.  The two electrodes are positioned
symmetrically across the LDV sensing volume.  Figure 2 shows a block
diagram of the high voltage drive circuits, the laser Doppler velocimeter,
and the associated electronics.  Figure 3 shows the electronic data
processing and the microprocessor based storage and display system.

     A signal generator at  40 kHz,  an RF power amplifier  and a step up
transformer provide the high voltage drive input to the electrodes.  A
voltage divider is used to  monitor the applied voltage and to serve as  the
reference for measuring the relative phase difference between the particle
motion and the electrical drive.

     As the aerosol sample  passes through the charger,  the particles
acquire an electric charge.  The charged particles experience the
oscillating electric field  while passing through the LDV sensing volume.
Light scattered by particles transiting the sensing volume is detected  by a
photomultiplier tube whose  output is connected to the signal processing
circuits.  These circuits,  shown schematically in Figures 2 and 3,  recover
the phase and amplitude of  each particle's induced velocity.

     The phase measuring circuit measures the relative  phase difference
between the particle motion and the electrical drive, and assigns a channel
number based on particle size corresponding to one of 128 microcomputer
memory locations.  During the phase measurement, an analog-to-digital
converter (ADC) is used to  measure the velocity amplitude of the particle
and its output is processed through the microcomputer,  which determines the
particle charge based on the measurement of V  and the size information
given by channel number^
                           RESULTS  AND  DISCUSSION
     To determine the size resolution of the electrical SPART analyzer,
monodisperse aerosols containing polystyrene latex spheres (PLS)  were used
as standards.  Figure 4 shows the typical size resolution of the
instrument, giving the distribution dN/d(log d )  of PLS aerosols  containing
0.82, 1.09, and 2.02 ]im particles.  The distributions show that the
resolution obtained was similar to that obtained  from the acoustic SPART
analyzer (4).  Figure 5 shows charge distributions measured on PLS
particles of different sizes.  The experimental results on charge indicate
that particle charges were higher for larger particles, but because of the

                                    164

-------
 Figure 3.  Signal  and data processing circuits for  E-SPART
      analyzer.
           0.10
                                         2.02
                  O.JO
                           0.50     1.0     i.O
                        AERODVNAmC DIMETER UIKDBIETERS)
Figure 4.   Size distributions of polystyrene latex
     sphere (PLS) aerosols  as measured by  the
     E-SPART analyzer.
                              165

-------
                                     2.02
           it.
                  tt.
                                         100.
                           CHMCE (ELEnENTMlY UNITS)
                                                  500-
                                                        1000.
Figure  5,   Charge distributions  of polystyrene latex
     sphere (PIS) aerosols as measured by the E-SPART
     analyzer.
                               166

-------
lack of any instrument which could be used to measure the particle charge
independently, it was difficult to perform a comparative test.  Experiments
are now in progress to use a charge analyzer having a Millikan cell similar
to the one developed by Gooch et al. for comparative studies.
                                 CONCLUSION
     Preliminary test results indicate that the electrical SPART analyzer
can be conveniently used for measuring particle charge and size simulta-
neously in real time.  It will also be possible to use the electrical SPART
analyzer to determine the fraction of particles of a given size that
acquire charge by noting the number of particles that pass through the
sensing volume without any oscillatory motion.
                               ACKNOWLEDGEMENT
     The authors thank J. D. Wilson, K. Kalb, and R. Chen for their
 technical assistance and constructive criticism, and P- Roberson, M. Elms,
 D. Watson, and P. T. Archer for their assistance in preparing the
 manuscript.  This work has been supported by the U. S. Environmental
 Protection Agency under Grant No. R806192.
                                  ENDNOTES
 1.  Masters, J. I.  An Aerosol Analyzer.  Rev. of Sci. Instrum.
    24:586-588, 1953.
 2.  Langer, G., J. Pierrard, and G. Yamak.  Further Development of An
    Electrostatic Classifier for Submicron Airborne Particles.  Int. J.
    Air Wat, Poll.  8:167-176,  1964.


 3.  McDonald, J. R., M. H. Anderson, and R. B. Mosley.  Charge
    Measurements of Particles Exiting Electrostatic Precipitators.  EPA
    600/7-80-077, 1980.


 4.  Mazumder, M. K., R. E. Ware, J. D. Wilson, R. G. Renninger, F. C.
    Killer, P. C. McLeod, R. W. Raible, and M. K. Testerman.  SPART
    Analyzer:  Its Application  to Aerodynamic Size Distribution
    Measurement,  J. Aerosol Sci.   10:561-569, 1979.
                                     167

-------
5.  Mazumder,  M.  K.   Laser Doppler Velocity Measurement Without
    Directional Ambiguity by Using Frequency Shifted Incident Beams.
    Appl. Phys. Letts.   16:462,  1970.
6.  Renninger,  R-  G.,  M.  K.  Mazumder,  and M.  K.  Testerman.   Particle
    Sizing by Electrical  Single  Particle  Aerodynamic Relaxation Time
    Analyzer.  Rev-  Sci-  Instrum.   In  press.
                                    168

-------
    APPLICATION OF LASER DOPPLER INSTRUMENTATION TO PARTICLE TRANSPORT
              MEASUREMENTS IN AN  ELECTROSTATIC PRECIPITATOR
                                    By
                M.  K.  Mazumder, W. T.  Clark III, R.  E. Ware
          P. C. McLeod, W. G.  Hood, J.  E.  Straub, and S. Wanchoo
               Department of Electronics and Instrumentation
          University of Arkansas Graduate Institute of Technology
                               P.  0.  Box 3017
                        Little Rock, Arkansas 72203
                                  ABSTRACT


     The two-color frequency biased laser Doppler velocimeter (LDV)  and the
single particle aerodynamic relaxation time (SPART) analyzer are two laser
based instruments.  Their development and application to electrostatic
precipitators are briefly discussed.  The LDV is designed to provide
simultaneous measurements of two orthogonal velocity components:   one along
the flow direction and the other along the electric field direction.  The
instrument is mounted on a three-dimensional traversing stage which permits
measurements of flyash trajectories at any desired point inside the
precipitator.  The SPART analyzer has been developed for measuring,  in real
time, the aerodynamic size distribution of aerosol particulates in the
respirable range 0.3 to 10.0 ym in diameter.  The instrument is capable of
measuring and storing the size distributions of aerosols sampled from the
two ends of the precipitator and computing the fractional efficiency as a
function of particle size.  Experiments are in progress on the measurements
of particle trajectories and penetration  inside a flow model precipitator.
The preliminary experimental data are presented.
                                INTRODUCTION


     Recently there has been considerable interest in understanding the
 particle transport properties, particularly the role of turbulence, that
 affect particle migration velocity  inside an electrostatic precipitator
 (1,2).  Both mathematical modeling  and experimental observations are needed
 to understand the complex interaction of the Coulomb force on the particle,
 the  turbulent diffusion, and the  electric wind that govern the particle
 motion.  Experimental studies on  the particle transport and fluid flow
 properties are difficult to perform because ambient conditions inside the
 precipitator prohibit conventional  instruments, such as the hot wire
 anemometer, to be placed inside the precipitator  The high voltage applied
 to the discharge electrodes, the  high electrical  field, and the high

                                     169

-------
particulate concentration inside the precipitator are not conducive to
placing any physical probe in the precipitator for studying the flow
properties.  The laser Doppler velocimeter (LDV) overcomes most of the
difficulties associated with making particle transport studies inside the
precipitator.  The LDV can measure the localized particle velocity at any
given point inside the precipitator; however, such measurements may not
yield the properties of fluid flow since the electrical field acting on the
charged particle deviates the particle motion considerably from the flow of
fluid.  The objective of this project was to develop a two-dimensional LDV
and particle seeding techniques for measuring:  a) electrical migration
velocity components, b) turbulence, and c) electrical wind inside a
precipitator.

     By sampling aerosol from both upstream and downstream of  the
precipitator, a single aerodynamic relaxation time (SPART) analyzer was
used to measure in real time the fractional efficiency of an electrostatic
precipitator.  A brief discussion of the experimental design and
preliminary data are presented.
                             EXPERIMENTAL  SETUP
     Experiments are being carried out using a flow model electrostatic
precipitator designed and built in such a manner that the particle
transport and turbulence properties can be studied along the entire length
of the precipitator using a two-dimensional LDV-  The experimental
arrangement consists of the precipitator, the LDV mounted on a traversing
stage, an aerosol generator for seeding the precipitator, and a SPART
analyzer for measuring the fractional efficiency.
Flow Model Electrostatic Precipitator
     A laboratory model electrostatic precipitator (Figure 1)  made of
Plexiglas was used with the following dimensions:   46 cm in height,  2.7 m
in length with wire-to-distance variable from five to 13 cm and a wire-to-
wire spacing of 15 m.  The collection electrodes were made of  six mm-thick
plates of NESA glass which had a conductive coating for electrical
connection but were transparent, thus permitting optical probing with an
LDV.  The mean velocity inside the precipitator was adjusted to
approximately two m/sec with the free stream turbulence intensity
approximately three percent.  The high voltage applied to the  discharge
electrodes, with wire diameter of 0.25 cm,  could be varied from zero to -
50kV.  The applied high voltage and the corona current collected by each
plate were measured (Figure 2)•
                                    170

-------
Figure  1.   Schematic diagram of the  laboratory  model
     electrostatic precipitator.


i
It!
1
in
1
Ul
§
M
£
2.50
2.25
2. P0.
1.75

1.50
1.25
1.00
0. 750
0. 500 >
            O SIT PBECIPITflTOR
                                     SRI COWUTE8 MODEL
         (.250
           25.0  27.5   30.0  32.5  35.0   37.5  10.0  12.5  «. 0  47,5   50.0
                            flPPLIEO UOLTOGE IKILOUOLTS)
Figure 2.   Experimentally determined V-I  curve of  the
      flow model  precipitator compared with  the computed
      results  from SRI  mathematical  model.
                                171

-------
Laser Doppler Velocimeter
     An LDV (Figure 3) for measuring two-dimensional velocity components is
used to measure the velocity distribution along the flow direction (x-axis)
and the direction of electrical migration velocity (y-axis) of the
particles.  The transmission optics of the LDV contain two dual-beam,
frequency-biased (3) LDV's with their optical axes at +45  to the y-axis.
The receiving optics consist of two light-collection and detection optical
systems that are mounted on a platform opposite the transmission optics.
This arrangement allows the forward scattered light to be collected by the
receiving optics.  A three-dimensional translational stage is used for
mounting the LDV transmission and receiving optics; the stage also
facilitates velocity mapping in the x-y-z-directions.

     The optical system is designed for two-color,  two-dimensional,
simultaneous velocity measurement on a given particle.  The system
currently is used with only one color (514.5 nm) while velocity
measurements are performed at +45  and then at -45° sequentially.  The
experimental data are then processed to determine the x-y velocity
components on a statistical basis (4).  The measurement process requires
the use of monodisperse aerosols with no alteration of process variables
between two successive measurements.
Aerosol Generators
     Currently two aerosol generators are used to seed the flow model
precipitator.  One, a Rapaport-Weinstock aerosol generator, produces bis
(2-3,ethyl hexyl) sebacate (BES) aerosol in the size range 0.1 to 1.0 ym in
aerodynamic diameter by a vaporization-condensation process.  The second is
a Wright Dust Feed mechanism which is used to generate flyash and talcum
powder aerosol in the size range 0.5 to 5.0 Urn in aerodynamic diameter*
SPART Analyzer
     The SPART analyzer determines the aerodynamic diameter (d )  by
measuring the relaxation time (T ) of individual particles and3droplets in
real time.  The aerodynamic diameter is defined as the diameter of a unit
density sphere having the same aerodynamic properties as the particle in
question.  The aerodynamic diameter takes into account the size,  shape,
density, and surface properties that effect the behavior of a particle in
air.  The principle of operation and the description of the SPART analyzer
(5) were previously discussed.

     The performance of the SPART analyzer can be summarized as follows:
The dynmamic size range of particles measured by the analyzer is currently

                                    172

-------
        COLLECTOR
Figure  3.   Schematic of  the LDV transmission  optics
      for measuring two-dimensional  velocity components.
            10.


            9.0

            e.o


            T.O


            e.o


            s.o


            +.0


            3-0


            2.0


            1.0


            0.
                   -7.50   -s. oe   -z.so    o.     J.BO   B.JO   i-.so    10.0
                              DISTANCE FROM CENTER (IN)
Figure  4.   Velocity profile inside  the flow model
      preci pita tor as measured by a  hot wire anemometer.
                               173

-------
0.3 to 5.0 ym in aerodynamic diameter measured in 119 channels or in 30
channels.  The resolution of this instrument is dependent upon the
frequency of acoustic excitation.  The current instrument, operated at a
frequency of 23 kH, has a maximum resolution in the range 0.5 to 2.0 ym.
The sensing volume of the instrument is approximately 5 x 10   cc.  The
instrument samples at a rate of 200 cc per minute.  The highest count rate
is approximately 400 counts per second, although in actual measurements the
count rate is never allowed to exceed 100 particles per cc in order to
reduce the coincidence error below five percent.  The maximum number of
counts in any channel is approximately 1,000,000.  The instrument can store
two size distributions, such as one for aerosol at the upstream and the
other for aerosol at the downstream of the precipitator.  It can operate in
automatic or in manual modes.  Data from the SPART analyzer can be obtained
on a line printer or on punched paper tape in the following format:
     1)  the total time in seconds during which the aerosol has been
         sampled,
     2)  the channel number,
     3)  the diameter in micrometers,
     4)  the total number of counts in that channel,
     5)  dN/d(log d ), and
     6)  dV/d (log 3a).
Items 2 through 4 are repeated for all channels.  The output of the
analyzer is connected to a microcomputer which reads  the data directly from
the microprocessor memory and computes 1) the count median aerodynamic
diameter (CHAD), 2) the mass median aerodynamic (MMAD),  3) the standard
deviation (a ), and mass concentration in yg/m .   The microcomputer  is
programmed to compute the fractional efficiency of the precipitator  using
the size distributions of the aerosol present at the  upstream and the
downstream of the precipitator.
                           RESULTS AND DISCUSSION
Velocity Profile
     Only preliminary experimental data are presented here.   Figure 4 shows
a typical high velocity profile that was measured with a hot wire
anemometer across the precipitator.  No high voltage was applied and the
precipitator was not seeded with aerosols.  The velocity profile was mapped
at different values of x from the inlet of the precipitator  and it was
found to be fairly uniform.
Migration Velocity
     Measurement of migration velocity is currently performed on aerosol
particles inside the precipitator while the mean velocity of air along the
x-axis is maintained at or near zero.   In this experiment,  the precipitator

                                    174

-------
is seeded locally with a line source of aerosol placed at an upstream point
with respect to the point of measurement and close to a discharge wire.
The blower, used to maintain the air flow through the precipitator, is
turned off during this experiment.  The aerosol particles, after injection
inside the precipitator, acquire charge in the electrical field and migrate
immediately in the direction of the wall.  Effective velocity is determined
by the electrical migration velocity and by the electrical wind.  The
velocity components in the x and y directions are measured at several
points.  Figure 5 shows a map of velocity measurements at a given applied
voltage.  Figure 6 shows the change of velocity at four given points as a
function of applied voltage.  It is of interest to note that there is a
velocity (w ) reversal in Figures 6 and 7 which suggests a strong effect
from the electric wind inside the precipitator.
Measurement of Aerodynamic Size Distribution and Fractional Efficiency
     The aerodynamic size distribution measurements of the test aerosols
 used in the precipitator were performed with the SPART analyzer.  For
 measuring the fractional efficiency  of the precipitator, a polydisperse
 aerosol in the size range 0.5 to 5.0 ym was used and the size distributions
 at  the upstream and downstream were  measured.  The calculated fractional
 efficiency is shown in Figure 8.   Since the measurements are made in real
 time, it is possible to determine  the fractional efficiency as functions of
 high voltages applied to the discharge electrodes and the physical
 characteristics of test aerosols.
                                 CONCLUSION
     Preliminary experimental  data  indicate  the feasibility of applying the
 LDV based techniques  to measuring particle transport inside the
 precipitator and to measuring  the fractional efficiency of the precipitator
 in real  time.   It appears possible  that  such measurements could yield data
 which may augment the recently developed mathematical models incorporating
 turbulent diffusion.
                               ACKNOWLEDGEMENT
     The authors wish  to  thank R. W. Raible, R. L. Bond and R. A. Sims for
 their  technical contributions in the project; P. Roberson, M. Elms, P. T.
 Archer, and D. Watson  for editorial assistance.  The authors appreciate the
 valuable guidance and  several comments  received during the entire phase of
 this work from L. E. Sparks, Project Monitor.  This work was supported by
 EPA Grant R806192.
                                     175

-------
               ELECTRICAL MIGRATION PARTICLE VELOCITY PROFILE
                              Hx « Electrical Dov.-natream Velocity
                              Hy - Electrical ;El8ratlon Velocity (
     Vortical i»rray
     of Horizontally
rfj
v*^1
ectrode




t
1"
t



^
0

0



',
-*— r— ».
1
i.^i

!3 0
1.23
74 0
X5^


)5 0
1
13
1.19

24 0
1.15
13 0
1.33 p
'

OE c

22 0.
1.43

15 0
0.12
37 0.
1. 11 m


33 i).

06 0.
1 \

31
1,08
2G 0
.02


2 0
"

I

1.25 ^
43 -0
1.37 j.


11.07 ^

32 -JU 0.

""Pi-charge Electrode

1-13 ^ R
32
.36 ^ r


no
01 ^ n

3
                             Collection Electrode •
Figure 5.   Two-dimensional  migration  velocity components
      (Wx and Wy)  measured  at 2.5cm x  2.5cm  grids inside
      the precipitator.
   6
   t-i
   i
      -t. sot
         25.0
              27.S  30.0  32. S  35.0  3?.S  40.0  42.S  15.0  47.S  50.0
                             APPLIED UQLTAtE CKU)
Figure 6.   Variation of Wy with applied  voltage measured
      at  different points  fsee  Figure 5)  inside the
      precipitator.
                                  176

-------
     *  0. TSO
       O.SOO
       0.250
       -O.SSO
                              30.     SS.
                            APPLIED WH.TMC 001)
                                                         so.
Figure  7.   Variation of  Wx and Wy  with applied voltage.
     Wy was negative (particles movina away  from the
     wall) at  the point  6D (see Figure 5).
      100.


       90.


       10.


       70.


       so.


       50.


       40.


       30.


       20.


       10.


       0. r
                       2  SECTIONS,  50 KU
        0.10
               0.20
                        0.50     1.0     2.0
                          DIWETER CIIIttPIETERS)
                                                5.0
                                                       10.
Figure  8.   Fractional efficiency  of the flow model
     precipitator as measured by  the SPART  analyzer.
                               177

-------
                                  ENDNOTES
1.  Robinson, M.  Turbulence in Electrostatic Precipitators.  Minerals
    Processing.   9:13,  1968.
2.  Leonard, G.,  M. Mitchner,  and S.  A.  Self.   Particle Transport in
    Electrostatic Precipitators.   Atmospheric  Environment.   14:1289,
    1980.
3.  Mazumder, M. K.   Laser Doppler Velocity Measurement Without
    Directional Ambiguity by Using Frequency Shifted Incident Beams.
    Appl. Phys. Letts.   16:462,  1970.
4.  Johnson,  D.  A.  and W-  C.  Rose.   Measurement  of  Turbulent  Transport
    Properties in a Supersonic Boundary-Layer Flow  Using Laser
    Velocimeter  and Hot Wire  Anemometer Techniques.  AIAA  Paper 73-1045,
    1973.
5.  Mazumder,  M.  K.  and K.  J.  Kirsch.   Single Particle Aerodynamic
    Relaxation Time  Analyzer.   Rev.  Sci.  Instrum.   48:622,  1977.
                                    178

-------
                      THE APPLICATION OF MEASUREMENTS OF_
                     AEROSOL CHARGE ACQUISITION BY BIPOLAR
                      IONS TO THE PROBLEM OF BACK CORONA

            By:  R.A. Fjeld
                Environmental Systems Engineering
                Clemson University
                Clemson, SC   29631

                R.O. Gauntt, G.J. Laughlin, A.R. McFarland
                Air Quality Laboratory
                Texas A&M University
                College Station, TX   77843

                                   ABSTRACT

     Measurements relevant to the impact of bipolar ions on particle charge
acquisition are presented.  Experiments were conducted to determine charge
acquisition by aerosols exposed simultaneously to positive and negative ions
of unequal current  densities.  Initially uncharged, highly monodisperse
aerosols in the micrometer to submicrometer diameter size range were subjected
to counter currents of positive and negative ions in the presence of an
electric field.   Measured charge is found to be significantly larger than that
calculated by classical bipolar field charging theory.  Substantial charge
degradation can occur under bipolar conditions, but the extent of degradation
exhibits a strong time dependence.  The time behavior could be an important
factor in assessments of the effect of back corona on charge levels in elec-
trostatic precipitation.

                                 INTRODUCTION

     Back corona occurs in electrostatic precipitators when a large voltage
drop develops across the dust layer on the collecting electrode.  The voltage
drop is due to highly resistive particles or high ionic current densities,
and it causes dielectric breakdown in the dust.  A counter current of ions of
opposite polarity results, transforming the formerly unipolar charging region
to a bipolar region.  A reduced voltage drop across the inter-electrode region
accompanies the change in charging currents.  Lowered particle charge levels
and depressed electric fields which result from these changes in electrical
conditions cause precipitator collection efficiency to decrease (1).

     A thorough understanding of back corona and its impact on precipitator
performance thus requires knowledge of the electrical conditions (electric
field strength,  ionic current densities) which exist and the fundamental pro-
cesses which occur.  These processes include dielectric breakdown and ion
generation, ion transport to the charging region, and particle charge acquisi-
tion in the presence of bipolar ions.  Pioneering work was conducted in some
of these areas by Pauthenier, Penney, Cooperman and White.  Present day efforts,
motivated in large part by the relatively recent trend toward low sulfur coal
and the high resistivity ash accompanying its combustion, are building on the
foundations provided by the early studies.


                                      179

-------
     Bipolar charge acquisition and its role in precipitators experiencing
back corona was first examined by Pauthenier in the 1950's.  In this present
paper the results of recent experimental measurements on charge acquisition
by aerosols exposed to positive and negative ions of unequal current densi-
ties are presented and compared to calculations based on field charging
theory.  These charging data, in combination with recent measurements of  ionic
current densities in back corona, are then utilized to infer the potential
impact of back corona on particle charge levels.

                                    THEORY

     The bipolar field charging theory of Pauthenier (2) is identical to
classical unipolar field charging, with the exception that the transport  of
ions of opposite polarity to the back side of the particle is also considered.
The mean charging rate is calculated as the difference between the rates  at
which ions of opposite polarity collide with the particle.  When the two  rates
equalize, net charge acquisition ceases; and the particle maintains a steady
state charge level given by

                                                                       (1)

where nss is the steady state charge level, y i-s the ion current density
ratio, and nsat is the unipolar saturation charge.  This latter quantity  is
given by

                         ngat=(TreoDp2/qe)[3K/(K+2)]E;                  (2)

                                                                        -12
where D  is the particle diameter (m), eo is the permittivity (8.85 x 10
coul^/j.m), qe is the elementary charge unit (1.6 x 10~19 Coul), K is the
particle dielectric constant, and E is the electric field strength (V/m).  In
field theory it is assumed that diffusional transport of ions is negligible
and that the predominant mechanism for charge acquisition is ion drift along
electric field lines.

     The bipolar field charging equations have been used to predict charge
levels under back corona conditions, but the theory has not been thoroughly
tested experimentally.  Measurements (3,4,5) on large particles (Dp>20um) lend
support for the  calculations, but are limited in scope and do not provide un-
qualified experimental verification.  Experimentally measured charge on sub-
micrometer diameter particles in low fields is substantially higher than  that
predicted by the Pauthenier theory; the failure of the Pauthenier theory  being
attributed to its neglect of ionic diffusion (6).  This observation is con-
sistent with unipolar comparisons (7), where it is found that field theory
yields charge levels significantly lower than experimental measurements.  This
is especially true for particles of micrometer size and smaller.  Thus, there
is some question as to the applicability of Pauthenier's theory to the pre-
diction of charge acquisition during back corona.

                                  EXPERIMENT

     An experimental program has been established to provide a data base  on

                                      180

-------
bipolar charge acquisition  (8).  Initially uncharged, monodisperse  aerosols  in
the micrometer and submicrometer diameter size range are  exposed  to counter
currents of positive and negative ions in the presence  of an  external electric
field.  Control variables are particle size  (Dp),  electric field  strength  (E),
ion current density ratio (y), and uNt product; where N is the  ion  density of
the dominant ion species, y is ionic mobility and  t is  charging time.

     The experimental apparatus  (Figure  1) includes an  aerosol  generator for
polystyrene latex spheres,  a bipolar charger, and  a particle  charge analyzer.
The aerosol generator is comprised of an air blast atomizer and *"Kr charge
neutralizer.  The charger consists of three regions (A,B,  and C)  bounded by
parallel plate electrodes.  The  outer plates (1 and 4)  contain  radioactive
sources (24lAm) which emit  alpha particles into regions A and C.  Ions gener-
ated in these regions by the radiation are driven  by applied  electric fields
to electrodes 2 and 3, where a fraction  penetrate  wire  screens  and  enter the
charging region, B.  For the electrode polarities  indicated in  Figure 1, posi-
tive ions enter the charging region through electrode 2 and negative ions
enter through electrode 3.  The  potential to electrode  3  is applied in the
form of a square wave.  During: the first half cycle particles are exposed to
the ions and may experience lateral movement in the applied field.   During the
second half cycle the field in the charging region is reversed, preventing
ions from entering the region and reversing the lateral direction of particle
travel.  This prevents particle  migration to an electrode. Charging condi-
tions, ie. ionic current densities and charging fields, are chosen  such that
space charge perturbations  of the applied field and recombination are small.

     Particles extracted from the charger are routed to an integral mobility
analyzer.  Those which pass through the  analyzer are counted  by an  optical
particle counter, and mean  charge is determined by integral mobility analysis
 (9).

     In the results presented below positive ions  are dominant; that is, posi-
tive ionic current densities are greater than negative  current  densities.
Tests were also conducted with the negative ions dominant and,  within the
limits of experimental error,  the absolute value of charge was  found to be
independent of the dominant species.  Thus, the results should  apply equally
well to either case.

                                    RESULTS

     Data presented in Figure 2  illustrate the effect of  particle size and
ion current density ratio on charge.  In this case the  electric field strength
is 50KV/m and the yNt product is 1.9 x 109/Vm.  A  current density ratio of
infinity corresponds to unipolar conditions.  For  particles of  a  given size,
it is seen that charge levels are reduced as the current  density  ratio de-
creases.  For a given current density ratio, charge increases with  increasing
particle size.

     Comparisons of the data with classical field  charging theory (Figure  3)
show the measured charge to be significantly higher than  that predicted by
theory.  These data are for 1.09ym diameter particles with yNt=6.3  x 10°/Vm,


                                      181

-------
Y=10 and electric fields ranging from 25 KV/m to 300 KV/m.  An interesting
aspect of this figure is that it reveals the experimental and theoretical
slopes to compare favorably, yielding an approximately constant difference of
25 to 30 charges.  If the charge level calculated from continuum diffusion
theory (approximately 21 for this case) is added to the charge calculated from
field charging theory, the result is only about 20 per cent less than the
experimental measurement.  For other ratios this calculation yields even bet-
ter results, being well within experimental uncertainty for Y=3 and Y=100, and
approximately ten per cent less than experiment for the unipolar case.  This
empirical practice of summing charge obtained from independent field theory
and diffusion theory calculations has been found to compare reasonably well
with unipolar measurements also (7), although in the above referenced compar-
ison White's theory was utilized rather than the continuum model used in the
present calculations.

     In Figure 4 the ratio of bipolar charge to unipolar charge has been plot-
ted as a function of yNt.  This ratio is an indicator of the impact of bipolar
ions on charge levels.  The curve was obtained by plotting time dependent
data for various bipolar ratios and for the unipolar case, graphically fitting
a smooth curve to the data  and computing ratios based on the curves.  It is
seen here that the fractional bipolar charge decreases with time.  This de-
crease occurs because unipolar charge continues to increase in time while the
bipolar charge reaches a constant steady state level.  These curves illustrate
that time dependent behavior is an important consideration in assessing the
degradation of charge under bipolar conditions.

     To determine the impact of back corona on charge levels it is necessary
to know the initial charge state, the electrical conditions (ion current
density ratio and electric field strength), and, as evident from Figure 4, the
yNt product.  Such data are not generally available, although a mathematical
model of electrical conditions in wire-duct precipitators is being developed
(10), and measurements of ion currents under simulated back corona have re-
cently been reported (11, 12).  The data of Masuda et al (12) may be utilized
to make a preliminary assessment of charging in precipitators experiencing
back corona.  They report a current density ratio of approximately three for
what is termed severe back corona.  Using the data in Figure 4 it is seen that
a neutral particle entering a uniform region of back corona with J=3 would
acquire about 60 per cent of the charge it would have acquired under unipolar
conditions for yNt = 5 x lO^/Vm and 15 per cent for yNt = 5 x 10'/Vm.  For
Y=10, the corresponding percentages are approximately 95 per cent and 40 per
cent, respectively.  In these estimates of charge degradation the charging
field is assumed to be the same for unipolar and bipolar conditions.  If the
field were depressed due to large voltage drop across the dust layer, the
charge degradation would be substantially greater.

     For the case of localized back corona, the above calculation would ob-
viously not be appropriate.  In such a situation a highly charged particle
might move from a unipolar, high field region to one characterized by a re-
duced field and bipolar ions.  Assessing charge degradation under these con-
citions is not possible with the data presented here.
                                      182

-------
                                    SUMMARY

     Measurements relevant to the impact of bipolar ions on particle  charge
acquisition have been presented.  Measured charge levels are significantly
larger than those predicted by classical bipolar field  charging  theory.  How-
ever, the empirical practice of adding the charge obtained by field theory to
that obtained by diffusion theory yields a good estimate of the  measured
charge, at least for 1.09 ym diameter polystyrene.

     Substantial charge degradation can occur in bipolar regions, especially
for ionic current density ratios less than ten; but the extent of degradation
exhibits a strong time dependence.  This time behavior  could be  significant
in assessing the effect of back corona on unipolar charge levels.

                               ACKNOWLEDGEMENT

     This material is based upon work supported by The  National  Science
Foundation under Grant ENG-77-01130.

                                   ENDNOTES

1.   White, H.'J. "Resistivity Problems in Electrostatic Precipitation."
     J_. Air Pollut. Contr. Assoc. 24., 314  (1974).

2.   Pauthenier, M., R. Cochet, and J. Dupuy, "Probleme General  de la
     Charge Acquise par une Particule Spherique dans en Champ Electritrique
     Bi-ionise," £.R. Acad. Sci. Paris 243. 1606 (1956).

3.   Phillips, B.B. and R. Gunn, "Measurements of the Electrification
     of Spheres by Moving Ionized Air," ,J. Meteorol. 11. 348 (1954).

4.   Gunn, R., "The Hyperelectrification of Raindrops by Atmospheric
     Electric Fields," J_. Meteorol. 1_3, 283 (1956).

5.   Fjeld, R.A., R.J. Heinsohn and S.H. Levine, "Nonequilibrium Bipolar
     Charging of Aerosol Particles: Theory and Experiment," £. Colloid
     Interface Sci. 62, 69 (1977).

6.   Fjeld, R.A., R.O. Gauntt, G.J. Laughlin and A.R. McFarland,
     "Measurement of the Charge on Submicrometer Aerosol Exposed to
     Bipolar Ions," Conf. Rec. IEEE 1980 Annu. Mee_t. Ind. AppJ^. Soc_.,
     1063  (1980).

7.   Smith, W.B., L.G. Felix, D.H. Hussey, D.H. Pontius and L.E. Sparks,
     "Experimental Investigations of Fine Particle Charging By Unipolar
     Ions," J_. Aerosol Sci. 9, 101  (1977).

8.   Fjeld, R.A., R.O. Gauntt and A.R. McFarland, "Aerosol Charging
     By Bipolar Ions of Unequal Current Densities: Experiments in Low
     Electric Fields," accepted for publication by J_. Colloid Interface
     Sci.
                                      183

-------
9.   Gauntt, R.O., R.A. Fjeld and A.R. McFarland, "Characterization
     of Aerosol Mobility Distributions by Integral Mobility Analysis-
     The Method of Mobility Moments," .J. Aerosol Sci. 12 (1981).

10.  Lawless, P.A. and L.E. Sparks, "A Mathematical Model for Calculating
     Effects of Back Corona in Wire-duct Electrostatic Precipitators,"
     j;. Appl. Phys. 51, 242 (1979).

11.  McKay, R.B. and I.I. Inculet, "Bi-ionized Space Charges Generated
     by Means of Corona," Conf. Rec. IEEE 1977 Annu. Meet.  Ind.  Appl.
     Soc., 717 (1977).

12.  Masuda, S. and Y. Nonogaki, "Detection of Back Discharge in
     Electrostatic Precipitators," Conf. Rec. IEEE 1980 Annu.Meet.  Ind.
     Appl. Soc., 912 (1980).
                                     184

-------
                  A 24'
                  Am,
                              MONODISPERSE

                                 AEROSOL

                                GENERATOR
                                          rJ"
                          s>
                                 PARTICLE

                                  CHARGE

                                 ANALYZER
0
      A 241
      •Am
                                                  BIPOLAR


                                                  AEROSOL


                                                  CHARGER
Figure  1:  Schematic of  the experimental apparatus (6).


                                    185

-------
     80
     60
     40
  oc
  
-------
  140
  120
CO
"c
3
  100
0)
LU
O
<60
O
U)
o
F40
   20
y- 10
Dp=l.09ftm
      6.3xl08/Vm
                 EXPERIMENT
                         o
               50        100        150       200        250
                     ELECTRIC FIELD STRENGTH, E  (KV/m)
                                                        300
Figure 3:  Comparison of experimental charge to field theory calculation.
         Dp = 1.09 urn, yNt = 6.3 x 108/Vm and y = 1°-

                                 187
-------
UJ
IE
X
o
_l

o
o
u_
oe
          1.0
         0.8
         0.6
         0.4
      2 0.2
      m
           10
                                             E = 50 kV / M
                                             Dp = 1.09/Am
                          io8                  io9
                      ION CONDUCTIVITY  TIME-/iNt (VM)"
                                                                           10
                                                                            io
Figure 4:   Ratio of bipolar charge  to  unipolar charge as a function of
            yNt for y =  3  and y = 10.
                                        188
-------
                 IDENTIFICATION OF BACK DISCHARGE SEVERITY

               By:   Senichl Masuda and Yutaka Nonogakl
                    Department of Electrical Engineering
                    University of Tokyo
                    7-3-1, Kongo, Bunkyo-ku, Tokyo 113
                    Japan
                                   ABSTRACT

   The effect of back discharge on collection performance of ESP's are examin-
ed, and a quantitative expression of its severity based on the measurable
parameters is presented.  A linear relationship is observed to exist between
the measured charge-to-mass ratio and apparent migration velocity.  In this
case the charge-to-mass ratio measured at the outlet can be assumed to repre-
sent approximately its average value inside the collection field.  It is
evident that its outlet value should become greatly different from its in-
field value with increasing length of collection field because the highly
charged, larger particles are more rapidly collected whereas the smaller par-
ticles, difficult to collect, tends to raise the level of an overall charge-
to-mass ratio at the outlet.  In order to avoid this difficulty in the diag-
nosis of back discharge severity a bipolar current probe has been developed
by the authors which enables a separate measurement of negative and positive
ionic current density in a bipolar ionic atmosphere such as in the back-
discharging field.  These two quantities make possible a theoretical predic-
tion of the charge-to-mass ratio in the back-discharging field, and thus the
collection efficiency of an ESP, too.
                                 INTRODUCTION

   The effect of back discharge on the collection performance of an ESP is
twofold.  One is the reduction of field intensity, E, in charging and collect-
ing fields owing to the lowering of sparking voltage itself (excessive spark-
ing) or an abnormal increase in current at a lower voltage which cannot be
raised because of a limited current capacity of a power supply.  Another
effect of back discharge is the production of false  (positive) ions which
reduce the particle charge, q. Since the apparent dust migration velocity, w,
in Deutsch's equation is proportional to qE, the reduction in q and E impairs
the performance through the lowering of w.  As a result an impairment factor
of migration velocity for identifying the overall back discharge severity may
be estimated from two measurable parameters, the impairment factors of E and
q-
   In the ordinary collection field with a longer residence time the charge-
to-mass ratio measured at the outlet mostly differs much from its value domi-
nating in the collection process.  Theory predicts, however, that the particle
charge in the back-discharging field can be calculated, provided the quanti-
ties of negative and positive ionic current density can be measured.  So, a
particular probe (bipolar current probe) is developed to enable this measure-
ment.  This provides a possibility of identifying the back discharge severity
through the current density of positive and negative ions, i+ and i_, and its

                                      189
-------
ratio, i+/i_.  These quantities are the more meaningful parameters which pre-
cisely describe the bipolar ionic structure of the back-discharging field.

1. Theoretical Considerations

1.1 Expression of Back Discharge Severity

   The electrical collection efficiency of an ESP after correction of mechani-
cal contribution can be given, even under back discharge condition, by a well-
known Deutsch's equation (1):

               n = 1 - exp(-wf)                                      (1)

where          w = KqE = apparent migration velocity of dust         (2)

and  f = specific collection area, q = average particle charge, and
K = constant.
   The quantities, q and E, are lowered by back discharge from their ideal
values, q0 and Eo, under no back discharge conditions to become

               q = aq0 ,          E = BE0                       (3), (4)
and we get
               w = Cw0                                               (5)
where  wo = ideal value of w, and a, 8, and ?(= «8) are constants.  The magni-
tude of w0 is a function of all the variables in a ESP other than pd, whereas
the constants, a, g, and £, are specific to each back discharge condition
which is determined by pd, gas composition, gas temperature, particle size,
dust composition etc.  It is evident from equations (1) and (5) that overall
effect of performance impairment by back discharge is represented by £, which
can be numerically grasped from the measurable quantities, a and 3.  So, we
name £ the "impairment factor of apparent dust migration velocity" to specify
the severity of respective back discharge condition.
   Among the quantities, q and E, needed to estimate £ the former is not only
difficult to measure accurately, but also represents a secondary quantity
determined by the more fundamental quantities to specify back discharge:   the
current density of negative and positive ions, its ratio, and field intensity,
E.  Based on these quantities it is possible to predict theoretically the
magnitude of particle charge in the charging process in a bipolar ionic atmo-
sphere, its saturation value, and its charging speed (charging time constant),
not only in the case of a conventional ESP, but also in the case when a pre-
charger is being used.  This supports the use of these intrinsic quantities as
the most essential parameters to identify the back discharge severity.

1 .2 Particle Charge in Bipolar Ionic Atmosphere

   The two different cases are considered separately:  one being the case  when
the particles enter into the charging and collecting field of a conventional
ESP without any initial charge (Case I), and another being the case when the
particles are highly precharged (Case II) .  The field charging is assumed  in
both cases as the charging mechanism.  The case of diffusional charging is
considered in reference (2) .
                                     190
-------
1.2.1. Case I (field charging without  precharging)

   The magnitude of charge of a  single spherical particle in this case is
given by Pauthenier (2) as
                               1 - exp(-t/T)
                qf = qfoo	  (C)                (6)
                         1 - {(l-Y)/(l+Y)}2exp(-t/T)

               qfro = qfoo* {(l-y)/(l+Y) } = saturation  charge  under bipolar(7)
   where                                  condition
              "If oo* = 4TC0{3es/(es+2) }&2E = saturation charge under mono-  (8)
                                           polar condition
                 Y = AI+P+/U-P-  =  /L+/1- = current parameter             (9)
                 T = e0//u_|_p+u_P~ =  e0E//i+i_ =  A+T_/4   (s)             (10)
   and
                 t = time   (s)

                                                     ~ 1 2
where EO = dielectric  constant  of vacuum = 8.842x10   (F/m) ,  e   = specific
dielectric constant of particle, a = particle radius (m) ,  y+ ana  y_ = mobili-
ties of positive and negaive  ions  (m2/Vs) , p+ and p_ =  space charge density
of positive and negative ions (C/m3) ,  i+ and i-  = current  density of positive
and negative ions  (A/m2) and

                TO = 4e E/i = charging time  constant under monopolar    (11)
                              condition at current  density i = i+ + i_
                T+ = 4e0E/i+  =  „  „   at current  density i+               (12)

                T_ = 4eQE/i_  =  ,,  „   at current  density i_               (13)

It can be seen from equation  (7) that the  saturation charge  in a  bipolar ionic
atmosphere lowers with increasing current  density ratio,  i+/i_.   Figure  1
shows the particle charge normalized with  its ideal value  in a monopolar ionic
atmosphere, If^/lf^*,  as a  function  of the current  density ratio, i+/i_.  It
should be noted that the particle charge is  almost  halved  (52%) by a slight
supply of positive ions at  i+/i_ = 10%.  Equation (7) also indicates that par-
ticles can be positively charged in  the local area  where  i+  exceeds i-.   Fig-
ure 2 shows for different values of  i+/i-  the increase  of particle charge,
denoted by q/qf^*, as a function of charging time, t.
1.2.2 Case II  (field charging with  precharging)
   The highly  precharged particles  lose their charge in a bipolar ionic atmo-
sphere owing to charge elimination  by  positive ions, finally to obtain the
saturation charge  (7) specific  to the  back discharge condition.  This process
differs depending  on a level of the initial charge,  q0,  in relation to the
value of ideal saturation  charge, qf^*.
i) Range where qfoo>qf ,,<,*:   The charge lowers to qf^*  according to the following
   relation derived by the authors  (5):
                qf = q  exp(-4t/T+) = qfoo*  exp{4(<^t)/T+}               (14)
                                      191
-------
   where
              0 = (T+/4)ln(q0/qfoo*) = time necessary for qo -> qf^*       (15)

   The charge elimination proceeds in this range without being affected  by
   negative ions because the particle potential is excessively higher than  the
   original local potential at the particle position.
ii) Range where  qf * > qf = qf :   In this range the particle potential  is
   lower than in range i), so that negative ions interfere with the charge
   elimination process by positive ions, and the following relation holds for
   charge reduction (4) :
                       1 + {(HT)/(l-Y))exp(-t/T)
             q  = q   - — — -                       (16)
   Figure 3 shows for different values of i+/i_ the time decrease in particle
charge from qfOT* to q^ according to equation (16) .
   Curves in f igure :4 shows, for the cases of i+/i- = 0.02 (low severity of
back discharge) and i+/i- =0.10 (medium severity) respectively, the time
decrease in particle charge from its initial value qo ( >qf00*) to qfOT* and
further to the saturation value, qf^, plotted in its actual value.  A great
difference can be seen in both the saturation value, qf^, and the charge re-
duction  speed , depending upon the difference in the magnitude of i+/i_.  The
lower curves in this figure represent the charging process starting from zero
initial charge according to equation (6) depicted for the purpose of reference.

1.3 Relationthip between Charge Impairment Factor and Bipolar Current Density

   It is evedent that the charge impairment factor,  , defined by equation (3),
can be easilly calculated using the time averages over the residence time of
the values (qf in equation  (6)) and (qf* given by the following equation) for
the case I :
            qf* = qfoo* { (t/T0)/(l+t/T0)} = time increase of charge       (17)
                                           under no back discharge condition

   In the case II where precharging is made for enhancing the collection per-
formance, the charge enhancement factor,  = q/qo  (>1), and the enhancement
factor of apparent dust migration velocity, ^ = w/wo, come into question.  As
the particle charge lowers in time in the succeeding collection field because
of charge elimination, its time average must be taken in calculation of (j> and
ty using equations (6), (14) and (16).
   Since equations (6), (14) and (16) for the field charging are the functions
of the residence time, t, field intensity, E, and the magnitudes of positive
and negative ionic current density. i+ and i_, as in the forms of its ratio,
i.f/i-, and the square root, ^i+/i_, it is theoretically possible to estimate
the values of a, 8, ?, , and ^ from these quantities.

1-4 Theory of Bipolar Current Probe

   Figure 5 shows the schematic of the bipolar current probes of both spheri-
cal and cylindrical types, each having three electrodes A, B. and C.  The
probes are inserted into a bipolar ionic atmosphere so that the central  plane
D becomes perpendicular to the local electric field, as illustrated in figure
6.  The measuring electrodes A and B, and the detection electrode C are

                                      192
-------
connected through a current meter and wires  to  a variable  dc  high voltage
source to be applied with the same voltage,  Vo  (figure  7).  When the magnitude
of V0 is adjusted to become equal to the  original  local potential of the probe
center point, the electric lines of force  in its surrounding  takes  a symmetric
configuration respect to the central plane D (figure  6).   Then, the measuring
electrodes, A and B, accept respectively  only the  negative and positive ionic
current, I_ and I+, which can be easily calculated as  (6):

   (a) Spherical Probe:   I_ = -37ra2i_cos2(d/2a),  I+  =  3Tra2i+cos2(d/2a) (18)

   (b) Cylindrical Probe: I_ = -4al i_cos(d/2a),   1+  =  4al i+cos(d/2a)  (19)

where  a = radius of probe,  d = width of  electrode C,  and 1  = length of C in
a cylindrical probe.  Under this "balanced condition",  the number of electric
lines of force entering into the electrode C is equal  to that leaving from it.
But, a slight current is detected by C owing to the difference in the mobility
and space concentration of positive and negative ions.   It is evident that
this probe can also be used in the usual  monopolar ionic atmosphere.  The val-
ues of I- and 1+ at the balanced condition can  be  derived by  the following
methods:

i) Graphical Method;  Figures 8 and 9 shows,  for the  monopolar and  bipolar
   ionic atmosphere respectively, the values of I+, I_  and Ic (= current flow-
   ing into electrode C) as the functions  of probe potential, Vo.   Theory
   predicts that each curves in both monopolar  and bipolar cases consists of
   three parts, two straight lines with different  gradients (L and  N) and a
   curved line connecting them (M).  In the  monopolar  case the second straight
   branches conincide with the horizontal  axis.  Theory further tells that the
   extrapolation of these two straight line  branches  provides a crossing point
   p on the horizontal axis, which corresponds  to  the balanced potential.
   This method requires a tedious procedure,  and the  ionic field may subject
   to disturbances in the mean time.
ii) Use of Correction Factors at "Quasi-Balanced Potential":  It is much eas-
   ier in the bipolar ionic atmosphere to  determine the probe potential at Ic
   = 0.  This potential, called "quasi-balanced potential", slightly differs
   from the true balanced potential owing to the difference in positive and
   negative ionic current density (point  q in figures  8 and 9).  But its
   deviation as well as deviations of 1+  and I- can be  calculated throretical-
   ly  .  Thus, the values of 1+ and I_ at  the true balanced potential to be
   used in equations (18) and (19) can be  obtained by dividing their values
   at the quasi-balanced potential by the  correction  factors  k+ and k_ indi-
   cated in figure 10.  These correction  factors are  calculated for different
   values of the probe parameters, d/2a,  and depicted as the  functions of
   I+/I- at Ic = 0.
   In the monopolar case the method as described above  fails  because the exact
determination of the quasi-balanced potential becomes  extremely difficult  (see
point q in figure 8).  However, theory indicates that,  at  the balanced poten-
tial in this case, the ratio of IC/I_ takes  a definite  value  depending upon
the type and geometry of the probe:

   (a) Spherical Probe:   IC/I_ = tan2(d/2a)                           (20)

                                      193
-------
   (b) Cylindrical Probe:  IC/I_ = sec(d/2a) - 1                          (21)

   In the use of this probe in a back-discharging field the negative ionic
current from the discharge electrode is partly obstructed by the probe  to pro-
duce its shadow on the dust deposit on the collecting electrode.  As a  result
no back discharge occurs (figures 11 and 12).  Then, the positive ionic cur-
rent detected by the lower electrode becomes extremely low or almost zero.
This error can be avoided by providing a gas flow perpendicular to the  field
lines at a low velocity, the magnitude of which must be determined by the probe
dimension and its distance from the collecting electrode.

2. Results of Experiments

2.1 Migration Velocity and Charge-to-Mass Ratio
   The effect of charge-to-mass ratio of dust, q/m, on the apparent migration
velocity of dust is investigated at a short collection field (two ducts in
parallel, plate-to-plate spacing  = 30 cm, height = 70 cm, length in gas flow
direction = 50 cm) located in a laboratory race-track system, using fly-ash at
gas temperature of 130°C.  The dust resistivity, pa, is 1011- 1012 £>cm.  When
the collecting electrodes are cleaned by thorough rapping, back discharge dis-
sappears at this resistivity level resulting in a normal current density and
a high charge-to-mass ratio.  Then, the current rises gradually with increasing
dust deposit owing to the increase in back discharge severity to cause the drop
in charge-to-mass ratio and collection performance.  The charge-to-mass ratio
is measured at the outlet of the collection field.  The inlet and outlet mass-
loading of dust are measured at the same time to evaluate the apparent migra-
tion velocity of dust.  It is considered in this short field that the charge-
to-mass ratio measured at the outlet is approximately representing its "in-
field" value governing the collection performance.  In the present system ca.
50 % of the inlet-dust in collected mechanically, and the apparent migration
velocity is derived from the electrical collection efficiency, r\e, after
subtracting the effect of mechanical collection.  The magnitude of the apparent
migration velocity, w, is considered to be propotional to ln(l - ne) •
   Figure 13 indicates the relationship between the measured values of charge-
to-mass ratio and those of apparent migration velocity, both plotted in the
normalized form, a = (q/m)/(q/m)o and t, = w/wo.  The subscript o denotes the
reference quantities under no back discharge condition just after the electrode
cleaning.  As a first approxomation a linear relationship can be seen to exist
between a and ? .  However, a closer examination of the plots shows that the
collection performance becomes superior to the linear line with the decrease
in charge-to-mass ratio which is resulted by increasing the severity of back
discharge.

2.2 Tests of Probe Diagnosis  ol  Back Discharge Severity

   Prior to the probe diagnosis of the collection field, the magnitudes of i+
and i_ are measured under back discharge condition in a point-to-plane  elec-
trode system (figure 7) using different dust samples where the dust resistivi-
ty is altered in air by changing ambient temperature(5).  Figure 14 shows the
results obtaines for fly-ash with the spherical and cylindrical probes, both
providing the same results in this particular case.


                                      194
-------
2.2.1 Charge-to Mass Ratio and Current Density Ratio

   The bipolar current probe of cylindrical type  (a = d = 1 mm, 1 = 10 mm) is
inserted into the short collection field in the race-track system as previous-
ly descrived.  The probe is located throughout the present experiments between
the 6th wire and a plate.  The probe can be moved vertically along the axis y
parallel to the wire, with the distance from the plate kept constant at z =
10 cm.  The charge-to-mass ratio of dust, i+ and i_ are measured under differ-
ent back discharge severity.  Figure 15 indicates   one of the probe diagnosis
results under back discharge condition.  It must be noted that the values of
1+ and i_, as well as i+/i_, are extremely non-uniform in space.  This fact
strongly suggests that the multi-point measurement must be made in its
application for predicting the collection performance.
   Figure 16 indicates the normalized charge-to-mass__ratjio,_a = (q/m)/q/m)0,
plotted against the average current density ratio, £ = i+/i_, measured at the
points along y-axis.  The solid curve represents a theoretical curve according
to equation  (7).  The agreement between the experimental and theoretical values
is satisfactory.

2.2.2 Migration Velocity and Current Density Ratio
   The inlet and outlet mass-loading of dust are measured at the same time
when the  probe diagnosis as in 2.2.1 is being made.  The collection efficiency
is calculated from  the mass-loading  , and  t, = w/wo derived.  The normalized
charge-to-mass ratio  is  calculated from the measured value of 5 = i+/i- using
equation  (7).  Figure  17 shows the relationship between the values of t, and a
thus obtained. A propotionality exists approximately between these quantities.

3. Conclusions
    The following conclusions  are derived from the present  investigations;
 1)     Charge-to-mass  ratio of dust provides an usable parameter  to  identify
    back discharge severity.
 2)     The magnitudes  of positive and negative ionic current density,  i+ and  i-,
    and its  ratio,  i+/i_,  can  also be used as the most essential  parameters for
    specifying  back discharge  severity,  provided the distribution of back
    discharge be considered.
 3)     The bipolar current probes of three electrode type provides a usable
    means for diagnosing a bipolar ionic field.

 References
 (1) S.Masuda,  A.Mizuno and H.Nakatani,  Proc. US-Japan Seminar on Measurement
     and Control of Particulate Generated from Human Activities,  Kyoto(1980),
     p.193.
 (2) S.Masuda,  Y.Nonogaki, H.Nakatani and T.Oda, ibid., p.66.
 (3) M.Pauthenier, La  Physique des Forces Electrostatiques  et leurs Applications
     ,  Centre National de la Recherche Scientifique, Paris(1961), p.279.
 (4) S.Masuda and Y.Nonogaki,  Memorandum presented at Committee Electrical Engi-
     neering for Pollution Control, Inst. Elect. Engrs. Japan (July 1,1980).
 (5) S.Masuda and Y.Nonogaki,  Conf. Rec., IEEE/IAS 1980 Annual Meeting, p.912.
 (6) S.Masuda and Y.Nonogaki,  to be presented at the 4th International Congress
     on Electrostatics (1981,  Hague).


                                       195
-------
 01
 ee
 a 0.5
 a
 01
 •o
 
-------
Needle electrode
                 Bipolar current probe
                       (spherical  type)
                                                            L±(true) = rr i±(roeasured) -
Fig. 7   Diagnosis of Needle-to-Plane

         Field by Bipolar Current      Fig.  10  Correction  Factors K+  and K_

         Probe


      [nA]
    % -50
    a
     M
     3
     CJ

     a
     .n
     o
                           Vm=30kV
              i	i
                                N
             -11   -12   P -14  -15   [kv]
              Probe  Potential  Vo
          Shadow         Shadow


(a) without airstream (b) with airstreara
                                      Fig.  11  Formation of Negative  Ion  Shadow
Fig.  8   Probe Currents vs. Probe              and Ifcs  Movement by Air  Flow

         Potential  (monopolar ionic

         atmosphere;  spherical probe)
              Probe Potential  Vo  [kV]
 Fig.  9  Probe  Currents and  Probe      Fig.  12  Photograph of Probe  Diagnosis

         Potential (bipolar  ionic               and Negative Ion Shadow

         atmosphere;  spherical probe)             (cylindrical probe)
                                       197
-------
                •  V = 43kV
                  V = 40kV
                o  V = 35kV
    0              0.5
          a =  (q/m)  / (q/m)o
                                 1.0
Fig.  13  Apparent Migration Verocity
          vs.  Charge-to-Mass Ratio
     50
   J 30
      10
         Pd (n-crc)
       * 8.0 x 10'
       • 8.0 X 10
        • 1.4
        a 1.4
   11
X 1011
X 1011
                Type
Spherical
Cylindrical
Spherical
Cylindrical
                        i Flashover
         10  20  30  40 50       [kV]
             Main Voltage  Vm

Fig. 14   Current  Density Ratio vs.
          Applied  Voltage (figure 7,
          fly-ash,  50-80°C)
                                    [cm]
Fig. 15   Spacial  Distribution of
          Ionic Current Density along
          y-axis  .V = 38kV
                                                                     V = 35-43 kV
                                                              0.5     _      1.0
                                                  current density ratio i+/i_
                                             Fig. 16   Charge-to-Mass  Ratio vs.
                                                       Current  Density Ratio
                                                             0.5            1.0
                                                    a  calculated from i+/i_
                                           Fig.  17  Apparent  Migration  Velocity
                                                     vs.  Calculated Charge-to-
                                                     Mass Ratio
                                        198
-------
           MODELING OF ELECTROSTATIC  PRECIPITATORS  WITH RESPECT
               TO RAPPING REENTRAINMENT  AND  OUTLET  OPACITY

          By:  M. Greg Faulkner,  William E.  Farthing,  Jack  R.  McDonald
               Southern  Research  Institute
               2000 Ninth Avenue,  South
               Birmingham,  Alabama 35255

               Leslie E. Sparks
               Industrial Environmental  Research Laboratory
               U.S. Environmental Protection Agency
               Research  Triangle  Park,  N.C.  27711

                                  ABSTRACT

     Revisions of the  Environmental Protection Agency's mathematical model of
electrostatic precipitation  which allow a dynamic, time-dependent representa-
tion of rapping reentrainment and  predictions of outlet opacity are discussed.
The new rapping reentrainment scheme allows different rapping schedules  for
the various independent  sections,  reentrainment due to rapping of a specified
percentage of the total  mass collected in a given increment of length with a
specified particle size  distribution, recharging and recollection of re-
entrained particles,  a representation of hopper boil-up, and a time history
of dust layer thickness.   The addition of opacity calculations based on Mie
theory allows the prediction of total and fractional opacities based on pre-
dicted outlet mass loadings and particle size distributions, a specified
complex index of refraction, particle density, and stack diameter.  Simu-
lations demonstrating the predicted effects on outlet particulate emissions
and opacity are presented.


                                INTRODUCTION

     The Environmental Protection  Agency's mathematical model of electrostatic
precipitation is undergoing revision  and improvement at the Southern Research
Institute.  One recently completed modification alters  the calculation pro-
cedure for rapping reentrainment  and  another  provides the mechanism to cal-
culate outlet opacity.   In  the  present model,  the outlet dust loading con-
tributed by rapping reentrainment  is  determined by  an empirical process which
estimates the reentrained mass  in the last section  of the precipitator, fits
that mass to a size distribution,  and adds the result to the effluent size
distribution of  the precipitator.   The revised version  introduces  the re-
entrained mass into the  gas flow  at the  point where it  was  rapped  loose and
then follows the dust through the collection process  to determine  the outlet
dust loading due to the  rap.  Opacity is not calculated in  the present model.
In the revised version,  opacity calculations utilize Mie theory  to predict
total and fractional  opacities  based  on  the  outlet  mass loading  and particle
size distribution.

                            RAPPING REENTRAINMENT

     In  order  to properly model the rapping  reentrainment  process it is
necessary  to understand  what takes place when the rap occurs.  When the
                                      199
-------
plates are rapped, the collected dust layer fractures and separates from the
collection plates.  Some of this dust is reentrained directly into the gas
flow but most of it falls into a collection hopper where it is broken up by
the impact.  A large dust cloud then billows up out of the hopper and is re-
entrained into the gas flow.  A study of a pilot precipitator demonstrated
that this dust cloud contributes most of the reentrained particles.1  The
amount of particles carried aloft in this cloud will be a function of time
as the cloud builds up and subsides or is carried away by the gas flow.  The
particle size distribution in the dust cloud is also expected to vary with
time as large particles have a higher settling rate than smaller ones.  Ex-
periments have shown that the particle size distribution due to rapping found
at the top of the precipitator is different from that near the bottom, where
a higher concentration of large particles is found.

     In the computer model, each electrical section of the precipitator is
divided into several increments to allow a finer calculation of the efficiency
in each section and of the dust layer growth rate.  The rapping subroutine
calculates the thickness of the dust layer in each increment at the time of
each rap so that the amount of dust to be reentrained can be determined.
Figure 1 shows a record of the dust layer thickness for a hypothetical three-
section precipitator before and after two raps.  The rapped sections are
shaded for easy identification.  When a rap occurs, a portion of the mass
removed from each rapped increment is fitted to a size distribution and the
resulting information is stored in the form of numbers of particles in each
size band.  The program is then returned to the beginning of the efficiency
calculation procedure for the calculation of collection efficiency during the
rapping puff.  During this calculation, at the beginning of each increment,
the computer checks the storage register to obtain the number of particles
in each size band reentrained by the preceding increment.  These numbers are
added to the numbers of particles already present at that point in the pre-
cipitator and the calculation of particle collection in that increment pro-
ceeds in the usual manner.  This mechanism implies instantaneous recharging
of the reentrained dust particles to the value of charge that the same size
particles in the regular flow have at that point.  This is an approximation
made to simplify the calculation.  In the actual process, the charges on
these particles, and therefore the rate at which they recharge, are unknown.

     In order to more closely approximate the actual rapping puff, the re-
entrained dust is shaped according to a rapping profile.  This is an array
of numbers which describes the quantity and the size distribution of the
reentrained dust as a function of time.  This will allow the shaping of the
reentrainment puff to closely approximate an experimentally observed puff.
An example of this is shown in Figure 2, in which an actual rapping puff
observed by Spencer1 on a pilot precipitator has been duplicated.

     This model has been used to examine the collection efficiency of a power
plant precipitator.2  The particular unit studied had three electrical sections,
each having a collection area of 2500 m2 (26,000 ft2).  This gives a specific
collection area of 34.5 m2/(m3/sec) or 175 ft2/1000 acfm for the design volume
flow of 214 m3/sec (453,000 acfm).  There are 42 lanes spaced 28 cm (11 in.)
apart with 12 wires per lane per field.  The first section is rapped 10 times/
hour, the second field is rapped 6 times/hour, and the third field is rapped
1 time/hour.  The plates are rapped two at a time.
                                     200
-------
     0.2 ir
E



co"
CO
u

X
I-

cc
111
>-
co


Q
     0.1
     0.0
        BEFORE   AFTER


    RAP SECTION 1 AT 10 MIN
                                          BEFORE  AFTER


                                      RAP SECTION 2 AT 15 MIN
                                                          4172-7A
       Figure 1.  Thickness of dust layer before and after rap.
                            201
-------
    10'3
            ACTUAL
            IX
    10'4
I
    10-5
                            SIMULATED
            I      I     I     I
                            I      I     I
0    10     20
                            30          0



                                TIME, sec
10     20    30
                                                           41 72-8
                      Figure 2. Simulation of rapping puff.
                                  202
-------
     Figure 3 shows the experimental data for the plant and a comparison
between the new rapping simulation and the old  simulation:  the new calcula-
tion more closely approximates the experimental curve.  Overall collection
efficiencies are 99.55% by actual measurement,  99.20% using the old rapping
calculation, and 99.68% using the new procedure.  The new procedure was run
using an assumed 10 ym mass median diameter  for the reentrained particles
which were introduced into the gas stream in the form of a dust cloud which
lasted 4 seconds.  Each of the 42 lanes was  considered to be independent from
its neighbor during the rapping process.  Seventeen separate rapping calcula-
tions were performed to simulate the complete rapping scheme, resulting in an
increase in computer time from 4 minutes  (no rap case) to 40 minutes.  If only
the rap of the final ESP section is simulated,  similar results can be obtained
with 7 minutes of computer time.

     The flexibility of the new rapping procedure is much greater than that of
the estimation procedure found in the present precipitator model.  However,
before the procedure can be widely used, a data base of size distributions and
quantities of reentrained dust measured at the  point of reentrainment must be
obtained.

                                   OPACITY

     The reduction in the transmission of light passing through suspended
particles occurs by scattering and absorption by these particles.  The degree
of reduction is a function of particle size, concentration, and composition,
and the optical path through the aerosol.  For  particles with diameters greater
than about 1 ym, the extinction of light  is  essentially determined by the
total projected area of the aerosol particles.  For smaller sizes, particle
composition becomes important.  Consequently, for general applications, the
calculation of opacity must be performed using  a mathematical solution to
Maxwell's equations.  The theory used in  this development is that of Gustav
Mie3 which  gives, among other parameters,  the  extinction  efficiency, Q    ,

 in terms of  the  particle  refractive  index, n,  and  size parameter rrD/X for
 isotropic homogeneous  spheres.  D  and A  are  the particle  diameter and the
wavelength  of  the light beam, respectively.  The  extinction efficiency re-
presents the effective projected area of a particle  for removing light from
a light beam compared  to  its  geometrical projected area.  This  is related to
opacity by

                     Opacity  =  l-exp(-EL)  =  1-exp (-NAQext;L),


where

          Extinction coefficient  (m"1),
                                        (m),
                                               and
L = Pathlength of the light beam  (m),
N = Particle concentration (number/m3),
A = Particle projected area (m2).
 Figure 4  shows  the  extinction efficiency as a function of  the particle size pa-
 rameter for four  complex  refractive indices.   The actual computer  code for Mie

                                      203
-------
   102
    101
       \-   A'
z
O
oc

Ul

g   100
Q.
O
oc
UJ
a.
    10'
    10-
T	1   I  I I  IM|	1		l|


      a,
      ^"^
                             •   EXPERIMENTAL

                             O   STANDARD RAPPING CALCULATION  ~j

                             A   REVISED RAPPING CALCULATION
                                              v
                                             -111
                           10°                  101


                           PARTICLE DIAMETER, jum
                                                    I  I II
                                                                     0.0
                                                                     90.0
                                                         99.0
UJ

H

ui
O
OC
UJ
a.
                                                         99.9
                                                                     620-242
 Figure 3. Comparison of standard and revised rapping reentrainment calculations.
                                 204
-------
                          E = E(Df)

E a
uj U
  Z
UJ
  UJ
00    2

P li-
CC u.
                         PARTICLE DIAMETER (^m) FOR WAVELENGTH 0.55

                    0.5         1.0          1.5          2.0         2.5
                     I            I


                      (b) n = 1.5
     I             I           1


(a) n =1.33 (H2O)


   d) n = 1.5-0,1!

        (c) n = 1.96-0.66! (carbon)
                                         3.0
                                                  SCATTERING COMPONENT OF (c)



                                                 ABSORPTION COMPONENT OF (c)




                                                  I	I
                                                  10                  15

                                   PARTICLE SIZE PARAMETER x = ?rD/X
                                                  20


                                              4172-48
             Figure 4.  Particle extinction efficiency as a function of particle size parameter.
                                             205
-------
theory used in the ESP model is that of Moore4 which employs the criteria
of Deirmendj ian5 to determine a sufficient number of terms to achieve con-
vergence of Mie's series solution.

     The ESP model predicts the size distribution of the aerosol exiting the
ESP for the no-rap case and for the rapping puff.  The extinction coefficient
for each size band of each case is calculated and printed along with the total
extinction coefficients.  The value of extinction efficiency, Qext> normally
used for each size band is the average of Q    at 10 wavelengths weighted
according the the photopic color response of the human eye.  EPA approved
opacity monitors must simulate this color response which is maximum  at
X = 0.55 ym and has a width at half maximum of 0.1 ym.  The option to per-
form the calculations at X = 0.55 ym is available to reduce the required
computer time.  The resulting error depends upon the size distribution.
Typical errors found thus far are about 5% in the total extinction coeffi-
cient.  Computation time required for the calculations at all wavelengths
is about 1 minute compared to about 45 seconds at one wavelength using the
DEC 2020 computer.  This is small relative to the hour typically required
for the entire model.

     The opacity calculations can be performed for several refractive indices
for each run of the model if desired.  The two default values of 1.5-O.Oi
and 1.5-O.li will be used if the user does not supply this information.  In
addition, the variation of refractive index with wavelength can be taken
into account by the program if these data are available.

     The program output lists the contribution of each particle size band to
the extinction coefficient, dE/dlogD, for the no-rap condition, the time-
averaged rapping puff contribution, and the combined emissions.  Following
the fractional extinction coefficients, the total extinction coefficient, E,
is given for the no-rap, rap, and combined cases.  The extinction coefficient
per unit mass loading, SPE, and the K parameter initiated by Pilat and Ensor6
are also given.  These quantities are related to opacity by

                  Opacity = l-exp(-(SPE)WL) = l-exp(-WL/PK) ,

where

     W = outlet particulate mass loading,
     P = particle density,
and the other quantities are as previously defined.

     The opacity calculation has been performed for the power plant described
in the section on rapping reentrainment.   Since no information on the refrac-
tive index was available for this plant,  the default values were used.  The
opacities calculated for a refractive index of 1.5 were 13.5% for no-rapping,
2% for the rapping puff, and 15% for the combined case.  For an index of
1.5-O.li, the values were 12%, 2%, and 14%.  The actual plant opacity at the
time that these data were taken was estimated to be between 10 and 15%.

     The opacity calculation developed for the ESP model is a highly flexible
approach to the calculation of opacity.  Based on the ESP emissions predicted
                                     206
-------
by this model, the new routine can calculate  the opacity of  the plume when
only the pathlength of the light is available.  However, for more exact cal-
culations, more exact data may be supplied.   In addition to  the total opacity,
this routine furnishes the contributions  to the opacity from each particle
size band.

                                  ENDNOTES

1.  Spencer, H. W., III.  Rapping Reentrainment in a Nearly  Full Scale Pilot
    Electrostatic Precipitator.  EPA-600/2-76-140  (NTIS PB255  984), U.S. EPA,
    Industrial Environmental Research Laboratory, Research Triangle Park,
    NC, May 1976.

2.  Gooch, J. P., and G. H. Marchant, Jr.  Electrostatic Precipitator Rapping
    Reentrainment and Computer Model Studies.  EPRI Contract RP413-1, The
    Electric Power Research Institute,  Palo Alto, CA, June 1978.

3.  Mie,  G.  Beitrage zur Optik  truber  Medien, Speziell Kolloidaler
    Metallosunge.  Ann. Physik.  25:377-445,  1908.

4.  Moore, J. D. M.  1968.  Tests of Approximations to Electromagnetic
    Scattering by Spherical Particles.  M.S.  Thesis.  Auburn University,
    Auburn, AL.

5.  Deirmendjian, D., and R. J.  Clason.   1962.  Light Scattering on Partially
    Absorbing Homogeneous Spheres of Finite Size.  Paper R393-PR, The RAND
    Corporation, Santa Monica, CA.

6.  Pilat, M. J., and D. S. Ensor.  Plume Opacity and Particulate Mass Con-
    centration.  Atmos. Environ. 4:163-167, 1970.
                                     207
-------
            NEW PRECIPITATOR TECHNOLOGY FOR  PARTICULATE CONTROL

                    By:   J. R. Zarfoss
                          Environmental Elements Corporation
                          Subsidiary of Koppers Company,  Inc.
                          Post Office Box 1318
                          Baltimore, Maryland 21203

                                  ABSTRACT

All  precipitator  designs  must  accommodate  the  requirement  of adequately
cleaning  the  internal  components.   However,  the  collection of  highly
resistive  particulate  matter   from boilers  burning  low  sulphur  coal  is
considered  to  be  one of  the more demanding applications.   The fundamental
capabilities  needed  to  meet  this  challenge  are the  collecting  surface
response  to rapping  and the  electrical  characteristics of  the  discharge
electrodes.    This  paper  outlines  a  development  program  on  this  subject
which  spans six  years, beginning  with  laboratory  studies  and concluding
with  the results  from working  installations.   One  of  the  basic  studies
quantifies  and compares the rapping  response,  full size,  of three  popular
collection  surface  designs.   No evidence has been found  to indicate  that a
comparison  of  this  type has ever been attempted before.  Levels of  surface
acceleration,  frequency  and  uniformity  are  described.   Techniques  for
altering   the   voltage   and   current   characteristics   of   the  discharge
electrodes  are also presented.   This  is not a  theoretical  study.   All of
the  information  relates  to  actual   measurements and  is  useful  in  all
precipitator designs.

                                INTRODUCTION

     During  the  evolution of  the electrostatic precipitator as a  control
device   for  particulate  matter,  each  variable  pertinent  to  collection
efficiency  has been reviewed many times.  The objective  each time has been
to  achieve  higher efficiency and,  in  particular,  accomplish  this  under the
most difficult circumstances.   Currently,  manufacturers  are  faced with the
problem  of  collecting  high  resistive  particulate  matter  from  boilers
burning  low  sulphur  coal.    One  of  the  variables  of interest  in  this
application  is  the  proper dislodgement  of the particulate matter  from the
collection  surfaces and discharge  electrodes.   The basic  motivation is to
minimize  the  influence  of  the collection  and  removal  of  this ash  on the
voltage  and current characteristics of  these two  fundamental precipitator
components.

     It  is  generally  accepted  that  if  the  collected layer of high resistive
material  is  thin,   it  will  not  unduly  diminish the  level  of  charging
current.   As a result,  collection  surface  rapping response  specifications
often  require 50  G's  acceleration  as a  minimum acceptable level.    When
these   requirements  were   first  released,   precipitator   manufacturers
responded  with  claims  for  their  equipment  that were disimilar.    This
apparent  inconsistency,  all  based  on  credible  records, gave  rise  to the
question as  to whether there was a single  rapping technique more desirable
than any other.

                                     208
-------
     To resolve  this question  we initiated  a multi-phase program to  study
the dynamics of  rapping and  develop  appropriate hardware.  The project  was
comprised  of a  literature  search,  a  field  pilot  study,  a  comparison  of
rapping response  on three  different  full  size  collection surfaces and  the
design and testing  of new hardware.

     This paper is  a brief  report on  selected subjects from these studies.

Background

     The  published  literature  related   to  rapping  was  collected   and
reviewed.  This  information was  compared  to  our knowledge  of rapping, which
had  its  beginning  in  1962  with  our  first  full  size  mechanical  test
facility.   Following  are  some  of the  subjects  on which  there is general
agreement.

     Ninety  percent of the  authors  agree with our experience that the most
relevant directional component  of  acceleration for consistently predicting
the removal  of dust  is  the  one  perpendicular to the surface.

     Measurements  made during  the  subject project  confirm published test
results  and  therories that  the  sheet  metal acutally buckles when struck  in
the plane of the metal.  It  physically moves in a serpentine manner.

     The  rapping   shock  and  resulting  wave  transmission  is   similar   to
dropping a  pebble  into a  pond of water  surrounded by irregular boundaries
in that  the  resulting ripples and their  reflections  add,  subtract or cancel
at different locations.  It  should  also be understood that the shock waves
on a  collection surface  are comprised  of a  spectrum  of  frequencies.    The
resulting  sheet  metal motion is  therefore extremely complex.   As a result,
even   under   controlled  conditions,   repeatability  of  measurement   is   a
problem.   Add  to  this  the  fact that the  impact energy  diminishes  as   it
moves  away  from the source  and  it is  easily understood that the collection
surface response to a  rapper cannot be uniform.

     As  a  rule  of  thumb,   the  acceleration of  the  metal  controls  the
thickness  of the  collected  layer  and the  rapping rate  is  related  to  the
rate of deposit.

     When   a  rapper   strikes    a   collecting  surface   in  an  energized
precipitator,   only  a  small   amount  of   material   breaks   free.      The
agglomerated  dust  breaks  free only in  those  places where  the shock wave  can
upset  the  balance  between  the  mechanical  forces  within the  layer,  the
electrostatic  forces  and  gravity.    Visual  and  photographic  observations
show that  the removal of material  looks like a miniature  explosion in that
the  particulate  matter  is  violently  projected  horizontally.    The  large
agglomerates  fall  and  the  finely divided material moves downstream until it
is recollected.   If too much material  is  removed  at  one time, the resulting
surge  in  dust  concentration   can  influence  the  downstream  voltage  and
current  characteristics,  and from an outlet  field,  the   result  will  be  a
sudden increase in  stack opacity.
                                      209
-------
Pilot Phase

     In  order  to  quantify the  level of  acceleration  best suited  to  ash
removal, a pliot  precipitator was installed  at  a power station burning  low
sulphur  coal  and  equipped with  a hot  side precipitator.   The  pilot  was
connected in a manner which permitted a  portion of the flue gas to be  taken
from either  before or after  the  large  precipitator.   In  addition,  a heat
exchanger was installed  ahead of the pilot unit  to  vary the temperature  as
desired.   The collection  surface response  to  rapping was  measured before
operation began making  the unit a calibrated instrument.   The results show
that rapping requirements  can vary with  temperature.  At both cold side  and
hot side temperatures the  unit performing  as an inlet field was best rapped
at between 15 and 25  G's (average).   When performing  as  an outlet field  at
the  hot  side  condition,   no  change was  required  in  rapper  intensity,
however, at  cold  side temperatures,   a range  of 65 to  85 G's  (average) was
required for  best performance.   Raising the rapping  intensity above  these
levels  always caused  excessive  reentrainment  and   created a  significant
increase in the scatter of efficiency data.

     The  conclusion  is  that  50 G minimum  is acceptable   as  a  design
specification.  That  is,  the  equipment  should be capable  of achieving this
level, but the actual operating values must  be  selected wisely because even
high resistivity ash can be rapped excessively.

Full Size Test Tower Evaluation

     In  order to  establish an understanding  of  accepted rapping practice  in
the  air cleaning industry,  three different types  of collection  surface
configurations were  selected   for  comparison.  One  configuration consisted
of separated vertical panels  joined  only at  the  top  and bottom by a common
member  and  rapped at the  bottom.   The  second  was  a  structural  frame with
many  sheet  metal  panels  attached to  it.   The  third was made  of vertical
panels  interlocked along  their adjacent edges  and  attached  to a stiffener
at the  top and bottom.   The latter two  styles are top rapped.   Each of the
three  designs  has  a  long history  of good  performance and  in combination
they  are  believed  to   adequately  establish  a  typical  range  of  usable
characteristics.

     Test  specimens  of  types  one  and  two  were  direct  copies of   other
manufacturers  equipment  and   the  third  was  our  standard design.    The
corresponding suspension  system was  also  duplicated  for each  system.   All
were  installed  in the largest  and newest  of our two  test  towers.   All  of
the  collection  surfaces  were  the  same  size,   3.94 meters  wide by   14.76
meters high.

     The  analytical  experience used  for   this  phase of  the project was
accumulated from  the  many tests conducted over  the  years  in our mechanical
test  facilities.    This  background   provided  a  complete  awareness  of  the
complexity of  this kind of measurement  and  a concerted effort was  made  to
control  as many  variables as  practical.    Instrumentation  and  techniques
were selected with care,  in fact,  as many  as fifty accelerometers were used
in   order    to    eliminate   variations  resulting   from    relocation    and
                                    210
-------
attachment.   In  this  way  a variety  of tests  could be  conducted  without
disturbing  the  instruments  or  the test  specimen.   The measurement  system
had a  flat frequency  response  to SO KHz.   The accelerometers weighed  0.4
grams  and  were  attached with  a  quick curing  adhesive.   The  acceleration
from  each   location  was measured  with a  peak  reading  voltmeter (zero  to
peak).  A frequency spectrum analyzer was among  the instruments used.

     Rapping  for  all  tests was of the  single  impulse type;  specifically,  a
mass falling  by  gravity.  Each surface received an  impulse  of  13.56  joules
(10 ft.  Ibs.).   This value was selected  arbitrarily  and does not represent
field practice.   To be consistent,  a 27 joule  (20  ft. Ibs.)  impact was used
for the two systems which rap two plates simultaneously.

     The  two  top  rapped systems typically  use two  rappers,  one near  the
leading  edge  and one  near  the  trailing edge.    Previous testing  experience
demonstrated  that  the  two  rappers essentially  act  independently,  therefore,
the  response  to  rapping was assumed  to be symetrical  about the vertical
centerline.   The bottom rapped plate  was  fully instrumented because only
one  rapper was  used.   This  technique is consistent with  the understanding
of  dust removal  as  described  earlier.    That  is,  the  layer thickness  is
determined  by  the rapper  closest  to  the  location  in  question (maximum
acceleration).

     The  patterns of  acceleration  are shown  in Figures  1,  2,  and  3.  The
differences are  the  result  of  the methods of construction  and the influence
this  has  on  the  characteristics  of wave  propagation and the  frequency
composition of  the response.  The frame  or  drum head construction produced
the   strongest   low   frequencies,  the  lowest  acceleration  levels  and,
therefore,  the   greatest  displacement.     The  separated  slender  panels
contained  the  highest  frequencies,  highest  acceleration and,   therefore,
smaller metal displacement.   The interconnected panels produced  a response
characteristic between and overlapping the other two  responses.   The three
styles are  similar in  their  lack of uniform distribution.

     The significant  frequencies  ranged from 10's of Hz to  10,000 Hz.   One
frequency  spectrum for each style is shown below.  Actually, the frequency
composition   is  not   consistent  over  the  surface  of  any  of  the  test
specimens.    The  cause  of  this  is  the complexity  of the  wave  motion  as
mentioned  in  the background section  of this paper.   Therefore,  these data
samples  serve only to graphically display our  analysis and  cannot indicate
the variety of characteristics  that can be found.

                         FREQUENCY  SPECTRUM
                               0  TO  10KHz
  SEPARATED PANELS
PANELS ON FRAME
  ENELCO  COLLECTION
        SURFACE


WWWwJ^
                                     211
-------
ENELCO COLLECTION SURFACE   PANELS ATTACHED TO  FRAME

                  RAPPING RESPONSE  (G)

                 SIZE  - 3.94m X 14.76m
               FREQUENCY RANGE  TO 30 KHZ

              RAPPING ENERGY-13.56 JOULES

    RAPPER      RAPPER
       t            •

315
325
263
183
187
177
167
+
138
4
120
•f
173

863
782
587
320
340
338
350
+
233
+
198
•f
305

465
380
423
260
220
258
203
+
218
•f
175
•f
197

338
220
188
253
135
107
193
4
140
+
127
•f
122

338
220
188
253
135
107
193
•f
140
4
127
+
122

465
380
423
260
220
258
203
+
218
+
175
•»•
197

863
782
587
320
340
338
350
+
233
+
198
+
305
•
315
325
263
183
187
177
167
+
138
•f
120
•f
173
         FIGURE 1
RAPPER RAPP
t . .

+
195
+
133
+
107
+
83
•f
70
•f
100
+
88
+
193
+
132
+
78
•»•
65
•f
75
•f
68
•f
90
•f
193
•f
132
+
78
+
65
+
75
•f
68
4-
90
•f
195
•»•
133
•»•
107
+
83
+
70
•»•
100
•f
88
:R

FIGURE 2
                           212
-------
     SEPARATED  PANELS
   RAPPING  RESPONSE  (G)
RAPPING ENERGY-13.56 JOULES
 FREQUENCY RANGE TO 30 KHZ

223
192
187
220
180
318
+
258
300
+
420
•f
577

217
183
198
268
203
252
+
218
•f
255
•f
443
+
572

223
207
200
182
182
242
+
252
•f
233
+
413
+
535

228
235
250
242
253
203
+
312
+
313
+
467
+
816

213
190
192
213
210
243
•f
288
•»•
343
+
488
+
650

295
232
285
250
^
265
323
+
332
+
420
+
555
•f
980

275
257
317
273
A
298
440
+
420
+
467
+
713
+
1067











ENELCO COLLECTION SURFACE
   RAPPING  RESPONSE  (G)
RAPPING ENERGY-6.8 JOULES
FREQUENCY RANGE TO 30 KHZ
                                    RAPPER
                RAPPER
          FIGURE 3
                      RAPPER
413
333
268
275
198
215
243
222
208
223
167
228
170
130
195
127
15$
146
!S2
182
140
102
1 52
155
142
130
155
202
123
533
387
297
417
447
240
343
317
300
357
218
257
337
318
283
278
27-7
$80
so5
235"
265
2*08
240
277
297
267
232
278
?SO
683
423
423
313
267
233
283
377
270
303
245
285
252
242
303
188
285
278
172
19*
163
238
2*30
193
170
140
222
220
?47
283
303
392
397
220
185
283
267
203
205
142
223
230
150
138
138
"90
235
18*8
158
183
14*2
ISO'
168
148
202
160
173
195
283
303
392
397
220
185
283
267
203
205
142
223
230
150
138
138
190
235
1§8
1*58
183*
14-2
ISO
168
148
202
160
173
195
683
423
423
313
267
233
283
377
270
303
245
285
252
242
303
188
28£
2^8
172
19*
16*3
238
230
193
170
140
222
220
247
533
387
297
417
447
240
343
317
300
357
218
257
337
318
283
278
257
$80
305
235-
20*5
208
240
277
297
267
232
278
•
250
413
333
268
275
198
215
243
222
208
223
167
228
170
130
195
127
15$
146
152
18*2
146
102
172
155
142
130
155
202
123
WG.=243.3G DISTR.=37.5*/
FIGURE 4
                            213
-------
     Realizing that  all  three styles of  collecting surfaces are  capable  of
providing good performance  and  discounting the influence of the mass  of the
dust  on the  frequency content,  it could  be  conjectured  from  their  wide
range  that  frequency  and  displacement  are  variables  of  no  practical
value.   However,  if there are desirable  frequencies they would most  likely
be found below  3,000 Hz  which is the upper  limit  of the common frequencies
on these three clean surfaces.

     The analysis  of the  data below 3000 Hz  does  not  permit a selection  of
one  combination   of   acceleration   and   frequency  as  being  typical  and
therefore  the  best.    In  fact,   based   on  the   variations   found  on  an
individual  surface,  it  appears  that  a   variety  of  combinations  must  be
accepted.    Even   if  a  good  combination  exists  it  is  probable  that the
practical  limitations  for  design and fabrication  leave acceleration as the
only  controllable  variable;  which is  what  the   single specification  of
minimum  acceleration  implies.    This  is  disappointing from a  technical
viewpoint  but  it  does  relieve  the  difficulty  and added cost  of precision
engineering.

     Additional testing  of the  interconnected  panel design  revealed that a
stronger  top stiffener would improve  rapping  transmission  and durability.
Test specimens using a new stiffener were  installed  for further testing.

     The  improved  interconnected   panel  specimens  were evaluated with  a
higher   density  of  measurement  locations.    The  locations  were  spaced
uniformly  vertically,  but,  one of three positions  was randomly selected
horizontally  to  lessen the influence  of  instrument location.  Two sets  of
data  were  collected.   The first  contained  all frequencies  up  to  30  KHz.
The  second  contained  only those  frequencies  below 3  KHz  because  of the
observation  described  previously.   See Figures 4  and  5.   It is interesting
to  note that the  level  of indicated  acceleration was  changed  by altering
the  frequency  content but  the  distribution  was not.   (Distribution  is
defined  as  the   standard  deviation  expressed   as   a percentage  of  the
average.)

     The 3 KHz  test  was  continued  in order to determine the energy required
to meet  two  logical  forms  of  a  50  G specification.  One form is the minimum
and  the other  is  the average.    Because  of  the  transmission  efficiency
resulting  from  directly  striking   the  new  top  stiffener,  only  4.3 joules
(3.2  ft.  Ibs.)  were required to provide a 50 G minimum.  The corresponding
average  value,  however,  is greater  than the  average value  indicated in the
pilot  study  inplying  that  this  amount of  energy  is  not  always  necessary.
See Figure  6.   An input of 1.15 joules (0.85 ft.  Ibs.) produced  a response
pattern  with an  average  of 46  G  and a  minimum  of 26 G.   Which of  these
forms  is the best  is not  clear  but  knowing the range of dynamic response  is
fundamental  to  selecting  operating parameters  and maintaining a history  of
performance.

     Additional  information,  for  use  when  designing  rappers for  the top
rapped  interconnected  panel,  was obtained by comparing the  response from a
magnetically, pneumatically  and gravity  driven cylinder  of  metal.    There
was no  significant  difference in response characteristics.   Following  this,

                                     214
-------
             ENELCO COLLECTION SURFACE
                RAPPING RESPONSE  (G)
              FREQUENCY RANGE  TO  3 KHZ
RAPPING ENERGY-6.8  JOULES
     RAPPER   RAPPER
{87
24*3
118*
120
l6o
i6s
uJ
•
92
110
112
•
72
88*
9*7
•
92
•
98
60
82*
270
246
1*85
217
200
126
257
122
138
140
145
158
150
125
123
105
108
75* 147
• •
70 92
88 145
• •
77 145
70 138
• c
75 85
83* 118
*80 147
8*0 105
• •
83 98
*82 110
9*0 137
293
208
170
213
?30
14*7
153
120
160
137
128
175
130
140
117
87
105
100
102
105
•
98
115
•
85
9*0
97*
*63
108
*98
102
222*
233
220
185
143
115
130
160
120
102
•
92
107
115
•
97
•
83
100
*98
90*
8*0
•
87
90*
7*8
•
83
•
77
8*0
98*
*88
7*7
*75
222
233
220
185
143
115
130
160
120
102
•
92
107
115
•
97
•
83
100
•
98
90*
8*0
•
87
90*
•
78
*
83
*77
8*0
98*
*88
•
77
*75
293
208
170
C
213
130
147
153
120
160
137
128
175
130
140
117
*87
105
100
102
105
•
98
115
*85
9*0
97*
•
63
108
*98
102
27*0
240
1*85
217
200
126
207
122
138
140
145
158
150
125
123
105
108
147
92*
145
145
138
85*
11*8
147
105
9*8
110
137
1 	 J
1*87
243
118*
120
100
108
117
•
92
110
112
•
72
88*
•
97
*92
•
98
6*0
82*
•
75
*70
88
•
77
•
70
•
75
83*
*80
8*0
•
83
•
82
9*0
RAPPING ENERGY-4.3 JOULES
     RAPPER   RAPPER
                                              r
178
183
133
128
8*7
95
105
*97
118
8*7
*78
88*
6*9
*80
*9S
5*0
59*
61*
*52
63
*
67
*67
58
69*
*76
7*7
69*
*69
6*7
228
168
188
185
185
120
180
103
152
133
118
163
*120
107
108
*92
110
*95
85*
115
105
102
81*
102
101
*87
9*4
91*
114
217
162
117
170
105
125
142
107
103
132
102
*152
123
in
8*6
*63
77*
7*7
*69
89*
7*0
91*
*93
7*2
73*
*55
94*
*74
8*1
182
157
125
150
118
112
105
140
102
*93
6*4
87*
*92
•
79
56*
7*7
*63
63*
6*5
65
73*
6*7
81*
61
7*6
65*
*80
6*1
*77
182
157
125
150
118
112
105
140
102
*93
6*4
87*
*92
7*9
56*
7*7
*63
63*
6*5
65
73*
6*7
81*
*61
7*6
65*
*80
6*1
*77
217
162
117
170
105
125
142
107
103
132
102
152
123
117
86
*63
77*
7*7
*69
89*
7*0
91*
*93
7*2
73*
*55
94*
*74
8*1
228
168
188
185
185
120
180
103
152
133
118
163
*120
107
108
*92
110
*95
85*
115
105
102
81*
102
101
*87
9*4
91*
114
178
183
133
128
87
95
105
97
118
*87
*78
88*
.
69
*80
*95
5*0
59*
61*
*52
6*3
67*
*67
58
69*
*76
7*7
69*
*69
6*7
 flVG.=122G   DISTR.=38.17.
        FIGURE 5
 AVG.=102G   DISTR.=37.17.
        FIGURE 6
                         215
-------
the gravity  device was used  to  evaluate the influence  of velocity at  time
of impact.   Using a constant 27 joules  (20  ft.  Ibs.), various weights  were
dropped from corresponding heights up to 3.28 meters  (10 ft.).  In general
the  response  had  the  same  frequency   components  under  all  conditions.
However,  the low velocity  impact  slightly  raised the  signal  level of  some
of the  low  frequencies and the high velocity  strengthened the upper end  of
the spectrum,  but the difference is considered  to  be  insignificant.  It  is
therefore possible  to  conclude  that  the  single impulse rapper mechanism for
the interconnected panel  style  may be  selected  on the  basis  of  economics
and preference,  assuming  the energy  output is  acceptable.   The  required
energy is of course a  function of each system design.

Discharge Electrode Development

     The  challenge  of  collecting high  resistivity  dust  with very   high
efficiency  has also influenced  the  requirements for  discharge electrodes.
These now include  the  following.

    1.  Be  suitable for  tall systems, which are  becoming more prevelant  in
        this application.
    2.  Possess  cleaning   characteristics   like  those   of  the  collection
        surfaces.
    3.  Enhance the collection efficiency as compared  to  wire  electrodes.
    4.  Have high  spark over voltage.
    5.  Have selectable voltage/current characteristics.
    6.  Be tolerant of in  service abuse.

     The  electrode  developed to  meet  these  requirements  is  based on  a
tubular  shape.    This  characteristic  alone   responds  to  many   of  these
specifications.     The   tubes   provide   good   vibration  characteristics,
significnt  additional  collecting  surface,  durability, a strong electrical
field and high spark  over voltage.   Pointed corona emitters attached to the
tube  provide early corona  onset  and the number  of emitters per  length  of
tube determine the  voltage/current characteristic.

     The   variability   of   the   voltage/current   relationship   has   a
significantly  wide range  considering  that  the  large  diameter  tube extends
the voltage  capability by raising  the  spark over limit.  An example of  this
flexibility  is shown  in Figure 7.   It is evident that  the  slope of the  El
curve  can be  varied  to  achieve the  two best characteristics,  i.e.,  early
corona  onset combined  with reasonable  current  usage at high voltage.  These
data  were obtained in air  in the  laboratory but the  relationship  has  been
confirmed in actual operation.

     It  is   typical of the breakdown  voltage  of  air  in  the  laboratory  to
vary  with temperature and  humidity.   Figure 8 demonstrates  the spark  over
advantage of the  tubular  shape during a  time when  the wire electrodes  were
limited by  sparking.   The full  size installations containing this  electrode
exhibit  little or  no  sparking  after  the inlet  field,  which,  compared  to
past  experience,   confirms  the  lab  prediction.     From  earlier  field
observations   and   this   small   amount  of   evidence,   it  is  possible  to
conjecture that a  significant amount of sparking  in weighted wire systems
                                    216
-------
collecting high  resitivity ash occurs  in  the  vicinity of where  the  wires
pass the top and bottom of the collection surface.
                        DISCHARGE  ELECTRODE
              VOLTAGE  AND  CURRENT  RELATIONSHIP
                                                                POWER SUPPLY
                                                                LIMIT
TUBULRR
RIGITRODE
                         SMOOTH WIRE
      10  20  30  40  50  60  70  80
             VOLTflGE CKV)
               FIGURE 7
                                                              TUBULAR
                                                             RIGITRODE
                  10  20  30  40  50  60  70  80
                        VOLTflGE  (KV)
                          FIGURE 8
     The  cleaning characteristic of  these  electrodes  is interesting.   In
 addition  to being  capable  of  a  50  G  minimum  surface  acceleration,  the
 vibration  is sustained.   In a clean state,  the motion is not totally damped
 for  as  long  as  2  to  4  seconds.    This  action is  similar  to a  musical
 chime.   The  pointed  pins,  being attached  to  the  highly  active  surface,
 remain  dependable corona  emitters because they do not retain excessive dust
 coatings.

 Total System Capability

     Three  precipitators  containing  this new technology are in service and
 three  are  under construction.   The  collection surface  rapping  is  the same
 for  all collection  surfaces  in  the  operating units.    A  midrange  value  of
 3.05  joules (2.25 ft. Ibs.)  per  collection  surface was selected.   It will
 be difficult  at best, and  certainly premature at this time, to quantify an
 enhancement  factor  over weighted wire  for  the  new system.   The only firm
 data  in hand  is  a 25%  improvement  in  drift velocity  determined  by pilot
 tests.   Nevertheless, the  test  results from the first three installations
 support the   expectation   that  an  exciting  improvement   in  precipitator
 technology  has been achieved.

 Acknowledgement

     The  test  measurements  and  analysis  were  supervised  by  K.   Teel,  a
 member  of  the  Technical  Department  at  the  time,   and  his  fine  work
 contributed to  the success  of the project.
                                     217
-------
          AN APPLICATION SUMMARY OF HIGH ENERGY SONIC CLEANING
                 APPLIED TO ELECTROSTATIC PRECIPITATORS

                                   By:

                          Michael J. Berlant
                          KVB,  INC.
                          18006 Skypark Blvd.
                          Irvine, CA  92714

                                ABSTRACT

     Acousticlean sonic sootblowers, or "horns", are low frequency, high
energy acoustic devices that have demonstrated an ability to solve many
of the particulate buildup problems associated with the operation of
electrostatic precipitators.  Particles clinging to surfaces are dis-
lodged by sound wave vibrational energy.

     The paper describes four applications reducing particulate build-
up in the precipitator.  The areas cleaned are the inlet duct, turning
vanes, distribution plates, and wires and collector plates.  The result
of using the sonic cleaner has been an extension of precipitator on-
line times and an increase in efficiency.
                              INTRODUCTION

     Low frequency, high energy sonic cleaning provides an operational and
economic alternative to traditional methods of cleaning particulate build-
up in electrostatic precipitators.   Initially used by Inspiration Consoli-
dated Copper (ICC) company to mitigate particulate buildup on a precipi-
tator inlet distribution plate, the Acousticlean sonic sootblower is
currently used to clean particulate buildup throughout the precipitator.
The results of applications that followed ICC have been extended precipi-
tator on-line times and increased precipitator efficiency.  These results
are due to continuous particulate removal provided by short, but frequent,
blowing of the sonic cleaners.

Areas of Particulate Buildup in the Precipitators

     A wide variety of industries use electrostatic precipitators to re-
move solid particles from their flue gas discharge.  The particulate in
the flue gas varies from pulp mill salt cake to copper smelter metallic
oxides to utility coal ash.  The particulate may build up in one or more
of the following areas in the precipitator:

        inlet duct
        turning vanes
        distributor plates
     .  wire electrodes and collector plates
        ash hoppers
        outlet duct
                                   218
-------
     Particulate buildup in these areas can require  frequent  shut-downs for
cleaning and can reduce operating efficiencies.  Strategically  located sonic
horns can mitigate or eliminate buildup in these areas.

     Before discussing typical sonic horn applications in these problem areas,
the next section describes the horn.

The KVB Acousticlean Sonic Sootblower  (Horn)

     Figure 1 is a picture of the KVB  horn.  Horns are low  frequency, high
energy acoustic devices—250 Hz and 145 dB.  The bell is made of cast stain-
less steel and is good for temperatures of 1900°F  (1040°C).   There  is only
one moving part—a titanium diaphragm.

     The horns operate on compressed air  in the  60 to 80 psi  range.  Plant
air is satisfactory and  instrument  air is not  required.  Air  consumption
during insonation is about 60 SCFM. They are  turned on and off by  actuating
a 110 volt, normally closed, solenoid  valve.   In operation  this valve is
controlled by an automatic electric timer.

     Each horn weighs 55 pounds and measures about two feet in  length and
one foot in diameter at  the mouth of the  bell.  The  bell is bolted  to the
housing to facilitate installation. Diaphragms—which should last  six
months to one year—are  replaced by simply unbolting the back of the dia-
phragm housing.  An oil  mister  in the  air inlet  line can be used to extend
the life of the diaphragm.   In  operation  the diaphragm is cooled by a con-
tinuous air flow of three to  five SCFM.

     Soot and other particles clinging to surfaces are dislodged by sound
wave vibrational energy.  The wave  pressure  fluidizes particles by  breaking
their bond with other particles and the  surfaces to  which they  cling.  Once
 "fluidized," particles will  flow  from surf aces by gravity or gas stream
pressure.

     Sound pressure level is  a  physical measure  of the strength of  a sound
and is defined by the following equation:

L = 10 log


L = sound pressure level in dB

p = sound pressure in Pa
PO = reference sound pressure,  2 x  10   Pa

     Sounding two horns  simultaneously will  double the sound  pressure or
increase the energy level by  3  dB.

     The "fluidizing" effect is compounded  in  an enclosed space where large,
solid surfaces reflect sound waves, creating a homogeneous  vibrational  energy
field.
                                    219
-------
             PAPER AND PULP MILL RECOVERY BOILER EXIT DUCTS/
                        PRECIPITATOR INLET DUCTS

     Weyerhaeuser Company in Longview, Washington, is using ten horns  in its
recovery boiler economizer hoppers and exit ducts/precipitator inlet ducts.
Saltcake buildup in the ducts gradually restricted load on the boiler  and
channeled inlet air adversly to the precipitator.

     This recovery boiler is a 1,090,000 Kilograms per day  (1200  TPD)  low-
odor design  in which the flue gas flows directly from the economizer out-
'let to  an electrostatic precipitator.  The exit ducts to the precipitator
are inclined upward at a 45° angle, as shown in Figure 2.  Saltcake build-
ups tend to  occur in the rear sections of the economizer hoppers  and in the
inclined exit ducts.  The problem is believed to stem partly from high-sul-
fidity  pulping, which appears to produce unusually sticky ash, as well as
from the sharp turn the flue gas must negotiate as it exits the boiler.
The problem  was sufficiently severe that it gradually reduced maximum
load on the  unit as each scheduled maintenance and cleaning period
approached.

     The mill rejected the  addition of more steam lances to solve this
problem because of their relatively high initial cost and the continuing
 steam penalty required to operate them.  Vibrators were also rejected
because they would apply unwanted stress to the ductwork'and also be-
 cause they are intended to  flow the solids downhill, which in this case
 is against the gas flow.  Horns were selected primarily because they
would fluidize the solids so that they would be carried with the  flue
 gas into the precipitator.

     During  the Christmas outage of 1978, four horns were installed in
 the economizer hoppers and  exit ducts as shown in Figure 2.  Due  to the
 rapid rate of saltcake buildup and the sticky nature of the saltcake,
 it was  found best to sound  the horns rather frequently.  They each op-
 erate for 10 seconds at three minute intervals.  The horns have proven
 successful in reducing the  buildup problem to the extent that it  no
 longer  restricts load on the unit.

     Due to  the shallowness of the exit ducts, which are only 1.22m
 (four feet)  deep, the sound energy from the horns does not have an
 opportunity  to spread very  much before reaching the bottom of the
 ducts.  Therefore, each horn cleans only a fraction of the duct in
 which it is  located.  Each  horn appears to clean an area approximately
 1.53m  (five  feet) in diameter.  In order to keep larger fractions of
 the ducts open, and thus provide a more comfortable margin of pluggage
protection,  Weyerhaeuser installed four horns in these ducts.

             PAPER AND PULP MILL PRECIPITATOR TURNING VANES

     Horns have been used by ITT-Rayonier at Jesup, Georgia, to reduce
 saltcake buildup at the inlets of two recovery boiler precipitators.
                                     220
-------
Not to Scale
One Foot = 0.305m
                                         Side Elevation
                                         of Economizer
                                         Exit Duct/ESP
                                         Inlet Duct
           PLAN VIEW OF ECONOMIZER EXIT DUCTS/ESP INLET DUCTS
FIGURE 2.  Application of Horns to Recovery Boiler Economizer Exit Ducts/
                Precipitator Inlet Duct.
                                    221
-------
     A secondary precipitator at the mill treats half the  flue  gas  from three
recovery boilers having a combined rating of 682,000 kilograms  per  day (750
TPD) of pulp.  The precipitator has a history of saltcake  buildup between
turning vanes located in the inlet plenum.  These deposits were observed
at all scheduled maintenance outages and were normally removed  by washing.
During the Labor Day outage in 1979, two horns were installed on the  top
of the precipitator inlet plenum as shown in Figure 3.

     The horns were set to operate simultaneously for 30 seconds at ten
.minute intervals.

     At the Christmas outage in 1979, after nearly four months  of horn
operation, the turning vanes were observed to be free from saltcake build-
ups with the exception of those parts of the top vanes lying outside  of
the conical sound patterns produced by the horns.  The mill concluded
that the horns were effective in controlling saltcake buildups.

          COPPER SMELTER PRECIPITATOR INLET DISTRIBUTION PLATES

     Two horns have been operating in an electrostatic precipitator that
serves the flue gas from a copper smelter reverbatory furnace since July
 1979.  Prior to installation of the horns ash buildup in the inlet  nozzle
would bridge the gap between the inlet distribution plates.  The ash  would
build upon itself until the flue gas was effectively channeled, resulting
 in the nozzle being blanked off by plant personnel for cleaning every two
 to three months.  In January of this year the plant reported that after
 eighteen months of operation with the horns that the distributor plates
 have been clean and that the inlet nozzle has not been blanked  off.

     The copper smelter's reverbatory furnace is coal fired.  The flue
gas from the furnace contains coal ash and smelter dust  (metallic oxides) .
The flue gas from the furnace enters the precipitator at 330°C  (625°F)
at a gas flow rate of about 5700 m /minute  (200,000 ACFM) .  The flue  gas
is divided into three inlet nozzles one of which is currently blanked off.
Each nozzle has distribution grids of 7.3m x 7.3m  (24 feet x 24 feet)  and
9.2m x 9.2m  (30 feet x 30 feet), 76.2 centimeters  (30 inches) apart.

     The horns are installed forward of the first plate about mid-way in
elevation as shown in Figure 4.  The horns are operated simultaneously
for 30 seconds every 20 minutes.  The plant has not replaced the original
diaphragms during the eighteen months of continuous operation.

The overall results of utilizing the KVB low frequency acoustic horns have
produced several significant benefits.  Use of the horns has increased total
precipitator running time by eliminating the periodic shut downs to clean
the inlet nozzles.  The efficiency of the precipitator has increased  from
the even distribution of flue gas.  The plant has saved the labor cost
of cleaning the plates and the inlet nozzle.

                     COAL FIRED UTILITY PRECIPITATOR

     A coal-fired utility in Spain began using horns in their electrostatic
precipitator as a supplement to their rappers and vibrators about eighteen
                                    222
-------
                  FRONT
   SIDE
SECTION AA
                            r
                          JL	
                      34'
                                                                 30'
     Not  to  Scale
     One  Foot  =  0.305m
FIGURE 3.   Application of Horns to clean turning vanes  in Recovery
               Boiler Precipitator Inlet Plenum.
                                  223
-------
                                      TOP VIEW
                           24'
       FLUE GAS INLET
                     *-  30'
i!

                                                            30'
            Not to Scale
            One Foot = 0.305m
                                    SIDE VIEW
      FLUE GAS INLET    30'
FIGURE 4.  Application of Horns to clean Inlet Distributor Plates on
                Copper Smelter Precipitator.
                                    224
-------
months ago.  The utility's Puertollano  Station  installed the  horn  in  the
roof of one precipitator chamber  as  shown  in  Figure  5.

     The horns were installed after  the plant had  experienced problems  that
were related to the firing of low-sulfur coal (high  resistivity/low conduc-
tivity ash).  Heavy buildups of ash  would  occur rapidly on the plates caus-
ing sparking with the precipitator.   The sparking  would continue until  an
electrode would break.  Electrode breakage was  on  the order of one per  week.
Operation would continue with one field grounded but the loss of a second
field required shut down.  Shut downs would occur  every two weeks.

     The horns are installed in a French design precipitator  that  serves a
220 megawatt, oil-or coal-fired boiler. Typically,  the unit  fires a  low-
sulfur Bituminous coal with an approximate ash  content  of 37%.

     The precipitator at the Puertollano Station is  a two-chamber  unit  with
three fields per chamber as shown in Figure 6.   The  flue gas  velocity
through the precipitator chambers is about 2  meters  per second  (6-7 feet
per second) at 100°C  (212°F).

     Because of the frequent precipitator  shut  downs, the plant initially
tried steam injection to alleviate the  ash buildups. The steam lowered
the resistivity and increased the conductivity  of  the ash eliminating the
heavy buildup on the wires but at a  daily  cost  of  $3,000.  Due to  the high
operating  costs with steam, the plant installed the  horns.

     Since  initial horn operation began one and one  half years ago, the
precipitator has not had to come  down because of an  electrode failure.
In addition to the elimination of steam injection, plant personnel have
noted that  the precipitator chamber  with the  horns removes higher  levels
of fly ash. This is evident when they  compare  the ash  handling between
the chambers.

     The horns operate in  series, with  five seconds  of  operation and  a  two
second delay between each horn on a  continuous  basis.  The wire rapper  cycle
 is one minute and fifty five seconds every ten  minutes.  There are fourteen
collecting  plate rappers that rap eight seconds each in a ten minute  cycle.

     The horn diaphragms are replaced every four to  five months.   The plant
is not using an oil mister that could extend  diaphragm  life.  Plant per-
sonnel are  very impressed with the effectiveness of  the horns and  they
intend to  install nine additional horns in the  second chamber of the  pre-
cipitator.

                                CONCLUSIONS

     The use of low frequency, high-energy acoustic  horns significantly
reduced particulate collection problems in electrostatic precipitators.
The horns  are effective on a variety of particulate  types in  a variety
of industrial categories.
                                     225
-------
N3
r-o
               FIGURE
5.  Horn Installation  in Roof  of  Utility Precipitator,
-------
                                        TOP


                                        47'
                                                       A
                                                       ,P-6'

                                                       /.
                                                       "3.6'
                                                        6.6'
                                                H-H
      Not to Scale

      One Foot = 0.305m
                                       SIDE
FIGURE 6.   Application of Horns to Plates and Wires in Coal-
                Fired Utility Precipitator.
                                                                31'
                                     227
-------
FIGURE 1.  Acousticlean Sonic Sootblower.
                                    228
-------
     The horns effectively reduced particulate buildup around the precipitator
inlet ducts, turning vanes, distribution plates, and wire electrodes and col-
lection plates.  Plants using the horns have saved labor costs, reduced more
costly alternative cleaning methods, extended the on-line time for the pre-
cipitator and, in some cases, observed an increase in precipitator effici-
ency.
                                    229
-------
             THE IMPACT OF INTELLIGENT PRECIPITATOR  CONTROLS

                      By:   Dr.  Norman Z.  Shilling
                           Robert 0.  Reese

                           Jeff A. Fackler
                           Buell Emission Control Division
                           Envirotech Corporation
                           Lebanon, Pennsylvania  ]7042

                                  ABSTRACT

     This paper describes  both operational hardware and supporting software
which have been developed  for   microcomputer control of an electrostatic
precipitator.   Particular  emphasis is placed upon operator-computer inter-
action and the need to ensure acceptance of the control concept  at the plant
level.

     Specific  examples of  normal operator/control system interactions are
drawn from an actual microcomputer-controlled precipitator which is serving a
utility boiler.  The microcomputer provides a significant scope of flexibility
to obtain maximum precipitator performance consistent with low power consump-
tion.  Transformer-rectifier power disposition and rapping parameters are
controlled by opacity feedback and other key precipitator operating parameters.
The logical structure of the adaptive/recursive software which provides this
control is described.  Features of microcomputer control are outlined which
will provide increased reliability and nearly failsafe operation.

                                INTRODUCTION

     Intelligence is defined as the capacity for reasoning and  understanding.
With regard to a microprocessor/microcomputer-controlled (i.e.,  intelligent)
precipitator,  we would modify this definition to "the ability  to provide a
logical sequence of responses based upon a perception of operating conditions".
The endowment of mental faculties to equipment that sometimes is viewed as the
manifestation of black magic may bring shivers to some.  Hopefully, this paper
will put such fears to rest by rigorous definition of the benefits attainable
with microprocessor control.   The questions which are addressed here are, in
turn, (1) how is it implemented in terms of both hardware and  software, (2)
how is it structured to ensure acceptance and proper utilization by plant
personnel, and at the bottom line,(3) how is it expected to impact plant
operation?

     In answer to the last question,  and as a capsule summary of this presen-
tation, the Intelligent Precipitator will impact immediately on two fronts —
specifically,  improved maintainability (and ultimately availability) and re-
duced operating power consumption.  In plant operation, previously the sole
edict which came from "on high" was to keep the plant operating.  Now the
directive is to keep the plant running and do it efficiently.

     The hardware and interactive software described here are presently in

                                     230
-------
service at an operating power plant.  The high level software which provides
performance optimization is now being qualified.  There are no technology
barriers standing between this point in time and universal adoption of this
concept.  However, achievement of the full potential of computer control in
the plant will, in large part, depend upon acceptance in the "trenches"
through day-to-day operation, how it treats, and how it is treated by operat-
ing plant personnel.  This aspect of operation constitutes a very important
aspect.  Appropriately, it is discussed first.

              OPERATOR/INTELLIGENT PRECIPITATOR COMMUNICATION

     In the Buell Intelligent Precipitator, operational control proceeds on
two levels.  Monitoring, logging and direct operator control constitute the
primary level.  Here, detailed error and failure conditions are provided to
the operator, as well as key precipitator operational parameters via CRT,
typewriter, and control panel annunciation.  At the operator's discretion,
and similar to present standard precipitator control systems, he can choose
normal threshold of sparking control in which the microcomputer defers TR
control to the stand-alone microprocessors (AVCON  2000 Boards).  These would
normally have full responsibility for control in a standard system.  The
operator also has the choice of using default rapping parameters or entering
his own.  But this is where the functional similarity ends between standard
and Intelligent Precipitator controls.

     With Intelligent Precipitator controls, the operator has the option to
make the control of rapper parameters and TR set operation the responsibility
of the microcomputer.  Adaptive control decisions can be made by the Intelli-
gent Precipitator on the combined basis of opacity and long-term electrical
performance trends which are normally transparent to an operator during a
single  shift, or even a 24-hour period.

     First, however, we illustrate interactions at the primary level.

                                  THE CRT

     The first and most striking physical difference between standard and
computer controls begins at the CRT and chassis for the Intelligent Precipi-
tator  (Figure 1).  These controls will be located in the plant Control Room,
rather  than at an area remote from other data.  This simple physical change
in location will likely impact the operator's fundamental attitude towards
the precipitator, helping him to recognize it as an integral part of the plant.

     To give an idea of the control flexibility, available commands include
the following:

Rapper  System    1.  Change of a rapper row round trip time
                 2.  Change of a specific rapper frequency
                 3.  Change of a rapper row lift (ft/lbs)
                 4.  Change of a specific rapper lift  (ft/lbs)
                 5.  Obtain a graphic illustration of rapper configu-
                     ration by number and location
                 6.  Display round trip summary

                                     231
-------
                 7.   Choose a high level rapper system control

TR System        8.   Change TR control set point as a % of full KVA
                 9.   Display TR control operating point summary
                10.   Display electrical condition operating summary
                11.   Choose high level control of TR voltages (power
                     consumption minimization)

General System  12.   Display time
                13.   Set real time clock
                14.   Program tape copy
                15.   Copy from CRT or tape to keyboard

     Actual CRT displays which, for example,  are obtained for Commands #5,
#6 and #10 are shown in Figures 2, 3 and 4, respectively.  With these controls
and commands in closer proximity to both operation and plant data, it is
likely that anticipatory changes in control corresponding to operating plant
conditions will be made.  Also, precipitator operating trends leading to pos-
sible fault conditions will be more easily perceived due to detailed graphic
display and 24-hour data summary which can be called on the CRT.

                              FAULT DETECTION

     The primary control level detects and announces various faults on the
terminal screen, performs automatic initial action, and logs the type and
time of failure in hardcopy on the typewriter and system tape.  For a speci-
fic example of how major fault conditions are handled, consider the response
to a rapper short circuit:  This fault will cause the computer to interrupt
any present tasks and handle the shorted rapper in the following manner:

1.   Power is removed from the rapper upon completion of the present half
     cycle.
2.   A special program determines the rapper number and saves it.
3.   An alarm sounds, alerting the operator.
4.   The TIME, DATE and RAPPER NUMBER are displayed on the Terminal screen.
5.   The above sequence will be repeated until the rapper is taken out of
     service for repairs.

     Some other faults that are addressed by the system are:

                 Rapper Open Circuit
                 Rapper High Current
                 Rapper Low Lift

                 Field Box — cable open condition
                 Field Box — power loss
                 TR Short

                 TR Open Circuit

                 TR Over Temperature

                                     232
-------
                 Precipitator Below Energization Temperature

                 Precipitator High Temperature

                 Purge System Failure — fan or heaters

     And, of course, any other additional precipitator or plant paramaters as
necessity or whim requires.  The specific hardware configuration which accom-
plishes this is described in the next section.

                INTELLIGENT PRECIPITATOR CONTROL ARRANGEMENT

     The Intelligent Precipitator operates in a real time mode and implements
distributed rapper control, TR control, and analog/digital I/O control, all
connected to a central microcomputer via several data bus pathways.  Figure 5
shows the Intelligent Precipitator layout.  This arrangement reduces wiring
and installation cost while increasing precipitator control reliability.

     The control system for the rapper matrix are microprocessor-based and
function independently of the host computer.  They can be downloaded from the
host computer or connected to a local terminal via a serial data bus.  The TR
controllers can be programmed to send and write data to the host computer via
a serial data bus.  As previously mentioned, the T/R controllers can also
function independently of the host computer.

     The various subsystems that comprise the control system are:  (Figure 5)

I.   AVC Control Center

          The AVC Control Center is comprised of T/R automatic voltage control
     cabinets, each having a full complement of analog meters, start-stop
     switches, low voltage contactors, and power silicon control rectifiers
     (SCR's).  In addition, each T/R Automatic Voltage Control (AVC) cabinet
     contains two companion boards:

     a.   The AVCON 2000 Automatic Voltage Controller Board monitors T/R
          secondary current changes and closes the loop by phase-controlling
          the power SCR's in the T/R primary circuit.

     b.   The Communications Board provides a serial interface between the Host
          Computer and its associated AVCON 2000 board.  T/R primary and second-
          ary currents and voltages along with spark rate date are multiplexed,
          digitized and transmitted on the bus to the Host Computer.  The Host
          sends formatted control data to each T/R AVC board through the Com-
          munication Board interface.

          The present configuration is comprised of a three-wire serial bus.
     AVCON 2000 boards are daisy-chained on the serial bus up to a distance of
     500 feet from the Host Computer.  In the event the Host Computer is out
     of service, the AVCON 2000's revert to automatic mode.
                                     233
-------
II.  Rapper Controller

          Each Rapper Controller is microprocessor-based and functions inde-
     pendently of the Host Computer.  A precision zero crossing detector
     serves as the master clock signal to time the operations.  In addition, a
     hand-held portable terminal can be used in place of the Host Computer
     interface port to enter lift and timing data as well as display rapper
     status.

          Round trip times and lift values in foot-pounds can be downloaded to
     each Rapper Controller via the serial  interface bus at initialization
     time.  In the event of a power failure, the battery back-up mode of ope-
     ration is invoked and the Rapper Control continues to function.

          In the event of a rapper failure,  the RAPPER NUMBER, FAULT, TIME  AND
     DATE are locally displayed on the local rapper status display and the
     same data is transmitted to the Host Computer via the serial interface
     bus.

          The present configuration is comprised of a three-wire serial bus.
     Rapper controllers can also be daisy-chained on the serial bus to a dis-
     tance of 500 feet from the Host Computer.

III. Analog and Digital Input/Output Controller

          The Analog and Digital I/O Controller is microprocessor-based and
     functions independently of the Host Computer.  Under this control, analog
     signals such as temperature, pressure  and  opacity are multiplexed into
     local memory buffers to be accessed subsequently by the Host Computer via
     the serial interface bus.  Contact closures are monitored and controlled
     by the microprocessor controller, and  status information is formatted to
     be evaluated as required by the Host Computer.   In the event of a power
     failure, the battery back-up mode of operation is invoked and the I/O
     Controller will continue to function.

IV.  Fault Detection and Annunciation

          In the event of a contact closure or  loss of signal, the fault is
     displayed on the local status panel of the I/O Controller. Additionally,
     an error message is formatted and sent to  the Host Computer for further
     action.  The present configuration is  comprised of a three-wire serial
     bus.  Four I/O Controllers can be daisy-chained on the bus (700 I/O lines).

V.   Host Computer

          The Host Computer functions as the master control center, sending
     and gathering information to and from the  various remote control units
     distributed over the several serial bus pathways.  The Host Computer ope-
     rates as a single job monitor in a real time operating environment.  Tied
     to the Host Computer is a special peripheral serial bus containing the CRT
     display, input keyboard, data logging typewriter, and a dual cartridge


                                     234
-------
    tape storage unit.  Specific  equipment provided at the plant central room is:

    a.   Standard ASCII Keyboard  used as  the medium  for operational interface
         with the control  system.  At this  keyboard, the operator can initiate
         start-up or shutdown operations for all the T/Rs  and rapper controllers
         as well as the I/O Controller.  In addition, the operator  can request
         printouts or  invoke high level system algorithms  for power  consump-
         tion minimization or rapping control.

    b.   A CRT for the display  of all T/Rand rapper  status information. Also,
         alarm messages relative  to  the various precipitator controls are
         displayed.

    c.   A data-logging typewriter providing an hourly hard copy  summary of
         T/R primary and  secondary currents and voltages.  In addition, all
         alarms displayed  on the  CRT are  printed  out.

    d.   Cartridge Tape Storage

         A dual Cartridge  Tape  unit  provides the  following:

         1.  Unit #0 contains the Master  Precipitator Control Program.

         2.  Unit #1 contains a Scratch Tape to be used  as the medium to
             collect system status and faults on  an  hourly basis.

         Upon  initial  system start-up or in  the event of a power failure, an
    automatic  bootstrap loader  program begins to  read the  program contained
    on  the Master Tape Cartridge  into the Host Computer  Memory.   The Master
    Program has a self-start feature and,  through  prompting, the operator can
    enter, via the keyboard, precipitator parameters.

                        ADAPTIVE  CONTROL  SOFTWARE

    As  stated  previously,  the operator can  defer  rapping and TR control to
the Intelligent Precipitator.  The Intelligent Precipitator has two adaptive
feedback control loops  which control  overall TR power levels and system rap-
ping parameters.  This  level of  control adds another  dimension to  the capa-
bilities and benefits of Intelligent  Precipitator  control.

I.  Power Consumption  Minimization

         ESPs, although relatively efficient in terms of power consumption
    when compared to scrubbers, fabric filters or mechanical collectors, do
    expend power in excess of that minimally required to solely charge and
    collect particles. These losses are  primarily due to electrical  currents
    between collector  and  emitter which are not directly involved in the
    precipitation process. Such  losses are sensitive to  precipitator mechani-
    cal design (i.e.,  electrode design, plate spacing,  etc.) and  powering
    strategy  (i.e., high  electrical  sectionalization).   For ESPs  which are
    serving a  base loaded  process, the possible reduction  in TR power will


                                     235
-------
generally be small and proportional to preclpitator design margin for
statistical variability in coal and ash composition, uneven particle size
distribution, grinding mill variability, and temperature variations which
occur between regenerator cleanings.  In this case, infrequent manual
adjustment of the TR controls will probably be satisfactory to obtain the
major portion of potential power savings.   Power is reduced so as to
maintain the opacity within either regulatory opacity limits, or to
limits correlated with specific mass loadings.

     For cyclic operation, the increase in specific collection area (SCA)
corresponding to load reduction provides the greatest potential for power
savings with microprocessor control.  Generally, uncontrolled power in-
creases during boiler load reductions due to lowered flue gas tempera-
ture and corresponding flyash resistivity reduction.

     The magnitude of power savings possible is extremely site specific
so that it is misleading to quote general numbers.  However, power distri-
bution within a precipitator can have a significant effect upon the effi-
ciency obtainable for a given power consumption, or the converse problem
- power consumption required to obtain a given efficiency.

     This effect is demonstrated by test data shown in Table 1 obtained
on a Buell rigid frame precipitator. Even with only four fields in depth,
it was possible to obtain performance improvement by modifying the distri-
bution of power to the fields.  For example, in Test A, the power was
controlled at  threshold of sparking by automatic voltage control.   An
efficiency of  99-73% was obtained with 34.9 KW corona power.  When the
power distribution from front to back was changed in Test B, the effi-
ciency increased to 99.85% at an overall power consumption of 29.56 KW—
a net increase in performance with a decrease in overall power consumption.
This type of power distribution is consistently and automatically sought
by the Intelligent Precipitator controls.
Table 1
Test A
1A
2A
3A
Test B
IB
2B
3B
- Effect of Improved Power
Volume
(ACFM)
138,278
143,507
144,968
137,741
140,627
144,279
Distribution
Cabinet Power
ABC
2.12
2.55
2.34
Mean
5.60
5.00
3.91
Mean
8
9
13
.84
.00
.20
Total KW
9-72
10.08
10.26
9
7
12
6
6
6
Total KW =
.35
.70
.00
34
.24
.24
.36
29
(KW)
12
13
11
D
.42
.75
.44
-9 KW
8.48
8.48
8.32
.56

Precipitator Efficiency
%
99
99
99
Mean 99
99
99
99
Mean 99
.76
.78
.72
.73
.88
.84
.84
.85
      In  this particular case, power was reduced to electrical fields which
were  operating  in current limited mode, so that excessive current  could
be trimmed without a significant reduction in collection field  strength.
                                236
-------
    In some instances, it will be possible to achieve relative decreases in
    power consumption which exceed the relative load reduction.  The control
    loop capable of performing this function is  shown  in Figure 6b. Opera-
    tionally, there are various levels of complexity that could be chosen at
    the customer's preference.  Although the control loop in Figure 6b is
    relatively complex, this configuration  offers maximum flexibility and
    potential for power savings.  The key control parameters are suitable
    integrated opacity level and individual TR secondary voltages V and
    currents I.  Additionally, the Intelligent control  periodically perturbs
    the TR currents to determine the local slopes of each of the 'V-I curves.
    This latter feature of adaptive control—to specifically and continually
    test to determine where operation is and where it  should be to maximize
    collection efficiency for a given power consumption—is where the Intel-
    ligent Precipitator earns its name.

         Changes in overall power level are  made  so as  to maintain the outlet
    opacity within the limits of 0LL and 0LU (Figure 6a).  The band upper
    opacity limit 0LU is chosen to fall below the regulated opacity limit 0R
    by a margin e^.  This provides suitable reaction time when opacity
    exceeds 0LU.  Total TR power is changed  in discrete increments—positive
    when the opacity signal exceeds the upper limit and negative when the
    opacity falls below the lower limit.  The control then applies this power
    alteration to the TR which will yield the largest  gain in overall effi-
    ciency for a power increase, and the smallest decrease in overall effi-
    ciency if a power reduction is called for.

         Following the change in operating  voltage, a  pause period T^ is
    required prior to acquisition of the next opacity level. This allows the
    effects of any prior changes made by the control system to take hold.
    The  integration period is T£ and should either T^  or  1^ ^>e to° short,
    unstable operation will result.  If too long, significant drift of the
    signal may occur between corrections.

         Major activity in power disposition is  likely to be concentrated in
    the  outlet fields of the precipitator;  of course,  a major power reduc-
    tion, if optimally performed, will generally involve simultaneous power
    reduction over several fields.

         Finally, the Intelligent Precipitator is capable of discerning back
    corona and will automatically adjust powering to TRs in this condition so
    as to be operating at the maximum secondary  voltage achievable.

II.  Rapper Control

         Rapping parameters are adjusted based on secondary voltage charac-
    teristics.  Rapper adjustment routines  are invoked by the operator (after
    a change in coal, for example) or by the main routine periodically, or in
    response to a detected condition  (such  as excessive spark rate, back
    corona, or low corona current).

         The rapping adjustment routines measure and log corona onset voltage


                                    237
-------
     and voltage corresponding  to a given spark rate.  Corona onset voltage is
     functionally defined as that point where secondary TR current exceeds a
     specified threshold value.

          Measured corona onset  voltage is compared to a normal operating
     range.   If the measured value falls outside this  range,  emitter rapping
     is increased.  Similarly, measured sparking voltages are compared to the
     normal  ranges.  If the measured value falls outside this range, rapping
     to the  associated plates is increased or decreased, depending on whether
     the voltage is below or above this range.

          If any rapping change  is made, the parameter in question (corona
     onset or sparking voltage)  is measured ajain after a specific number of
     rapping cycles to monitor  response. This cycle may repeat several times
     until either deterioration  ceases, or rapping frequency, or intensity reach
     maximum allowable values.  At this point, the operator is  notified.  This
     situation could arise from  a change in ash or flue gas characteristics,
     invalidating the "normal" values of sparking and  corona  onset voltages.
     These values should, in this case, be updated by  the operator.

          Another routine is available to automatically control the precipita-
     tor during boiler startup.   This routine energizes only  portions of the
     precipitator as needed, starting from the precipitator back end, based
     upon opacity feedback.  Rapping intensity and frequency  are held at their
     maximum allowable value throughout startup while  operating temperatures
     are low.  This routine reverts to normal operating mode  when the flue gas
     is above the dewpoint range.

          Optionally, the Intelligent Precipitator can be configured to respond
     to unusual or dangerous operating conditions.  For example, boiler flame
     and combustible detectors can be monitored and used as inputs to initiate
     rapid precipitator shut-down.   Parameters, such as coal  feed rate and
     boiler  load can be input to the Intelligent Precipitator  to take advantage
     of its  data logging capability or can also be used as feed-forward sig-
     nals to the power control  algorithms.

                          SUMMARY AND CONCLUSIONS

     It has  been seen that preclpitators constitute a very fertile ground for
minicomputer   control.  The advantages which will accrue from the adoption of
such control include simplified  high voltage and rapper control wiring, more
convenient operator control, improved and more complete operator awareness of
precipitator operating trends and higher control system and precipitator
reliability.  Adaptive feedback control will reduce precipitator corona power
close to the minimum required for given operating conditions,  and will control
rapping to optimum values consistent with low emissions and maximum component
mechanical life.  Intelligent Precipitator control will result in reduced
operating costs and higher reliability.
                                     238
-------
Figure  1.    Intelligent Precipitator CRT and Chassis
I'KU.ici IAIUK CuLLtcTTSG ANU EnTTTtNG
RAPPER MATRIX CONFIGURATION
R9
R8
R7
R6
R5
R4
R3
R2
Rl
RO

NOTE:
144
123
112
96
30
54
43
32
16
0
CO
1)
(145)
(129)
(113)
( 97)
( 81)
( 65)
( 49)
( 33)
( 17)
( 1)
Cl
THE BRACKETED
146
130
114
98
32
56
50
34
13
2
C2
RAPPERS
INTENSITIES CAN '(EVER
147
131
US
99
33
67
51
J5
19
3
C3
ARE ^H
EXCEED
(148)
(132)
(116)
(100)
( 84)
( 63)
( 52)
( 36)
( 20)
( 4)
C4
I EMITTING
149
133
117
101
35
69
53
37
21
5
C5
rvpc _
5 FOOT-POUNDS.
              Figure 2.  CRT Rapper Configuration Display
RAPPEK PARAMEmR SUMMARY

ROW
0





NOTE



NO. RAPPER TO.
000
(001)
302
203
(394)
335
: THE ( ) ABOVE DENOT
"LIFT" VALUE GREATER
5 FT-L3S AT OUTPUT T
OPERATING
EVERY
12 MIN.
18 MIN.
13 MIN.
13 MIN.
13 MIN.
SM MIN.
ES "HE EMIT
^y.M 5 "-
IME.
-ibl KIG
LIFT
FT-L3S
5
20
22
15
15
23
TING RAPPERS
-BS WILL SE



RAPS
i
i
1

1
i
. ANY
SET T3

     Figure 3.  CRT Display Following Change of an Individual Raoper Time
                                   239
-------
               PRECIPITATOR  SUMMARY  DATA

               «ED  02-18-81 TIME 10:49:00
                                           UNIT MO. 1
                TR *1A   056 AMPS    270 VAC   213 MA
                TR «8   090 AMPS    244 VAC   313 MA
                TR #1C   076 AMPS    280 VAC   212 MA
                TR #10   081 AMPS    285 VAC   640 MA
           054 XV      16' MIN
           050 KY      14' MIN
           043 KY       3' MIN
           047 KV      10' MIN
                TR *2A  048 AMPS    327 VAC
                TR «8  070 AMPS    228 VAC
                TR i*2C  071 AMPS    232 VAC
                TR #20  090 AMPS    243 VAC
UNIT NO. 2

 306 MA
 306 MA
 540 MA
 500 MA
050 KV
050 KV
049 KV
046 KY
 5' MIN
 7' MIN
10' MIN
 7' MIN
                        PRIMARY     PRIMARY   SECONDARY   SECONDARY    SPARK
                        CURRENT     VOLTAGE   CURRENT     VOLTAGE     RATE      STATUS
               Figure 4.   CRT  Display  - Electrical  Operating Condition  Summary
c
    Figure 5.    Real-Time  Intelligent  Precipitator  Control Arrangement
CONfACT
5TMS1M6

rOMMCT
comix
           AMOOtCVIIAl tMFUT/OUTPUf COMTnOL
           EXPW-I^IOM CAPAfllLITIES
                      (STATUS H	'
                                                     AVC CONTROL CEMTtn
                                                   I	1
                                                   1
                                                                     T/d Hi
                                                 240
-------
     OPACITY
        0
                          TIME (HOU3S)      .

              6a.   Definition  of Opacity Deadband  Limits
      I SECONDARY 70LTAGS
      i AND CtTRRENT
      FEEDBACK:

     GAS FLOW
•
(
1 ; |V(,.V'-U'
I i
PRECIPIIATOS
i
FIELDS
1
                     6b.   Feedback  Control Loop


Figure 6.   Power Consumption Control Loop
                                  241
-------
                        AN ENERGY MANAGEMENT SYSTEM
                      FOR ELECTROSTATIC  PRECIPITATORS

                By: Robert R. Crynack and Martin  P.  Downey
                          Wheelabrator-Frye  Inc.
                      Air Pollution Control  Division
                             600 Grant Street
                           Pittsburgh, PA 15219


                                 ABSTRACT

     As the emphasis of air pollution control regulation  shifts  toward
maintaining daily compliance with stringent  standards,  the user  must
direct more attention to maintaining optimum performance  from his  control
equipment.  Modern electrostatic preicipitators have become  increasingly
larger with more power supplies and electrical auxiliaries which demand
monitoring by operating personnel to ensure  code compliance.  An Energy
Management System is proposed which can monitor precipitator electrical
equipment and optimize energy consumption while maintaining continuous
compliance. The Energy Management System is  a microprocessor-based
device which is connected to the power supply controls  and auxiliaries
and regulates electrical power in response to actual  on-line precipitator
operational needs.  A prototype Energy Management System,  which  controls
the corona power, has been operating for over one year with a high
degree of success on a precipitator serving  a utility boiler.


                               INTRODUCTION

     In recent years industry has been forced to purchase more conserva-
tively designed air pollution control equipment to assure  daily  compliance
with stringent standards.  In the area of electrostatic precipitators  on
large scale industrial processes, users have found it necessary  to
specify and purchase larger units, often with redundant collecting area
and a high degree of electrical sectionalization.   This has resulted in
higher maintenance and operating costs of the equipment.

     Good engineering design practice demands a reasonable amount of
conservativism for any equipment that must withstand the rigors  of a
heavy industrial environment.  This can be illustrated in the case of  an
electrostatic precipitator serving a utility coal-fired boiler.  Modern
power plants are generally designed for a 20 to 30 year life, during
which time the coal supply may come from a variety of sources that
demand a range of design tolerances.   When dealing with a range  of coal
characteristics, a boiler designer must size his furnace for the lowest
heating value, highest moisture and highest ash content. This in turn
presents the worst case sizing parameter for the precipitator because  of
the resulting largest flue gas volume and highest  dust loading.   The
precipitator and the boiler must be sized to handle the worst conditions
that can be anticipated over their projected design life.  Yet during
actual operation more favorable conditions may be  encountered.  For instance,

                                    242
-------
fuels may be used that produce  a more  collectable ash or the boiler may
not operate at its maximum continuous  rating  (MCR).

                         ENERGY MANAGEMENT CONCEPT

     The concept of the Energy  Management System (EMS) is to provide a
means of monitoring and controlling all sub-systems  of the electrostatic
precipitator.  The EMS will ultimately provide the user with operational
and equipment status  information to allow him to operate and maintain
good efficiency of the precipitator.   The EMS provides a single source
of precipitator control rather  than individual controls operating
independently.  More  effective  utilization of corona power is possible.
At the same time, power consumption can be reduced while maintaining
emissions below compliance levels.   The EMS can automatically respond to
changes in process conditions and  increase precipitator power when
operating conditions  demand  it.

     The transformer/rectifier  (T/R)  sets and controls which supply
corona power to the precipitator,  consume the most significant portion
of power. The primary function  of  the  BIS will be to monitor and control
corona power and still meet mass emission requirements.  The EMS obtains
an input signal from  a tranmissometer  (opacity meter) and adjusts the
precipitator power to maintain  a preset opacity level.  The opacity,
used as a measure of  the mass emission rate,  can be  maintained at a
specified level through a variety  of operating conditions.

                                CORONA POWER

     As defined by the Deutsch-Anderson equation, for a given migration
velocity the size of  the precipitator increases with increasing efficien-
cies and gas volumes. The relationship between corona power and collection
has been explored in  theory  and supported with field data by White (1).
The corona power density, expressed in watts  per square foot of collecting
area, has good correlation with efficiency up to 98  percent.  For higher
efficiencies, White's data has  shown that a higher rate of corona power
is required for increasing efficiency.  For these efficiencies corona
power was examined in watts per thousand ACFM, referred to as specific
corona power.  Figure 1 shows the  relationship of efficiency versus
specific corona power, where both  White's curve and  WFI data for one
specific installation are plotted.   The significance of this data is
that the rate of increase in efficiency diminishes with increasing
specific corona power.  A similar  trend can be seen  in the relationship
between outlet residual and  specific corona power for the same installation
as shown in Figure 2.

     Correlation between corona power and collecting area can be derived
mathematically by examining  the relationships among  corona power density,
specific corona power, and specific collection area  (SCA).  In high
resistivity applications where  corona power density  is low, a large SCA
is required to obtain the specific corona power necessary to obtain a
given efficiency.  For low resistivity applications  where corona power
densities are relatively high,  a small SCA is required to obtain the

                                     243
-------
same specific corona power and efficiency.

     For a given application, a specific corona power can be selected to
achieve a specified efficiency.  However, good practice demands providing
power supplies that are capable of delivering higher power densities
than are required.  When sizing a T/R, consideration must be given to
variations in operating conditions and ash resistivity.  At the same
time, present precipitator control logic centers around the concept of
seeking the highest level of corona power regardless of dust concentra-
tion or gas volume.  Standard precipitator controls automatically seek
the highest spark level or maximum rated current of the T/R.

     When all of these worst case conditions are present, all of these
sizing precautions are necessary.  However, if process conditions change
and operation becomes more favorable for precipitation, power in a
precipitator will tend to increase.  Process conditions such as a
reduction in boiler load and favorable particulate chemistry will create
this situation.  This increased corona power comes at a time when less
corona power is required to maintain emission levels.

                      PRECIPITATOR POWER REQUIREMENTS

     The design and electrical load requirements for a precipitator on a
coal-fired boiler unit will be examined. Table I shows a summary of
typical design parameters for a cold side ESP to serve a large size
utility boiler.  Present code requirements for particulate mandate a
maximum outlet of  0.03 Ibs/MBTU, which typically corresponds to an
outlet residual of 0.008 gr/ACF and an efficiency of 99.8% for a Western
sub-bituminous coal.  Redundancy in the collecting area is often
specified to allow for the loss of electrical bus sections due to
potential failure of the material handling equipment, high voltage
electrodes and electrical energization equipment.

     Table II shows an electrical load summary for a precipitator
system. No consideration has been given to power consumption of the I.D.
fans or of the ash removal equipment.  A significant point from the
summary is that there is a large number of subsystems which will require
service and maintenance.  In addition, with larger precipitators having
a  large number of T/R units, it becomes very difficult to understand the
operational status of the unit at any one time.  For instance, a plant
operator taking T/R control cabinet readings has a difficult time
interpreting any more than if the power is up or down in the precipitator
and has no concept of whether this indicates higher or lower particulate
emissions.  In the current state of the art design, all of these sub-
systems operate in a open loop mode, operating independently with no
central monitor or control.

     Regarding the electrical load summary in Table II, the connected
load tabulates power required at the distribution transformer, while the
expected power summarizes the operating power consumption that could be
anticipated at MCR.  Note that the T/R's and controls consume approx-
imately 791 of the total normal operating load.  A value of 500 watts/1000

                                     244
-------
ACEM was selected as a reasonable  specific  corona power required to meet
the outlet emissions.  This  calculates  to a corona power of 1500 KW for
the volume given in this example.   Then the corona power was converted
to line power using a factor of  1.25, which approximates the power
losses in the T/R unit, linear reactor  and  the thyristor package.   This
yields the final value of  1875 KW,  which represents the line power
required from the plant distribution transformers.

             OPACITY AND PARTICLE  CONCENTRATION RELATIONSHIP

     The opacity signal from a transmissometer is the key input  to the
EMS. According to Bouguer's  Law, the optical density is proportional to
the particle concentration in the  gas.   The constant of proportionality
differs with density, size and optical  properties of the particulate.
The constants of proportionality can be calculated if all parameters are
known, but these parameters  are  difficult to establish.   It is more
practical to determine the proportionality  constants experimentally.
Through a series of tests, the relationship between optical density and
particulate concentration  can be established along with its confidence
limits.

     A transmissometer measures  the ratio of the amount of light
transmitted through the gas  stream (I)  to the initial amount of  light
 (I0).  The transmittance  (T) is  defined as  the ratio of I/IO-  Opacity
is then defined as:
               Opacity = 1 - T = 1 - (I/I0)
The relationship between optical density and opacity is given as:
          Optical Density  (OD) = Iog10  (^-^-L^  = log10 (1/T).

     Although a transmissometer  measures transmittance  and thus  opacity,
the relationship between opacity and optical density is defined. Many
transmissometers will provide a  readout of  both opacity and optical
density.  Thus, particle concentration  can  be correlated with either
optical density or opacity.  This  relationship has been experimentally
shown in a variety of applications (2,3).

                               THE EQUIPMENT

     The EMS hardware can  take on  any level of sophistication, depending
on the number of inputs, complexity of  operation, and number of  outputs.
The equipment described here is  a  developmental prototype system used  to
test the applicability and capability of such a system  under actual
operating field conditions.  This  system monitors and controls the
corona power which is the  most significant  portion of the precipitator
electrical load.

     The EMS is structured around  a programmable controller and  an
interface panel.  The function of  the programmable controller is to
continuously monitor all incoming  signals (inputs), perform basic
operations based on these  inputs,  and provide outgoing  signals  (outputs)
on the basis of the programmed logic.   Using relay format (ladder
diagram) programming, this microprocessor based device  performs  the

                                     245
-------
logic, timing, counting, and simple arithmetic functions needed to
perform the desired control algorithm. The interface panel contains
hardware to (a) monitor and display opacity,  (b) provide system inputs
to modify EMS operation, (c) indicate power reduction levels  and (d)
digital to analog (D/A) converters to provide signals to regulate  the
power in individual analog T/R controls.

     Although the EMS program can be modified using a programming  panel,
the interface panel provides for changing three inputs to the EMS.   In
this prototype thumbwheel switches are used for this purpose.   These
three control inputs are (a) opacity setpoint, (b) response period and
(c) priority level.  The opacity setpoint can be adjusted to  the desired
value using a two-digit, 10 segment thumbwheel switch. The opacity
feedback signal from the transmissometer is compared to the setpoint  and
power is adjusted so as to force the two values to be equal.  The
response period is the parameter that controls the rate at which the  EMS
will react to an opacity error.  The rate at which power is changed
depends upon the opacity error as well as the setting of the  response
period.  Each controllable precipitator section is assigned a priority
level from 0 to 9, selectable using a single-digit thumbwheel switch.
This level assignment determines the order in which power is  increased
or decreased in the various precipitator sections.

     The EMS interface panel has a series of seven segment readouts to
indicate the corona current level of the individual precipitator sections.
A "9" indicates a 90% reduction while a zero indicates no reduction
(full power).

                               THE OPERATION

     The EMS samples the opacity signal and updates the stored  opacity
value using a digital filter.  The opacity value is compared  to the
preset opacity setpoint input from the interface panel.  The  EMS then
adjusts the precipitator power upward or downward to maintain the
opacity at the setpoint.  The EMS responds differently depending on
whether the opacity exceeds the setpoint or is less than the  setpoint.
The speed of response is determined by the response period input from
the interface panel and the magnitude of the opacity error  (difference
between actual opacity and setpoint opacity).

     Upon initialization of the system, if the opacity is below the  set
point opacity, the power is first reduced in 10%  (corona current)  steps
with "0" priority controls until the minimum power level is reached.
The EMS will then begin to reduce the power on priority level "1"
controls.  This will continue until all priority levels are completed or
until the opacity equals the set point.  If actual opacity is above  the
set point, the EMS will increase precipitator power in 10% increments.
The power will be increased in successively lower priority controls
until maximum power is reached or until the opacity setpoint  is reached.
The rate at which power is adjusted is a function of  (1) selected
response time,  (2) opacity error and  (3) opacity being above  or below
the set point.  The level of power reduction  is displayed on  the single

                                    246
-------
digit readout for each precipitator  control  panel.

                                THE TEST SITE

     A prototype Energy Management System for corona power was  installed
on a 156 MW coal-fired boiler  in November 1979.   The rated gas  volume is
751,000 ACFM at a temperature  of 300°F at the precipitator inlet.   The
design SCA is 716 and plate  spacing  is 10 inches.   This  is a two chamber,
six field precipitator.  Each  field  of each  chamber is powered  by one
T/R set rated at 45 kv and 2500 ma.

     The coal burned at this station is referred  to as Decker Coal,
mined in southern Montana.  The sulfur content of the coal is approx-
imately 0.6%.  When these units were sized in 1974,  it was thought that
this would be a very difficult ash to collect because of the low sulfur
content of the coal.  At that  time,  the effect of sodium in the coal was
not known.  The sodium in the  ash (reported  as sodium oxide)  is about
8.71. It has been established  that ash from  low-sulfur coal is  not
difficult to collect if the  sodium content is relatively high.   The
performance of these precipitators substantiates  this fact.

     In May 1978, a series of  reduced power  tests were conducted on
these precipitators.  Under normal operating conditions,  all fields
except the first field operate at full rated (2500  ma) corona current.
The inlet field generally operates in a sparking  mode at less than 1000
ma. Although precipitator voltage (KV) meters were  not supplied with
these units, measurements showed that the precipitator voltage  is  about
40-47 kv at full-rated corona  current.  It was determined that  precip-
itator performance remained  approximately the same  whether 6, 5 or 4
fields were in operation at  full corona power. The outlet emission
level was less than 2.3 mg/cu.m. (0.001 gr./cu.ft.).  Reducing  the
corona power to 75% and then to 50%  of full-rated corona current with
four field operation showed  almost no change in outlet particulate
concentration or in the opacity level of 0 to 1%.   It was when  only two
fields were energized at a corona power of 25% of rated  current value
that the mass emission level increased by a  factor  of 10 and the opacity
jumped above 10%.  This is the data  plotted  in Figures 1 and 2.  These
results indicate that an EMS would provide a substantial savings in
energy with no significant impact on emissions.

                           OPERATING EXPERIENCE

     The prototype EMS was installed and commissioned during the week  of
November 5, 1979.  The EMS has worked effectively in over sixteen  months
of operation.  The EMS automatically increases and  decreases corona
power in response to changes in opacity,  and thus mass emission.  The  EMS
has automatically responded to boiler upsets,  boiler load changes, and
electrical failure of the enegization equipment.  Little maintenance has
been required for the EMS, but periodic cleaning  of the  optics  of the
transmissometer is required to provide an accurate  opacity measurement.

     Before installation of the EMS,  plant personnel manually reduced

                                     247
-------
corona current to about 50% of rated values in order to conserve  on
power.  The installation of the EMS resulted in significantly more power
savings. Furthermore, the automatic response of the EMS eliminated the
need for operating personnel to monitor and adjust the power when
opacity levels began to increase.  The EMS is able to reduce the  corona
current in the entire precipitator to 10% of the full rated value and
maintain that level of power with opacity increasing to only 5%.  There
is a substantial energy savings at this installation.  One must bear in
mind that this Decker coal, together with the large SCA precipitator,
provide an extreme case but an excellent test facility.

                               COST ANALYSIS

     Based on statistical information published by the Edison Electric
Institute (4), the energy consumption costs of a precipitator can be
calculated. In 1979, the average cost of fuel for a coal-fired boiler
was approximately 1.33 cents per net kilowatt-hour (KWH). Based on
average operating expenses for investor-owned electric utilities, the
cost of generation (including fuel), was about three cents per KWH.
Bear in mind these numbers are for 1979 and the 1981 figures would be
higher.

     The amount of power consumption savings and cost per KWH are site
specific. Using the above costs of energy, a 10% reduction in the corona
power for one year would result in a savings of over $21,000 per  year  in
fuel costs alone and a savings of over $49,000 in production costs.
Based on an installed cost of approximately $100,000 for this EMS
equipment on a typical 750 MW boiler, the payback period would be about
five years if fuel costs are considered and only 2 years if production
costs are considered.  These figures are based on the average expected
power consumption of 1875 KW given in the example in Table II.  These
figures do not consider the advantages of central monitoring and  control
of the corona power that result in simplified operation and maintenance
of the electrostatic precipitator.

                                  SUMMARY

     Conservatively designed precipitators are necessary to maintain
daily compliance with stringent air pollution control codes.  An  Energy
Management System (EMS) is proposed to effectively monitor and control
corona power.  Over sixteen months of field operation of a prototype
EMS has verified its capability and reliability.  This microprocessor-
based device monitors opacity and automatically regulates corona  power
to maintain a maximum opacity level, and thus, a maximum emission level.
The relationship between power consumption and precipitator performance
and between opacity and particulate emissions are discussed.  Reduced
power tests were performed on a 156 MW coal-fired boiler which
demonstrated the need for a central management and power reduction
controller.

     While the EMS is monitoring and automatically responding to  varia-
tions in operating conditions, power consumption savings can be

                                     248
-------
significant on installations where worst case design limits are not
reached.  Although the amount of power savings depends on many factors, a
ten percent reduction in power consumption would offset the cost of the
EMS in a few years.  Although this paper is directed toward utility
applications where precipitators are larger and more conservatively
sized, the EMS is applicable to industrial applications as well.   In the
future the EMS will be expanded to monitor and control most of the
subsystems and auxiliaries of an electrostatic precipitator.

                                 REFERENCES

1.   White, Harry J.  Electrostatic Precipitation  of Fly Ash.
     Pittsburgh, Air Pollution Control Association Reprint Series,
     1977. p. 22-23.

2.   Beutner, Heinz P.   Smoke Density Measurement  With an On-Stack
     Transmissometer.  APCA Journal.  24:865-871,  September, 1974.

3.   Brennan, R. J., et  al., Review of Concurrent  Mass Emission and
     Opacity Measurements  for Coal-burning Utility and Industrial
     Boilers.  EPA Report  No. 600/7-80-062.  March, 1980.

4.   Statistical Year Book of the  Electric Utility Industry  -  1979.
     Washington, B.C. Edison Electric  Institute.   November,  1980.
                                      249
-------
  9995
g  99.9
UJ
5
   998-
m
UJ  99.5
U
oc
>-
o
UJ
U
U.
UJ
99

98


95

90
         FIGURE: 1    RELATIONSHIP OF SPECIFIC CORONA POWER
                                    AND EFFICIENCY
              200
                  400      600      800      1000     1200      1400
                      SPECIFIC CORONA POWER (WATTS/1000 ACFM)
                                                                        1600
                                                                                1800
           FIGURE: 2
     0.02

     0.01
 §  0.005
 Q  0.002
 55
 UJ
 tt  0.001

 £ 0.0005-
 o

   0.0002 •

   0.0001
                    RELATIONSHIP OF SPECIFIC CORONA POWER
                              AND OUTLET RESIDUAL
               200     400      600      800     1000     1200     1400

                             SPECIFIC CORONA POWER (WATTS/ACFM)
                                                                  1600
                                                                          1800
                                        250
-------
                TABLE  I  -  SPECIFIED DESIGN PARAMETER SUMMARY
Station Nominal Generating Capacity  -
Type of Fuel -

Air Pollution Code  Requirements for Particulate
    Maximum Allowable Particulate Emission
    Maximum Allowable Opacity
Maximum Flue Gas  Volume at Air Heater Outlet
Precipitator  Inlet  Grain Loading
Outlet Residual  (Equiv. to 0.03 Ibs/MBTU)
Design Efficiency
Design Specific  Collection Area (SCA)
Collecting Area  Redundancy
Total  Installed  SCA Required
750 MW
Western Sub-bituminous Coal
(Low Sulfur - Low Sodium)
0.03 Ibs./MBTU
20%
3,000,000 ACFM
4.0 gr/ACF
0.008 gr/ACF
99.8%
650 ft.2/!000 ACFM
10%
715 ft.2/1000 ACFM
                    TABLE II - ELECTROSTATIC PRECIPITATOR LOAD SUMMARY
Subsystem
1)
2)
3)
4)
5)
6)
n
8)
9)
10)
m
T/R's and Controls
Hopper Heating Units
Rapper Motors
Insulator Heaters
Insulator Purge Sys.
Insolation & Louvered
Dampers
Hopper Vibrators
Hopper Level Indicators
Control Room - HVAC
Roof Weather Enclosure
Ventilating Equip.
Area Liahtinq
No. of
Units
96
112
114
448
8
12
112
112
2
8

Connected Power
(KW)
4750
730
65
358
484
149
19
5
48
16
30
Expected Power
At MCR (KW)
1875
365
54
0
0
0
3
5
48
16
15
                                       TOTAL
                                               6654
                                                               2381
                                           251
-------
          RELATIONSHIP BETWEEN ELECTROSTATIC PRECIPITATOR PERFORMANCE
                          AND RECORDKEEPING PRACTICES

                            By:  S. P. Schliesser
                           PEDCo Environmental, Inc.
                          505 S. Duke St., Suite 503
                              Durham, N.C.  27701

                                   ABSTRACT

     This paper  discusses  the sensitive relationship between the performance
of  electrostatic precipitators (ESPs) to the  quality  of recordkeeping, spe-
cific to  the  aaalysis of voltage current  (V-I)  curves and corona power-mass
emission  correlations.   The  author  contends that   improved  recordkeeping
practices  will  mutually  benefit both  the  operator   and  regulatory agency.
Strategic  recordkeeping will permit  the operator  to  effectively  diagnose
restrictive performance problems, enabling corrective changes to be made on a
more timely and  regular basis.  For  regulatory agencies,  these records pro-
vide  a  means  other  than opacity  to  evaluate the  compliance  status  of the
unit.  V-I curves and corona power levels constitute the basis  for evaluating
ESP operation and maintenance (O&M) levels, along with continuous particulate
emission  levels.  A  fundamental characterization of ESPs is presented, limi-
ted  to  using  organized electrical  data as the  guideline for behavior and
performance.
                                 INTRODUCTION

     Qualitative  evidence  recently acquired demonstrates  a  strong relation-
ship between ESP  performance  and recordkeeping practices.(1)  The investiga-
tion  evaluated several  ESPs  by classifying  performance  and  recordkeeping
levels  into  five designated  categories.   A  detailed  inspection and  file
review  determined the performance  and  recordkeeping level  assigned  to  each
source.  Table 1  identifies the  criteria used in this  study to classify the
performance and  recordkeeping levels.   Figure  1 illustrates the  results of
the classification and  shows  the strong relationship between performance and
recordkeeping levels.  Note that the  correlation applies across the range of
recordkeeping/performance  levels  and industrial source categories  studied
(i.e.,  utilities,  portland cement,  arid kraft pulp  industries).   The  conclu-
sions drawn by the author are that a  relationship exists between performance
and recordkeeping  at  most  sources and that this  relationship becomes causal
when  records  are  used  for diagnostic: purposes.  These  conclusions are  sup-
ported by the author's experience with EPA's mobile  ESP.

                            STRATEGIC  RECORDKEEPING

     Strategic ESP recordkeeping  practices  consist  of obtaining, organizing,
and  comparing  transformer-rectifier  (T-R)  set  voltage and  current  levels.
The initial collection of  this electrical  data begins with the  design speci-
fications and the  set of air load V-I  curves conducted during equipment check
out.  The V-I curves  and corona power  levels experienced and collected during
tne initial performance  test  will constitute the technical  base for  initial
                                    252
-------
and  continuous,  compliance  determinations.   Continued  collection  of  this
voltage-current  data during  air load  tests  and regular operation will  con-
tinue to  serve  as  the measure of equipment readiness  and performance.  Thus,
a  complete  ESP  recordkeeping package  consists  of electrical data collected
from all  phases  of equipment design,  initial check-out  and performance test-
ing, and  continued  operation and maintenance periods.   Both V-I curves and
corona power  levels are the  supplemental  means  of  an  ESP evaluation, availa-
ble  for common  use by operators, regulators,  manufacturers,  and  consultants.

                 INSTRUMENTATION  AND KEY PARAMETER RECORDKEEPING

     The  most important ESP parameters  for regular recording are the second-
ary  electrical  levels.   The  secondary  voltage and  current  readings for  each.
transformer-rectifier (T-R)  set  should be  logged at least once per shift, and
spark rate  meter levels should  also  be recorded.   Even though primary elec-
trical meters are  more commonly used,  secondary meters  are  now  being recom-
mended  by vendors,(2)  enlightened users,(3)  and  consultants.(4)  Secondary
metering  is preferred since  it directly measures the electrical levels to the
discharge wires; primary meters monitor the  electrical  levels fed to the T-R
sets,  and obscure  the  diagnosis of  common problems within the  unit.  Other
essential data  sets to be regularly  logged  are  provided in  reference 2, and
include source  operating data,  fuel  analysis, ash  analysis,  electrode clean-
ing  equipment data,  ash removal system data,  internal component inspection.
data, and replacement parts  log.

               RECORDKEEPING - A MEANS OF  MINIMIZING MAINTENANCE

     Two  key engineering  tools  can  be incorporated  into  a  maintenance and
recordkeeping program to  improve  ESP  performance  with minimum  manpower re-
quirements.   These  two  tools are  1)  the voltage-current  curves and 2) the
corona power-emissions  correlation.

     The  key maintenance tool is the V-I  curve.  The characteristic profile
of the V-I  relationship can be used  during  off-line  (air  load)  and on-line
 (gas load)  conditions to understand the behavior of an ESP.   It will  serve as
an "eyeball" to ESP operators by  graphically illustrating  the conditions of
the  internal  wires  and plates.   Interpretation of V-I  curves  does not require
extensive training, and it  will accomplish  the  following:    1) assure proper
operating status,  2) define  high-voltage  power  loss from the T-R set to the
internal  bus bars, 3) indicate  short or open circuiting in  bus  sections, 4)
indicate  effective changes  in discharge wire size, either  wire  shrinkage or
ash  buildup, 5) indicate  alignment  problems between  wires  and plates, 6)
indicate  excessive deposits  on  collection plates,  7)  indicate sharp  edges on
plates, baffles, 8) indicate air inleakage,  9)  indicate ash  buildup  into the
energized region,  and 10) indicate  changes  in  gas  composition,  temperature,
resistivity,  conditioning   agents,  space  charge   effects  and  gas  through-
put. (4,5)

     A V-I  curve can be  obtained in less than 5  minutes, since a  minimum  of  4
voltage-current  data sets is sufficient  to  map a  curve.   The corona  initi-
ation  point  and the critical operating limit  need  to  be determined,  along
with at least two  intermediate points.   The controller needs  to be put  in the

                                     253
-------
manual, and not  the automatic, mode.  The  corona  initiation point  is  deter-
mined by  gradually  energizing the T-R set  until  a detectable secondary  cur-
rent level is achieved.  Generally, corona is initiated in the range of 16  to
22  kV.    Record  the  secondary voltage  and corresponding  secondary current
level.   Continue to  incrementally  energize  the  field  in  increasing power-
levels and record the respective voltage and current levels at a minimum  of 3
more  points.    As  each energization  level becomes  stable,  the  electrical
readings  are  recorded.   Any  of  three  critical   limits  will  determine  the
maximum  corona  point  achievable  for the prevailing  conditions:   1) T-R set
current limit, 2) T-R set voltage limit, and 3) spark limit.  Note and  record
the  limiting  case.   A standardized form with electrical  data formating and a
grid  with voltage  and  current graduations  should be  used  to expedite  this
exercise  (see reference 2).

     Because  of the  complexity  of ESP  units,  a  staged  troubleshooting ap-
proach  is necessary to isolate conditions and to assure  that each element  is
operational.  Air load V-I curves provide the first set of data necessary for
troubleshooting;  an  ESP before it goes on line.  A comparison of air load V-l
curves  taken after an  outage with  those  taken  during  a time  of suitable
credibility  (i.e.,  initial   installation),  one can  assess  the  operational
status  of an  ESP.  Reasonable agreement between the reference and outage air
load  V-I curve  is  evidence that the particular  field  is fully operational.
Lack  of  reasonable  agreement between the curves  is  evidence that  a problem
exists.   The  compilation and comparison of air  load  V-I  curves for each T-R
set  in  the ESP  system provide  evidence as to which fields are ready  and which
need  attention.   Figure  2  presents a  series  of  air load  V-I curves.  Each
shows  reasonable agreement  with the  other with  the  exception of  field  3,
which  shows a secondary power  leakage problem.  With reasonable experience  or
training  one can determine  the  nature of  the  problem  by  interpreting the
deviant  character of the V-I  curve.

     The  air-load procedure  includes:   1) preparing the  ESP  system  for regu-
lar  operation,   2)  providing  a nominal level of  ambient  airflow through the
ESP, 3) placing  the  controller in the manual mode, 4) energizing each T-R set
and  recording voltage  and  corresponding current  levels  over the achievable
range  of  electrical  operating  conditions, 5) graphing the tabulated  secondary
voltage-current  data  sets, and 6) comparing the graphed secondary V-I results
for  each  field  (or  T-R  set) with a) the other fields (under the same environ-
mental  conditions)  and  b)  the V-I  curves  taken  in  the  past (under similar
environmental  conditions).   Elaboration  of  this  sequence  of activities and
their  significance  is contained in reference  1.   It  is  strongly recommended
that  operators  follow the  manufacturer's  procedures  during this  exercise.
Substantial  levels  of  ozone  will be generated  from  corona formation  during
air-load  measurements,  and the ozone needs to  be  safely vented before reen-
tering  the unit.

     Figure 3  depicts the theoretical relationship between air load and gas
load V-I  curves.  The  uppermost curve, labelled "no mass  loading,"  represents
an air load V-I curve  in  that it describes the V-I relationship without the
presence  and effect of  particulate.   The  middle V-I  curve  represents  at
outlet  field  voltage-current  relationship.   The  bottom  curve represents  an
inlet  field V-I  profile.  Note the progressive shift in the V-I  curves
                                     254
-------
resulting from  the  relative amount of particulate flux experienced for these
three conditions.   Particulate  flux is equivalent to  the surface areal rate
of  particulate  and  is  dependent  on  particle  size  distribution   and  gas
throughput levels.   Generally  speaking,  then, the  air  load V-I curves will
resemble gas  load  curves,  but  typically will  be  shifted to the left because
cf the absence of particulate flux.

     One method  of  checking gas load V-I curves is  the same as that used for
air load  curves  (i.e., compare the reference  V-I curve with the one in ques-
tion).  Another method consists of common graphing of the V-I curves for each
ESP  field,  as 'shown  in Figure  A.   This figure shows  a  series  of V-I curves
taken during  normal  operation  of a  cold-side ESP  treating  high sulfur fly
ash.  Note the progressive shift in  the  V-I curves  resulting from the rela-
tive  amount of particulate flux experienced for the three fields.  Also note
the gradual decrease  in the operating voltage  and the gradual increase in the
current density  from inlet to  outlet.  The  slopes and sequences of these V-I
curves provide evidence  that this ESP  is  operating  and performing properly.
Deviations from  these slopes and sequencing are evidence that the respective
field is  experiencing some kind of  trouble.   Again,  one  can interpret the
problem from  the nature of  the  deviant V-I curve.

     This organized'approach to the  use  of  air load and gas load V-I curves
can  serve as  aa important  part of  an O&M plan for  any ESP and will apply to
all manufacturer types and vintages.  Note, however,  that  the V-I curves in
this  report demonstrate  the general  use of  this approach.  The practical use
of  this  approach  requires  obtaining site-specific  and manufacturer-specific
V-I  curves.   Once this V-I approach  is integrated into regular O&M plans, it
can be  used  to prioritize  the  fields warranting attention before an outage.
It  will  provide evidence as to which fields need repair  and which fields are
operationally sound.

                      CORONA POWER - EMISSIONS  CORRELATION

      Now  that voltage and  current  levels are  realized to provide evidence of
performance for each field, a  second engineering  tool,  corona power, can be
used  to  indicate the particulate removal  performance level of the total ESP
system.   Corona  power is  a measure of the  presence and  intensity of the
electrical  energy  effectively  used  in the precipitation process.  Effective
use  of  corona power excludes  secondary  power  leakage  levels  and reverse
corona  conditions,  both definable  from V-I  curves.   Corona  power  is the
product of the average voltage  and  ave;rage current from all  T-R  sets.  Corona
power  density is  related  to  collection  efficiency through the  equation:(6)


                         n  = l-e-°-°6kl(VV)                         (Eq.  1)

where

        n = collection efficiency

       k  = empirical parameter;  for  fly  ash,  typically 0.5  to  0.7


                                     255
-------
     V    = corona voltage, volts (secondary voltage)
      avg
     I    = corona current, amps (secondary current)
       P  = (V
        c     avg   avg
     P /V = corona power density, watts/ 1000 acfm


     For a  multifield precipitator,  P   is the sum  of  the individual corona
power levels  for  each T-R set.  Figure 5 shows actual performance results on
ESP efficiency  through  the  use of corona power density levels from many fly
ash studies.  Emission  levels  have been correlated with ESP collection effi-
ciency to illustrate that corona power density levels are directly indicative
of emission  levels.   Note,  however,  that the numbers  in Figure 5 are purely
illustrative  and  do not  represent all fly ash cases.   Again, site-specific:
and manufacturer-specific data would have to be collected to correlate corona
power  density  levels  with specific  levels  of ESP  fly ash  emissions.   The
corona power  density  relationship wit.h removal efficiency  is  not limited to
only  fly ash collection,  as characteristic relationships have been found for
other  industrial  applications. (7)  These industries include  pulp and paper,
cement, municipal incinerators, and steel source  categories.   As continuous
compliance  and  recordkeeping   practices  are  implemented,  a  corona  power-
emission level  relationship could become  a viable  method  of accounting for
continuous  ESP  performance levels  in  most  industrial  source  categories.

                                  CONCLUSION

     Strategic  recordkeeping   of  voltage and  current  data has  been demon-
strated to  be an effective method to characterize and diagnose ESP behavior.
Voltage-current data collected and organized into V-I curves and corona power
density units are the supplemental tools to characterize ESP O&M and perform-
ance.  Benefits to the operator and regulatory agency can be realized as this;
methodology is practiced and understood.

                              ACKNOWLEDGEMENTS

     The author  needs  to  acknowledge  that  the entirety  of  this  report is
referenced  literature,  merely  compiled  and limited to 10 pages.  The princi-
pal  investigators behind this  material are present or previous  members of
Southern Research Institute  and  Research-Cottrell,  Inc.,  partially repre-
sented by the authors listed in the endnotes.
                                     256
-------
                                  ENDNOTES

1.    Schliesser,  S. P., and  J.  R. Richards.  Development  of  Guideline  Docu-
     ment for State Operating and Maintenance Recordkeeping Programs (draft),
     December 1980.

2.    Bibbo,  P. P..  and  P.  Aa.  Increasing Precipitator Reliability by Proper
     Logging  and  Interpretation  of Operational  Parameters - An  Operator's
     Guide.   In:  Second Symposium on the Transfer and Utilization of Particu-
     late Control Technology, Vol. II,  Electrostatic Precipitators.  EPA-600-
     /9-80-039b,  September 1980.  pp. 219-241.

3.    Raymond, R.  K.  Electrostatic  Precipitators - Electrical  Problems  and
     Solutions.    In:  Second  Symposium on  the  Transfer  and Utilization  of
     Particulate  Control  Technology,  Vol.  II,  Electrostatic  Precipitators.
     EPA-600/9-80-039b, September 1980.  pp. 173-188.

4.    McDonald, J.  P.,  and  A. H. Dean.   A Manual for the Use of Electrostatic
     Precipitation  to Collect Flyash Particles.  EPA-600/8-80-025, May 1980.

5-    Banks,   S.  M. , J.  R.  McDonald, and L. E.  Sparks.   Voltage-Current Data
     from Electrostatic Precipitators  Under  Normal  and Abnormal Conditions.
     In:  Proceedings:   Particulate Collection Problems  Using ESP's  in  the
     Metallurgical  Industry.  EPA-600/2-77-208, October 1977.

6.   White,   H.   J.  Electrostatic   Precipitation  of  Flyash.   APCA  Reprint
     Series, July  1977.  pp.  22, 23.

7.   Oglesby, S.,  and  G.  B.  Nichols.  A Manual of Electrostatic Precipitator
     Technology,  Parts  I  and  II,  PB  196 380 and PB  196  381.   August 1970.
                                     257
-------
        TABLE 1.   CRITERIA FOR RECORDKKEPING AND PERFORMANCE LEVELS
    Level of Recordkeeping

    The  air  pollution control equip-
    ment and fan systems are devoid
    of monitoring instruments.  No
    operating records can be taken.
2.  Air pollution control systems
    have some of the necessary instru-
    ments, however, it is difficult
    to ensure compliance.  These in-
    struments are not always properly
    maintained and some will be ino-
    operative a large fraction of the
    operating hours.  Operating rec-
    ords are not taken.

3.  All the instruments considered
    necessary to evaluate compliance
    are available.  Because these in-
    struments are checked regularly,
    most are operational at any given
    time.  Although operating records
    are maintained routinely, they are
    often incomplete.

4.  All the instruments considered
    necessary to evaluate compliance
    are available.  These instruments
    are checked and calibrated on a
    regular schedule, therefore, most
    are operational and indicates
    correct data at any given time.
    Operating records are complete
    and these are reviewed routinely
    to identify possible compliance
    problems.

5.  Sophisticated instrument systems
    useful for evaluating compliance
    and optimizing performance are
    available.  These instruments
    are checked and calibrated regu-
    larly to ensure quality data.  Op-
    erating records are reviewed reg-
    ularly to ensure compliance and
    to optimize performance.  A diag-
    nostic repair log is maintained
    for each ccn.trcl device.
    Level of Performance

1.   The  air  pollution control device
    is in an  advanced state of deter-
    ioration.    Excursions  above  mass
    and  visible  emission  levels  are
    probably frequent and substantial.

2.   There  are  frequent  excess  emis-
    sion incidents  which  are  largely
    preventable.   Problems  are  not
    corrected   in a   timely  manner.
    Nevertheless, the units are  kept.
    in compliance or  near design lim-
    itations a  majority  (>50%)  of the
    time  by  corrective  maintenance.
    Excess   emission   incidents   are
    relatively   infrequent   and   at
    least  \ are  due  to  "nonprevent-
    able" factors.     Units  remain in
    compliance  or  near  design  limi-
    tations at  least  90%  of the oper-
    ating  hours and  malfunctions  are
    quickly identified.

    Air   pollution   control   devices
    remain  in  compliance  or  near  de-
    sign  limitations   a high  percent-
    age  (>98%)  of  the time.   Malfunc-
    tions  and/or  upsets are generally
    of   the   "nonpreventable"   type
    only.
 5. Operation of   control devices  is;
    optimized  to  levels  well  below
    applicable   mass   and   visisble
    emissions requirements.   Malfunc-
    tions  or upsets  are  very  infre-
    frequent  and  only  due  to  unpre-
    dictable,  "nonpreventable"  inci-
    dents .
                                     258
-------
                         LEVEL OF RECORDKEEPING
    Figure  1.   Relationship  between performance  and
        recordkeeping for  several ESP cases.(1)
                                       35   40   45   50
                       SECONDARY VOLTAGE, kV
Figure  2.   Examples of  air load voltage-current curves
                 for an  ESP off-line. (1)
                            259
-------
                     I     I    I     I     I     I

                 _  INLET MASS LOADING " 9.16 X 1
-------
   99.9
                                                                  0.02
c
OJ
   99  —
   95  —
o

£  90
o
o
   50  —
               100       200         300      400      500
                  SPECIFIC CORONA POWER, watts/1000 acfm
600
     Figure  5.   Efficiency versus specific  corona power density
         extended to high collection  efficiencies, based on
                          field test data.(6)
                                   261
-------
                  AN OPERATION AND MAINTENANCE PROGRAM FOR A
                   PHOSPHATE ROCK ELECTROSTATIC  PRECIPITATOR

                        By:  D. B. Rimberg, Ph.D.
                             North American PEMCO  Inc.
                             P.O. Box 655
                             Bardonia, N.Y. 10954

                                  ABSTRACT

     An extensive investigation was performed on a weighted wire electrostat-
ic precipitator  to improve its performance on opacity.  The application was
on a phosphate rock fertilizer calciner.  The program was  designed  to upgrade
precipitator performance without major investment in new pollution  control
equipment.  An extensive inspection, repair and replacement program was com-
menced to identify and rectify electro/mechanical problems.  Through a samp-
ling and testing program, it was determined that resistivity conditioning
by moisture control was required.  The program resulted in achieving compli-
ance with no capital equipment investment.  Sampled grain  loading were re-
duced by a factor of ten.  A regulated preventive maintenance program was in-
stituted, thereby maintaining continuous mass and more importantly, opacity
compliance status.
                                INTRODUCTION

     The principle purpose of the phosphate fertilizer industry is to convert
insoluble phosphate rock into water soluble phosphorus fertilizers for plant
uptake.  Florida possesses major phosphate rock deposits and accounts for ap-
proximately 78% of the nation's rock fertilizer production, the western
states about 14%, the remaining from Tennessee and North Carolina.

     Typically, phosphate fertilizer production initiates with crushing the
phosphate rock and mixing it with aqueous sulfuric acid to produce phosphoric
acid, insoluble calcium sulfate dihydrate (gypsum) and fluorine compounds.
This reaction is approximated by the equation:
               2H20 + Ca3(P04) + 3H2S04-» 2H3P04 + 3CaSO  + 2H 0

             Water+Phosphate Ore+Sulfuric Acid-»Phosphoric Acid +Gypsum
The phosphoric acid produced is 30-32% P 0 .   This acid is then used to make
various grades of fertilizers.

     The precipitated gypsum is filtered from the acid, sluiced with recycled
pond water and pumped to a gypsum pond.  Fumes from the attact vessel are
usually vented to scrubbers for fluoride removal before being passed to the
atmosphere.

     When utilizing western rock,  a preliminary processing step is required
before rock grinding, namely roasting (calcining).  This process is necessary
to handle undesirable organic matter (2%-8% by weight).  This organic matter
if not removed tends to stabilize  a foam layer on the surface of the acid
reactors.  The presence of the organic material also produces a slimy gypsum

                                     262
-------
filter cake which is slow filtering and difficult  to wash.

     Calcining also supports the decomposition of  carbonates which are pres-
ent in the rock.  Reduction in carbonate concentration reduces the release of
carbon dioxide gas which would otherwise add to  the foam problem.

     Current practices in calcining allow for the  roasting to occur at 1200°F
to 1600 F in a multi-stage fluid roaster.  The fluid bed roasters are norm-
ally equipped with cyclone collectors  to capture particles which are liber-
ated from the fluidized bed and returned to the  process (Figure 1).  However,
extremely small particles pass through the cyclone and require additional
collection equipment such as scrubbers or electrostatic precipitators.  In
the present application, an electrostatic precipitator was employed to col-
lect cyclone throughput.

Problem Background and Troubleshooting Program

     The subject electrostatic precipitator, although in compliance for mass
emissions, could not constantly comply with opacity regulations.  (Intermit-
tent puffing from the stack was prevelant).  Also, the precipitator often
exhibited excessive sparking and reduced operating voltage.  An expensive
automatic voltage control system was  installed and still the problem persis-
ted.  A replacement electrostatic  precipitator system was being considered as
a  solution for an installed cost of $500,000  (1978 dollars).

     Analysis of the problem was subsequently categorized into two areas:
 (1) characterization of the contaminants outlet  of the calciner, aftercooler
and electrostatic precipitator inlet  and outlet:  (2) electrostatic precipi-
tator equipment mechanical/electrical malfunction.

     A multiphase program was implemented to diagnose the various problems.
The phases were divided in  the following manner:
                     Phase  I:    Problem Identification
                     Phase  II:   Implementation
                     Phase  III:  Sampling & Testing
                     Phase  IV:   Preventive Maintenance

Equipment Description

     The design specifications for the electrostatic precipitator employed to
control the calciner emissions are provided in Table I.

     The precipitator was installed in 1966/1967 and subsequently became sub-
ject  to State emission regulations rate and an opacity of 40%  for existing
sources.   (Needless to say,  1980 regulations  are much more  stringent).


Phase I:  Problem Identification

General Observations/Interpretation^

     The pretreatment simple cyclones and aftercooler  cyclone  were  not signi-

                                      263
-------
ficantly reducing particulate loading to the precipitator.  This was  appar-
ent from previously obtained particle size sampling data,  (Table Test G)
which revealed particles approximately 50 percent by weight larger  than 10
microns downstream of the cyclones.  Accordingly, this size particle  should
be easily removed by the existing cyclones.

     Inspection of interconnecting material's handling system revealed that
the product elevator duct allowed ambient air into the system.  Aside from
the duct handling particulate emissions from the conveyor system, the ambient
air with entrained moisture was suspected to promote sporadic condensation
with subsequent contaminant buildup in the fan and precipitator.

     Previous information and onsite observations indicated correlation of
stream moisture and opacity.  A major portion of the moisture entered the
stream through cooling water injection nozzles in the aftercooler (Figure  1) .
The moisture was believed to affect particle resistivity.

     Frequent temperature changes in process were thought to be reaching dew
point in the precipitator.

     Rappers'-Nine of the 17 rappers (vibrators) were inoperative, malfunc-
tioning, or missing.  The complete rapper system contained many inoperable
components.
     Timer-Rappers were not sequenced or timed properly.
     Insulators^-One high voltage insulator was cracked.
     Wire Frame Stabilizer-Loose insulation.
     High Voltage Bus Ducts-Gasketing and several bolts on three bus  duct
hatches missing.
     Insulator Compartment Ventilation System-Contained air leaks and dirty
filters; blower was incapable of overcoming positive pressure in the  precipi-
tator housing.
     Additional Equipment Required-No temperature monitor was present at the
inlet of the precipitator.

Recommendations-

     Observe stack opacity while manually controlling water injection.
     Renovate, repair and replace old rappers.
     Replace broken insulator.
     Secure stabilizer insulators.
     Install new gaskets and insulator compartment blower and filter.
     Install thermocouple probe upstream of precipitator to determine if dew
point is reached.
     Inspect and clean cyclones.

     A sampling and testing program was devised to characterize the cyclones
and precipitator inlet and outlet streams.  Parameters measured included
grain loading (ASME-WP50), particle size (impactor and electron microscope),
moisture grainmetric), temperature, opacity, and pertinent process variables.
A prerequisite in acquiring the data was to minimize sampling duration ("five
minutes or less) to permit observations of peak concentrations.


                                    264
-------
Phase II;  Implementation
ma
     Rapper Units-Nine new rapper units were installed and adjusted.  The re-
  aning seven units were rebuilt and adjusted.  One of the two gas inlet dis-
tribution plate rappers (which had been missing since the precipitator was
originally installed) was not replaced since previous operating experience
revealed no contaminant buildup and therefore the single existing rapper was
considered sufficient.
     Rapper Mounts-All twelve of the plate rapper mounts were disassembled,
and reinstalled.  Three worn anvils and pins were replaced.

     The four wire frame rapper mounts were disassembled and cleaned.  Four
new shaft insulators were installed.  (It was suspected that these insulators
had probably never been inspected, cleaned or replaced since the precipitator
was installed).  A new cap for one of the insulators was machined and instal-
led.

     Rapper Timers-New contacts and cam followers were installed in the timers.
Cam timing and contact clearances were adjusted.

     A new starting capacitor was installed in the motor drive circuit.

     Insulators-All insulators were inspected, cleaned and coated with sili-
cone.  Two insulators which could not be cleaned properly were replaced.
     All insulator bushing gaskets were replaced.

     Wire Frame Stabilizers-All four stabilizers at the bottom of the wire
frames were cleaned and securely fastened to the wire frames.  New asbestos
pads were also installed.
     Bus Ducts Insulator Housings, Hatches-The bus ducts and insulator hous~
ings were cleaned.  New gasketing was installed on all hatches and all miss-
ing bolts were replaced; new studs were installed where required.
     Insulator Compartment and Ventilating System-New air filters were in-
stalled.  Heater and fan operation were checked.
     Duct air leaks around the heater was repaired.
     Air distribution ducting was unplugged and cleaned.  Baffling at the en-
trance to the insulator housings was modified to provide less restrictions
to airflow.
     Internal Inspection-An internal inspection of wires,, frames and housing,
was performed.  All wires and other components were found to be intact.
     The internal components of the precipitator were cleaned as required.
     Temperature Sensors-Thermocouple probes monitoring precipitator inlet
and outlet temperatures were installed.   A chart recorder and digital readout
were provided in the control room to continuously monitor the respective
temperatures.

Phase III:  Sampling & Testing

     To ascertain the effectiveness of the aforementioned mechanical service,
repair and replacement phase, a sampling and testing program was commenced.
At the outset, an airload test was performed to establish the baseline pre-


                                     265
-------
cipitator operating conditions.  Following this, the sampling and  testing
program was initiated.  These tests were designed to acquire inlet and  outlet
particle size distribution by both inertial impaction and electron microscopy
methods.  Also, sampling for grainloading, along the opacity observations,
were continuously performed during the entire sampling period.

     The test program was initiated and evaluated for a given production rate
(Test A) and then at a higher rate (Test B).  Opacity observations of the
frequency and duration of "puffs" were simultaneously recorded.  To examine
the effect of aftercooler water injection, Test C was performed at Test A
production rate.  The water injection rate during Test C was manually con-
trolled and maintained above lOgpm minimum while quantitative sampling was
performed.  The results of these tests are shown in Table II.  For comparison
purposes, previous sampling tests are also given in Table II.

     To expedite particulate sampling, single point isokinetic samples were
extracted after the average flow rate was determined.  Inlet and outlet samp-
ling was performed for five and twenty miiiutes respectively.

     Sampling for scanning electron microscopy (SEM) particle size distri-
bution was performed by "in stack" membrane (0.30 micron), filters.  Sampling
duration at the inlet was one second, and outlet twenty seconds.

Sampling and Testing Results-

     Grainloading, Particle Size (Impactor)-Quantitative and qualitative re-
sults from Table II for Test C indicate an outlet concentration of 0.014 GR/
AFC and opacity less than 5% with an efficiency of 99.7%.  These results
compared with previous outlet data (>. 109) GR/AFC) were the most gratifying
to date.  The reason for this was primarily due to the manual control of the
aftercooler water rate and the recently completed electrostatic precipitator
service and maintenance program.

     The apparent decrease in precipitator inlet particle size from 10.7
microns in Test G to less than four microns during Test A and C, reflected
the maintenance work performed on the cyclones.  However, the corresponding
inlet dust load still was less than anticipated.

     Since the variation of the production rates for Tests A and B were sig-
nificant, it was concluded that emissions reduction was strictly related to
improved precipitator performance.  Generally speaking, Tests A and B showed
substantial improvement in precipitator performance compared with  the pre-
vious Tests D-H given in Table III,

     Particle Size (Scanning Electron Microscopy)-Since visibility or light
scattering from a plume is largely determined by geometric particle size (and
refractive index) it was deemed necessary to characterize submicron size dis-
tribution by SEE.  in the range from 0.2-1.0 micron  (in intervals of 0.2
microns; Table III).  The cumulative frequency size distribution by number
and weight with their corresponding number (NMD) and mass median diameters
(HMD) and geometric standard deviation are given in Table III.  Larger  parti-


                                     266
-------
cles (>1.0 micron) were not counted since  they do not produce  significant
light scattering and would also require numerous SEM's  to be produced and
evaluated at low magnification.

     The size distribution obtained from the  SEM's  regardless  of  their inlet
and outlet conditions, reveal a number medium diameter  (NMD) range from 0.19-
.32 microns the data of which is given in  Table III.  These results are ex-
pected since electrostatic precipitators are  relatively inefficient for col-
lection of particles of less than one micron.  Since number concentration
measurements (particles per unit volume) could not  be obtained because of the
difficulty in controlling inlet sampling time (approximately one  second) a
valid comparison between submicron inlet and  outlet number concentrations
could not be made.

     The major value in acquiring the SEM's was to  verify that the plume
opacity was produced by high numbers of submicron particles.   One could
easily be mislead if the impactor data only was considered since  size distri-
bution obtained in this manner give an outlet MMD of 1.2. microns and greater.

Critical Service Items-

     The following items were critical to  proper precipitator  operation and it
was recommended that plant personnel check them regularly:

     Rapper System (Weekly)-Check and adjust  rappers in place  at  thirty per-
cent power  (30%).
     Check rapper shafts for freedom of movement where  it passes  through the
housing.
     Check timer for burned contacts.
     Insulators (when  excessive arcing occurs)-Check all insulators for
cracks, excessive dust buildup  (more than  1/8" thick),  and arcing tracks.
 (Be sure to check rapper rod insulators in insulator housings).
     Check lower stabilizers.
     Recoat insulators with silicqne after cleaning.
     Vent and Heater System  (Monthly)-Check heater  for  proper  temperature.
     Check air filters.
     Check system for  adequate airflow at  insulator housings.
     Bus Ducts (when open)-Be sure gasketing  and bolts  on all  hatches are in-
tact.

                                 CONCLUSIONS

Precipitator Capability

     When properly serviced and maintained, the existing equipment was cap-
able of adequately controlling the calciner emissions within the  environ-
mental requirements.

Calciner Liquid Flow Rates

     To minimize the periodic "puffing" of the exhaust  plume an automatic


                                     267
-------
water control system on the aftercooler was installed.  The controlled mois-
ture content of the airstream will increase the overall precipitator collec-
tion efficiency.

Temperature Monitors

     The temperature monitors on the precipitator inlet and outlet aided the
operators in maintaining the optimum temperature in the precipitator required
for efficient collection.  This also prevented moisture from condensing in
the precipitator.

Opacity

     Plume opacity occurred when the airstream contained high concentrations
of submicron particles less than 0.32 microns.
                                    268
-------
                   CYCLONE
FEED
       CALCINER
               WATER
             INJECTION
                               CYCLONE
r
                               AFTERCOOLER
                                                CYCLONE
                                                          ESP
                                                                 TO ASH
                                                                STORAGE
                         PRODUCT
        Figure 1.  SCHEMATIC DIAGRAM OF CALCINER SYSTEM
                                      269
-------
       TABLE I.  ELECTROSTATIC PRECIPITATOR - DESIGN SPECIFICATIONS
Normal Operating Conditions

       Flowrate:
       Temperature:
       Pressure:
       Inlet loading, approximate:
       Gas velocity thru collector:
       Retention time:
       Type of suspended matter:

Estimated Operating Requirements

       Draft loss  (pressure drop from the
       inlet to the outlet of the collec-
       tion equipment, not including
       nozzles):
       Total connected load, base
       offering:
       Electrical  characteristics:

Collection Equipment

       Number of electrostatic
       precipitators:
       Number of cells per
       precipitator width:
       Number of fields per pre-
       cipitator length:
       Design pressure, positive:
       Casing thickness:
       Collecting  surfaces -
          Gas passages per cell-field:
          Spacing  of gas passages:
          Surface  size:
       High voltage electrode diameter:
       Collecting  surface rappers per
       cell-field:
       High voltage system rappers per
       cell-field:
       Destribution device rappers, total:

Electrical Energizer and Control Equipment

       Energizer sets:
       Transformer ratings:
       Rectifier circuit:
       Rapper control panel:
70,000 CFM (nominal)
234 - 285 °F
Positive
6-7.5 gr/acf
3.26 fps
5.51 sec
phosphate dust
0.5" VWC

81 KW
440 v, 60 cy, 3
ph
15" VWC
3/16"

24
9"
9' x 20'
.1055"
 2
 2
 1
50 KV 1200 MA
 full wave
 1
                                     270
-------
                                       TABLE n.
                                                     RESULTS OF SAMPLING
               .   FLOWRATE
LOCATION   TEMP °F    ACFM
                               MASS         NO.
MOISTURE   CONC.  OUST LOAD  MED. DIA.   MED.  DIA.  PRODUCTION   EFFICIENCY
                             (MICRON)*   (MICRON)*  RATE (TPH)
OPACITY  NO. PUFFS  DURATION
   %       m.       SEC.
INLET
OUTLET
INLET
OUTLET
INLET
OUTLET
INLET
OUTLET
INLET
OUTLET
INLET
OUTLET
INLET
OUTLET
INLET
OUTLET
265^3
265i3
265-3
265^3
265i3
265i3
325
313
330
312
335
320
331
310
335
313
61940
65755
70120
65000
71000
-
63100
63600
56600
17
28
18.3
22.5
22.6
24.4
17.6
22.1
14.3
27.7
14.1
17.7
14.3
16.7
11.3
28.5
	 , 	
4.1
0.014
3.8
0.027
5.3
0,022
6.2
0.5
8.01
0.109
NA
0.192
6.430
0.107
4.590
0.121
2199
7.56
2160
15.2
3190
13.2
3454
278
4874
66.3
NA
103.8
3505
58.3
2227
58.7
4.0 (0.74)
1.2 (0.80)
2.7 (0.68)
1.8 (0.80)
4.0 (0.68)
1.2 (0.69)
NA
NA
NA
NA
NA
NA
10.7
NA
NA
NA
0.32
0.19 c
0.22
0.29 B
0.19
0.24 A
NA
NA D
NA
NA E
NA
NA F
NA
NA G
NA
NA H
99.7
<5 — 10 10 SEC
99.3
<10 . 	 30 10-30 SEC
99.6
<10 ~> 30 10-30 SEC
99.2
98.6


98.3 - - -
97.4
*   HMD AND NMB OBTAINED BY EEH BELOW  ONE MICRON ONLY
+   BY IMPACTOR
-------
                                 TABLE  III.      PARTICLE SIZE DISTRIBUTION
             TEST


                       - INLET



              A


                       - OUTLET






                       - INLET
                       - OUTLET
KJ
•-J
                       -  INLET
                      - OUTLET
NO %
CUM %
WT %
CUM %
NO %
CUM %
WT %
CUM %
NO %
CUM %
WT %
CUM %
NO %
CUM %
WT %
CUM %
NO %
CUM %
WT %
CUM %
NO %
CUM %
WT %
CUM 55
SCANNING
0.2
26.44
26.44
0.20
0.20
54.06
54.06
0.65
0.65
48.08
48.08
0.72
0.72
31.88
31.88
0.28
0.28
53.44
53.44
0.92
0.92
43.63
43.63
0.69
0.69
ELECTRON MICROSCOPY
0.2 - 0.4
26.44
52.88
5.29
5.49
22.10
76.16
7.19
7.84
28.02
76.10
11.25
11.97
33.15
65.03
7.79
8.07
28.07
81.51
13.09
14.01
32.67
76.30
13.90
14.59
0.4 - 0.6
28.81
81.69
26.68
32.17
11.72
87.88
17.65
25.49
15.93
92.03
29.62
41.59
19.49
84.52
21.21
29.28
10.80
92.31
23.31
37.32
17.29
93.59
34.04
48.63
0.6 - 0.8
10.85
92.54
27.56
59.75
6.89
94.77
28.46
53.95
4.87
96.90
24.83
66.42
8.20
92.72
24.48
53.76
5.13
97.44
30.38
67.70
3.68
97.27
19.91
68.54
0.8 - 0.10
7.46
100.00
40.27
100.00
5.24
100.00
46.06
100.00
3.10
100.00
33.58
100.00
7.29
100.00
46.24
100.00
2.56
100.00
32^9
100.00
2.74
100.00
31.46
100.00
NMF
MMD
4-
0.32
2.0
0.74
1.54
0.19
2.58
0.80
1.66
0.22
2.0
0.68
1.6
0.29
2.1
0.8
1.7
0.19
2.1
0.68
1.7
0.24
1.9
0.69
1.7
-------
                  ELECTROSTATIC PRECIPITATOR PERFORMANCE
                           WITH PULSE EXCITATION

                                    by

                                Donald  Rugg
                              Michael Durham
                               George Rinard

                Denver  Research Institute, Denver, Colorado

                                    and

                                Leslie Sparks

                Industrial  Environmental Research  Laboratory
                      Environmental Protection Agency
                   Research Triangle Park,  North Carolina


                                  ABSTRACT

     A laboratory  electrostatic  precipitator  was energized  by pulse  power
supplies.   The voltage-current characteristics  were  measured  under clean  and
high-resistivity  dust   conditions.    The  results,  which   illustrate   the
relationships between corona  current  and the peak and average  voltages,  are
shown.   Charge-to-mass  ratios,  particle-size   distributions,   and  mass
efficiencies were  measured.   The  data  were analyzed to  determine  the  major
factors  which  account  for  the  increase  in  mass  efficiency  for  pulse
excitation   compared  to  efficiencies  obtained  with dc excitation.   The data
and the results of the analysis are presented.

                                INTRODUCTION

     Pulse  excitation,  consisting of  an average voltage with a pulse voltage
superimposed upon  it, was  applied to a laboratory electrostatic precipitator
(ESP).  Measurements were made to obtain experimental data which would assist
in  understanding  and  evaluating the  basic  concepts  of pulse excitation.

     The  electrical  characteristics   of  interest  were   the  peak voltage,
average voltage, average current, and current distribution over  the collector
plates   These characteristics  were measured at  150°C   (300°F)  under both
clean  and  high-resistivity dust  (6  x 1012 ohm-cm)  conditions    The results
show  the changes  in the voltage-current relationship due  to  high-resistivity
dusts  and  also provide a basis  for  selecting the  operating levels of peak
voltage, average  voltage,  and  average current  for  optimum  ESP performance.

      Charge-to-mass  ratios and  mass  penetration measurements  were made to
determine the  charging  and collecting  characteristics of the ESP  wit.pulse
excitation.   Also,  a Meteorology Research  Incorporated  (MRI)   Plant Process
Visiometer  (PPV) was   used  to monitor the relative  outlet  opacity.  These

                                    273
-------
results were compared  to  measurements made using dc excitation  in an  attempt
to determine  if improvement  in  efficiency was  due to  higher charge on the
particles or to higher field strength for particle collection.

                      DESCRIPTION OF PULSE POWER SUPPLY

     A sketch of the pulse power supply is shown in Figure 1.  The circuit  is
similar to the ones described by Lausen [1] and Masuda [2].

     The controls  consist  of a variable pulse repetition  rate (PRR)  of 0  to
110 pulses  per  second (pps) and a variable pulse amplitude of 0 to 55 kV.  A
capacitor  in the  primary  of the  1:50  step-up transformer  is  charged  to  a
voltage v  which determines the pulse amplitude.   The SCR switch is triggered
by a  timfhg circuit which determines the PRR, and the capacitor is connected
to  the pulse  transformer  for  one cycle  of  primary current.   Figure 2 (a)
shows  that  the primary  capacitor discharges during the first half-cycle  of
primary current and recharges during the second half-cycle.  The output pulse
v  is  shown in Figure 2 (b).  The output  voltage  increases  during the first
ha"lf-cycle  and decreases during the next half-cycle.  This duration, which  is
considered  to  be  the  pulse  width,  is about 220 us for  one  power supply and
about  180 us for the other.

     The   output   pulse  was  capacitor-coupled   to   the   corona  discharge
electrode.   A  dc  bias voltage,  variable  from  0 to 60  kV,  was  added to the
pulse  voltage  through a  blocking diode.  Therefore,  the  pulse power supply
permitted independent  control over peak and average voltages and PRR.

                     ESP VOLTAGE-CURRENT CHARACTERISTICS

     Pulse  excitation consists of  an average voltage  with  a pulse  voltage
superimposed  upon  it.   Measurements were made to determine the  extent  to
which  the corona  current  can be  regarded as consisting  of two independent
components,  one due to the  pulse voltage, and the other  due to the  average
voltage.

     Wire electrodes  3.18 mm (1/8-in.)  in diameter were placed  in the ESP  at
a  spacing  of  22.9 cm  (9  in.).   Measurements were  made  at 150°C (300°F) for
both  clean  and dirty  conditions.  The  average  current vs.  average  voltage
(Vft/I.)  curves  were measured using dc excitation  and  pulse excitation.  For
puTse  excitation,  the  peak voltage, which  is the sum of  the pulse voltage and
the  average voltage,  was maintained  at 60 kV and the PRR was  set at 110 pps.

     The  results  of   these  measurements  are  shown in  Figure 3.   For  clean
conditions  and  dc excitation, the corona  onset voltage  was  26 kV.   For  clean
conditions  and pulse  excitation,  there was corona current  due to the  pulse
voltage  when the  average voltage was only 5  kV.   As the  average voltage was
increased   and  the peak  voltage  was held  constant  at 60  kV,  the  current
increased  until  the average voltage  was  about  15  kV.   This  current  increase
was  probably due  to  the  increase  in time that the  pulse voltage was  above
corona onset.   The current  density remained  constant at  12 nA/cm2  as the
average  voltage was increased from 15 kV  to  25  kV.   This indicates  that the
current  was due   to  the  pulse  voltage and  was  independent  of the  average
                                    274
-------
   CONTROLS
      PRR
   0-IIOpps
    PULSE
  AMPLITUDE
     0-55kV
                  T
                         SCR
                       SWITCH
50 M v0 ESP
t
{
BIAS
VOLTAGE

y
                FIGURE I. PULSE POWER SUPPLY
  500
  400
^300
5 200
   100
    0,
            0.05
             ms
o )  Primary Capacitor Voltage
                             0.1
                                                 b) Output Vollage
           FIGURE 2.  PULSE POWER SUPPLY WAVEFORMS
                              275
-------
                    1.6
                    1.4
                    1.2
                   1.0
                   0.8
K3
•^1
ON
                o

                §  0.6-t-20



                      -15
                   0.41

                      -10


                   0.2-
                      -5
                       -50
                       -45
                       -40
                       -35
                       -30
                       -25
	CLEAN

	DIRTY
        ONSET OF CONTINUOUS
        CORONA OR SPARKING
                                              10
                                     20
                            CORONA VOLTAGE, kV
30
40
                                FIGURE  3.
            AVERAGE CURRENT vs AVERAGE VOLTAGE FOR 3.18mm (l/8-in.)WIRE
            WITH/WITHOUT 60 kV PEAK PULSE.
-------
voltage,  which was  below dc corona onset.  For average voltages above 25 kV,
the  corona  current  increased  rapidly  as  shown  by  the  curve   for  clean
conditions.   This  indicates  that the  corona current  was  being sustained
between pulses  by the  average voltage.   Occasionally,  sparking rather than
continuous corona would occur when the average voltage reached 25  kV.

     For dirty  conditions  and  dc excitation,  the  corona  onset  voltage  was
26 kV  (the  same as  for clean  conditions).   However, as  the dc voltage  was
increased,  the  current  increased  abruptly,  as  shown  in  Figure  3.   As  the
voltage was  reduced,  the corona current  did  not  cease  until the voltage was
18 kV.   This  hysteresis  in the V./I.  curve  indicated the presence of back
corona.

     For dirty  conditions  and pulse excitation,  the  V./I.  curve was similar
to the clean-condition  curve  when the average voltage was below 21 kV.   It
appears that  the  current was due  to the  pulse voltage and  not to the average
voltage.   As  Figure  3 shows, when average voltage  was  increased above 21 kV
and sparking  did  not occur, the current  increased  abruptly, indicating that
the average  voltage  was sustaining corona current between  pulses.   Also,  the
hysteresis  in the  V./I.  curve for  dirty conditions  and  pulse  excitation
indicates that  back  corona was produced  and  did  not cease  until the average
voltage was  reduced  below 18 kV.  This was the  same voltage at which corona
ceased with dc excitation.

     The  effects  of  reducing  the PRR  were  observed.  For average voltages
below  the onset of continuous corona, the corona  current decreased as the  PRR
was decreased.  However,  corona current  could be controlled by  the PRR only
if the average  voltage was maintained  below  the  onset  of  continuous corona.

     Current  distributions on the collector plate  were  measured under dirty
high-resistivity conditions for dc and pulse excitation.    A 0.279 m2 (3 ft2)
area on one collector plate was divided into 96 isolated cells 29 cm2 in area
and the current  to  each cell was  measured.  For  dc excitation and an average
current density  of  5  nA/cm2,  there was  no  current  on 90% of the collector
plate  area and high  current densities on a few  cells.   This indicates that
back corona  was  established on these  localized areas.   For pulse excitation
without continuous  corona and at  about the same  average current density,  the
current was  uniform  over the collector plate.  When the  average voltage was
high enough  to  produce continuous corona current, the additional current was
not  uniformly distributed over the  collector plate.  This again indicates
that  under pulse  excitation,  back corona was  established when  continuous
corona current was present.

     The curves in  Figure 3 show  that  the corona current can be regarded as
consisting of a component  due to the pulse voltage and a component due to the
average voltage.  When corona current was due to only the  pulse voltage,  the
current distribution  was uniform and back corona did not exist for the high-
resistivity  dirty  conditions.    However, if  the  average  voltage  was  high
enough to  sustain corona current  between pulses, then  it  appeared that back
corona  existed.   With  dust,   the onset of   continuous  corona  occurred  at
average voltages of 18 kV  to 26 kV, depending upon the amount of pulse corona
current.                             277
-------
                 ESP CHARGING AND COLLECTING CHARACTERISTICS

     Charge-to-mass ratio,  particle size  distribution,  and mass penetration
measurements were  made to  determine  ESP  charging  and collecting character-
istics of high-resistivity fly ash with pulse excitation.  Since charging and
collection of particles  may be a function  of  both  peak and average voltages
[3],  tests  were  run  at  two different  average  voltage levels  to determine
their  effect  upon charging  and  collecting.  The test was  repeated using dc
voltage  in  order  to  make  direct  comparisons  between  the   two  types  of
excitation.

     The laboratory ESP  consisted of  one 22.9 cm (9 in.) wide channel, which
was 1.22 m  (4 ft)  high and 4.88 m (16 ft)  long.  The  corona electrodes were
3.18mm  (1/8-in.)  wires,  spaced  22.9cm  (9 in.)  apart,  and  the  total
collector plate  area  was  11.9 m2 (128 ft2).  The  gas velocity  was  2 m/sec
(6.65  ft/sec), which  resulted  in an  SCA  of  21 sec/m  (107 ftVlOOO acfm).
Mass  loading and  impactor measurements were made across the inlet and outlet
of  the ESP.   These measurements were  made at  three sampling  ports (top,
center,  and  bottom)   at  each  location.   Also,  the  charge  and mass  were
measured  at a  center port  located  1.22 m  (4 ft)  upstream from  the  outlet
sampling ports.   Mass  penetration,  inlet and outlet particle  size distribu-
tion,  and Q/M ratio were determined from these measurements.

     The experimental  results  of  the  three  tests are  listed in Table I.  In
all three tests,  the  dust at  the inlet  had a mass  median diameter of 4.9 [im
and a  geometric standard deviation of 3.   In Test No.  1, the peak voltage was
set  at 60  kV,  which  was  slightly  below  the  sparking voltage;  the  average
voltage was  set  at 20 kV, which  was  below the onset of continuous corona or
back  corona; and  the  current density was  adjusted  to  8.4  nA/cm2 by the PRR.
The  measured Q/M  was 19.6  uC/g  and  the   mass  penetration was  0.118.   The
modified migration velocity, iu.,  was 0.22 m/sec.  Several investigators [4, 5]
consider the empirically  derived  ID.  to be  a measure  of ESP performance, and
ratios of to. are referred to as improvement  factors.

     Test No. 2  was the  same  as  Test No.  1, except the average voltage was
reduced from 20 kV to 15 kV.  The Q/M ratio was 17.6 pC/g,  which was 10% less
than  in the first test.  The mass penetration of 0.115 and modified migration
velocity of  0.22  m/sec were essentially unchanged.   The mass median diameter
of  the particles  at the outlet was 2.5 urn compared to 1.75 urn in Test No, 1.
In  part,  this difference  was  probably a result of making  only one impactor
measurement  at  each of  the  three outlet sampling ports during  a test.   The
effects  of  lowering the  average  voltage were not  large enough  to be deter-
mined  from the results of Test No. 1 and Test No.  2.

      In Test  No.  3,  dc excitation was used.  The average voltage was  27.6 kV
and  the current density  was about 9 nA/cm2.  This operating  point,  and the
average  voltage  and current  operating points for  the first  two tests, are
shown  in Figure 3.   The  dc  operating  point was  kept near  the increasing
voltage  V./Ift  curve by  manually  decreasing and  then  increasing the  voltage
whenever  the  current  rose  above  12  to  14 nA/cm2.   This  procedure kept the
outlet opacity at  a minimum level.  Although  the average  voltage was higher

                                    278
-------
                       TABLE I:  EXPERIMENTAL RESULTS

                     Test No. 1     Test No. 2     Test No. 3

   din,  (jm               4-9            4.9            4.9

   °in                   3.0            3.0            3.0

   V , kV               60             60

   VA, kV               20             15             27.6

   JA, nA/cm2            8.4            7.6            8.9

   SCA,  sec/m           21             21             21

   Q/M,  nC/g            19.6           17.6            6.0
   Mout/Min              °-118          °-115          °'248

   uik*, m/sec            0.22           0.22           0.093

   dQut, [am              1.75           2.5            3.0


   "out                  2'°            2"4            2'7
Resistivity = 6 x 1012 ohm-cm

*w,  = [In (M  ,-M. )]2/SCA
  k         out in

than during the  two pulse excitation tests,  Q/M  was  only 6.0 pC/g, the mass
penetration was  0.248, and  10.  was 0.093 m/sec.  The  outlet  particles  had a
mass median diameter  of  3 urn, which was  larger than  in the first two tests.
The  larger  size particles  were  expected  since the  mass  penetration  was
larger.   In  comparison to  dc excitation,  the  improvement factor  for  pulse
excitation was 2.36.

     The  effect  of  operating with  the average  voltage  above the  onset  of
continuous  corona  was  determined by  observing  the   PPV  output which  is a
relative measure of  the  outlet opacity.  Initially,  the  average  voltage was
set at  27  kV  and the pulse  source was  off.   Then a 25 kV pulse was added to
the average voltage to produce a peak voltage of 52 kV and the outlet opacity
decreased.  The  peak  was  kept at 52 kV and the average voltage decreased to
10 kV.    Again,  the  outlet  opacity  decreased.   These  observations indicate
that the best performance was achieved with pulse excitation when the average
voltage was below the onset of continuous corona.

     The Q/M  ratio  depends  upon the particle size distribution, ion density,
treatment  time,   and  field  strength.    To  assist  in  understanding  this
relationship,  Q/M ratios  were calculated as a  function of applied dc voltage
on the  ESP for three different particle-size distributions, assuming constant
                                    279
-------
current  density  and  constant  treatment  time.    The  results  are  shown  in
Figure 4.   It appears  that a  dc charging  voltage of  50 kV to  60 kV  would
produce the same Q/M ratios that were  measured in the pulse excitation  tests.
This equivalent  de-charging voltage was  considerably higher than the average
voltage.   The  de-charging  voltage  of  27.6 kV  in Test No.  3  should  have
produced  a Q/M  ratio  of  10 pC/g.   Since  the measured Q/M  ratio was only  6
|jC/g,  it appears  that many particles were  never charged  or that particles
lost their charge  through  collection  and  reentrainment.

     The  ESP  performance for various  de-charging and collecting voltages was
calculated.  The charge  and penetration,  as a function of particle size, were
determined  using  field  and  diffusion  charging  theory  and  the  classical
Deutsch  equation  for  collecting.  The  results were  integrated to calculate
the  theoretical  Q/M ratio and  penetration.   The modified migration velocity,
u>. ,  was   calculated   from the  theoretical   penetration.   The  inlet   size
distribution,  current densities,  and SCA were  assumed  to be the same as  in
the  experimental tests.

     Column A   in   Table II  shows  theoretical   results  for  charging  and
collecting voltages  of 20 kV.   The  calculated  Q/M  ratio  was  low  and the
penetration was  high in comparison  to Test No. 1.  This  indicated that both
charging and collecting were   better  than for  an ESP operating  at 20 kV  dc
without  high-resistivity  dust  problems.    The theoretical  results  for  60  kV
charging voltage  and  20 kV collecting  voltage are  listed  in  Column B.   In
comparison to Test  No.  1, the  calculated Q/M was  high,  indicating that the
effective charging  voltage  was  less  than  the  60  kV   peak  voltage.   The
calculated penetration  was high,  indicating that the  effective collecting
voltage  was  higher than the 20 kV average.   The  dc conditions of Test  No.  3
were used to calculate  the  values in Column  C.   The  theoretical  penetration
was  lower than measured values.   This indicates that the ESP performance was
low  with dc excitation because of the high-resistivity dust.

                           TABLE II:  CALCULATED ESP PERFORMANCE
                       FOR  VARIOUS CHARGING AND COLLECTING VOLTAGES
           din, pm                 4.9          4.9          4.9
           
-------
   26

   24

   22

   20

   18

   16

P" 14

   12

   10

   8

   6

   4

   2

   0
o
0.
    0
               10
20
                                   I
30
kV
40
                                 = 30pm
                                 - 2.7
50
60
    FIGURE 4. CALCULATED Q/M RATIOS  FOR VARIOUS CHARGING VOLTAGES
                              281
-------
                             CONCLUDING REMARKS

     Maximum ESP efficiency was  achieved by operating with  the peak voltage
slightly below  sparking,  the  average  voltage below the  onset of continuous
corona, and the  PRR  at its maximum rate.  Under these conditions, the corona
current was uniformly  distributed  over the collector plate  and there was no
indication of back corona.   However, when the average voltage was high enough
to sustain corona  current,  there were indications of back corona and the ESP
outlet  opacity  increased.   Therefore,  in evaluating pulse excitation,  it is
important to know  under which of the  two  average  voltage conditions the ESP
was operating.

     The improvement in efficiency  with pulse excitation in  comparison to dc
excitation appears to be due to a higher charge on the particles and a higher
field strength for particle collection.

                                 REFERENCES

[1]       Lausen,  P.;  Henrikson,  H.;   and  Petersen,  H. ; "Energy Conserving
          Pulse   Energization   of   Precipitators;:   IEEE-IAS   Conference,
          Cleveland,  Ohio;  October,  1979.

[2]       Masuda, S.;  Obata,  G.;  and Hirai, J. ; "A Pulse Voltage Source for
          Electrostatic Precipitators;"  Proc.  IEEE-IAS Conference,  Toronto,
          Canada; October,  1978.

[3]       Masuda, S.; and Mizuno, A.; "Maximum Charge  of  a Spherical  Particle
          Imparted by  Pulse Charging;" Proc.  1978  International  Workshop on
          Electric  Charges  in  Dielectrics,   Kyoto,  Japan;   October,  1978.

[4]       Petersen,  H. ;  "New Trends  in  Electrostatic  Precipitation - Wide
          Duct   Spacing,   Precharging,  Pulse   Energization;"   IEEE  Cement
          Industry Technical Conference, Toronto, Canada;  May, 1980.

[5]       Feldman,  P.;  and  Milde,  H. ;  "Pulse Energization  for  Enhanced
          Electrostatic  Precipitation  in  High  Resistivity  Applications;"
          Symposium  on  Transfer  and  Utilization  of  Particulate  Control
          Technology,  Vol.  1,  (Denver, CO;  July, 1978);  EPA-600/7-79-044a
          (NTIS No. PB  295226); February, 1979.
                                    282
-------
        DEVELOPMENT OF A CHARGING DEVICE FOR HIGH-RESISTIVITY DUST
                    USING HEATED AND COOLED ELECTRODES

  By:   G. Rinard, M. Durham,  D.  Rugg, Denver Research Institute,  Denver,  CO
    L.  Sparks,  U.S. EPA, Industrial Environmental Research Laboratory,
                         Research Triangle Park, NC

                                  ABSTRACT

    An efficient cooled-pipe charger/collector which is capable of control-
ling back  ionization  and  allowing efficient charging of  particles  has  been
developed.   It has  also  been shown to be an excellent collector.   Feasibility
tests   were  conducted, utilizing parallel heated-pipes  and corona-discharge
wires  placed in a  plane  normal to gas flow, to determine if the dust resist-
ivity,   and  thus back ionization,  could  be controlled  by heating  a  small
collector surface area.   The success of these tests led to the development of
a more  practical  design  utilizing cooling  on  the collector  pipes.   Cooled
pipes  are more  practical for many applications since they do not require the
input  of  high quality heat.   The  results of the testing  of  the heated and
cooled electrodes that led to the present  design are given.


                                 INTRODUCTION

     The electrostatic  precipitator  offers  advantages in controlling parti-
culate emission for many  industries.  However, the operation of electrostatic
precipitators  in  collecting  fine,  high-resistivity dust  is  a problem.   The
size of a standard Cottrell-type precipitator necessary  for controlling these
problem  sources of  dust  is  often  not  cost effective.   The  most important
factor  in  the  operation  of a  standard  Cottrell-type  precipitator  is  the
electrical  resistivity  of  the  dust collected.   The  most  common problem
encountered  in the use  of Cottrell precipitators in industrial application
occurs  when the electrical resistivity  of the  dust is too  high  (greater than
 1011 Q-cm).   In the  presence  of  high-resistivity  dust,  the negative ions
produced  by corona generation  produce  an excessive  field in the dust layer
resulting  in back  ionization.   Although the exact nature of back ionization
is  not completely understood, the  electrical characteristics produced and the
deleterious  effect on particle  collection are well known.   Back  ionization is
characterized   by  an operating voltage  considerably  lower  and  operating
current considerably  higher  than those  obtainable under  clean  plate  operating
conditions.  Since back  ionization is produced  stably at a voltage  lower than
that   required  for   normal   operation,    effective  uni-polar  ionization is
greatly reduced.   This,  in  conjunction with the presence of positive  ions
produced  at the plate,  has  an  overall  effect  of greatly  reducing  the  effi-
 ciency of   charging  and collecting the dust  particles.   Back-ionization
effects can so  overwhelm the  operation  of an electrostatic precipitator  as to
make it effectively useless  as  a dust collector.

     The  electrostatic  precipitator would become more  practical for collec-
 tion   of  high-resistivity dust  if  it  could  be designed  to eliminate  the
possibility of back  corona being produced.   A survey of the technology shows
                                      283
-------
that  this  is  the goal  of  most novel  precipitator concepts  (Bush,  et al.,
1980; Gelfand,  etal.,  1980;  Masuda, etal.,  1978a; Pontius,  etal., 1979).
These  concepts  rely  on  means  that  attempt  to  separate  the  charging  and
collecting processes.   The  two processes  may be separated  and performed in
different  parts  of  the  precipitator,  in  which case the  unit  is  usually
referred to  as a  two-stage  electrostatic precipitator.  The  purpose of the
first stage  is to charge,  without collecting, the  particles, thus avoiding
production of back ionization.  The purpose of the second stage is to collect
the  particles  under  conditions  of high electric field and low current, again
avoiding production  of  back ionization.   These two  processes may  also be
separated  in  time by  means  of  varying  the precipitator   corona  voltage
waveform.  The high-voltage waveform used  for this purpose is  a  DC voltage
with an  impressed pulse of  long or  short duration  (Feldman and Milde, 1979;
Lausen,  etal.,  1979;  Masuda,  etal.,  1978b; White,  1952).   This  type of
waveform is applied to two or three electrode designs.

      In  order  to  study  a  concept  originally investigated by  Penny (1950,
1962) and White (1956-59 as reported by White, 1974), a precharger was tested
which consisted of parallel heated-pipes and  corona discharge wires.  These
initial  tests  were conducted to determine if  the dust resistivity, and thus
back ionization, could be controlled by heating only a small (pipe) collector
surface  area.   The primary  objective was to determine the characteristics of
a  precharger  when  operating with  high-resistivity  dust  but  not  in  back
corona.  The  results  were very encouraging in that  not only were the parti-
cles charged to  a high level, but  a considerable quantity of  the dust was
collected  by this configuration.  Further  tests were conducted to optimize
the   heated-pipe  charger/collector  and  determine  its practicality.   These
tests,  while  encouraging,  indicated that  a  considerable quantity  of high-
quality  heat   was  required  for  heated-pipe operation.   Therefore,  tests
were conducted with the  same wire/pipe configuration except  that the pipes
were cooled  to control resistivity.   For these tests the moisture content of
the  test-gas stream was  raised to  levels  normally encountered  in flue-gas
streams.   The  results  of  the tests  on both the  heated- and cooled-pipe
charger/collector  are given below.  This work has led to an important concept
in  prechargers.   The parameters necessary for good particle charging are the
same as those necessary for good  particle  collection.   Therefore,  if the
precharger  can  be  designed to  control back ionization  due  to collected
high-resistivity  dust,  the overall collecting efficiency of the two stage-ESP
can  be greatly  enhanced.

                             HEATED-PIPE TESTING

      Initial  testing of the heated-pipe system occurred  in the 0.472 m3/sec
(1000 acfm)   EPA pilot  precipitator in  Room H-300  of  EPA's Environmental
Research Center in Research Triangle Park,  North Carolina.   Two  heated-pipe
systems  were set up with one in the first section and the other in the third
section  of the pilot-scale  ESP.  Each heated-pipe  system consisted of two  6
cm   (2-3/8  in.) pipes  each  containing two  1500 W electric  heaters.  The two
pipes were mounted onto the  collector plates  directly across  from each other
in   the  front  third  of  the  section.   The  corona  electrode  consisted  of  a
single  0.32  cm (1/8-in.) or 0.16 cm (1/16-in.) wire  aligned with the center
of  the  pipes.  Two focusing  electrodes consisting  of  1.27 cm  (1/2-in.)  rods
                                     284
-------
were placed 7.62 cm (3 in.) upstream and downstream  of  the wire.  The purpose
of these electrodes was to focus the electric  field  in  such  a manner that all
of the current would go to the heated pipes and not  to  the unheated collector
plate.

     The  heaters were  connected  to  a timer/control  unit  so  that  the duty
cycle  of the  heater  could be  varied.  No  thermocouple was mounted  on the
outside of the pipe because it would interfere with  the electrical conditions
of  the  system,  and  because  it  would not  measure  the temperature  of the
outside surface of the dust layer  that was of  interest.

     The  laboratory-measured electrical  resistivity  for the  dust  used was
2 * 1012 fi-cm  for  the experimental  condition.  Figure  1  is a  plot of  clean
and  dirty voltage  current  characteristics  as  a  function  of  the  internal
temperature  of  the  pipes.   These tests  are  for  0.16 cm   (1/16-in.)  corona
wire.   Initially 0.32 cm  (1/8-in.)  wire was  used but  this produced only a
small  operating  range between the  corona onset and the  sparking voltages.  As
can  be seen  from  Figure  1,  the clean curves all  have about the same  shape
except that  as  the  temperature  of  the  pipe  is  increased,   the  current
increases  for a given voltage and the sparking voltage decreases.  This can
be  expected  from the change in  the air density in the  vicinity of the pipes.
The unheated  pipe,  T =  165°C (330°F), displays the  sharply  rising character-
istics of back  ionization.

     Mass-concentration  tests were run on the  inlet  and outlet of the precip-
itator to determine the  effect of  these improvements in electrical conditions
on  collection efficiency.   The precipitator was  set up with the heated  pipes
in  sections  1  and  3 and  parallel-plate electrodes  operating at 40  kV in
sections  2 and  4.   Three  simultaneous inlet  and  outlet tests  were run with
the pipes  unheated and the average penetration was 31.2 percent.  The  pipes
were   then  heated  to 595°C  (1100°F)  and  seven  tests were  run,  and the
penetration  was  reduced  to  7  percent.   The  field  strength  for  the  pipe
section was 3.7  kV/cm.

     Further  testing  was  done  with  the  pipe spacing  increased  to 27.9 cm
 (11 in.)  which  is  more  representative  of  a multichannel unit  with  6 cm
 (2-3/8 in.)  pipes  on 22.9 cm (9 in.) centers.   For this  spacing  the  field
strength  could  be  increased  to  4.5 kV/cm.   For  these tests  the  overall
penetration  was  reduced to  3.8 percent.   Charge-to-mass  measurements were
made  following the pipe section.  The average  of five tests was 7.22  Mc/8-
These  measurements also indicated that each  pipe  section  had a collection
efficiency of more than  50 percent.

     After  repeating the  tests  at Denver  Research Institute's Cherry  Creek
Field  Site on  a multichannel unit,  attempts  were  made to reduce the  input
power  necessary to keep the pipes hot.  Smaller  pipes  and various methods to
reduce heat loss were  tried.  None proved sufficiently  successful  to make the
heated-pipe concept practical.   Another problem anticipated  with  heated  pipes
is  that  they could exhibit the  same  deteriorating performance over a period
of time that  is  experienced in some hot-side precipitators.

                                      285
-------
     <
     E
     LJ
     CC
     K
     13
     O
     <
     z
     o
     
-------
                             COOLED-PIPE TESTING

     Because of  the practical  problems with heated  pipes  it was decided to
try to  take advantage of  the  decreased resistivity  on  the  cold side of the
resistivity  curve.    Super  cold-side  precipitators  have  been  demonstrated
successfully in  Japan  (Drehmel, et al., 1979).  In this case,  the collecting
plates  of  the  entire  ESP are  cooled  by means  of water-filled  compartments.
The  collected   dust is  removed  from  the  collection surfaces by  means  of
mechanical scrapers since the water-filled plates cannot be  rapped by conven-
tional  means.   This problem could be  overcome using the  wire/pipe design,
since  the  area  of  the  cooled  surface  would  be  greatly  reduced.   The
cooled-pipe  concept would have several advantages over  the heated pipes if
both  work  equally  well.   The  cooled  pipes  would be removing a low-quality
waste heat  from the  gas  stream rather than  adding  high-quality heat to the
gas.  The  cooled pipes  would  also  be operating  in the region of surface-
controlled  resistivity rather  than  bulk resistivity and would therefore be
more  a  function  of  the components  of  the gas  stream than the bulk material of
the fly ash.

      A  water-pipe  manifold  and a cooling  system were  designed to test the
cold-pipe  concept.    The  manifold consists  of  seven pipes  spaced 22.9 cm
(9 in.) center to center welded to an upper and a lower header.  Water enters
the bottom  header,  rises through  the  cooled  pipes, and exits  on  the opposite
side  of the upper header.  Water temperature  is measured at  the inlet and the
outlet  of  the  system  in order to monitor the temperature drop through the
system  and  calculate  the heat removed from the gas stream.   Two separate
systems  were built using 6 cm (2-3/8 in.)   and  3.81 cm (1-1/2 in.) outside
diameter pipes.

      The  dust  resistivity for  these  tests  was the same  as  that used in the
experiments  at  EPA's Environmental Research  Center.   The dust resistivity at
cold-side  temperatures  is   dependent  on moisture  content   of  the  flue gas.
Water was  sprayed  into  the system at a  constant rate and  wet-bulb/dry-bulb
measurements were  made  to   determine  the volumetric  moisture  content of the
gas  stream entering  the  precipitator.  This can be controlled in the test
unit  from 0 to 14  percent.  Most data  were  taken  at  10-percent  moisture,
which is typical of power plant flue  gas.

      A  series   of  tests were  run which  included voltage-current character-
istics  and  charge/mass data for both sets of cooled  pipes.   The data for the
3.81  cm (1-1/2  in.)  pipes  are  presented  in  Table  1,   and  the  curves are
plotted  in Figure  2  for  three  different  inlet-water temperatures.  With no
water in  the pipes  the VI curve is nearly vertical  at about 20 kV.  When the
water was  added to  the pipes and  the  inlet temperature  was adjusted to
29°C  (85°F),  38°C  (101°F),   and  47°C (117°F),  the electrical  conditions
improved and data  all fell  on  the same curve.  No  clean-pipe data  was  taken
at  this time but  later measurements  indicate  that  the cooled-pipe data lie
along the  clean-pipe  curve.    The   highest  operating  voltage,  49 kV, was
obtained for the intermediate  temperature of 38°C (101°F).   At  8.3°C  (15°F)
below and above  this temperature,  sparking occurred  at a lower voltage.   When
operating  with  an  inlet  temperature  of 29°C  (85°F),  the  lower header and
                                     287
-------
         10
00
CO
       
-------
vertical pipes  were  dry but condensation  was occurring on the  upper  header.
This  was  probably  due  to the  fact that  the upper  pipe  is in  an area  of
stagnant air  flow.   The  condensate dripping  from  the upper  pipe led to  the
lower operating voltage at this temperature,  and a spark was initiated each
time  a  drop   of  water   fell   from the  manifold  toward the  high-voltage
electrode.   At  both  of the other  temperatures the  collected  dust  on all pipe
surfaces was  completely  dry.   It  appears,  therefore, that the cooled pipes
operate best  at a temperature  above the water dew point of  the  gas  stream.


                 TABLE  1.  OPERATING CHARACTERISTICS OF THE
         3.8 cm (1.5 in.)  COOLED PIPES WITH  0.16  cm (1/16-in.) WIRES
Tin
°C
29
39
47
(°F)
(85)
(102)
(117)
V
kV
42
46
42

I
mA
4
3
6
.3
.9
.5
Q/M
MC/g
3
4
3
.7
.3
.6
m/v
g/m
1
0
1
3/min
.00
.91
.12
Heat
Out
W/m3/min
15
15
17
.5
.5
.0
     The  data  for the 6.0 cm  (2.375  in.)  cooled pipes, given in Table 2, is
 similar to  that for the  smaller  ones.   The VI  curves  are  given in Figure 3.
 Without  cooling,  severe  back  ionization,  resulting  in very high current,
 occurs with operating voltages between  15 and 20 kV.   Although no data were
 taken at  29°C  (85°F),  the data  at 4l.6°C (107°F)  and 45.5°C (114°F) both fall
 close to  the curve for clean  conditions,  with the higher  operating voltages
 being  obtained at  4l.6°C (107°F).   For both sets  of pipes the temperature
 drop from the  inlet to the outlet was maintained between 2.8  and 5.5°C (5 and
 10°F).   Comparing the data  between  the two  sets  of  cooled pipes  it can be
 seen  that  (as  with  the heated pipes),  although higher voltages  can  be
 obtained  with  the  smaller pipes,   taking into  account   the  increased  gap
 between the corona wire  and the  pipes,  the larger  pipes had  the higher field
 strengths.


                  TABLE 2.  OPERATING  CHARACTERISTICS OF THE
        6.0  cm  (2.375  in.) COOLED PIPES  WITH 0.16 cm (1/16-in.) WIRES
Tin
°C (°F)
27 (80)
41.7 (107)
42 (108)
V
kV
40
42
32
I
mA
6.6
6.8
2.85
Q/M
MC/g
3.9
4.6
3.7
m/v
g/m3/min
0.66
0.75
1.00
Heat Out
W/m3/min
24.7
21.2
21.2
     With  the  cooled pipes it  was  possible to operate at  a  voltage  approxi-
mately  30 to  40 percent  higher than  with the heated  pipes.   However,  the

                                    289
-------
          O TEMP. = 42°C(I07°F)
          + TEMP. =46°C(II4°F)
          D TEMR =!49°C(300°F)
                                      20        25       30
                                       CORONA VOLTAGE,  kV
Figure 3.  Voltage-Current Characteristics  of the  6.0  cm  (2.375  in.) Cooled-Pipe Electrodes
-------
charge/mass values were  not as high as those obtained at lower voltages with
the heated pipes.  This  cannot be  explained  at this time.  The mass loading
following the cooled pipes was slightly higher, but  the feed rate of dust was
much higher  for  these tests than for the hot-pipe tests.  It should be noted
that the opacity monitors were much lower  for  the cooled pipes when the dust
feed rate  was the  same  as  for  the heated pipes.   This  indicates that the
cooled pipes were better collectors than the heated  pipes.

     Besides  the  better  operating  conditions  of  the  cooled  pipes,  the
low-quality  heat removed from the  gas stream was less than half the quantity
of the  high-quality  heat needed to be added to the  heated pipes.   The cooled
pipes take less  than 1  W of  thermal  energy from each  ft3  of  air which has
an  effect  of lowering  the  flue  gas  temperature 1.0°C  (2°F)  per  section of
cooled pipes.

         EFFICIENCY TESTS WITH A SINGLE SECTION OF COOLED PIPES AND
                             HIGH-VOLTAGE PLATES

     The purpose  of this  series of  mass-efficiency tests was to determine the
collection efficiency of a single section of a  precipitator with cooled pipes
as  a  charger/collector  and high-voltage plates as a collector.  The charging
section  consisted of a  set of  seven 6 cm (2-3/8 in.)  diameter pipes spaced
22.9 cm  (9 in.)   center  to   center   with  3.18 mm (1/8  in.)   corona  wire
electrodes.   The  collector  section   consisted  of  seven  collecting  plates
spaced 22.9  cm  (9 in.) alternating  with six high-voltage plates.  The nominal
plate  size   was  1.1 x  1.6  m  (3.6  x 5.2 ft)  for a total  collection  area
 (excluding high-voltage  plates)  of 21 m2 (225  ft2).  Dust resistivity at the
operating  gas temperature was  1012  Q-cm.

     The  pipe  voltage  was  set at 42-43  kV.  The  plates  were  operated at
approximately 2  to 4  kV  lower  than  the point where back corona started.  This
was approximately 48  kV.  The  results  obtained  were  quite impressive.  With a
specific   collection  area   of  less  than  11 sec/m  (55  ft2/1000  acfm),  the
collection  efficiency was  84.4  percent.    This is  especially  good on a test
machine  designed for testing mechanical characteristics and therefore didn't
have any baffles to prevent  sneakage.

              TESTING OF  COOLED-PIPE CONCEPT ON ACTUAL FLUE  GAS

     After   successful  tests  of  the   cooled-pipe   charger/collector  in the
laboratory,  a  small  test  unit was  constructed  for  testing  under actual
flue-gas conditions.   The purpose  of  the  unit  was   to determine  if there was
any  component  in the actual  flue gas  which  would prevent the  cooled-pipe
concept  from working as it  did  in  the  laboratory.   Tests were made with this
"Tiny TEP" at the Coors  Power  Plant  in Golden,  Colorado, and at  the Valmont
Station near Boulder, Colorado.

     The chamber in the  Tiny TEP  is a  0.31  m  (1 ft)  cube.  The corona wire  is
3.18 mm  (1/8-in.) in  diameter and  uses  four  cooled  pipes  3.1  cm  (1.5  in.)  in
diameter with centers  11.43 cm (4.5  in.)   from the  corona wire.   The  corona
wire is  connected to a  negative  high-voltage power supply with  meters  which
measure voltage  and current.         291
-------
     The results of tests under actual flue-gas conditions at the Coors Power
Plant are shown  in Figure 4.   The clean VI curve and the dirty VI curve with
the  internal  cooled-pipe  temperature   at  118°C (244°F)  are  shown.  The
difference  between the  two curves  shows  that back  corona occurs  when the
pipes are  not cooled.   Sparkover voltages and current were measured as the
internal pipe temperature was  varied.   When the internal  temperature of the
cooled pipe was 31°C (88°F), the VI characteristics were essentially the same
as if the pipes were clean,  except sparkover occurred at about 49 kV as shown
in Figure 4.   As  the  temperature was increased to  36°C  (96°F),  the current
and voltage at sparkover decreased.  However, the sparkover point lies on the
clean VI  curve which  indicates that when the dust layer breaks down, a spark
(rather than  stable back corona)  results.   At about 43°C (109°F) the current
at  sparkover  started  to increase  indicating that  positive ions  from back
corona  were present.   As the  temperature was  increased to 53°C (127°F)  or
above, the  sparkover point lies near the 118°C (244°F) dirty VI curve showing
that  stable back  corona  was  established  on  the collector  electrodes.   The
dust  layer  thickness  on the  cooled pipes affected  the temperature  of the
cooling  water which was  needed  to  operate  the Tiny  TEP at  high  sparkover
voltages and currents.

     The test results  at the  Valmont Station were  similar  to  those obtained
in the  laboratory and at the  Coors  Power  Plant.  The Tiny  TEP  tests showed
that  the  cooled  pipes produced the  desired  electrical  characteristics under
actual flue-gas conditions.   They also showed that the laboratory conditions,
with  reentrained  Coors  dust,   accurately  simulate actual  power-plant stack
conditions.

                             CONCLUDING REMARKS

     The results  of  the  above  test program indicate  that  back corona can be
effectively eliminated  in  the pipe-type  precharger  by  either heating  or
cooling the pipes, the more practical device being when the pipes are cooled.
When  back  corona  is  eliminated,  the  pipes  can  be  operated  at  field
intensities and current densities considerably higher than typical wire plate
ESP's.  The charger also exhibits appreciable collection efficiency in excess
of 50 percent per  pipe section.

                                 REFERENCES

Bush,  P.V.,  D.H.  Pontius,  W.B.  Smith,   and  I.E.  Sparks   (1980),  "Field
Evaluation  of the SoRI-EPA Precharger-Collector  System," presented  at the
73rd  Annual  Meeting   of the  Air Pollution  Control Association,  Montreal,
Quebec, June 22-27.

Drehmel,  B.C.,  C.H. Gooding,  and G.B. Nichols  (1979),  "Particulate Control
Highlights:   Recent  Developments  in  Japan,"  EPA-600/8-79-031a  (NTIS  PB
80-148802), November.

Feldman,  P.L.,   and H.I.  Milde  (1979),   "Pulsed  Energization  for Enhanced
Electrostatic Precipitation in High Resistivity Applications,"  Symposium on
the  Transfer  and  Utilization  of  Particulate  Control  Technology  Volume 1
EPA-600/7-79-044a  (NTIS PB 295226), February, pp. 253-274.
                                     292
-------
UJ
        1.6
        1.4 -
        I43°C GAS TEMPERATURE

     -2.0  7% VOL. HUMIDITY

         -y VI CURVE

          * SPARKOVER VOLTAGE AND CURRENT
        1.2-
UJ
tc. l.O
1C

o

<

O 0.8
K
o
u
           -1.5
        0.6
        0.4
        0.2
           -1.0
           -0.5
                              IO
                                                        kV/cm
                                               2
                                               I
                                           20                 30

                                            CORONA VOLTAGE, kV
40
                                                                                                          50
      Figure 4.  Effect of Water  Temperature  on the Sparkover Voltage  of the Cooled-Pipe Electrodes

                  Operating on an  Actual Flue  Gas Stream.
-------
Gelfand,  P.C.,  D.  Norman,  W.V.  Piulle,  H.W.  Spenser,  and  O.J.  Tassicker
(1980),  "Results  of  High Intensity  lonizer/Precipitator Evaluation  at the
EPRI  Test  Facility,"  presented  at  the  73rd  Annual  Meeting  of the  Air
Pollution Control Association, Montreal, Quebec, June 22-27.

Lausen,  P.,  H.  Henriksen, and  H. Petersen  (1979),  "Energy  Conserving Pulse
Energization  of   Precipitators,"  IEEE-IAS   Conference,   Cleveland,   Ohio,
October.

Masuda,  S.,  M.  Washizu, A.  Mizuno,  and K.  Akutsu (1978a), "Boxer Charger - A
Novel  Charging  Device  for High Resistivity Powders,"  Proc.  IEEE-IAS  Annual
Meeting, Toronto, Ontario, October.

Masuda,  S.,  G.  Obata,   and  J.  Harai  (1978b), "A  Pulse Voltage  Source  for
Electrostatic Precipitators," Proc.  IEEE-IAS  Conference, Toronto,  Ontario,
October.

Penny, G.W.  (1950), British Patent 643,363, September 20.

Penny,   G.W.  (1962),  Industrial  Precipitator  with  Temperature  Controlled
Electrodes.  U.S. Patent 3,026,964,  March 27.

Pontius,  D.H.,  P.V.  Bush,  and  L.E.   Sparks (1979),  "A New  Precharger  for
Two-Stage Electrostatic Precipitation of High Resistivity Dust," Proceedings:
Symposium on the Transfer and Utilization of Particulate Control Technology:
Volume 1.  EPA-600/7-79-044a  (NTIS PB 295227),  February, pp.  275-285.

White,  H.J.  (1952),  "A  Pulse  Method  for  Supplying High-Voltage  Power  for
Electrostatic Precipitation," Trans. AIEE,  November, pp. 326-329.

White,  H.J.  (1974),  "Resistivity Problems  in  Electrostatic  Precipitation,"
J. Air Poll. Control Assoc.   24 4, p.  313,  April.
                                     294
-------
             THE EVALUATION OF NOVEL ELECTROSTATIC PRECIPITATOR
                   SYSTEMS USING A TRANSPORTABLE PROTOTYPE

              By:   G.  Rinard, M. Durham, D. Rugg, J. Armstrong
                    Denver Research Institute, Denver, CO
                            L Sparks, J. Abbott
           U.S.  EPA, Industrial Environmental Research Laboratory
                         Research Triangle Park, NC

                                  ABSTRACT

     A program  is  presently being conducted  to  advance  the development of
economically feasible, two-stage  electrostatic  precipitators for collection
of high-resistivity dust.  The objectives of this program are to:  1) evaluate
alternative precipitator  designs,  2)  optimize the downstream collector for
use with a precharger, and 3) evaluate the SoRI precharger on various sources
of high-resistivity dust.   These  goals will be achieved through the design,
fabrication, and testing  of a transportable  electrostatic precipitator (TEP)
on problem sources of high-resistivity dust.  In order to determine alternative
precipitator designs  for testing in the TEP,  a program was undertaken to
investigate novel  precharger and  collector technology.  The design specifi-
cations for the TEP and configuration  for  the  first series of testing are
presented.

                                INTRODUCTION

     Electrostatic  precipitators  have  historically  been the primary  air
pollution control  device  for controlling particulate matter from coal-fired
boilers.  Precipitators  are used  because they perform reliably  and require
minimum operating  and maintenance costs.  However,  as more industrial  and
utility boilers were  switched to coal  and  the  burning of low sulfur coals
increased, problems arose in the  operation  of precipitators.  The burning of
low-sulfur  coals  produces  a high-resistivity fly  ash  (greater  that 1011
ohm-cm).  High-resistivity ash  leads to back  ionization which greatly reduces
the operating performance of precipitators.   To compensate for this, it was
necessary to build precipitators many times  larger than was previously required,
and the use of precipitators  became less attractive.

     The  Particulate  Technology  Branch  of EPA's Industrial Environmental
Research  Laboratory at Research Triangle Park,  NC,  began performing  in-house
research  and  began funding  contracted  research  to  develop  an  economically
feasible  electrostatic device to  collect high-resistivity dusts.  From the
early stages of this  research, it was  apparent that the effort should be
directed  toward developing a two-stage  system in which the  main  functions of
charging  and  collecting were  separated.   The first phase of research was
directed  toward developing a device to  charge  high-resistivity  dusts.   As
part of this program, EPA and the Southern Research  Institute (SoRI) developed
a novel tri-electrode system to precharge particles before they entered a
collecting device.  This  precharger was tested by SoRI  in the laboratory on
reentrained fly ash and  in a pilot precipitator on  a slipstream  of a utility
boiler.

     To advance the research on two-stage  precipitation, EPA awarded a  grant
to the  Denver Research Institute.   The  main  objectives of  the grant  were  to:
                                    295
-------
     o    Evaluate alternative precipitator designs.

     o    Optimize the downstream collector for use with a precharger.

     o    Evaluate the SoRI precharger  on  other sources of high-resistivity
          dusts.

     To attain these objectives DRI first undertook a program to  investigate
novel  precipitator technology.   This  involved searching  the  literature,
personal contacts  with scientists  and  engineers working  in this  field,  and
continuously reviewing other related EPA funded projects.   Concepts that were
decided to be promising were  then tested in the laboratory and in  some cases
in the  field. In  order to thoroughly evaluate  these alternative precipitator
designs,  a  testing program will be conducted.  This  program consists of:

     o    Design  and  fabrication  of  a  transportable  electrostatic
          precipitator (TEP).

     o    Testing  of  the  TEP  on problem sources of  high-resistivity dust.

     The  design philosophy  of the TEP was  to develop a system which allowed
for  maximum  flexibility,  accuracy, and utility in  the  investigation of
advanced  two-stage electrostatic precipitator  concepts.   The mobility of the
precipitator is important in  order to be able  to try these concepts  directly
on  various  problem sources at  widely  spaced  geographical  locations.    In
addition, the  precipitator had  to  be  large enough to  provide  meaningful
engineering data  for scale-up to full-sized precipitators.  To fairly evaluate
the  novel precipitator design,  its  characteristics had to be compared with
those  of  a standard  Cottrell design.   The  electrical  parameters of the
collector segment of a two-stage unit  may  be  considerably different than
those  of  the standard single-stage design.   In particular the electric field
intensity should  be higher, and the corona  current lower than in the standard
single-stage design.  The  TEP was so designed that it could be reconfigured
from single-stage  to two-stage operation.

DESIGN  SPECIFICATIONS

     A  set of design specifications to define the parameter limits and opera-
tional  characteristics of the TEP were first established.   These specifications
were  then used  as guidelines  in the precipitator design.   Consideration was
given  to  the  type of industry at which the TEP would be tested; the type  of
precipitator technology to be tested;  the physical  size necessary  for producing
data suitable for full scale-up; requirements  for flexibility in reconfiguring
the TEP to different precipitator technologies; practical  operating pressures
and temperatures;  durability;  safety;  and ease of operation.   The  specifications
established are:

     o    The TEP precharger (PC)  and collector  sections  are  of modular
          construction for  ease of shipping and for  flexibility  in  section
          configuration at  various  test sites.  All disassembled  components
          of the  TEP are of  a  size  suitable for transporting on  standard
          flat-bed trailers.
                                    296
-------
o    The high voltage electrode and collector plate assemblies of the PC
     and collector sections are designed for easy installation, reconfig-
     uration, and removal from the TEP.

o    The collector  sections will  accommodate  collector plates 1.5 m
     (5 ft) long and 3.66 m (12 ft) high.  The high-voltage supports are
     designed to  accommodate various  corona-generating structures  as
     well as flat plates.  The channel spacing of a collector section is
     variable and consists  of six channels  at a  typical 22.9  cm (9 in.)
     plate-to-plate spacing.

o    The  precharger sections  are sufficiently  long  to accommodate
     different  precharger technologies.   The  tops  of the precharger
     sections are  independent and easily modified  to accommodate the
     various precharger technologies.

o    Each collecting and PC section is  a separate  electrical  section.
     Power  supplies and controls suitable  for operation with all  the
     various technologies  are:   1) for the collector section operating
     with wire  electrodes,  a minimum of 50  kV and  100 mA; 2) in the
     collector  section  with plates,  a minimum of 100 kV and  10 mA; 3)
     for the precharger section with pipe electrodes,  a minimum of 50 kV
     and 50  mA; 4)  for  the  SoRI  precharger  grid, a  minimum of 15 kV and
     200 mA; and 5) specially constructed pulsing supplies  for operating
     collector  sections with pulse excitation.

o    The maximum TEP operating temperature  is 427°C (800°F).

o    The TEP operating gas  flow range is 4.72-7.08 mVsec (10,000-15,000
     acfm) with average gas velocities of 0.91-1.5 m/sec (3-5 ft/sec).

o    At  a  22.9  cm  (9  in.) plate-to-plate spacing, the SCA per collector
     section is 14.2 sec/m  (72 ftVlOOO acfm) and 9.5 sec/m (48 ftVlOOO
     acfm)  for  flow rates of 4.72 mVsec (10,000 acfm) and 7.08 mVsec
     (15,000 acfm), respectively.

o    All TEP  components are so constructed  to be  able to  withstand a
     minimum  negative pressure  of 6.2 kPa  (25 in.  H20).  A centrifugal
     blower  is  utilized to overcome all pressure drops in the gas flow
     system.

o    Supplemental electrical heating is provided to maintain ducting and
     precipitator temperatures  during operation  and  to preheat the TEP
     system upon start-up.

o    Three  10.2 cm  (4  in.)  sampling  ports are  located on the  downstream
     side  of each precharger section:   one near the  top,  one in the
     middle,  and one near  the bottom.  Other ports  for temperature
     measurement and gas  sampling will be provided  as required.

o    A  portable sampling trailer is provided for  the purposes of gas
     sampling,  particulate  sampling, data acquisition, and data reduction.
                                297
-------
     o    A portable instrumentation trailer is provided  for  controlling the
          TEP,  automatically  recording TEP operating  characteristics, and
          reduction of all  TEP data.

     o    Since the TEP utilizes hazardously high voltages, safety  interlock
          systems and  procedures  have been  established  to  eliminate any
          electrical hazards to the  operating  crew or any other  personnel.

TEP DESIGN

     A formal Preliminary Design Review was held between the project officers
of EPA and appropriate DRI personnel.   A comprehensive presentation  of  the
conceptual  design  alternatives was  conducted  and  design recommendations
presented.   Joint  decisions were made concerning design  philosophy,  major
operational  objectives,  and data  acquisition  and processing.   Work on the
final design was then  started.  The final design configuration  is the  result
of the evaluation  of  the new technology, site surveys, provisions  for data
acquisition and processing requirements,  operator safety and convenience, and
the specifications of operating parameters with the  established design goals.

     The main TEP assembly is presented in Figure 1  which shows the elevation
of the TEP unit including the inlet and  outlet  duct  components.    In  this
drawing the  TEP  has four collector sections each preceded by a PC  section.
Each  collector section includes dust  hoppers  and  an  overhead  compartment
which houses the high-voltage feed-throughs and hanger supports for plate and
corona frame assemblies.  As stated above, the  collector plates will be 1.5 m
(5 ft) long  and  3.66 m (12 ft) high.   Each collector section  is 2.1 m  (7  ft)
long.  The final section will house a device to control rapping reentrainment.

     Each  PC section  is  designed so that  it can be  attached to  either  end of
a  collector  section or another precharger section.   The precharger sections
are  capable  of accommodating various  precharging concepts.   These  sections
also  include hoppers  and overhead  compartments for rappers, hanger supports
for  plate and/or grid  assemblies,  and HV  feed-throughs.   The precharger
segments are 0.91 m (3 ft) long.

     The direction of gas  flow in this drawing is from the left to the right.
Gas  is slipstreamed from  the  source being sampled and  is  conveyed to  the  TEP
through  a thermally insulated circular duct having an inside  diameter  of
0.61 m (2  ft).   This  duct is attached to the  first TEP  inlet cone segment
where the  dust-laden  flue gas begins  to  decelerate in a  controlled manner.
The  TEP  entrance cone  design  has been  established through air flow  modeling.
The  results  of the modeling effort show  that  good  quality gas flow  which
meets the  IGCI standards can be maintained within the full-size TEP using a
combination  of inlet  diverging cones  having half angles  of 15° and 20°  and
two 40 percent open diffuser plates.  Once the gas passes through the  TEP, it
enters two 45° half angle converging  cones which in turn are connected to a
0.61 m (2  ft)  ID duct which returns the gas to the  source being sampled.  The
exterior  surfaces  of  the inlet and outlet cones are thermally  insulated.  A
centrifugal  blower  is  located in the  return duct to overcome pressure drops
through  the TEP and  the  delivery  and return  ducting.   A bypass  branch
connecting,  the inlet  and outlet ducting  is located at the blower position.
An electric  heater in the bypass branch  is  used to preheat  the  TEP  during
                                     298
-------
Figure 1.  Elevation Views of the Transportable Electrostatic Precipitator.
-------
start-up to prevent condensation in the TEP.  Valves  are  used  to  control  the
gas flow through  the  TEP and to direct the  gas through  either the bypass
branch in a closed  loop mode or to sample gas from the source and  to return
it.  Atmospheric  air  ports  with control  dampers  in  the closed-loop system
allow for rapid cooling of the system during shutdown.

     A  scaffolding  support  structure extends outward  toward  the  front and
back of the main unit and is accessible by means of stairways.  This structure
is more clearly shown by the  end elevation (lower left).   On  the  back of  the
precipitator,   commercial  rolling scaffolding  is  erected  for  purposes  of
sampling and  access  to the  upper portions of  the precipitator.   The front
section of each collector section is accessible by means of doors as shown in
Figure 2.   A specially designed dolly rides along guide rails on the scaffolding
support at the front of the precipitator.  This  dolly may be  positioned  in
front of each  collector section with collector doors open.  Fold-down rails
from the dolly are then placed in position to allow scaffolding riding on the
dolly to be moved into the collector  section.  This  allows access  into the
collector  section for  the  purposes  of installing and  modifying  collector
plate and  corona  electrode  components.   A cantilevered outboard  support  is
used for installing and removing the internal components of the precipitator.
All  internal  components are mounted on slotted track.  With the doors open,
extensions of  the slotted track may be installed and supported by the outboard
support.  When installing the internal components of the precipitator,  each
component  is  brought into position  beneath the  outdoor support by means of a
fork  lift.  The component is  then attached to the support and  moved into  the
precipitator.    The  process  is  reversed when  removing internal components.

     Each PC and collector  section has a separate hopper to minimize sneakage.
Further reduction in  sneakage is provided by simple baffles.  Rotary air  lock
valves  are  used  at  the outlets of the hoppers so that collected dust can be
removed  independently from each  hopper  while the TEP is operating.  The
hoppers, which are thermally insulated, are equipped with dust level indicating
devices and vibrators.  The hoppers are emptied by means of two screw conveyors.

Figure 2 shows the collector section assembly.  The overall outside dimensions
of the collector assembly are 2.1 m by 2.1 m by 8.2 m  (7 ft by 7 ft by 27 ft)
tall.  The unit is constructed of standard structural  steel components.    Each
outside wall  of the  collector assembly contains 4 in.  of  internal fiberglass
insulation.   The  primary feature of the collector section shown in Figure 2
is the  access doors.   These doors are  suspended by specially  designed hinges
allowing the  doors  to close parallel  to  the  sides of  the precipitator  in
port-hole  fashion.   This simplifies  gasketing and  sealing of the doors.
These hinges  also allow the doors to open completely  and  lie  flat  along  the
side  of the  precipitator in the open  position.   The doors are secured  when
closed  by  means  of  quick release clamps  along  all sealing surfaces of  each
door.   Since the precipitator is designed for operation under  negative pressure,
these simple  door closures  are sufficient.   Sealing  is accomplished on all
sealing surfaces  of  the doors by means of a  heat  resistant,  covered, Inconel
mesh  (tadpole) gasket.   In  addition the  doors  are restrained by  means  of a
rigid plate  to prevent  complete collapse  of the  gaskets.  These types  of
gaskets were  chosen  to provide the sealing necessary  and  yet  allow for some
                                     300
-------
misalignment due  to  fabrication of the mechanical components.   All panels of
the collector section, as well  as  the  rest  of  the TEP  unit,  are  reinforced  to
withstand a  total  negative pressure of  6.2 kPa (25 in.  H20) and 44.7 m/sec
(100 mph) wind  loading.   Stress analyses have been conducted on all exterior
components to ensure that the TEP  is structurally sound.

     Figure 2 also shows the support mechanism for the high-voltage electrodes
and the  collector plate.   The  voltage  insulators are especially designed and
fabricated by Coors  Porcelain of Golden,  Colorado.   High  voltage  is conducted
to  the  high-voltage carrier channel by means of  the  metal  insulator rod.
Support for the collector plates is very  similar to that  for the  high-voltage
electrode except  that no high-voltage  insulation is required.   The collector
plate suspension  is  extended through  the top of the precipitator in such a
manner that  load  cells may be  used to measure the weight of each collector
plate and thus determine the total  accumulated dust loading.  The  high-voltage
electrode and  collector-plate  carrier channels  are so designed as to allow
completely  variable  spacing of the  high voltage and  collector  electrodes.

     The  construction of the precharger section is very similar to that of a
collector  section.   The  various  components of  the precharger  section  are
assembled  by bolting  them to  the ends of the  collector  section  or flow
straightening  units.   Access to the precharger sections  is by means of the
doors  on the  collection sections  and removable panels  on  the  precharger
sections.   This  is necessary due  to the geometric shape  of the  precharger
section.  With all  high-voltage electrode and collector plates removed from,
and the  scaffolding  inserted  into, the  collector  section,  the  precharger
section  on  either side of  this collector section  is completely accessible.
The precharger assembly  is  designed to allow  complete  flexibility  for the use
of  various precharger concepts.  The only component of the precharger assembly
requiring redesign for a new type  precharger  is the top section.   This  section
is  easily removable from the rest of the precharger section.  Except for the
shape, the  hopper of  the  precharger assembly is identical  for the collector
section.

     As  stated above,  the  TEP  is designed to  operate to temperatures  of 427°C
(800°F).  This will  allow  the  system  to be operated in either a cold-side,
149 to  204°C (300 to  400°F),  or a hot-side,  371 to 427°C (700 to 800°F),
mode.  In addition,  the system is capable of operating at sites  located at
sea level to altitudes of  1980  m  (6,500  ft).

     The  rapping  system which will be  employed in  the collector sections for
the collector  plates and the high-voltage wire frames or high-voltage plates
will be  series of hammers which strike the individual  plates or wire frames.
The selection  of  the  hammer system is based  on  the results of laboratory
tests which  evaluated the  effectiveness  of using pneumatic impactors to rap
complete  fields  of dust laden plates versus the individual hammer rapping of
each plate.

     The  collection  of reentrained dust  is the reason for the provision of
having  PC sections  preceding  each collector  section.   Dust from upstream
collector and  PC  that is  returned to  the gas stream due to rapping must be
recharged by downstream  PC  sections so that it can be  effectively recaptured.
The TEP can be operated, however,  with selected PC sections either de-energized
or  empty.                            301
-------
                          IT*
                                               A
Figure 2.  Side and Front Views of a Typical Collector Section.
                                    302
-------
TESTING AT THE VALMONT POWER PLANT

     Work on  the  task of selecting  appropriate test sites commenced early in
the project.   A  survey of selected  industries was  made  for determination of
design parameters  for the TEP.  Design parameters  were  selected which would
allow the operation  of the TEP in  virtually all  industries except  the cement
industry.  The excessively high  negative  pressures  encountered  in  this industry
imposed an unreasonable  constraint on  the design  of the  TEP.

     Contact was made with officials of the  Public  Service Company of Colorado
(PSCo) concerning  the possibility of  the utilization  of one of PSCo's power
generating  stations  for the first TEP test  site.   All of PSCo's plants were
considered and the 180 MW  Unit 5 at  the Valmont Station  in Boulder,  Colorado,
was selected.  This  is a base-loaded unit.   Both  in-situ and laboratory tests
have  been made on unconditioned fly ash  at  Valmont.   The ash resistivity at
cold-side temperatures  is  approximately 1013Q cm.

      Site preparation at Valmont is nearly  complete.   The major portion  of
the  TEP  has  been  constructed.  Final assembly  of the  TEP at  Valmont  is
scheduled for July 1981.  The slipstream at Valmont will be taken  before  the
preheater and cooled to cold-side temperatures by means of a heat  exchanger.
This  will  allow very flexible control of the operating  temperature of the
TEP.

      The plan is to remain at Valmont for a series of evaluation and optimi-
zation testing.   The first series of tests will  consist of a SoRI  precharger
followed by two collector sections  with  parallel  plate  electrodes,  a second
SoRI  precharger,  again  followed  by  two collector  sections with  parallel plate
electrodes,  and  finally  a  third  SoRI precharger as  the last TEP section.  The
TEP  configuration for the second  series  of tests will consist of  replacing
the  first  and second SoRI prechargers with cold pipe charger-collectors.   In
this  manner  a direct comparison  will be made between the effectiveness of the
SoRI  and cold pipe  concepts.   It  is envisioned that two additional  units of
whichever charger  concept  proves to  be the better will  be built in the future
for  inclusion in  the  initially  empty charger sections  of the  TEP located
between  the  first  and second  and the third and fourth TEP collector sections.
The  final  series  of tests at Valmont will  be conducted to  optimize the
precipitator parameters.

FUTURE TESTING

      The configuration of the TEP  as  established during the Valmont program
will  be  tested at various  sources  of high-resistivity dust.   Present plans
are  that the next two  sites  will  be  one more power plant and a steel mill.
Testing  at  each  site will be  long enough  to collect  data and optimize
precipitator performance characteristics  for each source.

      Further operation at these sources will be conducted to demonstrate the
technology  for each application.  During this phase of the program, transfer
of this  technology  to potential manufacturers and users will  be  pursued.

                                     303
-------
           ANALYSIS OF THE ELECTRICAL AND CHARGING CHARACTERISTICS
                       OF A THREE ELECTRODE PRECHARGER

          By:  K.J. McLean
               University of Wollongong,
               Wollongong.  N.S.W.  2500, Australia.
                                  ABSTRACT

     The operation of the three electrode precharger is analysed using a
comprehensive set of electrical and charge/mass measurements.  The electrical
measurements are made with clean and contaminated electrodes, smooth and
perforated grids, and for a wide range of grid and corona voltages.
Particular consideration is given to the magnitude and direction of the grid
current flow.  It is concluded that biasing the grid reduces the back corona
component of the total current in the main gas stream for a restricted range
of voltages and that the main action of the biased grid is to suppress or
reduce the formation of back corona rather than to act as a collector of the
positive ions.
                                INTRODUCTION

     One of the most serious limitations of electrostatic precipitators is
that their performance deteriorates when the particulate being collected has
a high resistivity.  A phenomenon, commonly called 'back corona', is formed
on the deposited dust layer which degrades the precipitator's performance by
reducing the sparkover voltage, producing reverse charging positive ions,
distorting the electric field pattern and probably increasing particle re-
entrainment.

     A wide range of techniques have been investigated to overcome this high
resistivity prolem of which one important group is the precharger (1,2,3,6).
In the precharger, the effect of back corona is minimised by either reversing
the field direction (1), introducing a biased third electrode to suppress
back corona (2), using high gas velocities to prevent particle deposition (3)
and reducing the resistivity of the ash on the collecting electrode by
changing its temperature (6).

     One of these units is a pilot scale precharger,  designed by Southern
Research Institute and installed at the U.S. Environmental Protection Agency's
Industrial Environmental Research Laboratory, Research Triangle Park  N.C.
The unit is described elsewhere (2) and some of its overall performance
characteristics have been published (4).

     The purpose of this paper is to establish a technique by which the
operation of this type of precharger may be analysed from measurements of the


                                     304
-------
electrical parameters.

     The specific results recorded in this paper are  applicable only to this
precharger, to the particular high resistivity fly ash  tested and other
operating conditions  stated in this paper.  However,  the  technique of
analysis described should be universally applicable and the basic mode of
operation for other units collecting different ashes  can  be expected to
follow a similar pattern.

                            CURRENT FLOW PATTERN

     The main currents  that can be expected to flow in  the precharger are
shown in Figure 1.
           v
                           T
                           t
                             / 1
                                    lc
                                                NEGATIVE SUPPLY
                               PRECHARGER WALLS

                             "I
                                    '9   \  GRID

                                   Vg"P^
                                        hb
             GROUNDED
PLATE ~f  ^77    DEPOSITED DUST LAYER
         Figure  1:   Diagrammatic Representation of Currents Flowing
                     in Precharger.
                                     305
-------
Where:  Ic       Total corona current measured by an external meter
        Ig       Total grid current measured by an external meter
        Ic-      Total negative ion corona current to the grid
        Ix-      Stray negative ion current
        Ig-      Corona negative ion current terminating on the grid
        Ip-      Corona negative ion current passing through the grid to
                   the plate
        Ix+      Back corona positive ion current due to stray current
        Ig+      Back corona positive ion current from grid
        Ip+      Back corona positive ion current from the particle layer
                   on the plate
        Ipg+     Part of Ip+ collected by the grid
        Ipc+     Part of Ip+ which passes through the grid openings into
                   the gas space
        Igb      Grid to plate breakdown current
        Igbc     Proportion of Igb injected into gas space
        Igbg     Proportion of Igb collected by the grid

The resistance Rfo is necessary to provide an impedance path for grid current
when  it flows from the supply to the grid.  If this is not present, the grid
power supply loses control of the grid voltage level.

      The voltages present are:

        Vc       Voltage of corona wire to ground
        Vg       Voltage of grid to ground
        (Vc-Vg)  Gas gap voltage

Some  important current relationships are:

        *§ = V + Jg+ - ^Pg+ + Xgbg)                                   W

        Ic = Ix- + Ic- + Ipc+ + Ig+ + Igbc + Ix+                         (2)

It is of particular importance to note the direction of the current flow in
the grid circuit.  If there is only corona negative ion current Ig- and back
corona off the grid Ig+, an ammeter connected in the circuit as shown will
give  a positive reading. The presence of Ipg+ or Igbg will at first reduce
this  value and then reverse the direction of the meter reading.  Hence, the
sign  and the direction of variation in magnitude of this grid current can be
used  to determine which back corona currents are present.

      The relative magnitudes of these currents will vary over a wide range of
values and in some instances, certain currents will be insignificant and may
be neglected.

                           EXPERIMENTAL RESULTS

Electrical Transparency

      This is one of the most important parameters that affects the operation
of the precharger.  It is defined as the ratio of the current passing through

                                     306
-------
the grid to the plate, divided by  the  total  current approaching the grid.
Since it is not possible to measure  the plate current directly, the electrical
transparency is given by:
             Ic - Ig
             Ic - Ie
                                                         (3)
where Ie is an estimate of the current  flowing  to the smooth edges of  the
grid.  Measurements were made with  clean electrodes at 300°F and  the results
are shown in Figure 2.  The characteristics  show a dependence on  current
density.
                100
                80
               cc.
               <
                60
               OL
               \-
               O
                20
                  0.1    0.2      0.3      0.4
                                   vg/(vc - vg)
                               0.5
0,6
                                              0.7
          Figure 2:  Electrical Transparency.
     A.  Vc-Vg = 11 kV  Ic
     C.        = 15 kV
             130 yA  B.  Vc-Vg = 13 kV  Ic =  370 yA
             650 yA            = 17 kV     = 1000 yA
     A careful set of laboratory  scale  measurements  and a theoretical
analysis (5) shows that the electrical  transparency  for a grid  of  constant
optical transparency is a function of Eg/E0  and is given by:
                                                 If*
where:  E0
        K
                       tan
                    fig tairl
                                                      cot
the electric field just above the grid
the electric field between the grid and plate
constant
angle at which the field direction changes
                                      307
-------
In an ideal case, all the results plotted with Eg/Eo  as  an abcessor should
fall on a single curve.

      The reason for the electrical transparency  curves  of the precharger not
coinciding for all current densities is probably  due  to  variations in the
values of Eo and Eg caused by the changes in space charge  with current
density.

Smooth Grid Electrical Characteristics

      The perforated screens of the precharger grids were  replaced with  solid
smooth aluminium sheets and the corona characteristics measured for both clean
and contaminated grids.  One set of results are plotted  in Figure  3,  for
Vg = 5 KV.
             3,0
                       10
11      12      13

 GAS GAP VOLTAGE  (Vc - vJ ,  KV
15
16
          Figure  3:   Total  Corona Characteristics  -  Vg = 5 kV.
          A.   Clean  smooth  grid.             B.   Clean perforated grid.
          C.   Contaminated  perforated  grid.  D.   Contaminated smooth grid,

                                     308
-------
Perforated Grid Electrical Characteristics

     The perforated screens were refitted to the grid and the clean  electrical
characteristics measured.  Dust was  injected with the grid biased  as strongly
as possible so as  to reduce the amount  of dust deposited on the grid itself.
Conditions were allowed to stabilise and a range of electrical measurements
made.  One set of  typical results are plotted in Figures 3 and 4.
            1500
            1000
             500
          o:
          o:
          13
          u
            -1000
            -2000
            -3000
                                   INCREASE DUE TO GRID PLATE
                                   CIRCULATING CURRENT
                                                  •T-.
                                                   6  i   7
GRID VOLTAGE, Vn
                                 GRID TO PLATE BREAKDOWN ZONE
          Figure  4:   Typical Total Corona and Grid Current - Vc-Vg =  13 kV.
                      A.  Total corona  current with clean electrodes.
                      B.  Total corona  current with contaminated electrodes.
                      C.  Grid current  with contaminated electrodes.
                                       309
-------
 The characteristics for other (Vc-Vg) values had this same general shape.
 Once the grid-to-plate breakdown occurred, the circulating current could be
 easily increased to about 10 mA and was independent of  (Vc-Vg).  This voltage
 could even be reduced to zero and the breakdown current would  still flow if
 Vg is maintained.  In all cases, the total corona current increased with the
 magnitude of the circulating current.  The total back corona current appears
 to be controlled by grid biasing for (Vc-Vg) up to a maximum value which is
 approximately 14 kV for the conditions of the test.  At higher values (i.e.,,
 16 kV) the back corona current is substantially higher than the negative ion
 current.

      Conditions were then established at ambient temperature to allow the
 grid to become very dirty.   The electrical characteristics were then measued
 for various conditions at 300°F.   A typical set of results are plotted on
 Figure 5.
               2000
               1500
             £1000
                500
                                 GRID VOLTAGE, Vg, KV


          Figure 5:  Typical Corona and Grid Currents for Very Dirty
                     Electrodes - Vc-Vg = 13 kV.

Charge/Mass Measurements

     The Southern Research Institute probe was used to make these charge/mass
measurements.  The input to the probe was located 50 mm off the centre line
of the middle electrode, halfway up the duct immediately following the
precharger.  The total charge was measured by a 610 C, Keithley Electrometer
and samples were taken over a three minute period.
                                     310
-------
     In order to measure the effect  of  (Vc-Vg)  and hence the relative
magnitude of the back corona current, the grid  voltage was fixed at 5 kV and
Q/M measured for increasing values of  (Vc-Vg).   The results are shown plotted
in Figure 6.

     Similar measurements were also  made with the grids in a very dirty
condition corresponding  to the electrical conditions which are also shown
plotted in Figure 6.
                            10       12      1H       16
                             GAS GAP VOLTAGE (Vc - Vg), KV
          Figure 6:  Variation of Q/M with Gas Gap Voltage - Vg
                     A.  Moderately dirty grid.
                     B.  Very dirty grid.
= 5 kV.
                                   DISCUSSION

      From  these  tests  it may be concluded that the biased third electrode
  placef c°lose  To  thf c^llectin-g plate reduces the magnitude of the back corona
  compared with that in a conventional two electrode precipitator.  This
  conclusion is supported by the following:

  (1)  Comparison  of the total corona currents of the contaminated smooth
      and perforated grids as shown in Figure 3.  For all values of
       (Vc-Vg)  the total current is less with the perforated biased grid.

  (2)  Variation of the total corona current with Afferent grid biases
      as  shown in Figure A.  For a limited range of  (Vc-Vg) and to a
                    voltage Vg, the total corona current is not much

                                        311
-------
     greater than the clean electrode negative ion current,
     indicating a significant reduction in the generation of back
     corona current.

(3)  Variation of total corona current with grid bias for very
     dirty grids, as shown in Figure 5.

(4)  The Q/M measurements show that this ratio increases with
     increasing magnitudes of (Vc-Vg) to a maximum at about 14 kV
     and then begins to reduce.   The turnover corresponds exactly
     with the value at which the back corona current begins to be
     a significant proportion of the electron current as shown in
     Figure 3.  Charging is significantly lower when the grids are
     contaminated.

     Although the formation of back corona can potentially take place from the
grid surface and the plate, a careful examination of the magnitude and
direction of the grid current, for a relatively clean grid, indicates that the
grid is not emitting a significant proportion of positive ion current nor is
it receiving a significant positive ion current from the plate.  Hence, what
positive ion current is being measured by Ic, must be coming direct from the
plate and passing through the grid aperature to the discharge electrode;
following the same flux lines that bring in the negative ions.

     When breakdown occurs between the grid and the plate, a substantial
current circulates in the grid loop.  It is suggested that this breakdown is
initiated when the electrical transparency is high, about 80%, and some of
the positive ion back corona current is sufficiently close to the grid to be
captured by it.  Once this occurs a low impedance path is generated and a
large current is circulated by the grid power supply.

     No significant part of this current is generated by the incoming corona
negative ions.  The measurements however do indicate that some proportion of
this current escapes into the gas space and flows to the discharge electrode.
This breakdown occurs just on one grid and it is not unreasonable to specu-
late that it is very localised.

                         CONCLUSIONS AND COMMENTS

     For the successful operation of the three electrode precipitator as a
precharger it is necessary to carefully adjust the gas gap and grid voltages.
The tests show that there is a mode of operation in which the back corona is
suppressed and good charging of the inlet particles is possible.

     An alternative application of the triode precipitator is to use it in a
hybrid system as the final collecting stage.  Its main function would be to
provide a high collection efficiency zone which can charge and collect the
reentrained fly ash.   The aerodynamic shielding and electric field pattern
produced by the grid should reduce any reentrainment when this last zone is
rapped.
                                     312
-------
                               ACKNOWLEDGEMENT

     This project was funded by the U.S. Environmental Protection Agency,
Contract No.68-02-3143.  The authors are grateful for the assistance given
by G.H. Ramsey, B.E. Daniel and R.E. Valentine in carrying out the tests.

                                   ENDNOTES

1.  Masuda, S., Nakatani, H., and Mizuno, A.  Boxer-charger, A Novel
    Charging Device.  (Presented at the EPA-Symposium on Transfer and
    Utilisation of Particulate Control Technology, Denver, Colorado, 1979).

2.  Pontius, D.H., Bush, P.V. and Smith, W.B.  Electrostatic Precipitators
    for Collection of High Intensity Ash.  Environmental Protection Agency
    Report No. EPA-600/7-79-189, August, 1979.

3.  Air Pollution Systems, Inc.  Development Program for an Ionizer-
    Precipitator Fine Particle Dust Collection System as Applied to Coal Fired
    Utility Steam Generators:  Final Report, Vol.1:  Technical and Economy
    Summary.  EPRI FP-291.  Palo Alto, C.A. 1976.

4.  Sparks, L.E., Ramsey, G.H. and Daniel, B.E.  Collection of Fly Ash with
    High REsistivity in a Pilot Plant Electrostatic Precipitator Preceded by
    the EPA/SORI Precharger.  Air Poll. Cont. Ass., 29: 745-7, 1979.

5.  McLean, K.J., Herceg, Z. and Boccola, R.I.  Electrical Transparency
    of a Corona Triode.  Jn. Electrostatics, 9:  211-222, January, 1981.

6.  Rinard, G. , Durham, M. , Rugg, D.  Development of a Charging Device for
    High Resistivity Dust using Heated and Cooling Electrodes.  (Presented
    at the 3rd EPA-Symposium on Transfer and Utilisation of Particulate
    Control Technology, Orlando, Florida, 1981).
                                      313
-------
PARTICLE CHARGING IN AN ELECTROSTATIC PRECIPITATQR BY PULSE AND DC  VOLTAGES

   By:  L. E. Sparks, G. H. Ramsey, R. E. Valentine, and J. H. Abbott
        Industrial Environmental Research Laboratory
        U. S. Environmental Protection Agency
        Research Triangle Park, N. C. 27711

                                 ABSTRACT

     Measurements of particle charge as a function of particle radius
for dc and pulse energization were made at the exit of a pilot plant
electrostatic precipitator.  A Milliken cell was used to make the measure-
ments.  Measurements were made at three values of dust electrical
resistivity—low (2 x 10   ohm-cm), moderate (2 x 10   ohm-cm), and high
(2 x 10   ohm-cm).  All experiments were conducted using a resuspended
flyash.  In order to compare the charging characteristics of dc and
pulse supplies, the average current densities for both cases were kept
identical.  The results are compared with theory and previous experimental
data.  The data show that there is no difference in particle charge for
pulse and dc when the dc is operated without back corona.
                             INTRODUCTION

     Pulse power for electrostatic precipitators has been offered as a
possible solution to the problems associated with collecting dust with
high electrical resistivity_in electrostatic precipitators.  Petersen and
Lausen,  Feldman and Milde,  and others have presented data to show a
significant improvement in electrostatic precipitator performance for
pulse versus dc power supply.

     Because pulse power appeared to have significant performance
advantages over dc, an experimental program was developed to investigate
pulse power in cooperation with Denver Research Institute.
Denver Research designed and built the pulse power supplies and then
delivered them to the Particulate Technology Branch of EPA's
Industrial Environmental Research Laboratory at Research Triangle Park,
North Carolina (IERL-RTP).  All of the experiments were conducted in the
IERL-RTP in-house electrostatic precipitator.

     The experiments discussed in this paper were designed to compare
the particle charging characteristics of pulse and dc when the electro-
static precipitator was collecting dust with high, moderate, and low
electrical resistivity.  The results of these experiments are presented
in this paper.

                        EXPERIMENTAL EQUIPMENT

     The experiments were conducted in a single-lane pilot-plant electro-
static precipitator located at IERL-RTP.  The electrostatic precipitator
has been completely described by Lawless et al.  so only a brief descrip-
tion will be given here.

                                     314
-------
     The electrostatic precipitator  is  shown in Figure  1.   The electro-
static precipitator is a single-lane, four-electrical-section electrostatic
precipitator.  For these experiments Sections 1 and  2,  and  Sections  3
and 4 were electrically connected  so that  only two power  supplies were
needed to energize the electrostatic precipitator.

     The pulse power supply was  designed and constructed  by Denver
Research Institute.  The pulse height and  pulse rate were adjustable
from 0 kV to about 50 kV and  from  0  to  about 110 pulses per second.  The
pulse width was about 100 ysec.  A  plot  showing the pulse  shape and width
is shown in Figure 2.

     A of the pulse power supply is  shown  in Figure  3.  A capacitor  in
the primary of the 50:1 step-up  transformer  is charged  to a voltage  V
which determines  the pulse amplitude.   The SCR switch is  triggered by°a
timing circuit which determines  the  pulse  repetition rate and the capacitor
is connected to the pulse transformer for  one cycle  of  primary current.
The primary capacitor discharges during the  first half  cycle of the
primary current and recharges during the second half cycle.

     The output pulse was capacitor  coupled  to the corona discharge
electrode system.  A dc bias  voltage, which  was variable  from 0 to 60
kV, was added to  the pulse voltage through a blocking diode.   Therefore,
the pulse power supply permitted independent control over peak and
average voltages.

     For these experiments the pulse height  and pulse rate  were adjusted
to give the highest allowable current density at the plate  that did  not
give rise to sparking or back corona.   The dc bias voltage  was set below
the corona onset  voltage.  The exact conditions depended  on the dust
resistivity and are given later.

     The aerosol  used for the experiments  was flyash which  was redispersed
using sandblast guns fed with a  vibrating  screw feeder.   Previous experience
has shown that this is an excellent  method of producing an  aerosol.

                          MEASUREMENT APPARATUS

     The charge and diameter  of  individual particles were determined
using a technique and apparatus  developed  and described by  McDonald  et
al.  The technique consists of extracting  a  sample gas  volume from the
ESP and directing part of the gas  flow  into  a modified  Milliken measurement
cell.  A single particle is then isolated  and its upward  and downward
motion under the  influence of a  uniform electric field  can  be measured.
The particle charge and diameter can be determined by knowing the time
required to move  up and down  a given distance in the electric field.

     Figure 4 shows a drawing of the apparatus inserted into an ESP.
The gas sample is extracted through  5 cm diameter tubing.  Because the
technique is limited to particles  with  diameters less  than  4 ym, isokinetic
sampling is not required.  The tubing leading to the test chamber contains
a butterfly valve which controls the flow  of aerosol into the test

                                     315
-------
chamber.  The valve is opened when a sample is acquired  and  then closed
while the measurements are being made.  When  the valve is  open,  the pump
is turned on, and an aerosol sample is obtained.  When a sufficient
number of particles are in the measurement cell, the pump  is turned off
and the valve is turned off.  A very short time after the  valve  is
closed, particle motion due to the gas flow ceases and the motion of the
particles is controlled solely by the gravitational field, viscous  drag,
and the electric field in the measurement cell.

     Measurements are made only after all motion due to  the  gas  flow has
ceased.

     After the particle motion due to gas flow has ceased, the time
required for an individual particle to travel a known distance up and
down in the electic field is measured.  Typically the up and down time
for a single particle is determined 3 to 4 times.  A new particle is
then selected and the process is repeated.  Eventually,  the  particle
concentration in the cell becomes very small, and a new  sample is obtained.

     Data from the Milliken experiment are reduced on a  TRS-80 micro-
computer.  The computer calculates the particle charge and diameter for
each measurement.

                            TEST PLAN

     The experiments were designed to provide data on particle charge as
a  function of particle diameter for dc and pulse under three conditions
of electrical resistivity of the dust—low resistivity (2  x  10    ohm-
cm),,, moderate resistivity (2 x 10   ohm-cm), and high resistivity  (2 x
10  ohm-cm).  The electrical resistivity was varied by  changing the
operating temperature of the electrostatic precipitator.   The low
resistivity runs were made at ambient temperature (23°C),  the moderate
resistivity runs at 218°C, and the high resistivity runs at  150°C.

     The high resistivity runs were made with matched current densities
at the  plate.  The current density was about  5 na/cm  .   Operation above
this current density resulted in back corona  and poor performance for
both the pulse and dc cases.  It was found that stable operation at this
current density allowed both the pulse and dc to operate in  optimum
fashion.

     The moderate resistivity runs were made  again at matched current
densities at the plate.  The exact current density to use  was determined
by adjusting the pulse height and pulse rate  to give  the lowest  reading
on the  opacity monitor.  The current density  at the plate  for this  case
was 12  na/cm  which was found to be the best  operating point for the dc
also.   The low resistivity experiments were conducted to determine  if
pulsing significantly affected the charge on  the particles either due to
the high pulse voltage or due to the pulse of ions produced  by  the
pulse.  Consequently, the experiments were conducted at low current
densities similar to those used in the high resistivity  experiments.  By
conducting the experiments with low resistivity dust, we were able

                                     316
-------
to compare the effects of pulsing without  the confusing  factors  of  back
corona that existed when high resistivity  dust was  used.

     It was quite easy to operate the  pulser  at  the desired  current
densities for all cases.  The dc was extremely difficult to  operate at
the desired current density  for the high resistivity case.   Even with
extreme care, the dc tended  to go into back corona  at unpredictable
times.  Whenever the dc went into back corona, particle  collection  was
reduced.  The effects on particle charge are  discussed later.

                               RESULTS

High Resistivity Dusts

     The results for the high resistivity  runs are  shown in  Figure  5.
The most striking feature for the data is  that there is  no difference
between the charge on the particles under  conditions of  pulse and dc
energization of the electrostatic precipitator.  The charge  on the
particles is in very good agreement with the  charge predicted by a  model
based on the sum of the charging rates due to diffusion  and  field charging.

     It is important to note that the  results shown in Figure 5  are the
charge on those particles that have charge.   Whenever the dc went into
back corona, significant numbers of uncharged particles  were observed.
But even under these conditions of back corona,  those particles  that had
charge were charged to the expected values of charge.  This  behavior is
consistent with the idea that the dc current  density under conditions of
back corona is very nonuniform.  Consequently, particles  are randomly
exposed to areas of current  or no current.  Those particles  that are
exposed to current are charged as expected, while those  particles that
are not exposed to current are not charged.

     All of the particles for the pulse charging experiments were charged
all of the time.  This is consistent with  the data  that  showed that the
current distribution at the  plate was  very uniform under pulsing conditions.
Thus all of the particles are exposed  to charging current.

     The particle charge data that show that  the particle charge is the
same for both dc and pulse are consistent  with particle  collection  data.

     The main value of the pulse power supply appeared to be that it
allowed stable operation for long periods  of  time with the high  resist-
ivity dust.  Such operation  was not possible  with the dc.  It should be
emphasized that as soon as the dc was  allowed to operate in  back corona,
the particle collection efficiency of  the  electrostatic  precipitator
dropped and significant numbers of uncharged  particles were  observed.

Moderate Resistivity

     The results of the moderate resistivity  experiments are shown  in
Figure 6.  Again note that there is no significant  difference between
the dc and pulse data.  The  standard deviation for  the particle  charge

                                    317
-------
at a given particle diameter is about 25 percent as was the case  for  the
high resistivity data.

     These particle charge results are consistent with the measured
collection efficiency of the electrostatic precipitator as shown  by
Figure 7 which is a plot of penetration (1-efficiency) versus particle
diameter.  There is no significant difference between the data for dc
and pulse.

     In this case there was no benefit to pulse power even as a control
system.  Stable operation was possible for both the dc and pulse  under
these conditions.

Low Resistivity Runs

     As mentioned earlier, the low resistivity runs were conducted to
see if there was any effect of the pulse on particle charge.  The experi-
ments were conducted at the low current densities used in the high
resistivity experiments.  The results are shown in Figure 8.  Again note
that there is no difference between the dc and pulse data.

                              DISCUSSION

     The main conclusion we draw from these data is that there is no
difference between the performance of an electrostatic precipitator
using either dc or pulse power provided that the electrostatic pre-
cipitator is not allowed to go into back corona.  As soon as the
electrostatic precipitator is allowed to operate in back corona (most
likely with dc), the particle collection efficiency decreases and
significant numbers of uncharged particles are found.

     The main advantage of using pulse power supplies is that good
control of the electrical operating conditions of the electrostatic
precipitator is possible, even when dusts with high electrical resist-
ivity are being collected.  Such good control is often not possible with
dc.  The value of good control should not be neglected and the fact that
pulsing allows one to control the operation of an electrostatic pre-
cipitator is significant.

     There is still the problem of reconciling these results with the
data reported in the literature which show a major improvement with
pulsing.  It seems likely that much of the data comparing particle
collection with pulse and dc were generated under conditions where the
dc was operated under conditions of back corona.  We observed similar
results when pulse power was compared with dc in back corona.  However,
significant improvements in particle collection were obtained when the
electrostatic precipitator was operated with dc out of back corona.   For
this reason we took extreme care to operate the dc so that back corona
was avoided.  The fact that the dc could not be operated for 'long times
without going into back corona shows that pulsing for control of  the
electrical conditions is a significant advantage.


                                    318
-------
                                 ENDNOTES
1.
2.
3.
4.
Petersen, H.  H.  and P.  Lausen, "Precipitator  Energization Utilizing
an Energy Conserving Pulse Generator," in Second Symposium on the
Transfer and  Utilization of Particulate Control  Technology:  Volume II
EPA 600/9-80-39b.   (NTIS PB81-122210), September 1980.

Feldman, P. L.  and H. I. Milde, "Pulsed Energization for Enhanced
Electrostatic Precipitation in High Resistivity  Applications,"
in Symposium  on the Transfer and Utilization  of  Particulate
Control Technology, Volume I, EPA-600/7-79-044a  (NTIS PB 295  226)
February 1979.

Lawless, P.,  G.  Ramsey,  and B. Daniel, "Characterization of the
EPA/IERL-RTP  Pilot-Scale Precipitator," EPA-600/7-79-052 (NTIS
PB 292-820),  February 1979.

McDonald, J.  R., M. H.  Anderson, R. B. Mosely, and L. E.  Sparks,
"Charge Measurements on Individual Particles  Exiting Laboratory
Precipltators with Positive and Negative Corona  at Various  Temper-
atures." J. Appl.  Phys.  51,3632 (1980).
              Sampling
               Ports
                A
      Aerosol
     Injection
      Pores
                          Sampling
                           Ports
                            B
Sampling
 Ports
  C
                          Hopper
Gas ' 1 I
. , Rapper 1 i
irrtetB Injection B 1 1
Ports m I 1 n 1
O

; ;

O
0 o 0
o
o ° O
o
o ° 0





Section 1




O

o

o
Section 2




n





O

O

O
Section 3




                                  Diagram of ptJot-:iCiiJe F.SP.
                                     319
-------
500
400
£300
§> 200
100
°C




•
i i
) .05 0.1
ms
      a ) Primory Capacitor Voltage
                                           KV
                       0
                      -10
                     -20
                     -30
                     -40
                     -50
                                                 0
                                    ms
                                    0.5
1.0
                            b) Output Voltage
              FIGURE  2. PULSE POWER SUPPLY WAVEFORMS
 CONTROLS
    PRR
 0-IIOpps
  PULSE
AMPLITUDE
  0-55 KV

  SCR
SWITCH
50 M v0 ESP
I
t
BIAS
VOLTAGE


1
     Figure 3.  Pulse power supply circuit.

                                        320
-------
                                     DISCHARGE ELECTRODES
                            CONTROL PAD
                         VOLTS
                           POWER SUPPLY
Figure 4.  Experimental apparatus for measuring the diameter and
           charge  of particles.
                                 321
-------
FIGURE 5.  CHARGE VS RADIUS FOR HIGH RESISTIVITY
TEMPERATURE - 150 C
10001
z
0
a
i -

LU _
U.
O
,£
u 100:
«• ^
rfb _
1 :

LU
a:
<
5 -
10-
A
_Ov
ceo
^
CD
CD
O
a
B°
e
e
§ 0 DC
a PUL
_ _ . — —





1 i 1 i 1 1 ii i i i i 1 1 1 it 1
.1 i 10
  PARTICLE RADIUS  
                  322
-------
FIGURE 8. CHARGE VS. RADIUS  FOR MODERATE RESISTIVITY
TEMPERATURE - 218 C
1000!
fc I
IER OF ELECTRON!
!-•
§
II 1 III
LU ™1
§ :
Z_
NX —
g -
K
5
10-
*
O
-------
   <
   3  -
    1  -
I-


I 3
UJ

UJ
Q.
  10  -
  30  -
100
       .3
                            i

1            3              10

  DIAMETER -  MICROMETERS
30
                                                   111
                                                   1-1
                                                   U
                                                   »-i
                                                   li.
                                                   U.
                                                   Ill
     Figure 7.  Penetration versus  particle diameter for moderate

     resistivity.
                                   324
-------
             FIGURE S.  CHARGE VS RADIUS FOR LOW  RESISTIVITY EXPERIMENTS

             TEMPERATURE - 24 C
CO
NJ
Ul
(A

O
CK
d
U.
Q

OS  100 -j
      g
      £K
                             §
                           8
                    e
                        8
                            1 1
                             1
                                    O DC

                                    D PULSE
                                     i m
                                       10
                PARTICLE RADIUS  (MICROMETERS)
-------
    Particle Collection in a Two-Stage Electrostatic Precipitator  with
                           Various Collector Stages
                                      by
          I.E. Sparks, G.H. Ramsey, R.E. Valentine, and J.H. Abbott

Introduction

     One of the major goals of the Environmental Protection Agency's
electrostatic precipitator research and development program is  the
development of improved electrostatic precipitators for collecting dusts
with high electrical resistivity.  Fairly early in the program  it  became
apparent that some two-stage electrostatic precipitator design  was most
likely to give the desired improvement.  Consequently a major program
was funded to develop and demonstrate a new two-stage electrostatic
precipitator for collecting dusts with high electrical resistivity.

     Early work concentrated on the development of a suitable precharger
for the system.  The precharger that came out of this work is a three-
electrode precharger, shown in Figure 1, developed by Southern Research
Institute under EPA's sponsorship.  The three-electrode precharger is
described by Pontius and Sparks-^.

     The three-electrode precharger has been tested as part of a two-
stage electrostatic precipitator system and the results reported by
Sparks et al*  and Pontius et al^.  The collector stage used in the previous
work was a first generation collector and was essentially a conventional
electrostatic precipitator with discharge electrodes consisting of either
wire mesh or closely spaced wires.

     The experimental results in References 2 and 3 showed that the first
generation system was a significant improvement in electrostatic precipitator
technology.  However, the collector stage was not the optimum.  Theoretical
analysis and limited experimental data showed that additional improvements
in electrostatic precipitator technology were possible if the collector
stage could be improved.

     Work to develop an improved collector stage, begun by EPA/IERL-RTP's
Particulate Technology Branch using inhouse capabilities, is discussed in
this paper.

Characteristic of Optimum Collector Stage

     Because the collector stage is designed to collect previously charged
particles, the highest possible electrical field should be maintained in
the collector.  If there were no reentrainment, the collector stage would
not require any current.  Unfortunately, reentrainment is present  in any
dry system.  Thus, some current is necessary in the collector stage to
minimize the reentrainment problem.

     The major functions of the collector stage are to collect  the charged
particles from the precharger and to charge and recollect particles that
                                  326
-------
are reentrained.  Because  the  size  of  reentrained  material  is  large,
the particle charging requirements  of  the  collector  stage are  much less
severe than the particle charging requirements  of  the  precharger.  This
means that low current densities are acceptable in the collector stage.

     The mesh and close-spaced wire discharge electrodes used  in the
first generation collector stage were  selected  to  provide fairly high
electric fields and low current densities.   Pilot  plant experiments with
the mesh and wire discharge electrodes showed that,  when the discharge
and collection electrodes  were clean or covered with low resistivity
dust, the applied voltage  (and thus the achievable electric field) was
much higher than that obtained with conventional discharge  electrodes.
However, when the discharge and collection elecrodes were covered with
high resistivity dust, there was no significant difference  between the
applied voltage obtained with  the mesh or  close-spaced wire discharge
electrodes and the applied voltage  obtained with conventional  discharge
electrodes.  The stable electric operating points  for  all discharge
electrode configurations were  essentially  the same^.   The only exception
was that flat plate discharge  electrodes could  be  operated  at  high applied
voltage and no current.

A downstream collector with flat plate discharge electrodes would be the
ideal collector stage except for the problem of reentrainment.  A system
could be designed with precharger collector, precharger collector, etc.
But reentrainment from the last collector  stage would  still be a problem.
Also the use of the precharger to charge the large particles from reentrainment
appears to be a non-optimum economic solution to the problem.

     Analysis along the lines  outlined above led to  the idea that the
optimum downstream collector would  consist of sections with flat plate
discharge electrodes in combination with something to  handle the reentrainment.
A pilot plant experimental study was begun to study  various combinations
of collector stage configurations.  The configurations which were studied
are shown in Table I.

              Table I.  Collector Stage Configurations.
1.  Flat Plate, Wire, Flat Plate, Wire
2.  Flat Plate, Flat Plate, Wire, Wire, dc power.
3.  Flat Plate, Flat Plate, Wire, Wire, pulse power.
Experimental Electrostatic  Precipitator

     The experiments were conducted  in a  one-lane  pilot  plant electrostatic
precipitator located in EPA's  Industrial  Environmental Research Laboratory,
Research Triangle Park, NC.  The  electrostatic  precipitator was operated
with plate-to-plate spacing  of 22.9  cm.   The  discharge electrodes were
0.318 cm diameter rods spaced  22.9 cm apart.
                                  327
-------
     The aerosol used for this study was fly ash which was injected  into
the electrostatic precipitator through two sand blast guns fed by a
vibrating screw feeder.  The electrical resistivity of the fly ash was
about 2X10-L2 ohm-cm for the experiments discussed in this paper.  This
corresponds to an electrostatic precipitator operating temperature of
150°C.

     Additional information on the pilot plant electrostatic precipitator
and the aerosol injection system is given by Lawless et al^'

     The overall mass particle collection efficiency of the system was
determined using mass filters.  The collection as a function of particle
diameter was determined using Meteorology Research, Incorporated (MRI)
cascade impactors calibrated by personnel of EPA/IERL-RTP's Particulate
Technology Branch.  Data reduction was done on a TRS-80 microcomputer
using software developed by Denver Research Institute.

Results

     The penetration as a function of particle diameter for the various
configurations studied is shown in Figures 2-4.  The results of the
previous experiments with the mesh discharge electrode are shown in
Figure 5.  The performance of the pilot electrostatic precipitator as a
conventional electrostatic precipitator is shown in Figure 6.

     The results shown in the various figures show that the initial or
first generation collector stage was not the optimum.  The various flat
plate configurations offer improved particulate collection when compared
to the first generation collector.  The results also show that changing
the rapping cycle in the precharger had a major impact on the particle
collection efficiency of the system.

Discussion and Conclusions

     The results of the pilot plant experiments demonstrate that improved
electrostatic precipitator technology requires both a good precharger
and a good collector stage.  The difference in particle collection
between the best of the collector stages studied and the first generation
system is almost as much as the difference in particle collection between
the first generation system and the conventional electrostatic precipitator.

     The experimental data were used to estimate the penetration versus
specific collector area curve for a typical fly ash with a log-normal
particle size distribution (mass mean diameter = 20 ym and geometric
standard deviation = 5.0).  The penetration for a given specific collector
area was estimated by using the experimental particle diameter specific
migration velocities and the specific collector area to calculate the
penetration for each particle diameter from:

          Pt(d)=exp[-w(d)SCA]                (1)
                                 328
-------
where Pt(d) is the penetration  of  particles  of  diameter d,  w(d)  is  the
migration velocity of  particles with diameter d,  and SCA is the  specific
collector area.  The overall  penetration was calculated from:

          Pt0 = Z±  Pt(di)f(di)              (2)

where f(d^) is the fraction of  the particles with diameters between d^
and d^+ dd.

     The migration velocity versus particle  diameter for the conventional
electrostatic precipitator, the first generation  system,  and the  second
generation system is shown  in Figure 7.

     The results of the  calculation are  shown in  Figure 8.   Note  that
because of particle size distribution effects,  the performance gain of
the two-stage system increases  as  the overall penetration decreases (as
the overall efficiency increases).

     The major unknown about  the improved collector stage is its  cost.
Work is now underway to  determine  the costs  of  using all plate collector
stages.  Preliminary results  are promising and  show that the added  cost
of all plate stages is more than compensated by the improved performance.
Additional work  is necessary  to improve  the  cost  estimates  before any
quantitative information is released.

     Also additional experimental  work is planned to improve the  performance
of the system as a whole.   This means that optimization of  both  the
precharger and the collector  stages is necessary.

References

1.  Pontius, D.H. and  L.E.  Sparks.  "A Novel Device for Charging  High
Resistivity Dust."  J. Air  Pollution Control Association, 28, 698 (1978).

2.  Sparks, L.E., Ramsey, G.H., and B.E. Daniel.   "Collection of  Fly Ash
with High Electrical Resistivity in a Pilot  Plant Electrostatic  Precipitator
preceded by the  EPA/SoRI Precharger."  J. Air Pollution Control Association,
29, 745 (1979).

3.  Pontius, D.H., P.V.  Bush, and  L.E. Sparks.  "Field  Evaluation of a
Two Stage ESP for High Resistivity Dusts." Staub. (in press) (1980).

4.  Lawless, P., G.H.  Ramsey, and  B.E. Daniel.  "Characterization of the
EPA/IERL-RTP Pilot-Scale Precipitator."   EPA-600/7-79-052 (NTIS  PB  292-
820) (1979).
                                  329
-------
Discharge Wires
 High Negative
Voltage (-20kV)
                       Discharge
                         Wire
                 Screens
              High Negative
              Voltage  (12kV)
                                              Note:  all dimensions
                                                     are approximate
                                                Plates
                                              to ground
       Figure 1.  Schematic drawing of EPA/SoRI precharger.
                       o o o o
c
o
                  o o
o  10 Sec Rap EFF = 69.44%
*  3 Min Rap EFF = 86.3%
o
tfl
n
M-l
^x
C3 " *
0
•H
4-1
ts
j-j
4J
01
S
Q)
P-(

X * *^ °
*? A. y
* *** *** Precbarger Experiments
* * Plate-Wire-Plate-Wire
* 150 C Resistivity = 2E12
^ OHM-CM
*







                    1                10

            Physical Diameter  (micrometers)
                                                        '100
     Figure 2.   Penetration versus  particle diameter for precharger
                followed by plate-wire-plate-wire collector.
                                 330
-------
u
ctf
M
H-l
a
o
•rl
•U
0)
a
-------

i
*-<
i
§ .1:
UJ
s
   ,01-
      .1
                    FIG.  S 1ST GENERATION PRECHASGER SYSTEM
-I - 1 — riii  r i
                                                                i  i  it
                        PHYSICAL DIAMETER CMICROMETERS)
                      FIGURE 6. PENETRATION FOR PILOT ESP
                 RESISTIVITY - 2E12 OHM-CM
    .01-
       .1
                 I   I  I I I I I I
       1


    PHYSICAL DIAMETER (MICROMETERS)


                  332
1   I  I  i i i i i
          100
-------
-
-
-
Crt
§
•U
•rl
O
O 10-
H
a
>
o
•H
4-1
to
M *"
00
•rl
a
i -
6 First
Generation System
* Conventional ESP
* Plate-Plate-Wire-Wire Pulse On
Comparison of first
and second generation systems
with conventional ESP for resistivity -2 E12 OHM-CM

* *
* *
if
00 * *
- 2 * A c 0 0 o
* A
A A
A





* ***
*•* «°*
* * « o °
— L 
-------
                     HIGH INTENSITY IONIZER DEVELOPMENT
                   By M.  H.  Anderson
                      J.  R.  McDonald
                      J.  P.  Gooch

                      Southern Research Institute
                      2000 Ninth Avenue South
                      Birmingham, Alabama 35255

                      and

                      D.  V.  Giovanni
                      Electric Power Research Institute
                      3412 Hillview Avenue
                      Palo Alto, California 94303

                                  ABSTRACT

     Results from parallel laboratory and field studies designed to character-
ize effects associated with a double-bellmouth, high intensity ionizer (HII)
are presented.  These studies include measurements after the HII of particle
charge as a function of particle diameter in the fine particle size range for
various voltage levels, particulate mass loadings, fly ash compositions,
charge to mass ratio measurements after the HII for various HII and gas condi-
tions, measurements of the gas velocity distribution after the HII for various
main gas flows, temperatures, and anode purge air pressure drops in the HII,
observations of effects in transition cavities containing highly charged parti-
cles, and observations of effects in small-scale HII/ESP combinations.  The
results of the studies show that the HII charges fine particles in accordance
with field charging theory,  the gas velocity distribution after the HII can
be adjusted by varying the anode purge air conditions, space charge precipita-
tion effects can be significant in transition cavities following the HII, and
the HII provides improved performance of an ESP for small-scale HII/ESP
combinations with the degree of improvement depending on ESP operating condi-
tions.

                                INTRODUCTION

     The electric utility industry is faced with the problem of finding cost
effective procedures for upgrading the performance of electrostatic precipi-
tators which are not in compliance with particulate matter emission regula-
tions.  One approach for accomplishing this objective is to provide a pre-
charging device at the inlet of a conventional precipitator for the purpose
of imparting a higher than normal value of charge to fly ash particles entering

                                     334
-------
the precipitator.  Theoretical calculations have  indicated  that  the precharging
approach can provide significant improvements  in  the collection  efficiency of
a downstream precipitator.

     Promising laboratory results and pilot plant trials with a  precharging
device known as the high-intensity  ionizer prompted EPRI to initiate a research
program in 1975 to respond to the need  for finding practical methods to upgrade
precipitator performance.  As shown in  Figure  1,  the basic  design of the
ionizer consists of a disc electrode centered  in  a venturi  shaped anode assem-
bly equipped with a purge system.   An array of ionizer assemblies is placed at
the entrance to an ESP.  For the combined HII-ESP system to be successful,
three technical goals must be achieved:

     1.  The HII must impart an enhanced charge to the fly  ash particles.

     2.  The particles must retain  this high charge as they pass through
         the HII assembly and transition cavity into the ESP.

     3.  The ESP must respond by efficiently collecting these highly
         charged particles without  deterioration  of performance  over
         time.

     The initial research approach  consisted of evaluation  of the charging
capabilities of a prototype HII system  under controlled operating conditions
at a scale representative of utility ESP installations.  The evaluation was
performed in 1978 on a 10 MW prototype  ESP at  EPRI's Advanced Emissions Control
and Test Facility, at the Arapahoe  Station of  Public Service Company of
Colorado.  A schematic diagram of the prototype system is given  in Figure 1.
The HII system tested at Arapahoe in 1978 did  not achieve the goals listed
above.  The collection efficiency of the 10 MW ESP prototype at Arapahoe did
not improve sufficiently to warrant further scale up.  As a result of these
findings, the research effort was redirected towards identifying HII per-
formance limitations and developing hardware and  process improvements to
overcome these limitations.  The redirected research effort involved experi-
ments on laboratory test stands in  the  APS laboratories and diagnostic tests
and analyses on the 10 MW prototype system and a  smaller (^5000  acfm) pilot
unit at the Arapahoe Test Facility.

     Diagnostic testing on the 10 MW prototype indicated that ionizer per-
formance was limited by electrical  breakdown associated with build-up of high
resistivity ash on the anode vanes.  Ionizer voltage vs. current measurements
were performed, using ambient air,  and  using simulated flue gas  corresponding
to both natural gas and coal-fired  boiler conditions.1  Operating conditions
were established at ambient and at  elevated temperatures.   It was found that
the initial sparking point dropped  from about  11  kV/cm to about  6 kV/cm after
eight hours of operation (See Figure 2), unless the moisture content of the
anode purge gas was increased to about  5 to 12% by volume.   Typical required
operating conditions for the anode  purge gas were established at 6% by volume
moisture, at a temperature of approximately 43°C  (110°F), and purge air gas
flows equal to about 8% of the main gas flow.
                                      335
-------
     The necessity for conditioning the thin layers of ash deposited on the
anode vanes can be explained by considering the electric field intensity in
the ash layer that results from the high current fluxes at the anode surface,
which are typically ^2500 nA/cm2.   Although the electric field intensity
(breakdown strength) required to produce electrical breakdown in the ash layer
can vary widely, values in the range of 10 to 20 kV/cm are considered typical.
Since the maximum value of resistivity which can be tolerated without break-
down is equal to the breakdown strength divided by the operating current
density, it can be shown that, if breakdown strength is in the 10 to 20 kV/cm
range, the resistivity of ash on the anode must not exceed 4-8xl09 ohm'cm.  If
the resistivity does exceed 4-8xl09 ohm*cm, some means of reducing the resis-
tivity will be required.  It is therefore apparent that the effectiveness of
the anode purge/conditioning system will depend upon the response of fly ash
resistivity to the moisture content and reduced temperatures of the anode
purge gas.

     Although the addition of water vapor to the anode purge gas alleviated
the degradation of ionizer electrical operating characteristics at Arapahoe,
this remedial approach did not result in improved system performance for
ionizer field strengths above 7 kV/cm.  It was determined that 7 kV/cm was
not an electric field value of sufficient intensity to accomplish the desired
level of particle charge and that additional diagnostic testing and develop-
ment efforts were required.

     Further diagnostic testing provided the following conclusions:

     • Charge loss within the ionizer assembly was occurring downstream
       of the discharge electrode, and the performance of the ionizer-
       precipitator system was limited by the charge loss.

     • The charge loss was induced by the high space charge of the
       particles and was severe at higher mass loadings.

     • The fiberglass exit cone enhanced the electric field at the
       fiberglass-metal interface where most of the discharge occurred.

     As has been previously reported,1 the original design of the HII evolved
following a series of optimization tests to obtain both a high intensity
electric field in the ionizer throat and a low gas pressure drop through the
ionizer system.  The inlet cone section was designed with a bellmouth
configuration, but the outlet cone section was designed to recover kinetic
energy from the main gas stream.  Since charge loss in the ionizer assembly
was found to be a major limiting factor in system performance, an experi-
mental effort was initiated in the laboratory to minimize the charge loss.
It was found that simply replacing the fiberglass cones with metal cones
would not significantly improve particle charging performance because of
resulting high space charge effects within the cone.

     Promising results were obtained, however, by eliminating the fiberglass
exit cone and metal diffuser and by using a double bellmouth design which has
symmetric inlet and outlet geometries.  A schematic drawing of this design  is


                                     336
-------
given in Figure 3.  With this new geometry,  ionizer  field  strengths  of  10 kV/
cm and current densities on the order of 3000 nA/cm2 have  been obtained.

     This paper will summarize results obtained with this  ionizer geometry in
a parallel effort, which is still in progress, conducted at the APS  labora-
tories and at the Arapahoe Test Facility.  The objectives  of the research
efforts are:

     1.  To quantify the particle charge values produced by the ionizer
         on fly ash particles under various  conditions.

     2.  To determine the degree to which the fly ash particles retain
         charge imparted by the ionizer.

     3.  To quantify the increase in collection efficiency obtainable
         by precharging under various conditions.

Initially, Air Pollution Systems was primarily responsible for the development
of the HII, and Western Precipitation was responsible for  its field evaluation.
When it was decided in the spring of 1979 that a return to smaller test units
and the initiation of a diagnostic program were required, a Task Force composed
of members from EPRI, APS, Western Precipitation, KVB and Kaiser Engineers was
formed to direct the program.  In November of 1979, Western asked to be
relieved of their evaluation responsibilities, and Southern Research Institute
(SoRI) was placed under contract to EPRI.  The responsibilities of SoRI
included the performance and analysis of the testing at Arapahoe and providing
assistance in data acquisition and interpretation for the work underway at the
APS laboratories.

Experimental Technique and Procedures

APS Laboratory

     As was indicated earlier, the outlet geometry of the ionizer was changed
by replacing the metal diffuser and fiberglass cone by a diffuser bellmouth
in an attempt to alleviate the discharge activity associated with the diffuser/
cone assembly.  In order to evaluate the charging characteristics of both
outlet geometries under similar test conditions, two test stands were set up
in the APS lab.  One of the test stands was  equipped with a diffuser/cone
outlet, and the other had a bellmouth outlet.  The main gas flow for each test
stand consisted of hot air from a natural gas fired combustor combined with
ambient air.  The inlet gas temperature was  controlled by varying the ratio of
combustor air to ambient air.  A vibrating screw feeder was used to add the
desired amount of fly ash to the gas stream.  Each ionizer assembly consisted
of a 25.4 cm (10 inch) vaned anode and a 11.43 cm (4.5 inch) discharge elec-
trode.  Also, a wire-pipe ESP was installed  downstream of  the double bell-
mouth ionizer to investigate HII/ESP interactions.  The ESP could be operated
either wet (to eliminate high resistivity problems) or dry at simulated flue
gas conditions.
                                      337
-------
Arapahoe Test Facility

     At the Arapahoe Facility, two nominally 5000 acfm mobile support units
(MSUs) were used to obtain the data presented here.  The first,  (designated
MSU-I) which was used to study the charging characteristics of the ionizer,
consisted of a multiclone, a two field electrostatic precipitator, a two
ionizer array, and a single-field precipitator.  The ionizers had the
double bellmouth geometry with 25 cm anodes, with the anodes and bellmouths
isolated from ground, so that current flow from these components could be
monitored.  The precipitator fields were of the Western Precipitation design
and each had 40.1 m2 (432 ft2) of plate area.  Several isolated metal strips
were placed near the wall of the cavity between the ionizers and the last
field of precipitation.  These "current strips" were used to get a qualitative
feel for the manner in which charge migrated in the cavity and, also, to
determine if particulate charge was being lost in the cavity.

     MSU-I was adequate for the ionizer charging study, but it was not well
suited for the evaluation of an HII-ESP combination.  It was limited by low
temperature (200-220°F) at the ionizer outlet and low SCA (80-90 ft2/1000
acfm) in the downstream collector when operating at design flow.  Therefore,
it was replaced by a mobile unit (designated MSU-II) shown in Figure 4, which
consisted of a two ionizer array followed by a three-field electrostatic pre-
cipitator.  Each ionizer was of the double bellmouth design and had a 30.5 cm
(12 inch) vaned anode with a 13.7 cm (5.4 inch) discharge electrode.  Again
the anodes and bellmouths were isolated from ground.  The precipitator fields
were of the Western design with 43.2 m2 (465 ft2) of plate area per field.

     The charging capability of the ionizer was studied as a function of:
inlet mass loading, main gas flow rate, ash resistivity, ionizer field strength
and particle size.  The charge as a function of particle size was determined
using a modification of the Millikan Technique developed by Southern Research
Institute.2  The ratios of charge to mass and charge to volume were determined
by measuring the charge which accumulated on a filter substrate while iso-
kinetically sampling the mass concentration of the flue gas stream.  Mass
loadings, particle size distribution and velocity profiles were determined
using standard techniques.

Experimental Results - APS Laboratory3

Charging Effectiveness

     The work done to compare the charging capabilities of different ionizer
geometries will be summarized briefly.  For both geometries charge/mass  (Q/M)
was measured as a function of ionizer field strength for 0.5, 2, and 4 gr/acf.
For the diff user/cone geometry the values of Q/M were in fair agreement with
predicted values for the low grain loading case but fell below predicted
values for the higher grain loadings.  The degradation in Q/M could be cor-
related with an increase in the current to a ring at the exit of the cone.
The values of Q/M measured at the exit of the bellmouth for the  same con-
ditions displayed little or no dependence on grain loading and agreed with
values predicted for the particle size distribution and ionizer  setting.  All
subsequent work was performed using the double bellmouth geometry.

                                     338
-------
     Since Q/M results can be seriously misleading because of their sensitivity
to particle size distribution changes, considerable effort was directed toward
quantifying the charging characteristics of the HII as a function of particle
size.  The charge given to individual particles was measured as a function of
particle size at several different HII current-voltage settings for three
different mass loadings; .5, 2.0, and 4.0 gr/acf.  The results obtained for
2.0 gr/acf at the ionizer voltage setting of 71.5 kV are shown in Figure 5.
Also shown in Figure 5 is a theoretical prediction of the charge as a
function of particle size for particles exiting a highly efficient cold-side
ESP and a theoretical prediction of charge for particles exiting the ionizer.
It can be seen by comparison that, for the 71.5 kV case, particles exiting
the HII obtain significantly more charge than would be obtained by traversing
the ESP.  These data also give direct proof that the ionizer is effective in
charging submicron size particles.  Also, the agreement shown between measured
values of charges and those predicted by theory from conditions in the ionizer
is considered to be acceptable.

     In summary, the ionizer was found to charge particles to levels predicted
by theory if the anode purge system was operating properly.  If the moisture
content of the purge air dropped below 5% by volume, sparking occurred in the
ionizer and the charge levels measured were reduced.  Also, when the moisture
level rises above the saturation point, plugging of the anode vanes occurs.
This eventually leads to back corona in these plugged regions with a decrease
in the effectiveness of the ionizer.

Effect of Back Corona

     In addition to the particle charge measurements and comparison with theory,
it was of interest to determine in the laboratory the effect on outlet
emissions of back corona in an ESP following an HII.  Therefore, experiments
were performed using the double bellmouth - HII and wire-pipe ESP laboratory
test apparatus.  Tests were conducted using fly ash from the D field of the
pilot unit at Arapahoe.  It was found that with this fly ash, the ESP went
into back corona even at fairly low current levels.  The presence of back
corona was determined by evaluation of the current-voltage curves, by changes
in electrical conditions with increasing ash thickness, and by ESP current-
outlet optical density measurements.  It should be noted, that in this work,
optical density is used only as a qualitative indication of system performance
and is not related to mass emissions or mass efficiency.

     As mentioned previously, the ESP could be operated either wet or dry.
The wet operation insured the absence of back corona.  The test results for
both wet and dry operation show a decrease in the outlet optical density as
the HII current was increased.  For the wet ESP, the effect of the HII on the
outlet optical density appeared to decrease as the current level in the ESP
was increased, which is consistent with theory.  For the dry ESP case, the
baseline optical density increased with time, and as the ESP deteriorated, the
reduction in optical density produced by the ionizer decreased.

     The results from these experiments with Arapahoe fly ash (dry ESP case)
are shown in Figure 6.  To determine if reentrainment emissions were influ-
encing the results shown in Figure 6, the ash feeder was turned off after  3

                                     339
-------
hours of operation, and the optical density was monitored.  It was found  that
the optical density dropped to zero even though the gas flow was maintained,
indicating no significant reentrainment problems in the ESP under the condi-
tions of the experiment.  Although the ESP was in back corona at all current
levels tested, the electrical characteristics of the ESP deteriorated as  a
function of time during the experiment.  This indicates that the extent of
the back-discharge was increasing as the dust layer thickness increased.

     These results are significant because they illustrate that the HII will
not restore an ESP in back corona to the performance levels which could be
achieved in the absence of back corona.  This observation is also consistent
with results reported by Masuda,1* where positive ions produced by back corona
reduce the limiting negative charge which can be obtained.

Flow Distribution

     Since the gas flow uniformity across the inlet of an ESP has a signifi-
cant impact on performance, an experimental program was performed to study the
interaction between the purge gas and main gas flow rates.  The first series
of tests were performed with no purge air flow.  The main gas temperature was
maintained at 300°F and four different flow rates from 1600 to 3200 acfm were
used.  The test results show the gas velocity profile to be in the jetting
pattern for all flow rates with the degree of jetting decreasing as the main
gas flow rate was decreased.  Here "jetting" refers to a plume shaped velocity
profile at the exit of the ionizer, and "non-jetting" is defined as a velocity
profile in which the flow separates and follows the curvature of the bell-
mouth.

     Next, the effect of purge air AP (which is related to purge gas flow
rate) on the gas velocity profile downstream of the double bellmouth HII was
studied for main gas flow rates of 1600, 2000, 2800, and 3200 acfm.  Again
the main gas temperature was maintained at 300°F, while the purge air tempera-
ture was maintained at 110°F.  For the 1600 acfm flow rate the gas velocity
profile was fairly uniform at the AP of 1.0" H20 and switched to a non-jetting
profile when the purge air AP was increased to 1.5" HjO.  For the 2400 acfm
main gas flow rate, the gas velocity profile switched from jetting to a uniform
profile when the purge air AP was increased from 1.5" to 2.0" H20, and to non-
jetting between 2.0 and 2.5" H20 AP.

     In general, it was found that for all the main gas flow rates tested, the
gas velocity jet downstream of the HII steadily broadened as the purge air AP
was increased until it switched to the non-jetting profile.  Also, it was
observed that the purge air AP at which the gas velocity profile switched from
a jetting to a non-jetting condition increased as the main gas flow was
increased.

     Figure 7 illustrates the interrelationship between main gas flow rate,
purge air AP, and outlet velocity profile in the double bellmouth HII.  The
solid curve and the area above it is the jetting profile regime.  The dotted
curve and the area below are in the non-jetting profile regime.  The gas
velocity profile downstream of the double bellmouth HII switches from jetting
to non-jetting or vice-versa in the area between the two curves.  The practical

                                     340
-------
significance of these results is  that  the  gas  flow profile  downstream of  the
double bellmouth ionizer design is  reasonably  uniform when  the  purge  gas
flow rate is about 16% of the main  gas flow rate.

Experimental Results - Arapahoe Test Facility

Results from MSU-I

     After the size-dependent charge measurement efforts  in the laboratory
produced encouraging results, it  was decided to obtain similar  data in the
field at a point directly behind  the ionizer for the  three  possible MSU
configurations:

     1.  Multiclone - AB field -  HII - C field (MABIC),

     2.  Multiclone - HII - C field (MIC),  and,

     3.  HII - C field (1C).

For the 1C configuration the ionizer inlet  grain loading was on the order of
1.5 gr/acf, for the MIC case it was on the  order of 0.4 gr/acf and for the
MABIC case less than .1 gr/acf.   In Figure  8,  data  obtained  for the MIC con-
figuration are compared to the charge  as a  function of particle size predicted
by field charging theory for the  three voltages at which data were obtained.
This figure shows that reasonable agreement  was obtained between the measure-
ments and theory.

     Figure 9 is a plot of charge as a function of  ionizer electric field for
several particle sizes for the ^.4  gr/acf data.  Due  to the  statistical nature
of the data, there is some scatter  present;  but, in general, the charge in-
creases with increasing electric  field over  the range of particle sizes.

     These data confirm the results obtained with the laboratory test stands,
and show that the double bellmouth  ionizer design is  capable of producing
charge levels in approximate agreement with  theoretical predictions.  Agree-
ment with theory was also obtained  for the higher grain loading case  (1C
configuration), but it was found  that  the distribution of charged particles
in the cavity was more highly dependent upon position.  This is thought to
result from higher space charge fields existing in the downstream cavity
which accompany the higher mass loadings.  The space  charge  field tends to
force the charged particulate matter to the  wall region of the cavity.

     For all cases, the ability of  the ionizer to achieve theoretical levels
of charge was contingent upon sufficient conditioning of anode ash deposits
by the steam-air mixture comprising the anode  purge gas.  Malfunctions in the
steam generator system which produced  either low water vapor concentrations,
(below 6% by volume) or plugged anode  vanes  due to moisture  carry-over, were
accompanied by reduced charge levels on particles passing through the ionizer
under these conditions.

     As previously indicated, size  distribution has a major  effect on Q/M
values measured downstream from an  HII.  This  observation is illustrated by

                                     341
-------
the data and calculated charge values shown in Table 1.  The size distribution
data were obtained in the cavity region downstream from the HII in the MIC
configuration, and the particle charge values were calculated from field
charging theory using the HII applied voltage of 75 kV.  The calculated sum
of 35.7 yC/gm is in fair agreement with a measured value of 45 yC/gm obtained
under similar conditions.  Note that over 60% of the calculated charge is
contained in the sub-micron diameter particles.

Test Plan and Results for MSU-II

     Since the charge vs. size measurements at both Arapahoe and at the APS
laboratory indicated that the double bellmouth ionizer design produced levels
of charge in approximate agreement with theory, EPRI initiated a paper study
designed to evaluate the market potential for HII technology.  This study is
described in another paper,5 but some of the methods will be discussed briefly
in this paper to illustrate the logic used in formulating a test plan for the
Arapahoe MSU unit.

     The evaluation of the market potential for HII-ESP systems is based in
part on simulations performed with a modified version of the EPA/SoRI computer
model.  Typical results from simulations of a precipitator system with and
without the ionizer are shown in Figure 10 for high (5 nA/cm2)  and low
(40 nA/cm2) resistivity conditions.  The "with ionizer" curves do not con-
sider the effect of purge gas, and therefore the indicated SCA for a given
efficiency must be increased by approximately 16% for the with ionizer cases
if purge gas is used to obtain uniform flow.  The projected plate area gain
as a result of ionizer energization for these hypothetical cases is approxi-
mately 17% at the 99% efficiency level for the low resistivity case,  and 50%
for the high resistivity case at the 95% efficiency level with the effects of
purge air considered.  These percentages are obtained by dividing the equiva-
lent SCA decrease by the required SCA for the "without ionizer" cases.  These
examples are cited to illustrate the following point:  The ionizer provided
greater theoretical equivalent plate area gains for high resistivity cases
than for low resistivity cases, if no charge loss is assumed to occur.

     Since models for predicting the extent of back discharge and the resulting
charge loss are not available, it is essential to determine whether the en-
hancements indicated by the computer simulation can, in fact, be obtained under
realistic conditions.  Therefore, the MSU-II system was installed at Arapahoe,
and a test plan was designed to evaluate the effect of the HII on precipitator
performance for both high and low resistivity conditions.  Preliminary results
from this effort are discussed in the following paragraphs.

     During the time period that the MSU-II system has been available for
testing, (approximately 2 months), brief experimental programs have been
conducted with a moderate resistivity ash (Amax Coal) and with a high resis-
tivity ash (Energy Coal).  Resistivity vs. temperature predictions obtained
from Bickelhaupt's6 correlation, along with in situ data from the HII inlet
area, are shown in Figure 11.  At the operating temperature of 300°F, the
resistivity of the Amax fly ash was 1-2x10*J ohm-cm.  As was discussed
earlier, resistivities of 4-8xl09 ohm-cm can lead to breakdown of the ash
layer on the anode vanes.  The estimated temperature of the anode is 'V200°F.

                                     342
-------
From Figure 11, it can be seen  that  the  predicted  resistivity would be ^1  to
2x10   ohm*cm with Amax ash, and  1 to  2x10 u with  Energy  ash.  We  conclude
from these observations that some degree of back discharge  is inevitable with
the Energy coal at these conditions  and  is possible with  the Amax  coal.

     The anode and bellmouth currents  with Energy  coal  show evidence of some
discharge occurring at higher ionizer  current densities (2000 nA/cm2).  This
discharge activity was not observed  for  the Amax Coal until a current density
of 3000 nA/cm  was reached.  It is important to note that the earlier charging
data obtained with MSU-I were taken  with anode temperatures estimated to be
^170°F.  Therefore, the more realistic temperatures achieved in the MSU-II
unit require more resistivity reduction  to be accomplished  by the  anode con-
ditioning system.

     With the Amax Coal, preliminary results from  testing of the MSU-II system
at Arapahoe indicate  that the overall  collection efficiency of the system was
increased from 99.11  to 99.53 as  a result of energization of the ionizer.
These data were obtained for a  specific  collecting area of  300 ft2/1000 acfm
and include both mass train and impactor results.  Fractional efficiency data
obtained with impactors for this  test  series are given  in Figure 12, along
with calculated efficiencies using the test conditions  as input data to SoRI's
computer model.  These data are encouraging in that the theoretical predictions
of collection efficiency enhancement in  the fine particle range were approxi-
mately obtained for particle sizes smaller than 0.6 pm  diameter.   For particle
diameters larger than 1.5 um, size resolution was  not obtained with the
impactors because of  nozzle collection interference with upper stage cut
points.  This problem has now been corrected by using an  in-line nozzle geom-
etry.

     With the Energy  coal,  (high  resistivity M.012 ohm-cm)  the following
preliminary results have been obtained for overall efficiency:

     300 ft2/1000 acfm, with ionizer:  99.3
                      without ionizer:  98.8

     200 ft2/1000 acfm, with ionizer:  92.9
                      without ionizer:  92.8

The results for the higher flow rate (200 ft2/1000 acfm)  show that only small
or insignificant enhancements were achieved, which are  well below  predicted
values of enhancement in the absence of  back discharge  and  charge  loss.  At
the higher flow rate  significant  discharge activity was observed from the  cur-
rent strips in the cavity behind  the ionizer.  The particulate charge loss
associated with this  discharge  phenomena is thought to  be responsible for  the
lack of enhancement.

     Figure 13 presents the mass  efficiency data discussed  above,  along with
projections of collection efficiency using the computer model with the condi-
tions of the experiment as input  data.  This graph illustrates, that for all
cases, the enhancement achieved in overall collection efficiency by energiz-
ing the ionizer was significantly below  the values predicted by the computer
model.  The figure also indicates that the model was reasonably successful in

                                     343
-------
predicting the baseline collection efficiency of the MSU without the ionizer
energized.  This indicates that some charge loss still occurs in either the
ionizer assembly, the cavity, or the ESP for resistivities on the order of
10   ohm* cm.

Summary

     The following conclusions have been derived from the laboratory and field
studies with the double bellmouth ionizer geometry used in pilot-scale HII-ESP
system.

     • For high resistivity dusts, (>1011 ohm*cm), an anode conditioning
       system is required to reduce ash resistivity on the anode vanes
       to 109 to 1010 ohm*cm.  Water vapor additions to the anode purge
       air have been shown to be a marginally effective means of re-
       ducing anode ash resistivity when high resistivity dusts (up to
       1011) are being treated.  It is the opinion of the authors that
       steam purge air systems may not be practical in a power plant
       environment, and therefore another conditioning method should be
       developed if HII technology is to be successful for high resis-
       tivity dusts.

     • The HII achieves levels of particle charge, within + 10 to 20% of
       those predicted by field charging theory, when the anode condi-
       tioning system is conditioning the ash to the 109 ohm*cm range,
       and this charge is retained as the particles pass through the
       ionizer assembly.

     • Results obtained with moderate resistivity dust (VLO11 ohm*cm)
       indicate that model projections have the potential for predicting
       HII-ESP performance under conditions where detrimental effects
       due to high resistivity ash in the cavity and in the ESP are
       eliminated.

     Certain unresolved technical issues must be addressed if the ionizer
is to become an effective means of upgrading precipitators.  These issues
are:

     • To what extent does particle charge loss due to back discharge
       in the ESP or in the transition cavity negate the beneficial
       effects of charge imparted by the HII?  The results obtained
       to date suggest that if an ESP is operating in back corona
       without the ionizer, then the beneficial effects of energizing
       the ionizer may be substantially neutralized.

     • By what means does one obtain a good velocity profile downstream
       of the ionizer while minimizing the purge air requirements?
       With the double bellmouth design, the uniformity of gas flow
       downstream depends on the ratio of the purge gas and flue gas
       and flue gas flow rates.  Preliminary results indicate that the
       gas flow profile is reasonably uniform (standard deviation of
       25% in MSU-I) when the purge gas flow rate is about 16% of the

                                      344
-------
       flue gas flow rate.   This has the effect of offsetting the plate
       area gain by an amount approximately equal to 16% of the plate
       area requirement for an HII-ESP system.  An alternative means
       of producing uniform flow profiles will probably be needed for
       an economically attractive system.

Continuing Work

     Additional overall mass and size dependent efficiency data will be ob-
tained at high resistivity conditions with the two selected values of specific
collection area (200 and 300 ft2/1000 acfm).   Also, S03 conditioning will be
used to obtain resistivity values of the order of 109 to 1010 ohm'cm.  An
experimental program will be conducted with these lowered resistivities to
determine the enhancement which would be produced by the ionizer for a high
resistivity ash if back discharge problems in the ionizer, cavity and pre-
cipitator were not present.  Results from these measurements are expected to
determine the potential for commercial development of HII-ESP systems.

                                   ENDNOTES

1.  Spencer, Herbert W., Ill, et. al.  "Results of High Intensity Ionizer/
    Precipitator Evaluation Tests at the EPRI Test Facility", paper 80-23.3,
    73rd Annual Meeting, Air Pollution Control Association, Toronto, June,
    1980.

2.  McDonald, J. R., et. al.  "Charge Measurements on Individual Particles
    Exiting Laboratory Precipitators with Positive and Negative Corona at
    Various Temperatures".   J. Appl. Physics, 51(7):3632-3643 (July, 1980).

3.  Unpublished reports by APS, Inc. to EPRI.

4.  Masuda, S., and Y. Nonoyaki.  "Detection of Back Discharge in Electrostatic
    Precipitators".  IAS Annual Meeting, Cincinnati, Ohio, September 28 -
    October 3, 1980.

5.  Lagarias, John S., and Jack R. McDonald.   "Application of the High Intensity
    Ionizer to the Electric Utility Industry".  Presented at the Third
    Symposium on the Transfer and Utilization of Particulate Control Technology.

6.  Bickelhaupt, Roy E.  "A Technique for Predicting Fly Ash Resistivity",
    EPA-600/7-79/204, Environmental Protection Agency, Research Triangle Park,
    NC, August, 1979.
                                     345
-------
CURRENT-VS

Dia.
(UE)
.2
.3

.4
.5
.6
.7
.8
.9
1.0

2.0
3 0
4.0
5.0
6.0
7.0

8.0
9.0
10.0
14.0
18.0
22.0
30.0
60.0

Radius
(pm)
0.10
0.15

0.20
0.25
0.30
0.35
0.40
0.45
0.5

1.0
1.5
2.0
2.5
3.0
3.5

4.0
4.5
5.0
7.0
9.0
11.0
15.0
30.0

Cum
of M
2
3.

5.

. %
ass

5

0
6.7
a.
9
11
12
14
1
8
2
7
2

26 . 0
34.5
40.8
46.8
51.0
54.6

58
60

.0
.8
63.0
70.5
76
78
84
.0
.5
.5
100

Mass
dm)
.02
.015

.015
.017
.014
.017
.014
.015
.015

. 118
.085
.063
.060
.042
.036

.034
.028
.022
.075
.055
.025
.060
.155

Ma

9.
3.

7.
1.
2.
4.
6.
8
1

.-VOLTAGE

ss/Part.
(gm)
802x10""
308x10-"

841x10" '*
532x10-"
646x10" '3
203x10-"
273x10-"
932x10-"
225xlO"lz
_12
3. 308x10" ' '
7.841x10""
1.532x10""
2.646x10"'°
4.203x10"'°

6
8
1
3
7
1
3
2

273x10
932x10"'°
225x10"'
.362x10""
. 145x10"'
.305x10"°
.308x10"*
.646x10"'
DATA

No
Pa
2.
4.

1.
1.
5.
4


. of
rticles
040x10"
534x10"

913x10"
110x10 "
291x10"
045x10 "
2.232x10'°
1
1
679x10"
225x10'°
1 204x10 '°
2.570x10'
8.035x10*
3.916x10*
1.587x10*
8.565x10'

5
3
1
2
7
1
1
5

420x10'
135x10'
796x10'
231x10'
.698x10*
.916x10*
.814x10*
.£•58x10*




Rux-shaping-^
electrode/
T°tal Discharge electrode -, /
Charge/Part. Charge (cathode)/ /

3.
8.

1.
2.
3.
4.
5
7
9
3
(Cont) (VJC)
718x10"'* 7.5b5 /
HV cathode -, /
365x10"" 3.793 support/ /
Rectangular-^ / /
487x10"" 2.845 support lube / / / /
324x10"" 2.580 ~/ / j ^~/1
346x10- 1.770 / J /A
i? / ^amm//— —/
554x10 " 1.842 / -vmj^^j 	
948x10"" 1.328 d |J |[ J
S9B,in-l' 1 57B '

294x10"" 1.139 1 / I / /
/ !/ /
71«~1n-!6 /, rcT /I/ /
8.360x10"" 2.149 / [ ""-/—.
/ /

/ / "Cavity"

Js — ..
/ ' /h
/ I!; f
----^ 'Current . . 23cm
^^-- 	 sensors i
t
85 cm B~^, 1 * *
1 '|P
r~-^ b * v-
^~^71 ' '
/I ff r
\ .
/ / / — r ' ~
1.487x10-" 1.195 Inletcone— ' / / / ff,
It 1 W
2.324xio-" 0.910 Porous anode-J / Exhaust conej
3.346x10-" 0.531 Expander section-'
4.554xlO"ls 0.390

5
Porous Anode Ion
948x10"" 0.322
ESP
collecting A emitting
system arrangement
zer Assembly

Figure 1. Initial Ionizer and Collector
7.528x10"" 0.236 Field Arrangement, Unit 2
9
1
3
4
8
3
294x10"" 0.167
822x10"'* 0.406
OllxlO-"1 0.232
498x10"'* 0.086
.365x10"" 0.152
346x10"" 0.196






        Spark Points  •
    A = Ambient air
    O= Natural gas fired
    O = Coal fired
                                  Inltiat spark point t
Spark point after 6 hours
                                Coat firing
                   Natural gas firing
                         Spark
                          point
              4         6         8        10
                  Ionizer Becltric Reid (kV/cm)
                                                                                   HII-ESP WITH NEW GEOMETRY
CAVITY
100cm
                                                                                                                            • 23c
Figure  2.   Ionizer  current versus ionizer electrical
            field  (33,000 ACFM)  MICD configuration.
                                                                    Figure 3.   Double bellmouth high intensity ionizer  followed
                                                                                 by an electrostatic  precipitator.
                                                           346
-------
                                                                                                                            CHAflOE VS 'ARTICLE SIZE
  '    Figure  4,  APS lonizer-PrecipHator Mobile Unit  (MSU-II).
                                                                                                                               	PREDICTED CHAME FR01
                                                                                                                               	 PREDICTED CHARGE AT E
                                                                                                                                     Ef FIBENCV COL&4IDE B
                                                                                                                         PARTICLE  RADIUS (jim)
                                                                                                             Figure 5. l*bor»tory MMurwnu and prediction* of pureicle
                                                                                                                     chanj* atth« outlet of the donblc b«tlaouth high
0.0       0.5        1,0         1.5        2.0

                 HII    CUR RENT (mA)
                                                         2.5        3.0
                                                                                                3500


                                                                                          J

                                                                                          3    3000
                                                                                          <   2000

                                                                                          z

                                                                                          £    1500
Jetting profit*
   ragime
                        Non jetting profile
                             regime
                                                                                                   1.0      1.5      2.0      2J      3J3      3.5

                                                                                                                    PURGE  AIR  AP(in.HO)
   Figure 6.  Outlet optical density versus Ionizer current  for dry ESP operation.
              MEASURED CHARGE VS PREDICTED CHARGE FOR HUM
                  THEORETICAL CHARGE

      Figure 8.  Correlation of fi«ld BUauroMfltB and piadlctloni of
               particl* cbacg* at the outlet of the double b*lJuoueh
               high intemlty ioniz*r.
            ELECTRIC  FIELD (kV/cm)
                                                                          347
-------
 100                 200

   SPECIFIC  COLLECTING  AREA (ft2/iooo acfm)
                                                                                                                     OENERGY COAL
                                                                                                                     AAMAX  COAL
                                                                                                                     AAMAX I-V
                                                                                                      TEMPERATURE (°f)
                                                                                                         •als burned at the Arapahoe Facility.
                  PENETRATION-EFFICIENCY

21 -24 JAN. 81.  AMAX COAL
                                                                                          PREDICTED PERFORMANCE OF MSIMI AS A FUNCTION OF SCA
                 PARTtCLE  DIAMETER (urn)



             efficiencies without and with  the high
                                                                                         NON-IDEAL PARAMETERS
                                                                                         S " 0.05
                                                                                         og = 0.15
                                                                                       	AMAX COAL
                                                                                          li * 1 2K1011 ohm-cm
                                                                                       	ENERGY COAL
                                                                                          p = 1x1012 ohm-cm
                                                                                                                • MEASURED WITH Mil FOR AMAX COAL
                                                                                                                O MEASURED WITHOUT HII FOR AMAX COAL
                                                                                                                • MEASURED WITH HII FOR ENERGY COAL
                                                                                                                D MEASURED WITHOUT HII FOR ENERGY CO>
                                                                                                Si'ECinc CT.LECTiNG AREA, ft2/1000 acfm
                                                                   348
-------
     DEMONSTRATION OF AIR POLLUTION SYSTEMS HIGH INTENSITY
   IONIZER/ELECTROSTATIC PRECIPITATOR ON AN OIL-FIRED BOILER

               By:   Gary A. Raemhild, Anil Prem
                    Air Pollution Systems, Inc.
                    18642 - 68th Avenue South
                    Kent, Washington 98031

                    Fred Weisz
                    Long Island Lighting Company
                    Northport Power Station
                    Eaton's Neck Road
                    Northport, New York  11768


                            ABSTRACT

     A study was performed to evaluate the applicability of Air
Pollution Systems High Intensity Ionizer  (HII) to the electro-
static precipitation of particulate emissions from a high sulfur
oil-fired boiler.  The study was conducted on a 380 MW boiler at
Long Island Lighting Company's Northport Station, Unit 1.  The
APS High Intensity Ionizer/Electrostatic Precipitator (HII/ESP)
mobile pilot plant was utilized for this study.  Simultaneous
particulate measurements were initially performed at the inlet
and outlet of the south side precipitator on Unit 1 to character-
ize the performance of the existing precipitator.  Testing was
done at 360 MW and 125 MW and included total mass samples, cas-
cade impactor tests and ultra-fine particle analysis.  The
electrical and physical parameters for the three field precipi-
tator in the mobile HII/ESP pilot plant were adjusted to simu-
late the south side precipitator on Unit 1 as closely as possi-
ble.  Simultaneous inlet/outlet particulate sampling for the
pilot unit was performed using the same methods as on the main
precipitator.  Performance tests were conducted with the HII
both energized and de-energized at 14,273 M3/hr.(8400 ACFM),
11,894 M3/hr.  (7000 ACFM), and 5947 M^/hr.  (3500 ACFM).  All
tests were  performed with the boiler at a baseline load condi-
tion between 340 and 360 MW.  The results show a significant de-
crease in particle penetration when the HII was energized.  Due
to certain delays in data reduction and analysis, the results
have not yet been released.  Upon release, the data shall be
available to the public from ESEERCO, or Air Pollution Systems.

                          INTRODUCTION

     The Empire State Electric Energy Research Corporation
(ESSERCO) awarded a contract to Air Pollution Systems, Inc. to
conduct a turnkey pilot plant program that would demonstrate the
applicability of Air Pollution Systems' High  Intensity Ionizer
(HII) technology on oil-fired boiler emissions.
                                349
-------
     This program was carried out at Long Island Lighting's
Northport Power Station.  The first stage of the program was the
characterization of the existing Unit #1 main precipitator by
simultaneous inlet/outlet testing.  The HII/ESP pilot plant was
then evaluated with identical testing techniques using a slip
stream off the same exhaust emissions.  Comparative simultaneous
tests were performed with the HII energized and de-energized.
This provided the data necessary to define the reduction in par-
ticle penetration with the utilization of an HII for particle
pre-charging upstream of an existing conventional precipitator.

                       TEST OBJECTIVES

     The objectives of the test program are listed below:

     1.  Characterize the operating parameters and performance
of the south side main electrostatic precipitator on Unit #1 at
LILCO's Northport Power Station.

     2.  Using the APS HII/ESP pilot plant, determine quantita-
tively the improvement in performance of the ESP when an HII is
used to precharge the particulate.  This involves studying the
effect of gas flow rate and specific collecting area (SCA).

     3.  Utilizing the pilot plant and main ESP data, predict
the level of reduction in particulate emissions from the main
 (Unit #1) ESP with the addition of an HII array.

               HIGH INTENSITY IONIZER TECHNOLOGY

     The APS High Intensity Ionizer, due to its unique electrode
geometry, is successful in maintaining a highly stable intense
corona discharge.  The cathode is a solid metal disc supported
by a structurally reliable tube section centered in a cylindrical
anode arrangement.  An HII assembly is illustrated in Figure 1.
The electrode configuration produces a substantially uniform
three dimensional field which is a principle factor in the great-
er electrical stability.  Typical operating field strengths of
10-13 kV/cm have been obtained compared to 3.5 kV/cm for wire
electrode geometry in an industrial flue gas.

     In addition to the very high electric fields, the HII elec-
trode geometry produces a concentrated field with ion densities
of 1C)9 - 10-LO ions/cc, many times that obtained in the wire elec-
trade geometry.  As a result of higher fields and ion densities,
both the level of charge acquired by the particles and the rate
of charging is significantly higher.  For example, even with the
significantly lower residence time in an HII compared to an
electrostatic precipitator, the level of charge obtained by the
particles exiting an HII are two to three times higher compared
to the particles exiting an ESP.

     The HII effectively charges particulate at velocities 7 to

                               350
-------
GAS
FLOW
                         ANODE  PURGE AIR
         ELECTRODE
         MAST-
         ELECTRODE-
FOCUSING
NOSE-
               -INLET      LEERING  OUTLET
               BELLMOUTH                BELLMOUTH
   FIGURE 1 - HIGH INTENSITY IONIZER ASSEMBLY
                              351
-------
10 times higher compared to the more conventional electrostatic
geometries.  Gas velocities in excess of 100 fps can be main-
tained through the HII with the charging time of a few milli-
seconds.  This means particles can be charged to a high level in
a very small volume resulting in a very compact system that can
be retrofitted to the existing control devices.

     A more detailed description of the HII technology can be
found in the EPRI Report FP-291 (1).

                   HII/ESP MOBILE PILOT PLANT

     The HII/ESP Mobile Pilot Plant combines an HII module with
a conventional ESP for field demonstration work.  The designed
nominal flow rate for the pilot unit is 11,894 M3/hr.(7000 ACFM).
At the nominal flow condition the specific collection area (SCA)
is 37.23 M2/(M3/sec)  (189 ft2/1000 ACFM).  Flow conditions for
determining flow rates and SCA are measured at the pilot plant
outlet.  This is due to the HII anode purge air which is injected
upstream of the first field in the ESP.  A schematic diagram of
the pilot unit is shown in Figure 2.

     The pilot plant is a trailer mounted unit.  It contains a
conventional wire and plate precipitator with three electrically
separate fields of equal size.  The ESP is a solid harp frame
design with flail hammer rappers for the collection plates and
electrical vibrating rappers for the H.V. harp frame and wires.
Heated purge air is injected into each insulator compartment to
keep the high voltage insulators clean and dry.  The flyash
collected in the hoppers is removed by a screw conveyor and
rotary air-lock valves.

     Located just upstream of the first ESP field is the HII
module.  It is a two ionizer array with the throats located
vertically one above the other.  Anode purge air is injected to
keep the anode vanes clean.  The purge air was heated for this
application to prevent any possible 803 condensation.

     The controls for the mobile pilot are located in a separate
control trailer.  The pilot is equipped with the capability of
changing its configuration to match the main ESP under study.
The three ESP fields and the HII field have separate controls
 (both manual and automatic) which allows for variation in current
densities and field strength.

     Operating parameters and test conditions for both the pilot
unit and the main ESP will be given in a future report to
ESEERCO.

     The gas is taken from the boiler flue by a slip stream and
enters the pilot through an insulated duct.  Inline heater con-


                                352
-------
           Insulator Compartments
                                                       HII Module
C Field
B Field
A Field
          Figure 2 - Mobile Pilot HII/ESP
-------
trols monitor the slip stream temperature and maintain the gas
temperature entering the pilot approximately the same as that
exiting the boiler flue.  At this installation, the slip stream
was taken off just upstream of the main ESP south compartment by
an isokenetic sampling scoop.  The flue gas passed through the
pilot and was reinjected into the boiler flue approximately 10
feet downstream of the takeoff location.

     The relative location of the pilot unit is illustrated in
Figure 3.  A detailed description of the Unit #1 at the Northport
Power Station will also be included in the later report to
ESEERCO.
                          TEST PROGRAM

Main Precipitator Testing

     Performance testing for the main ESP consisted of simul-
taneous inlet/outlet particulate measurements.  Simultaneous EPA
Method 17 tests for total mass determination and overall particle
collection efficiency were made.  Particle size distributions at
both the inlet and the outlet were determined by three methods.
Low pressure cascade impactors, electrical mobility analyzers
and electrical aerosol samplers (scanning electron microscopy)
were used to determine the particle size distributions.  Parti-
cle diameters measured ranged from .05 to 20.0 microns at both
the inlet and outlet of the precipitator.

     Testing was performed as described above at two boiler load
conditions.  A baseline sustained load was established at 362 MW
to 368 MW.  Testing was also done at a reduced load of 128 MW.

     Precipitator operating parameters were monitored during the
test program.  Boiler operating conditions were also monitored
with no testing being performed during an upset condition. Stable
boiler operation was insured by continuous measurement of 02, CO
and NOX emissions at the economizer exit.  Periodic measurements
for SOX were also performed on both the main ESP and pilot unit
using the Goksyor/Ross controlled condensation method.

HII-ESP Pilot Plant Testing

Determination of Pilot Plant Operating Parameters

     Establishing the ESP electrical operating parameters first
involved characterizing the main precipitator.  Voltage and
current characteristics were averaged for each electrical field
over the period of one month of continuous operation prior to
pilot testing.  Average current densities and field strengths
were calculated.  Since the V-I relationship for the main preci-
pitator fields did not exactly match those of the pilot unit, the
current densities in the pilot ESP were adjusted to equal those
for the main ESP.  The field strength at these current densities

                               354
-------
Air
Air
Heater
  I.D. Fan
                        Boiler
                               Economizer
                                           Air
                                           Heater
                                                 Air
                Flue
                Gas
                              Flue
                              Gas
                — Main Precipitator
                 Inlet Sampling Port-
                Electrostatic
                Precipitator
                                                 Inlet
                                             Sampling Port-
                                                      Hi I/ESP
                                          I.D. Fan |  |Pilot
                                                      Unit
                                            Outlet
                                        Sampling Port

                                         Electrostatic
                                         Precipitator
                          Stack
                        o
                              Main Precipitator
                              Outlet  Sampling Port-
                 Figure  3  -  Overall  Schematic
                              355
-------
was maintained at both conditions of HII energized and HII de-
energized.

     Physical characteristics such as wire to plate spacing,
total collecting area, and number of lanes were established and
set prior to operation of the pilot plant.  Such physical para-
meters were adjusted to simulate the main precipitator as close-
ly as possible.

     Testing on the pilot unit was performed at three gas flow
rates; low flow, baseline flow and high flow.  The baseline flow
was established by matching the face velocity in the last field
of the pilot ESP with the face velocity at the exit of the main
ESP.

Performance Testing

     Test methods for evaluation of the pilot plant performance
were identical to those used on the main ESP.  At  each of the
three gas flow rate  (SCA) conditions, simultaneous inlet/outlet
testing was performed with the HII energized and de-energized.

     Pilot plant operating parameters and boiler conditions were
monitored throughout the testing.  The boiler maintained a base-
line load condition within a range of 340 MW to 360 MW.  A sys-
tem load limitation towards the end of the testing required some
tests to be performed at a slightly lower load condition.

Presentation of Results

     A later report to ESEERCO, when released, shall include the
results of the testing program.  Particle size distributions will
be presented for both the main ESP and the pilot plant.  Particle
collection efficiency and penetration (both overall and fraction-
al) will also be presented.  Utilizing this data, the level of
reduction in particulate emissions from the main ESP will be pre-
dicted with the addition of an HII array.

                            ENDNOTES

Acknowledgements

     This pilot demonstration pilot project was funded by the
Empire State Electric Energy Research Corporation under ESEERCO
Project EP80-3.  Fred Weisz of Long Island Lighting Company's
Northport Station was the ESEERCO Project Manager.  Special
appreciation is given to the LILCO personnel for their coopera-
tion in this program.

References

1.  Schwab, J., et.al. (1976) , "Development Program for an Ionizer
    Precipitator Fine Particle Dust Collection System as Applied


                               356
-------
to Coal-Fired Utility Steam Generators", final report sub-
mitted to EPRI on Project 386-1, by Air Pollution Systems,
Inc., Seattle, Washington.
                            357
-------
                   PRIMARY AND SECONDARY IONIZATION IN AN
                    ELECTRON BEAM PRECIPITATOR SYSTEM*
           By:  W.C. Finney, L.C. Thanh, J.S. Clements, R.H. Davis
                Department of Physics
                Florida State University
                Tallahassee, Florida  32306

                                 ABSTRACT

     Recent investigation into the possibility of using energetic electron
beams for generating high ion current densities for particle charging in
electrostatic precipitators has shown exceptional promise.  Copious ion cur-
rent densities, at least 500 times that in a conventional corona driven pre-
cipitator, have been reported earlier.  Experimental results, however, have
indicated a secondary ionization phenomenon in a parallel plate system which
affects the charge density stability.

     The results of a study of the onset and extent of secondary ionization
are presented here.  Ion current vs. voltage experiments were performed using
the parallel plate system used in previous ion current density investigations.
The electron beam flux delivered to the system was restricted in intensity to
determine the transition from primary ionization to the saturation plateau,
then the onset and extent of secondary ionization was explored.  An analysis
of the; plate system's variable parameters was made to quantify the conditions
for initiation of secondary ionization.  This secondary ionization is thought
to be the result of the combined effects of plate spacing and plate voltage
in the precipitator section.

                               INTRODUCTION

     Electron beam ionization is the process of delivering an electron beam
into air, thereby creating a bipolar plasma of charged air molecules or ion
pairs, the number of pairs being dependent upon the energy and current of the
electron beam. Past experiments using electron beam ionization in coal fly ash
precipitator geometries have demonstrated measured ion current densities in
excess of two orders of magnitude greater than those generally obtainable in
a conventional corona-wire driven electrostatic precipitator (1,2).   These
experiments showed that the observed ion current vs.  applied voltage rela-
tionship can be approximated by a straight line using linear scale axes.
No saturation current plateau, which marks the total extraction of charge
from the ionization zone,  was observed.

     A preliminary investigation of the factors influencing the onset of
ion current saturation was initiated and the results were reported recently
(3).  Saturation of ion current density was defined as the process whereby
the "cloud" of ion pairs produced by the ionizing electron beam is totally
extracted from the bipolar interelectrode volume by an electric field
strength larger than a certain value.  It was found that the ion current vs.
voltage characteristics exhibited a well defined ion current saturation
plateau which depends on the electrode spacing, the beam energy and current,
                                    358
-------
and the type of beam collimation device used.

     This study will focus on primary and secondary ionization processes by
refining and extending the experimental measurements done in the preliminary
study (2).  Primary ionization is defined in this study as the plasma pro-
duced directly by energetic primary electrons from the electron beam  which col-
lide with gas molecules.  Secondary ionization in a parallel plate system occurs
when free electrons created in the primary ionization process gain energy
from high collecting plate electric fields and, in conjunction with space
charge phenomena, cause further ionization in transit to the electrode.

     The effect that separation distance, beam current, and beam geometry
have on the saturation plateau and secondary ionization will be addressed.
A comparison between the calculated and measured ion current saturation values
will be made.  A correction for recombination of opposite polarity ions which
is caused by incomplete separation and collection of the bipolar plasma
between the electrodes is calculated.  Finally, a quantitative analysis of the
amount and extent of secondary ionization will be made at different plate
spacings.

                            EXPERIMENTAL DETAILS

     The 3 MeV Van de Graaff electron accelerator at Florida State University
was used to generate an electron beam which was brought out into air from
vacuum through a thin aluminum foil window.  A movable beam current measure-
ment probe was located downstream from any beam restriction devices present,
if any.  The probe beam current is referred to as the "actual" beam current,
and represents the amount of flux available for ionization.  Electron beam
energy remained constant but beam current was varied in one of two ways:
either by controlling it at the accelerator cathode source or by mechanically
restricting the amount of beam allowed to leave the window.  In this investi-
gation, the accelerator was operated to deliver a beam of 1.2 MeV (million
electron volts)  energy and actual beam currents of 10, 20, 50, 100, 200,  and
500 nA, and 1.0, 2.0, 4.0, and 10.0 yA.  All experiments were run at 25°C,
70% relative humidity, and 760 mm Hg.

     A parallel-plate electrostatic precipitator test system incorporating
three electrodes each on an upper and lower tier was used to collect the ion
current  (Figure 1}.  Plates b, b1, d, and d1 served as guard electrodes while
plates c and c1  were ion current measuring electrodes.  The distance from the
foil window to the center of plates c and c1 was 50 cm.  All of the plates in
the system were biased at the same voltage; the upper set was at positive
(anode) and the lower set was at negative (cathode) electric potential.  To
determine the effect of plate spacing on ion current saturation, separation
distances of 2.5, 5, and 10 cm (1, 2, and 4 inches) were used.  This labora-
tory scale system is analogous to a conventional precipitator except that
the corona wires which produce an avalanche of unipolar working ions are re-
placed by an electron beam with bipolar working ions.

     Saturation phenomena were studied by mounting thick stopping baffles on
the end of the accelerator beam tube.  The baffles were 4" x 4" x V4" thick
aluminum plates each with a different diameter circular aperture drilled in
the center.   According to the range-energy relation of energetic electrons
passing through aluminum (4), all electrons not passing through the aperature
                                     359
-------
are stopped at the baffle.  Diameters (in inches) of the apertures are as
follows:  Vl6, 1/8, X/4, 5/16, 3/8, 7/16, 1/2, 1,  When no baffle was used,
the beam passed unobstructed out of the beam tube, the end of which is 2.0
in. diameter.  The purpose of using stopping baffles with apertures was to
restrict the ionizing flux entering the test volume between plates c and c'
to attempt to achieve ion current saturation.  The baffles allowed a rela-
tively disperse region of low intensity ionization to be delivered to the
test volume  as compared  to  a  tightly collimated  beam.  To determine what  role
beam geometry might play, several baffles with different sized apertures were
used at one actual beam current.

                                  RESULTS

Electrode Spacing Comparison

     Ion current vs. voltage curves for various experimental conditions were
plotted and analyzed.  Only the negative ion currents are shown since they
are approximately equal in magnitude to the companion positive ion currents
in every case.  The ion current characteristics of three electrode spacings
at two electron beam currents, each an order of magnitude apart, were com-
pared.  Figures 2 and 3 show ion current and ion current density plotted
against electric field at 2.5, 5, and 10 cm plate spacings.   The electric
field is an average value, i.e., it is the applied voltage divided by the
electrode spacing.

     At 10 nA beam current  (Figure 2), the 2.5 cm curve has a short straight
section but levels off above 500 volts/cm as the ion current saturates.  The
ion current increases at about the same rate from 0.5 to 19 kV/cm with very
little upturn at the end indicating secondary ionization.  This is in con-
trast to the 5 cm and 10 cm curves which do exhibit ion current enhancement.
The saturation bend on the 5 cm curve is more gradual and the plateau has a
greater slope than the 2.5 cm curve but the ion current saturation knee is
apparent.  Ion current enhancement effects begin at about 11 kV/cm and con-
tinue to increase until  sparkover.  At  10 cm spacing, there is  almost  no
plateau with the saturation knee almost immediately giving way to substantial
secondary ionization.

     A two order of magnitude increase in electron beam current to 1.0 yA
produces a different arrangement as shown in Figure 3.  The 2.5 cm curve
levels  off at 2  kV/cm but a slight upturn between 19 and 21 kV/cm may
indicate the onset of secondary processes.  A very gradual turnover marks the
transition from the linear to the saturation section of the 5 cm curve which
is terminated by a secondary ion current rise.   An almost linear 10 cm'curve
does bend slightly but no other features are distinguishable.  Secondary pro-
cesses undoubtedly occur at low electric field and completely obscure any
saturation plateau.

     The observation that more absolute ion current is collected at any single
beam current when the electrode spacing is larger is qualitatively accounted
for in several ways.  First, the  larger volume means  that more  of  the  ionizing
flux will fall between the electrodes and will be collected.  Additionally,
fewer energetic electrons still capable of initiating ion formation will be

                                    360
-------
absorbed by the plates.  Volumetric considerations alone, however, are not
sufficient to explain the shape differences of the curves.  Electrons pro-
duced in the primary avalanche must travel farther to reach a widely-spaced
electrode and therefore they encounter a greater number of molecules than at
smaller separation distances.  The additional electron-molecule collisions
increase the probability of ion pair production.  This explains the reduction
in electric field necessary to initiate ion current enhancement when separa-
tion distance is increased.

     In contrast, the onset of ion current saturation occurs at a higher
electric field at larger spacings because more of the primary ionization cone
falls within the interelectrode space.  Therefore, a higher applied field is
required to collect all of the primary ion current.  The combination of higher
electric field for the onset of saturation and lower electric field for ini-
tiation of secondary ionization as plate spacing is widened causes an increase
in the saturation plateau slope and a reduction in the absolute size of the
plateau.  At large spacings saturation is not observed because the rise in
copious secondary ionization compensates and then overrides the tapering off
of primary ionization.  A nearly linear plot makes it difficult to separate
the relative contribution of primary and secondary effects.

Electron Beam Current Comparison

     Ion current vs. voltage relationships for 5 electron beam currents at
a plate spacing of 2.5 cm  are  shown in Figure 4.  The curves are plotted on
linear axes to indicate their relative magnitude although the shapes of the
lower beam current curves are somewhat indistinct.  All 5 curves show the
characteristic linear, saturation, and secondary ionization regions.   The
onset voltage of saturation increases with higher beam current but secondary
multiplication begins at the same applied voltage.  The magnitude of the
saturation ion current increases and the asymptotic approach is more gradual
as the electron beam current is increased.  Raising the beam current creates
more primary ionization, requiring a higher applied electric field to com-
pletely separate the bipolar ion flux.  Plate spacing does not affect this
result for a given primary ionizing current.  The combination of a higher
onset voltage for saturation and the same onset voltage for secondary amplifi-
cation results in a smaller and steeper saturation plateau as beam current is
raised.

     Table 1 shows maximum ion current densities and sparkover voltages for
the entire set of experimental conditions.  A larger absolute amount of
collected ion current was obtained with higher electron beam current because
of a more densely populated ionization region.  Doubling the plate spacing
increases the maximum amount of ion current collected,  probably due to higher
plate voltages obtainable before sparkover and a larger interelectrode volume.
At 10 yA beam current and 10 cm spacing,  the highest ion current density
obtained was 82.5 mA/m2.  This value is in agreement with previous experi-
mental work (1), and is about 400 times greater than that generally found in
conventional corona wire precipitators (5).  Electrical breakdown alone limits
the maximum ion current value and larger values can be obtained by increasing
the separation distance, raising the electron beam energy or current, or
restricting the angular dispersion of the beam using collimation.

                                    361
-------
     Each of the experimental ion current vs. voltage curves was extended un-
til cathode-anode sparking halted the measurements  (Table 1).  When no elec-
tron beam was admitted to the test volume no ion current was measured until
plate-plate breakdown occurred.  An initial abrupt decrease in the sparkover
voltage is attributable to ionization in the test volume.  Following this
drop sparkover voltages remain constant except at high beam currents.
Widening the plate spacing increases  the voltage required to sustain a break-
down  as expected.   The effect,  however,  is non-uniform.

     Sparkover voltages measured in this parallel plate electrode system with-
out the electron beam are somewhat lower than other published values at  eash
plate spacing.  The electrode perimeters were rounded to suppress edge effects
but a strong field exists at the lateral edges of the center electrodes
(plates c and c') , which is where sparkover usually takes place.  At high
beam currents, a large electric field exists near the plates due to space
charge effects.  A high localized electric field is the major cause of pre-
mature sparkover.  When the electron beam is admitted to the test volume,
primary ionization reduces the effective interelectrode gap by producing a
disperse plasma at approximately ground potential between the plates (1).
Spark propagation becomes easier as the plasma-electrode distance decreases.

Baffle Aperture Comparison

     As described earlier, mechanical electron beam current variation by sub-
stituting baffles with different sized apertures also altered the beam geome-
try, producing a more disperse electron beam as aperture size was enlarged.
A comparison of ion current densities using different baffle apertures while
keeping electron beam current constant was performed.  Figure 5 shows ion
current vs. voltage plots at 10 cm plate spacing and 2.0 yA beam current
using aperture diameters of  /2", 1", and 2" (= no baffle).  Although similar
in shape, the plotted curves reveal that less ion current is collected as
aperture size increases.  The curves are characterized by a smooth rolloff
of ion current to 35kV followed by a very slight upturn.

     For a constant electron beam current, increasing the size of the baffle
aperture allows a more diffuse beam and a greater amount of scattering leading
to a lower beam density in the test volume.  Less delivered beam means a lower
ionization rate and therefore a smaller measured ion current.

                                DISCUSSION

Analysis of a Generalized I-V Curve

     The form of the collected ion current vs. applied plate voltage plots
reveal the complex nature of the interaction of several ionization phenomena.
A generalized I-V curve exhibits three regions (6).  From the origin of the
plot, the ion current rises linearly with increasing voltage because a larger
fraction of the ion flux produced by the electron beam is drawn to the elec-
trodes.  As the voltage is increased further the ion current asymptotically
approaches a saturation plateau, and at still higher voltages the current
increases rapidly until sparkover occurs.  Although each of these character-

                                    362
-------
istic regimes can be discussed separately, at high beam currents or large
plate spacings they often overlap, obscuring their precise delineation.

     During irradiation of air by an electron beam, both positive and nega-
tive ions are produced resulting in a bipolar plasma at nearly ground poten-
tial midway between the electrodes.  The negative ions are formed by electron
attachment to electronegative molecules while positive ions are formed by
electron-neutral molecule collisions which strip  off  other electrons.  At
zero applied voltage, the ionic charges in the interelectrode volume simply
recombine in a continous process of ionization and recombination.  When an
electric field is imposed, separation of charge occurs and as the electric
field increases, the fraction of ions which recombine decreases.  The current
value that is measured if all of the ions formed between the plates by the
electron beam are collected is the saturation current.

     The extent of the saturation plateau is limited by the onset of secondary
ionization  (7).  At sufficiently high electric fields an electron gains enough
energy in a single mean free path to ionize the next molecule with which it
collides.  This releases other free electrons so that a rapid multiplication
of electrons and ions occurs.  This explanation is consistent with the rapid
rise of the ion current as the electric field is increased.

     An example of the calculations which have been performed with the data
is shown below for the experimental conditions of 1.2 MeV electron beam
energy and 20 nA beam current.  A low value of beam current is used so that
the effects of space charge are minimal.

Saturation Currents and Recombination

     The actual saturation currents are obtained from the 3>V curves using
the method of Boag  (6):

     j =  (f)jg =  (1 ~ R)j  =  (1 - (2S2)maj)Js                            (1)

                                     6V2

Where:  j   = measured ion current density (statamps/cm)
        js  = saturation current density (statamps/cm)
        f   = collection efficiency
        R   = percentage of charge which recombines
        m2  — 35.7 V2-sec/esu-cm = a  constant characteristic of the gas
        a   = width of ionization region measured in the direction of the
              electric field  (cm)
        V   = applied voltage = E(2S)
        2S  = plate spacing (cm)

A plot of j vs.  ^2 should be a straight line with a slope of - (l/6)m2a(2S) 2Js
and a Y-intercept equal to js.  The saturation current can be obtained by
extrapolating the line backward to the Y-axis.  The saturation current plots
for 2S = 2.5, 5, and 10 cm are shown in Fig. 6.  The data points were fit with
a straight line using equation (8).  The best fit line with slope = (1/6)(36.7)2
(2) (2S)  was used for each value of 2S, and the electron beam was assumed to
have an effective diameter of a = 2 cm.  Theoretically, the saturation ion
currents should be the same for different plate spacings if the ion pair pro-
                                   363
-------
duction in the test volume remains constant  (Eq. 7).  As  seen  in Figure  6,  the
saturation current increases with increasing plate  spacing.  Because  of  the
angular dispersion of the primary ionizing flux, the configurations with
larger volume receive more electron beam and utilize it more completely,  and
there is less absorption of energetic electrons by  the plates.

   The saturation ion current densities  (js) calculated from Figure 6 are
converted to ion currents  (Is) and compared below to the  values of the satura-
tion plateau currents found from the I vs. E plot of the data,

                             Is. C|a A)      Plateau Current (yA)

                              1.8              1,8

                              3.2              3.0

                              5.3               ?

    The amount of recombination for 2S = 5 cm is given by:

    R = 1 - !_  =1- 3.0  = 6.25%, so 6.25% of the  ions recombine.        (2)
            Is       3.2

Secondary lonization

    The rapid increase of current at high electric  fields can be described
using the equation for secondary ionization (8):

    I - Ioe(a~n)d  for large d.                                           (3)

where:  Io = Is = saturation plateau ion current
        I  = measured ion current during secondary ionization
        T|  = attachment coefficient
        a  = Townsend's first coefficient
        d  = average distance primary charge travels to collecting plate  = S
        2S = plate separation

The primary charge consists of the charge liberated symmetrically about the
center of the test volume by the electron beam.  The primary charge liberated
by the electron beam forms the saturation current.   At high electric  fields
some of the electrons do not attach to the molecules with which they  collide
but instead ionize the molecules releasing more free electrons.  The  coeffi-
cient (a-T)) gives the relative degree of multiplication and attachment.   The
value of (a-r|)  is determined by plotting In (I)  vs.  S for a constant electric
field, as shown in Figure 7.  The coefficient  (a-T))  is given by the slope of
the line, which is a function of E/p, where E=energy and  p=pressure.   The
primary ion current  (Io) is given by the value of the Y-intercept.

    In(I) = (a-n)S + ln(I0)                                               (4)

    Since the primary currents (saturation currents) were not constant for
the different plate spacings, the 2S = 10cm currents (I)  were normalized  so
that Is(10cm)  = Is(5cm).  The following values of (a-T)) were obtained from
the slopes.
                                    364
-------
                        E(kV/cm)      (a-n)      (a-n)/P=76Q mm Hg

                           8          0.36           0.00047

                           6          0.236          0.00031

                           4          0.19           0.00025

The value of the primary ion current  from the Y-intercept in Figure 7 is
I0 = 3.4 pA.  This is consistent with the previous value of Is for 2S = 5 cm
of 3.2 yA.  As can be seen in Figure  7, secondary ionization occurs in the
region E = 4 to 8 kV/cm.  In the absence of an electron beam, the normal
range (9, 10) of electric field which produces secondary ionization is E =
20 to 30 kV/cm.  The electron beam causes the primary electrons to have a
higher average energy; therefore, they are accelerated to the ionization energy
of the molecules at lower electric fields.

                                CONCLUSION

    A parallel plate electrostatic precipitator test system using electron
beam ionization instead of corona wire energization was employed in a study
of primary and secondary ionization and ion current saturation.   A series of
experiments were conducted to investigate the nature of the ionization pre-
sent in the interelectrode volume.  Large ion current densities of 80 mA/m2,
or 400 times that in conventional precipitators, were found.  Sparkover vol-
tages were lowered  somewhat when the  test  volume was irradiated, but high
applied electric fields on the plates were  still possible.

    At lower electron beam current, the ion current vs. voltage curves have
linear, saturation plateau, and secondary ionization regions.  At higher
beam currents, the ion current saturation plateau region is not established
before secondary ionization begins.   The experimental conditions required
for saturation are low beam current or narrow plate spacing.  As the electric
field and plate separation distance were increased, a  greater amount of
secondary ionization  occurred.   Beam geometry, varied  using a series of stop-
ping baffles with apertures of different diameters, did not greatly affect
the saturation of ion current or secondary ionization.

    While the purpose of this study was not to focus on a complete mathema-
tical description of ion current saturation and secondary ionization,  many of
the mechanisms involved were quantified.   For all three plate spacings,  the
measured saturation currents are in agreement with those calculated from the
electron beam intensity and from the recombination equations.  Secondary ioni-
zation was described using the Townsend equation which yields values for
(a-n) .   Secondary ion multiplication occurred  at electric  fields between 4  and
8 kV/cm which  is much lower  than the  normal range  (E  = 20 to 30 kV/cm) with
no electron beam.

                              ACKNOWLEDGEMENTS

    The authors gratefully acknowledge the assistance of Mr. Kenneth J.
Schafer and Mr. R.  Mark Whitton in all phases of this study.  We also acknow-

                                    365
-------
 ledge Mr.  Dempsey Lott and Mr.  Walt Phillips for their operation of the 3
 MeV electron accelerator.

 *Work supported in part by the U. S. Department of Energy, Contract No.
 DE-A521-78EP11316.

 •{•Present address;  Research-Cottrell, Inc. ,  Somerville, New Jersey.

                                  ENDNOTES

 1.  Finney, W.  C., L.  C. Thanh, and R.  H. Davis.  Ion Current Densities Pro-
     duced by Energetic Electrons in Electrostatic Precipitator Geometries.
     In:  Proc.  of the Second Symposium on the Transfer and Utilization of
     Particulate Control Technology.  Volume  II, Electrostatic Precipitators.
     pp. 391-398, EPA-600/9-80-0396, September 1980.

 2.  Thanh, L. C., W. C. Finney, and R.  H. Davis.  The Effects of Back Corona
     Discharge on Ion Current Density in Electron Beam Precipitators.   In:
     Conf.  Record of IAS-IEEE Annual Meeting, Cleveland, Ohio, p. 126, October
     1979.

 3.  Thanh, L. C., W. C. Finney, and R.  H. Davis.  Total Separation of Charges
     Produced by Electron Beam lonization in  a Precipitator Development System.
     In:  Conf.  Record of IAS-IEEE Annual Meeting, Cincinnati, Ohio. pp. 935-
     940.  October 1980.

 4.  Berger and Seltzer.  Tables of Energy Losses and Ranges of Electrons and
     Positrons.   NASA SP 3012.  p. 125.   c.a. 1960.

 5.  White, H. J.  Industrial Electrostatic Precipitation.   Reading, Mass.,
     Addison-Wesley, 1963.

 6.  Boag,  J. W.  lonization Chambers.  In:   Radiation Dosimetry, Volume II.
     Attix, F. H. and W. C. Roesch (eds).   Academic Press,  pp. 11-17,  1966.

 7.  Loeb,  L. B.  Basic Processes of Gaseous  Electronics.   University of
     California Press,  pp.  647-727, 1961.

 8.  Loeb,  L. B.  Basic Processes of Gaseous  Electronics.   University of
     California Press,  p. 712, 1961.

 9.  Prasad, A.  N. and J. P.  Craggs.  Measurement of lonization and Attachment
     Coefficients in Humid Air in Uniform Fields and the Mechanism of Break-
     down.   In:   Proc.  Phys.  Soc. London.   76_:223, 1960.

10,  Dutton, J.   Electron Swarm Data.  J.  Phys.  Chem. Ref.  Data.  4:577, 1975.
                                      366
-------
       TABLE 1.  MAXIMUM ION CURRENT DENSITY AND SPARKOVER VOLTAGE
Electron Beam
Current
CnA)
0
10
20
50
100
200
500
1000
2000
4000
10000
FOR VARIOUS
BEAM CURRENTS AND PLATE SPACINGS
Maximum Ion Current
Density (mA/m2)
2.5 cm 5 cm 10 cm
-
0.2
0.4
0.5
0,9
1.2
3.8
5.7
12.0
19.5
36.3
-
0.6
1.0
1.5
2,7
4.0
8.5
13.0
20.3
33.3
76.3
-
1.0
1.2
1.8
3.3
7.0
12.8
20.8
30.8
50.0
82.5
Sparkover Voltage
(plate-to-plate) (kV)
2 . 5 cm 5 cm 10 cm
56
50
50
50
50
50
50
50
50
50
50
84
70
70
70
70
70
70
66
62
62
58
110
82
82
82
82
82
82
82
82
82
72
                      Baffle with

                      Aperture
      Accelerator
         Tube _  ..
               Foil
             Window
          Electron

            Beam
                                                    Plates
Figure 1.  Schematic diagram of the experimental apparatus.  The electron

    beam passes through the baffle aperture and into the interelectrode

    region.  L = 10 cm, foil window to center of plates c,c' = 50 cm.  Upper

    plates are anode, lower plates are cathode.


                                  367
-------
               Electrode Spacing Comparison  -
                     Electrode Spacing Comparison
                     Electron Beam Current -\u. A
                                           Figure  2.  Comparison of ion current  vs.
                                           applied electric  field curves.
                          Figure 3. Comparison of  ion current vs.
                          applied electric field curves.
   750'

   700


   600



   500
<

~  400
o
3
(J
a>  300
-o
o
c
<
   200


   100
Beam Current Comparison
Electrode Spacing =2.5cm
35


30





   S
20 S
   >\
   "to
   c
   o>
15 2
   C
   £

10 f
         Figure 4. Ion  current vs.  plate voltage
         comparison for five electron beam currents.
                                                       TOO,
                                                              i    I    i    i    I   i    i    r
                                                             Baffle Aperture Comparison    |/2"
                                                       600- Electron Beam Current =
                                                             Electrode Spacing=IOcm
                                                       soo'
                                                     -400
                                                      = 300
                                                      !200
                                                        100
                                                                  10   15  20  25  30   35  40
                                                                     Plate Voltage (kV)
                                                                35

                                                                30

                                                                25 N£
                                                                    x

                                                                20  -f

                                                                 15  I
                                                                    Q
                                                                    C
                                                                    O
                                                                 101
                                                                    I


                                                                 0
        8    12    16   20
        Plate Voltage (kV)
2426
                                                           Figure 5. The  effects of  three different
                                                           sized baffle apertures on the ion  current
                                                           vs.  plate voltage characteristics.
                                          368
-------
                                                                     -5.
                                       Saturation Current Calculation
                                       Electron Beam Current = 20 n A
                                                 16   18  20   22  24
                                         cm2V2


Figure 6.  Straight line fits to  j  vs. j/V2  plots used to calculate the

     saturation  currents.

                                             , - , - ,2.0
                       20
                     IB
                     o
                     o
                     c
                              Townsend Coefficient Calculation
                              Electron Beam Current =20 nA

                                                    kV/cm
                                                    E=8
                                                          1.0
                                                         0.6
0.4 E
    in
    c
    a>

0.2 °.
    c
    o>
    k_
    h.

    o

O.I  o
                                     2.5    r .   .   5
                                           S(cm>

Figure  7.  Ion  current vs.  S comparison used to calculate Townsend's first

    coefficient.
                                        369
-------
           INFLUENCE ON PARTICLE CHARGING OF ELECTRICAL PARAMETERS
                          AT DC AND PULSE VOLTAGES

                By: Hans Joergen Joergensen
                    Jesper Toexen Kristiansen

                    High Voltage Laboratory, Building 329>
                    Technical University of Denmark,
                    DK-2800 Lyngby, Denmark.

                    Preben Lausen,
                    F.L. Smidth & Co. A/S,
                    Vigerslev Alle 11,
                    DK-2500 Valby, Copenhagen, Denmark.

                                   ABSTRACT

     Enhanced particle charging is considered to be one of the reasons for the
improved performance of pulse energized precipitators.  For a pulsed condition,
the charging levels obtained are influenced by several  electrical parameters
such as underlying DC voltage, pulse level, and pulse repetition frequency. In
the investigation reported, these influencing factors are examined. Themeasure-
ments are performed by a Faraday cage method on a conducting ball of 3 mm dia-
meter, charged within the interelectrode space of two circular cylindrical elec-
trodes. The investigation includes DC as well as pulse  energization. The re-
sults of the DC situation are compared with the saturation charges  expected
from the field charging theory. In the case of pulse energization, a simple model
for the field charging under pulsed conditions is suggested and is  used for a
similar comparison.

                                 INTRODUCTION

     For the DC situation, the basic theory of field charging was  established
by Pauthenier (l), while the pulse situation has been treated by Masuda (2).
Masuda's theory covers the case where ionic space charge is present during the
pulse as well as in the period between pulses.

     This paper deals with the situation where the DC level is kept below co-
rona onset, and the pulse repetition frequency is low enough to allow a com-
plete removal of ionic space charge before application  of the following pulse.
Both calculations and measurements are performed with negative voltage, but
when discussing the results, numerical values are used  for convenience.

                         THE FIELD CHARGING PROCESS

     In the DC situation, free electrons are constantly created in  the ioniza-
tion zone near the discharge electrode and attached to  electronegative gas mo-
lecules to form negative ions just outside this region.  Because of the constant
flow of ions from the discharge electrode, the electric field strength and the
ion density do not vary with time at a fixed position in the interelectrode
space.  For this quasi-static case, Pauthenier derived the classic equation for
the field charging of a spherical conducting particle (l):


                                     370
-------
                          q(t) = 12-rre Ea2 t/lt +——                       (1)
where
                     charge of particle as function of time (C)
                     electric field strength                (V/m)
                     radius of sphericaJ particle           (m)
                 E
                 a
                e
                     vacuum permittivity                    (F/m)
                     ionic space charge density             (C/m3)
                     ion mobility                           (m2/Vs
     With pulse energization where short duration pulses are superimposed on  a
DC voltage, the corona emission is limited to the pulse period if the DC  level
is below corona onset. If in addition the time between successive pulses  ex-
ceeds the transit time of the ions, the interelectrode space will be free of
ionic space charge when a pulse is applied.

     During the negative pulse voltage, electron avalanches are initiated.  The
avalanches leave a positive space charge behind, and the electrons attach to
gas molecules in regions of lower field strength to form negative ions. The po-
sitive ions quickly reach the cathode, while the negative ions travel towards
the anode and only traverse part of the distance in the pulse period.

     A particle situated at a fixed position outside the region close to  the
discharge electrode thus experiences variations with time in both electric
field strength and ion density. In the pulse period, a high field strength is
reached, but no ions are present, while later when the negative ions arrive at
the position of the particle, the ion density is high but the field strength
lower than in the pulse period. As field charging requires the presence of ions
the saturation charge of the particle is determined by the maximum field strength
during the passage of the space charge cloud.

               MATHEMATICAL MODEL OF THE FIELD CHARGING PROCESS

     A simplified mathematical model has been established for calculation of
the charge attained by a spherical, conducting particle in a pulse energized
concentric-cylinder electrode arrangement. The model is based on the following
assumptions:
     - The corona discharge occurs instantaneously at a certain moment in the
       pulse period and is uniformly distributed over the emitting electrode.

     - The electrons generated in the corona attach to gas molecules and  form
       an ionic space charge cloud limited by two cylindrical surfaces.

     - The positive ions immediately travel to the inner electrode.

Temporal Variation of Field Strength and Space Charge Distribution

     The spatial and temporal variations of the electric field strength and the
space charge distribution are determined by successive calculations of the
electric field and the movement of the space charge.
                                      371
-------

4 R

-
S-, >
' "f :
Em!
elec

2r0
SJ
tr
•

•
>ion
ode
r2(t)
ri 0) k.
A 	
Ion
cloud
'•: '-.. '•:
«yl
Grou
electi

nded
•ode

                                      ' I ^^
Figure 1. Negative space charge cloud between concentric-cylinder electrodes.

     Figure  1 shows the situation at a time t.   The radii of the concentric-
cylinder electrodes are rn and R, respectively, and the negative space charge
cloud is situated between r.(t) and r (t).  Using the rotational symmetry and
Gauss' law, the electric field strength at  radius r is given by:
E(r,t) =
      2irE()r
                                               for
                          rQ < r < r (t)
[^(t
        'r^t)

        (•r2(t)

        'r/t)
p(r,t) 27rrdr]/2-rre r   for   r (t)  < r  <
                                                (2)
                         p(r,t)
                    for   r (t)  < r < R
where
               E(r,t) :  electric field strength                   (V/m)
                a (t) :  charge per unit length on inner electrode (C/m)
      r (t) and r (t) :  inner and outer radius of ion cloud       (m)
               p(r,t) :  space charge density of the ion cloud     (C/m3)
                   e  :  vacuum permittivity                       (F/m)

     The negative space charge per unit length, q , is given by the integral:
ci  =
                                 r2(t)
                           p(r,t)  2-rrrdr
                                                                           (3)
The mean value of q  is:
                                   1C = i/f                                (1;)

where i is the average corona current per unit  length (A/m)  and f is the pulse
repetition frequency (s"1).  To calculate E'(r,t)  from equation (2),  the space
charge distribution p(r,t)  has to be known.  In  the model,  p(r,t)  is assumed to
                                      372
-------
 be of the form:
                        p(r,t) = qc/[27rr(r2(t) - r  (t))]
                                   (5)
      Using equations (3) and (5), E(r,t) becomes:
                                                  for
                              r-r (t)
                                                        rQ < r <
                             ft) -
    E(r,t) =
 Integration of equation (6) yields:
                                                  for
                     < r < r (t)
               r2(t) < r < R
                                                                   (6)
= |2ire0U(t)-qc|l + loglFTn
                                  R
   r/t)

r (t)-r (t)
                                                     log
                                                         r,(t)
/log MM    (7)
 where U(t) is the potential of the inner electrode. Hence, if  U(t) and the posi-
 tion of the space charge cloud is known, E(r,t) can be calculated us ing equations
 (T)  and (6).

      Subsequently, the movement of the space charge in a small time  increment
 dt is calculated. Only the movements of the boundaries of the space charge cloud
 are  determined,  while the distribution of the space charge within the cloud is
 assumed to maintain the form given in equation (5).  The inner radius of the space
 charge cloud  at  the end of a small time increment dt is:

                          r^t + dt) = rt(t) - bE(r](t),t)dt                 (8)

 where b is the mobility of the negative ions.  The mobility is  taken  as  a con-
 stant.  Using  equation (8) and a similar expression  for r  (t + dt), the new lo-
                      -10
                     E -8
                     u
                     oi -6
                     I
                      -4
                     u
                    in _
   UDC=-32.6kV
    UP = -30.0kV
    q_ = -l23jiiC/m
                               2468
                                   Radial distance, cm
                10
Figure 2. Calculated temporal and spatial variation of electric field strength
     with concentric cylinder electrodes (r  = 1.5 mm, R = 100 mm).

                                      373
-------
cation of the space charge is found.  In the computer program used, the space
charge cloud is divided into several  parts the movements of which are calcula-
ted individually.

     A result of such step by step calculations is shown in figure 2. The in-
ner and outer radii of the concentric cylinders are 1 . 5 nun and 100 mm, respec-
tively. The pulse wave form used is shown in figure 5,  and the initial posi-
tion of the space charge is assumed to be between the inner electrode and a
cylindrical surface of radius 20 mm.

Charging of a Spherical, Conducting Particle

     The calculation  of the charge acquired by a spherical, conducting par-
ticle situated at a fixed position in the inter electrode space is performed
treating the passing of the ion cloud  as a series of quasi-stationary DC states.
In a short time interval dt, the electric field strength and the ion density
can be taken as constant, and the classic equations of field charging are ap-
plied. If the particle of radius a is placed at the radius r , the increase in
particle charge q(t) in the time interval dt is (l):

              dq(t) = 3ira2E(r  t)p(rp,t)b(l - _4||__^ 2dt           (9)
                                         v        0     p   '
     Initially, the particle is assumed to be uncharged. As the particle ac-
quires a charge, however, the repelling field from this charge is of sufficient
magnitude to prevent ions from reaching the particle during part of the passage
of the ion cloud. Thus charging takes place only if the following condition is
fulfilled:                               .
     The saturation charge is given by the Pauthenier limit  for the maximum
field strength in the presence of space charge at the position of the particle:

                           q  ,  = 12ir£na2E(r ,t )                           (1 l)
                           ^sat       0     p' max                       v

     The temporal variations of the field strength and the ion density at  the
particle position are determined using the mathematical model described above.
Then the passage of the ion cloud is divided into short time intervals, and the
charge flow to the particle in each of these intervals is  calculated by means
of equation (9) if condition (10) is satisfied. This procedure is repeated to
yield the particle charge as a function of the number of applied pulses.

     In figure 3, a comparison is made between the calculated charge acquired
by a conducting sphere of 3 mm diameter under pulsed and DC  conditions, respec-
tively. The DC curve is based on equation (1) using the formulas for the elec-
tric field strength in a coaxial cylinder geometry given in  (1). The saturation
charge is considerably higher with pulse than with DC energization, and de-
spite the slower convergence, the calculated pulse charge  exceeds the DC charge
at all instants.
                                      374
-------
                0)
                1
ra
o
T
4
w
"o
0>
E>
OS

i
a
1
                   -0.3
                   -0.2
                   -0.1
        	Saturation^ Charge, pulse
        Pulse (upper curve):
        UDC
        U
                                       U
                                        DC
                                         (lower curve):
                                          =-37.4kV
                                          = -123(L(A/m
                                                   I
                                         1
                    0.01        0.1
                                    Time, s
Figure 3. Charging of  a  conducting sphere
                                   10   30
                                           [a = 1.5 mm) in a concentric cylin-
     der geometry  (r   =  1.5 mm,  R = 100 mm). Particle position: r  = 60 mm.
                    U                                             p

                            EXPERIMENTAL ARRANGEMENT

     The measuring principle is  similar to the ones used by Smith (3) and Ma-
suda (^), where a  large,  spherical particle is suspended in the interelectrode
space by means of  an insulating  string, and the particle charge is measured in
a Faraday cage. In this arrangement, however, two Faraday cages are used: one
measuring the particle charge plus the charge on the adjacent part of the string,
                                       0
Figure U. Experimental arrangement.  1: Part of Faraday cage. 2: Pneumatic cy-
     linder. 3: Pneumatic motor.  U:  Suspended particle.
                                       375
-------
the other measuring the charge on part of the string above the particle. The
particle charge is found as the difference between these measured values.

     The experimental setup is shown in schematic form in figure h. A grounded
metal case containing the two Faraday cages and an air motor is placed above
the concentric cylinders. The string to which the 3 mm brass ball is glued is
fastened to the pulley of the motor, and is weight loaded to minimize deflec-
tion. The cylindrical Faraday cages each consists of two movable parts mounted
on the insulated piston rods of two pneumatic cylinders.

     The brass ball is charged in the interelectrode space using the pulse
wave form shown in figure 5 and moved to the position between the two parts of
the lower Faraday cage. The two cages are closed by activation of the pneuma-
tic cylinders, and the charges are measured simultaneously with two electro-
meters. A contact spring inside the Faraday cage ensures complete charge trans-
fer from the ball to the cage.
o
              -70

              -60

              -50

              -40

              -30

              -20

               -10
                    Time constant of
                    decay: 260 JJLS
                                  UDC = -32.6kV
                                   Up=-30.0kV
                    45  85
                     200       300
                        Time, /JS
400
500
      Figure 5. Pulse wave form used in calculations  and measurements.

                            EXPERIMENTAL RESULTS

     The 3 mm brass ball was positioned at a distance of 60 mm from the axis
of the coaxial cylinder arrangement with radii 1.5  and 100 mm and height 170 mm.
The particle was charged by applying approximately 3000 pulses to the electrodes.-
In each series of measurements, 10 individual measurements were performed, and
the mean value and a 95 per cent confidence interval  were  calculated,  assuming
the measured values to possess a normal distribution.

Variation of Pulse Voltage

     Initially, the performance of the measuring apparatus was checked by car-
rying out a number of DC measurements  and comparing the  results to calculated
values. A few of these measurements are shown in figure  6  together with the
calculated curve. Though deviations from the expected values did occur, the

                                      376
-------
general  agreement was satisfactory.
                       -0.4
                     U
                     o>" -0.3
                     s.
                     u
                     •c
                     0)
                     Q.
                     in
                     o
-0.2
                        -0.1
Pulse (upper curve):
UDC =-32.6KV
I = 100 pps
  Up =-10 KV
  Up =-20 kV
  Up=-30kV
  Up ~-40kV
- Up=-50kV
- Calculated T
  curve    I
                        DC (lower curve):
                        v : Measurements
                        _: Calculated curve
                          -10   -20     -50   -100  -200
                                   Average current, pA/m
                                   -500
         Figure 6.  Measurements  with DC and with varying pulse  voltages.

      Measured values of the  charge acquired with pulse energization and vary-
 ing  pulse amplitudes are also  shown in figure  6. The calculated curve gives
 the  charge obtained after the  application of 3000 pulses. This value corre-
 sponds  to approximately 95 per  cent of the saturation charge,  cf.  figure 3. The
 particle charge  is seen to rise substantially  with increased pulse voltage.
 This is mainly due to an increase in the amount of emitted  space charge and
 hence,  a larger  field enhancement at the particle position.

 Variation of Pulse Repetition Frequency
                   -0.5
                 O
                 •
                 r -0.4
                 s.
                 in
                 •5 -0.3
                 fr
                 10
                 6
                           10pps
                     50 pps  100 pps   200 pps
                             UDC = -32.6 kV
                              U = -30kV
                     -10
               -50     -100    -200
            Average current, /uA/m
                                                                -500
Figure 7.  Charge acquired with 3000 pulses at varying repetition frequencies.
                                        377
-------
     As long as the time between consecutive pulses exceeds the transit time of
the ions, the average corona current is found to be proportional to  the  pulse
repetition frequency (5), and therefore the mean value of the space  charge emit-
ted per pulse is constant, cf. equation (k). Hence, a variation of the  repeti-
tion frequency should not cause any change in the acquired charge  provided
that the number of applied pulses is the same at all frequencies. An experimen-
tal verification of this is shown in figure 7- The full curves are the calcu-
lated values at the four repetition frequencies while the dotted line  is the
calculated charging level corresponding to the average emitted charge per pulse
for all the measurements.

Variation of DC level

     In order to determine the influence on the particle charge of the  DC volt-
age, a number of measurements were carried out with varying DC level. The total
maximum voltage (Up+U^) was kept at -60 kV and the average current  therefore
fairly constant. The results are shown in figure 8 with calculated values cor-
responding to the mean value of the corona current for each of the three se-
ries. The particle charge is seen to increase with increasing DC level.
                U
                I
                o>
                "5
                o»
                U
                    -0.4
-0.3
                    -0.2
                    -0.1
-128/JV/m —
                       -124juA/m —
                    UP+UDC=-60kV
                    f  100pps
                    o Measured value
                    • Calculated value
                                                DC Corona
                                                  Onset
                              -10      -20      -30
                                   DC Voltage, kV
                                   -40
      Figure  8. Charge acquired with pulse energization at varying DC
           levels and constant peak voltage.

                                  CONCLUSION

     The  simple mathematical model suggested in this paper yields results  in
good agreement with experimental values. Even though the model  is established
for concentric cylinders, the concept of an ion cloud traversing the interelec-
trode space in the time between consecutive pulses is  equally valid  for  more
practical electrode configurations and therefore qualitatively  similar results
would be  expected in these configurations.
                                      378
-------
     To obtain a high particle charge, the peak voltage should be as high as
possible, as both field strengths and ion densities increase with increasing
peak voltage.

     A certain peak voltage can be reached using a low DC level and a high pulse
voltage or vice versa. In the investigated range of DC levels below corona"onset
the particle charge increases with increasing DC voltage.

     The pulse repetition frequency itself is of no importance to the charge
acquired by pulse energization as long as the time between pulses is sufficient
for the ions to reach the collecting electrode. The particle charge is only de-
termined by the number of pulses applied.

     A comparison between DC and pulse energization with the same average cur-
rent  shows substantial improvement in particle charging by using pulse energi-
zation. Furthermore, under pulse conditions a high particle charge can be
reached at very low current densities.

                                  REFERENCES

 1. Pauthenier, M. and M. Moreau-Hanot. La charge de particules spheriques dans
   un champ ionise. Journal de Physique et le Radium, Paris, France,  series 7,
   vol. 3, pp. 590-613, 1932.

2. Masuda, S. and A. Mizuno. Maximum Charge of a Spherical Particle Imparted
   by Pulse Charging. Proc.  1978 International Workshop on Electric Charges in
   Dielectrics, Oct. 9-12, 1978, Kyoto, Japan.

3. Smith,  P.L. and G.W- Penney. The Charging of Nonspherical Particles in a
   Corona  Discharge. Trans AIEE (Commun. Electron.) Vol. 80 I, p. 3^0-3^6,
   1961 .

h. Masuda, S. and M. Washizu. Corona Charging of a Spherical Particle Having
   an Extremely High Resistivity. Proc. 1978 International Workshop on Elec-
   tric Charges in Dielectrics, Oct. 9-12, 1978, Kyoto, Japan.

 5. Petersen, H.H. and P. Lausen. Precipitator Energization Utilizing an Energy
   Conserving Pulse Generator. Presented at the 2nd Symposium on the Transfer
   and Utilization of Particulate Control Technology, Denver, Colorado, July
   1979.
                                      379
-------
              BOXER-CHARGER MARK III AND ITS APPLICATION IN ESP's

            By:  Senichi MASUDA, Hajime NAKATANI and Akira MIZUNO
                Department of Electrical Engineering, Faculty of
                Engineering, University of Tokyo
                7-3-1,  Hongo, Bunkyo-ku, Tokyo, Japan 113

                                  ABSTRACT


     The latest model of Boxer-Charger denoted by "MARK III" uses for its
electrode assemblies the double-helix electrode units and very fast-rising
pulse voltages with 40 - 200 ns duration time for their excitation to produce
plasma ion source.  The helical discharge wires of a double-helix unit are
insulated not by the ordinary insulators but by coils which effectively reflect
such a fast-rising, short-duration pulse voltage.  This is called "inductance-
isolation", and contributes much to the ease in construction and lowering of
costs of the electrode assemblies.  Its charging performance is presented.
Then, the effect of pre-charging on the collection performance of a conven-
tional ESP operated with a dc voltage under severe back discharge condition,
a pulsed ESP operated with a pulse voltage superimposed to a dc voltage, and
a conventional ESP operated under no back discharge condition, is presented
and compared to each other.  It is shown that the collection performance at
dust resistivity of 10-'--'- - 10   ohm-cm can be greatly increased to a level
equal to that obtainable under no back discharge condition, when the pre-
charging is made in front of each collection field repeatedly and the collec-
tion field of an ESP energized with dc-plus-pulse voltage.
                                INTRODUCTION

     Boxer-Charger "Mark III", its third generation, is one of the ideal pre-
chargers capable of imparting very high level of charge to high resistivity
particles within a short time [1]. This allows the exact evaluation of the
requirements to be met in the collection fields of an ESP to be located down-
stream.  Since the enhanced charge of the particles from the precharger should
be diminished in the collection field subjecting to back discharge, it is con-
sidered to be imperative to suppress its effect by using the "back discharge
free" collection field, such as the pulsed field or parallel-plate field.  In
the present paper is reported the results obtained with the pulsed field. It
is expected that a slight quantity of ionic current, lower than its threshold
to cause back discharge, must be supplied to avoid a severe rapping loss by
electrical adhesion force.  The pulse charging superimposed to dc voltage can
provide the ionic current at any desired level without losing its uniformity
in distribution and also the required level of collection field strength.
From the same reason the parallel-plane field is considered as vulnerable to
rapping loss although free of back discharge.  However, its test is to be made
soon, and its results will be reported later.  It should be added that special
precautions are taken in the present tests of pulse charging so as to keep the
dc voltage below a critical level of causing spontaneous lateral propagation
of back discharge.
                                      380
-------
1. Experimental Apparatus

1.1 Pulse-Energized Double-Helix Type Boxer-Charger  (Mark III)

     Figure 1 shows a  double-helix electrode assembly  to  construct  the  latest
model of Boxer-Charger "Mark  III"  used in the present  tests.   This  configura-
tion has its inherent  advantages in (1)  maintaining  the wire-to-wire  gap at a
small and constant value without being affected by thermal deformation  of  its
supporting system, (2)  avoiding  the edge-effect to cause  corona  discharge  in
the unexcited period,  and  (3)  ease in its support.   The conditions  (1)  and (2)
are highly important to raise  the  peak value of ac main field between two  elec-
trode assemblies so as to  increase the particle charge.   But a small wire-to-
wire gap does not allow the stable coronas to appear uniformly along the wires
owing to excessive sparking.   This difficulty is solved by using a  very short
pulse voLtage with 40  ns duration  time which proceeds  along the  helix wires in
a form of a travelling wave producing very active and  uniform streamer  coronas
as an ideal plasma ion source  without causing any sparking at all (see  Figure
l(b)).  The use of such a  short  duration pulse voltage makes possible the  "in-
ductance-isolation" to be  used for electrically insulating the two  helix wires
from each other (Figure l(c)).  An inductance element, reflecting the wave,
can be used instead of an  insulator.   This contributes much to simplifying the
electrode construction and reducing its  cost.   The general description  of
Boxer-Charger is given in  the  references[1-3].
     Figure 2 indicates a  photograph  of  Boxer-Charger  I of Mark  III type in-
stalled within the inlet duct  of a test  ESP.   This consists of two  parallel
electrode assemblies spaced at 10  cm (face-to-face), each  electrode assembly
consisting of 3 double-helix  assembly units arranged on a  plane.  Its effec-
tive length in the gas flow direction is 60 cm, and  the residence time  0.040 s
at gas velocity of 15  m/s. The  theoretical charge at  this residence  time  is
54.1 % of Pauthenier's saturation limit  whereas  its measured value being  47.4
%.  The sparking field strength  between two parallel assemblies  is  5 kV/cm at
100 deg C, and 4.5 kV/cm is used.
     Figure 3 shows a  photograph of Boxer-Charger II of Mark III type located
inside an inter-field  section  between the first and  second  collection fields
of the ESP.  Six double-helix  assembly units  are arranged  at 10  cm  spacing
(face-to-face) on a plane  perpendicular  to the  gas flow direction,  so that this
configuration is especially suitable  for being  inserted into a narrow space
inside an ESP.  The gas velocity is 0.82 m/s  to provide 0.073 s  residence  time
for an effective length assumed  at  6  cm.   The  theoretical  charge for this  res-
idence time is 68.2 %  of Pauthenier's saturation limit for  the charging time
constant of 34 ms.
     Figure 4 shows a  circuit  diagram of the  power supply  providing a main high
voltage with 50 Hz to  be applied between the  electrode assemblies to construct
the ac charging field,  and two pulse  high voltages with 40  ns duration  time
(Figure 5), each to be  applied to  one of the  two electrode  assemblies facing
each other between its helical wires  at  an instant when it  takes the negative
peak of the main ac voltage.   As a  result the plasma ion source  out of  contin-
uous streamers is produced alternately on one of these two  electrode assemblies
in synchronous to the main ac  voltage, and negative  ions are extracted  from
this plasma by the main ac field to produce the alternating current of negative
monopolar ions within  the  charging  field.   The  dust  particles are bombarded by
these ions from both sides alternately,  and charged  up very quickly.

                                      381
-------
 1.2  Race-Track System and Model ESP

      Figure 6 indicates schematically the laboratory race-track system com-
 prizing a  model ESP which has three collection fields in series.  Each collec-
 tion field consists of two parallel ducts having a plate-to-plate spacing of
 30 cm,  effective height of 70 cm and a length in gas flow direction of 50 cm.
 The  discharge electrodes are wires with 3 mm in diameter arranged on each cen-
 ter  plane  at 6 cm wire-to-wire spacing.  This small spacing is to raise the
 corona  starting voltage, which is adjusted by adding fine-shaped projections
 along each wire at 5 cm spacing.  The voltage-current characteristics of each
 field collecting high resistivity test fly-ash with IQH - 10-^ ohm-cm at 100
 deg  C gas  temperature indicates a very distinct hysteresis, owing to the lat-
 eral propagation of back discharge initiated at one point.  Its corona start-
 ing  voltage is 42 kV at 100 deg C in air under contaminated condition, whereas
 the  back discharge extinguishing voltage 23 kV.  The resistivity of test fly-
 ash  drops  under IQlO ohm-cm at 20 deg C.  The circulating gas is room air, and
 its  temperature is changed by an electric heater up to 200 deg C for varying
 dust resistivity.  The standard test conditions are 100 deg C gas temperature,
 1,350 m-Vh gas flow rate (15 m/s at the inlet-duct and 0.82 m/s at collection
 field), and 10-^ - 10^  ohm-cm dust resistivity, but test is made also at nor-
 mal  temperature to obtain the reference data under no back discharge condition.
 The  mass-loading of dust is measured at both the inlet and outlet of the ESP
 using two  sampling filters and two dust-loading monitors (Konitest).  The
 charge-to-mass ratio of dust is also measured with two suction-type Faraday
 cages at the inlet and outlet.

 2.  Experimental Results

      The tests are further made under three different operation modes of the
 ESP  with and without the prechargers operated:
  i)  a dc voltage is applied to each field of ESP under high dust resistivity
     condition to cause severe back discharge (dc operation mode under back
     discharge),
 ii)  a dc voltage not to cause back discharge propagation, ca. -35 kV, is appl-
     ied to each field as a base-voltage, and a pulse high voltage with a crest
     voltage of -20 kV, duration time of 1 ms and repetition frequency of 100
     Hz  is  superposed to this base-voltage (dc-plus-pulse operation mode), and
iii)  a dc high voltage is applied to each field, but under no back discharge
     condition at 20 deg C gas temperature for the purpose of reference (dc
     operation mode without back discharge).

 2.1  DC  Operation Mode under Back Discharge

      The test conditions are: gas temperature T = 100 deg C, dust resistivity=
 10-H - 10-^  ohm-cm, inlet dust loading w^ = 6.5 g/Nm^, applied voltage V = 41 -
 43 kV corresponding to average field strength in collection field E = 2.7-2.9
 kV/cm,  and average current density J = 0.3-0.9 mA/m .  A severe back discharge
 is  detected by naked eye and a very distinct hysteresis in voltage-current
 characteristics.  The current rises with increasing dust thickness on the col-
 lection electrodes  to become maximum just before rapping.
      Figure 7 indicates the mass-penetration of dust through the test ESP as
 a function of the effective length in gas flow direction of the collection
 field.   The effective length is varied by connecting the first, second, and
 third collection field successively from the upstream side to the high voltage

                                      382
-------
source, and it is indicated by  the number  of  the  connected  fields,  each being
50 cm in length.  The curve A represents the  blank  case where no  precharging
is made.  The penetration starts  from  50 % as a results of  purely mechanical
collection which is made mostly up to  the  end of  the  first  field  in this par-
ticular case.  It is interesting  to  see that  the  curve A  shows an exponential
decay following Deutsch equation  even  under the severe back discharge  condi-
tion.  The curve B is the same  result  obtained when only  the Boxer-Charger  I
in the inlet-duct is being operated.   The  curve C is  for  the case when the
Boxer-Charger I and II are being  operated.  It can  be seen  that the effect  of
precharging appears distinctly  in its  succeeding  collection field to raise  the
apparent migration velocity of  dust, w, but its effect is lost from the next
field on to make the curves B and C  parallel  to the curve A from  the second
and third fields.  The enhanced charge decays in  the  first  downstream  field
due to back discharge down to a saturation level  specific to its  charge neu-
tralization condition,  As is to  be  reported  separately,  this condition is
specified by the magnitudes of  the field strength,  E, current densities of
positive and negative ions, i+  and i_, their  ratio  i+/i_, and the space dis-
tribution of these quantities.  The  higher migration  velocity in  the first
downstream field is evidently resulted by  the higher  average charge of dust
during its decay process down to  the new saturation charge.  This saturation
charge, approximately equal to  that  in the blank  case (curve A),  will  control
the collection process in the succeeding fields to  provide  a decreased migra-
tion velocity close to its value  for the curve A.
     Figure 8 shows the magnitudes of  charge-to-mass  ratio  of dust  measured at
the outlet in the same tests as described  in  Figure 7.  The highly  charged
particles will be rapidly collected  in the active field without contributing
to the indicated values of charge-to-mass  ratio at  the outlet, whereas the
small particles, hard to collect  and to become dominant at  the outlet, tend to
raise the measured value.  This makes  a clear interpretation of the curves  in
Figure 8 difficult.  However, it  seems that the charge-to-mass ratio enhanced
by the first Boxer-Charger I is further raised by the second Boxer-Charger  II.
It is not clear whether the rise  of  the curve A with  increasing field  length
represents a time increase in particle charge or  the  size decrease  at  the out-
let as a result of increasing collection performance.
     At any rate it is expected that the penetration  will drop exponentially
with the field length following the  dotted line in  Figure 7 when  the Boxer-
Chargers are used successively  in front of each collection  field.   The migra-
tion velocities, wo and w^ derived from the curve A and dotted line respec-
tively, are indicated in Table  1.  The enhancement  factor of collection perfor-
mance by the successive precharging, expressed in terms of  wi/wo, is as high
as 2.44 in this particular case.  It should be added  that this enhancement
factor is a function of different variables,  including the  chaging  performance
of precharger, residence time and back discharge  severity in the  combined col-
lection field, particle size distribution  of  dust,  etc.                     v
     In order to take full advantage of precharging it is imperative to use
the proper collection fields which are free of back discharge, such as those
consisting of parallel plane electrodes or the pulsed field of twin-electrode
or tri-electrode types.  The pulsed  field  is  likely to be more suitable in
view of avoiding rapping reentrainment with electrical adhesion force. The
value of ionic current density  in this case should  be higher than needed for
providing a sufficient electrical adhesion to the dust deposit on the  collec-
tion electrodes, but lower than the  critical  level  of triggering  back  discharge,

                                     383
-------
The successive use of prechargers by inserting into each inter-field  section
provides a second alternative in the case when the conventional ESP fields
can not be altered.

2.2 DC-Plus-Pulse Operation Mode

     With the considerations described above in mind the same tests are  con-
tinued with the pulsed collection fields under the same high resistivity con-
ditions.  The dc high voltage of -35 kV  (2.3 kV/cm) is applied to the  discharge
electrodes of each field, and a negative pulse high voltage with a crest value
of -20 kV, duration time of 1 ms and repetition frequency of 100 Hz is applied
superimposed.  All other variables are kept identical to those in the  previous
section.  A noticeable feature of this operation mode is a greatly reduced
current density level at J = 0.04 mA/m , and no back discharge is detected by
naked eyes.  The pulsed corona occurs uniformly along the wires, and the uni-
formity of current over the collecting electrodes is distinctly noticed  from
the mode of uniform dust deposition.
     Figure 9 shows the same results as  Figure 7 where the curves A, B,  and  C
are corresponding to the same precharging conditions of A, B, and C in Figure
7.  Again in this case the similar tendency appears in the curves A, B,  and  C
with the increased decay rate.  It is noted that a substantial improvement is
obtained by the pulsed fields only owing to an effective suppression of  back
discharge, providing the enhancement factor of wp/w0 = 1.63 as indicated in
Table 1.
     As expected the effect of precharging becomes much more pronounced  in
this case, and the enhancement factor derived from the migration velocity for
the dotted line, wp^, becomes as high as wpi/wo = 2.91.  This migration  veloc-
ity, wi, specific to this particular combination of successive precharging
with the pulsed collection field exceeds a level obtainable under an ideal ESP
condition where no back discharge is taking place, as described later.
     The deviation of curves B and C from the dotted line occuring from  the
second field on, although greatly reduced in this case, indicates a slight
effect of charge neutralization due to back discharge which is still existing
even under the pulsed condition.  The magnitude of this deviation, which can
be expressed by the ratio of the migration velocity for the second branch to
that for the first branch, wP2/wpl> provides a good measure of back discharge
severity usable for design purpose of this precharger-collector hybrid system.

2.3 DC Operation without Back Discharge

     For reference purpose the tests are further performed at room temperature
T = 20 deg C where dust resistivity drops under 10-*-^ohm-cm and no back dis-
charge occurs.  The magnitude of main voltage applicable to the Boxer-Chargers
and collection fields becomes greatly increased.  As a result the maximum
value of ac main field at the Boxer-Chargers is raised from 4.5 to 5.4 kV/cm,
and the dc high voltage applied to the collection fields from 41 - 43 to  51 -
55 kV (from 2.73-2.87 to 3.4-3.7 kV/cm) .  This is to compare the advantage
of the precharger and its combination with the pulsed fieds to the best  possi-
ble standard of ESP operated under ideal conditions.  The applied voltage in
the collection fields is regulated so that a constant current density  J  = 0.15
mA/m2 is maintained.  The inlet dust loading is kept constant at 8 g/Nm^.
     Figure 10 indicates the results obtained, where the curves A, B,  and C
represent the operating conditions corresponding to A, B and C in the  Figures

                                     384
-------
7 and 9.  The curve A in this case consists of  three  different  branches.   The
first one corresponds to a lower migration velocity,  possibly owing  to  lower
average charge of particles at the first field.   Then,  the migration velocity
rises with increasing charge in the second field.   It drops again  in the  third
field, perhaps because of decrease in particle  size.
     The use of the first Boxer-Charger I acts  for  raising the  migration veloc-
ity at the first field, indicating the effect of precharging without  which  the
particle charge must rise from zero and, hence, the time-average of  the parti-
cle charge becomes lower.  Then, the curve B becomes  parallel to the curve A
in the succeeding fields as in the case in Figures  7  and 9.  A  striking fea-
ture in this case is that the curve C shows very  little deviation  from the
curve B, distinctly indicating no improvement to be obtained by successive
precharging.  It is considered that, once the particles have been  fully
charged by either the first Boxer-Charger or corona charging in the  first
field, the high level of charge will be preserved in  this case  since no charge
reduction by positive ions will take place in this  case.
     Thus, the effect of precharging in the case of low dust resistivity
appears only in the first field, and this only reduces the inlet dust loadings
of its succeeding fields.  Ths enhancement factor in  this case is  only Wbi/wbo
= 1.21 as indicated in Table 1.

3. Conclusions

     The following conclusions are obtained from the  present experiments:
1) The Boxer-Charger Mark III using double-helix units for its electrode assem-
   blies and very short pulse voltage for plasma excitation indicates a very
   satisfactory charging performance.
2) The dust mass-loading shows an exponential decay with distance in gas flow
   direction even under a severe back discharge condition, indicating Deutsch
   equation to hold also in this case.
3) The use of a precharger  provides  a great performance improvement in the
   collection field immediately downstream, but its effect is lost from the
   succeeding fields in the case when a severe back discharge is occuring.
   This suggests that the enhanced initial charge of  the particles can drop,
   by charge elimination of back discharge, fairly  rapidly to a new saturation
   level, even within a residence time in this field.  This  means that the
   performance improvement obtained in this field by  no means corresponds to
   the level of the enhanced initial charge at its  inlet, but to the effective
   charge averaged over the entire field during its elimination process.
4) The performance improvement can be expressed by  "enhancement factor" as the
   ratio of enhanced migration velocity to its blank  value.
5) The repeated precharging is necessary in front of  each collection field
   under back: discharge condition in the case when  the collection  fields can
   not be altered from a conventional ESP design.   This provides the enhance-
   ment factor as high as 2.44 in the present experiments.
6) The use of "back discharge free collection fields" is necessary for taking
   full advantage of precharging.  A parallel plane electrode system or pulsed
   field system of either twin- or tri-electrode type can be used  for this
   purpose.  The pulsed field is likely to be more  suitable because  it can
   provide an electrical adhesion force to the dust deposit on collecting
   electrodes so as to avoid rapping reentrainment.
7)  The figure of merit of the back discharge free field is given by  the devi-
   ation of decaying line of dust mass-loading from its initial branch in  the

                                     385
-------
   first field, expressed in terms of the ratio of respective migration velo-
   cities.  This ratio can be used for design of the precharger-ESP hybrid
   system in general.
8) The combination of repeated precharging with pulsed collection fields of
   twin-electrode type energized with the dc-plus-pulse operation mode pr°-
   vides an enhancement factor as high as 2.91 at dust resistivity of IO11 -
   1()12 ohm-cm.  The dust migration velocity in this case becomes slightly
   higher than that for a reference case of ESP operated at its maximum per-
   formance under no back discharge condition.
9) In the case when no back discharge is taking place, only the first pre-
   charger provides a distinct effect, but no further improvement can be
   obtained by the repeated precharging.

     The test of parallel plane electrode system as another type of back dis-
charge free collection field is being undertaken.  The development of the
Boxer-Charger Mark III and its pulse power supply is being performed for appli-
cation in the large scale ESP's of coal burning boilers.  Figure 11 illus-
trates its conceptual design of frame-supported construction using the induc-
tance-isolation to support each double-helix unit at its both ends, and also
at its center point for fixing and rapping.

References
 (1) S.Masuda, A.Mizuno, H.Nakatani and H.Kawahara: Application of Boxer-
    Charger in Pulsed Electrostatic Precipitator, Record of IEEE/IAS 1980
    Annual Meeting, pp. 904-911 (Sept. 29  -Oct. 3, 1980 in Cincinnati, Ohio).
 (2) S.Masuda, M.Washizu, A.Mizuno and K.Akutsu: Boxer-Charger - A Novel
    Charging Device for High Resistivity Powders, Record of IEEE/IAS 1978
    Annual Meeting, pp. 16-22 (Oct. 1978).
 (3) S.Masuda, A.Mizuno and H.Nakatani: Application of Boxer-Charger in Elect-
    rostatic Precipitators, Record of IEEE/IAS 1979 Annual Meeting, pp. 131 -
    138  (Oct. 1979 in Cleveland, Ohio).
           TABLE 1  DUST MIGRATION VELOCITY AND ENHANCEMENT FACTOR
Operation Mode of
Collection Fields
DC Operation Mode
under Back Disch-
arge
DC-Puls-Pulse
Operation Mode
at high pd
DC operation Mode
under No Back
Discharge
Tg
100
100
20
Pd
(ohm- cm)
io11
-io12
io11
-io12
-------
       V-pulse
                 D
                 a
                 b
                 c
30 mm
 3 mm
 8 mm
 1 mm
                               (b)  streamers
 (a) Double-Helix Unit              (c)  Inductance-Isolation
 Fig. 1  Electrode Assembly  Unit  of Boxer-Charger Mark III
Fig.  2  Boxer-Charger I
       (installed in inlet-duct)
                   Fig. 3  Boxer-Charger II
                          (installed in inter-field
                           section)
      (4- : gas flow direction)
                                     387
-------
Fig.4  Pulse and Main  Voltage
       Supply
                                      three double-helical
                                            electrodes
         D  Ri Cable-1 Gap  Cable-2
                                     ZXXXXXXX3
                                     booooooo
                                                   Fig.  5   Pulse Wave
                                                           Form
/


V




















^"-
/
HEATER



BLOWER

1 7.5 ™<% _ , 	 .
S f'Q^I

V.






	 	
^_^
c
0
'a!
li
c
dJ
0.








DUST

Ki%^! *
FEEDEK K !
* \

„ ' N
Charger
H /
100
en;



20

1 U
5


?

1

U.b

n 9
- 	 I




t
EP
Field)
1





0.82 "Vtj




1
Ftm
Cho
(





pr






t
EP
Field]


N
[(Q/M)0|







y
t
EP
iField)


N^>_
" V~-\
xT^^->
_%,
x^

: ^j







ESP : dc oper

; T : 100 °C



- — ]
^-1
--^ T
^~-S
N
N






ation
12 .




•*^,

^^













~-~.


N^




mode







~"~~-!

^2

X-
S ^
s






A

B '

r


X





:
L pd : 10 -10 s;-cm -|
(severe back
I
discharge)
i


1

CJ

o
~n
a:
ro
2
0
0>
?
t
u





\


)



-I


W^, W0 : dust-loading
monitor
(Q/M)if (Q/M)0:
charge-to-mass ratio
measurement

T: thermometer
Fig. 6 Race-Track System
f
4


-3
















0





• ; I
ESP : dc operation mode
T : 100 °C
°d : 1011 - 1012 n-ca
(severe back discharge)
j

^^L T
^^^1
' 1 i.

^-^ B_T-]^= 	 H
^"~ — 5 	 "^ ^^\ ^*
^k
T
.tr*"
-//
i i














0123
Number of Collection Fields

"0
                1       2       3
            Number of Collection Fields
Fig.
                                       Fig. 8  Charge-to-Mass Ratio
                                               (measured  at outlet;
                                                dc operation mode)
7  Dust Penetration vs.  Effective Length of Collection Field
   (dc operation mode under  severe back discharge condition)
                                       388
-------
                   ESP : dc-plus-pulse
                       operation mode
  100

   50

~  20

I  10
tj
                                            c
                                            &
                                            Q.
              1        2       3
          Number of Collection Fields
    9   Dust Penetration vs. Effective
       Length of  Collection Field
       (dc-plus-pulse  operation mode)
                                               2

                                               1

                                             0.5

                                             0.2
         	1	1	1	
         	 •  -- ESP : dc operation node
                 T : 20 °C
     0123
        Number of Collection Fields

Fig. 10  Dust Penetration vs.
         Effective  Length of
         Collection Field
         (dc operation;  no back
          discharge condition)
                                                          H:  double-helix unit

                                                          F:  frame
                                                          A:  supporting arm
                                                          I:  inductance-
                                                              isolation
                                                          C:  center supporting
                                                              arm for rapping
                                                              with inductance-
                                                              isolation
Fig. 11  Conceptual Structure  of Frame-Supported  Boxer-Charger
         (to  be installed  in  inter-field sections)
                                      389
-------
    THE PERFORMANCE OF AN EXPERIMENTAL PRECIPITATOR WITH AN ALL-PLATE ZONE

                                By J. Dalmon
       Central Electricity Research Laboratories, Leatherhead, England

                                  ABSTRACT

An 'All-Plate' ESP collector zone has been developed by the CEGB and
tested in a small research rig at atmospheric temperatures.  To provide data
more closely related to power station precipitator conditions a programme of
tests has been carried out by permission of the Environmental Protection
Agency on their experimental electrostatic precipitator facility at Research
Triangle Park.  This consists of a four zone, single lane unit and for the
tests the conventional discharge wires in one zone were replaced by a single
charged plate.  In all zones the spacing between the earthed plates was
200 mm.  The gas velocity was kept constant at 1.6 m/s and the gas temperature
ranged from 17°C to 150°C.  The test dust was fly ash having a mass median
diameter of 6 pm and a resistivity of 1 x 10^-2 ohm cm.

The performance was. evaluated as an 'Effective Migration Velocity' enhance-
ment factor, defined as EMV with plate-plate electrodes/EMV with wire-plate
electrodes.  The factor averaged 3.1 with all plates in zone three and 2.9
for zone four with a value of 1.4 over the complete precipitator.

The results indicate a clear advantage for a precipitator incorporating an
all-plate zone.

                                INTRODUCTION

As part of a programme of research into methods of improving the dust
collecting performance of large power station electrostatic precipitators,
the Central Electricity Research Laboratories of the Central Electricity
Generating Board carried out a pilot scale investigation in which the dis-
charge wires in the final zone of a three zone precipitator were replaced by
negatively charged plates.  The work, which was reported by Dalmon (1974)^'
showed that particles entering the final zone were already charged and were
collected in the high electric field between the parallel plates without the
necessity of a corona discharge.  The experiments, carried out under atmos-
pheric conditions using aluminium oxide and perspex test dusts to simulate
normal and highly resistive ash respectively, showed clear advantages for the
'All-Plate' final zone with a twofold increase in its effective migration
velocity.  Inter-electrode spacings of 50 mm and 150 mm were investigated
over a rnage of velocities from 0.9 m/s to 2.75 m/s and it was concluded that
a spacing of 100 mm would be the practical optimum for full-scale use.  These
early tests did not include the effects of rapping and further work showed
that although the emv fell for both wire-plate and all-plate geometries in
the final zone, there was still an emv enhancement by a factor of two.

During 1979, the Industrial Environmental Research Laboratory of the United
States Environmental Protection Agency offered the use of their 'hot'
experimental precipitator facility at Research Triangle Park for further
testing of the All-Plate concept under dust and gas conditions more

                                     390
-------
representative  of  power  station practice.   This led to a programme of  tests
aunng  1980 which  are reported here.

                             EXPERIMENTAL FACILITY

The EPA experimental  precipitator has been fully described  by  Lawless, Daniel
and _ Rams eyUJ,  so  only its  main features are summarised here.   It  consists of
a single  lane with four  zones  in series.  Each zone normally has a pair of
flat collecting plates 1.22 m  square  and a single row of discharge wires, but
this was  modified  as  described later.  Both the plate spacing  and  the wire
pitch can be readily  altered.   Rapping of  both plates and wires is by
pneumatically operated hammers acting vertically and the frequency of
operation^can be controlled separately for each zone.   Ambient air is drawn
from outside the laboratory directly  into  the inlet ducting and may be heated
by  three  propane burners.   The gas flow rate is measured by an orifice plate
upstream  of the ID fan and  is  variable between wide limits.  Dust  is injected
by  two  sandblast guns facing into the gas  stream and situated  ^2 m downstream
of  the  burners  and a  similar distance from the inlet of the first
precipitator zone.  Sampling ports are arranged in vertical groups  of three
at  the  precipitator inlet,  between zones and at the outlet.  Each  zone is
supplied  by a manually controlled 100 kV,  10 mA transformer-rectifier.  Kilo-
volts and milliamps are  displayed digitally and may be  printed out  at selected
intervals.  Voltage-current characteristics may be taken and plotted in real-
time on an X-Y  plotter.  Gas temperatures,  orifice differential and dust
emission  are also  displayed and printed with the electrical readings.

                         MEASUREMENTS AND  TECHNIQUES

Mass Burden, Efficiency  and Effective Migration Velocity

Mass burden was measured by standard  isokinetic sampling techniques using
47 mm dia. GF/F glass fibre filters.   The  sampling duration varied from 5 to
150 mins  according to location and dust burden and either three or four
sampling  trains were  used simultaneously.   The middle of  the three sampling
ports was used  as  previous  EPA investigations  had  shown this to give represen-
tative  results.  The  sampled weights  were  used to  calculate overall and zonal
efficiencies and effective  migration  velocities (emv's)  were derived using the
well known Deutsch equation.

Particle  Size Distribution

This was an in-situ measurement  using an MRI  seven stage  cascade impactor with
a 47 mm GF/F back-up  filter.   Data from the  tests  was processed by a desk
computer using  a programme  developed  by the EPA to give  cumulative distribu-
tion by mass and,  by  combining data from various  locations  overall and zonal
size-efficiency relationships  were provided.

Particle Charge  to  Mass  Ratio

This was measured by  a Faraday cage type probe  of  EPA design inserted into the
sampling ports  previously mentioned with the exception  that it could not be
used at the outlet  due to insufficient  clearance.   Problems with dust by-

                                      391
-------
passing the inlet nozzle seal prevented any successful measurements at
ambient temperatures.

Ash Resistivity

Resistivity was measured at the precipitator inlet using a point-plane probe
designed by the Southern Research Institute^3), and on collected samples over
a range of temperatures and relative humidities by the Denver Research
Institute using a laboratory cell.

Emission Monitoring

The opacity of the emitted flue gas was monitored by an MRI  'Plant Process
Visiometer (PPV)(4).  The instrument passes a continuous sample of gas
through an optical chamber where opacity is measured by the  total light
scattered by the particles.

                               TEST PROGRAMME

Zonal Arrangement

Throughout the test programme the inter-electrode spacing was maintained at
100 mm  (i.e. 200 mm between earthed plates) in all four zones.  Base-line
tests  (Tests 1 to 7) were conducted with six 3.2 mm dia. round wires at 200 mm
centres in all zones.  A single h.t. plate, 1.07 m long by 1.14 m high having
radiused corners and edged by a 17.5 mm dia. tube, then replaced the wires in
the final zone (Tests 8 to 18).  The h.t. plate was then fitted in zone three
and the wires refitted in zone four (Tests 19 to 27).  Final tests were
carried out with wires in all zones (Tests 28 to 31).

Test Conditions

Tests were conducted at ambient temperatures  (VL7 C) and within the range
107°C  to 157°C at a constant gas velocity of 1.6 m/s both with and without
rapping.  The normal rapping rates are given in Table 1 but  higher rates were
used on zone four, as indicated in Tables 4 to 7, to accentuate the effects of
re-entrainment.

                       TABLE 1.  INTERVAL BETWEEN RAPS

                       Zone         1      23      4
Plate (min)
Wire (min)
15
7.5
30
15
60
30
120
60
 NB.   (1)   Timing  arranged  so  that  raps  are not  coincident.
      (2)   On certain  tests  zone  4  was rapped  every  10 min.

 The  test  dust was fly ash  collected  from  the  hoppers of  a Detroit-Edison ESP
 when Illinois coal was being  burned, and  its  resistivity, as  previously
 measured  by the Denver method was  1  x 10" ohm cm  at 150  C.  The ash analysis
 is given  in Table 2 and of  the rig gas  in Table 3.   The  inlet burden was kept
 close to  8 g/m3 NTP throughout.
                                     392
-------
                 TABLE  2.   ELEMENTAL ANALYSIS OF FLY ASH (%)


            Al        Si       Ca      S       K       Ti      Fe

           26.44     28.60     1.74    0.27    2.54    0.85    39.56


                      TABLE  3.   TYPICAL GAS ANALYSIS (%)


                      H20     C02    CO      02      N2

                      3.4     6.0    4.0    10.0    78.0

Test Procedure

Following preparation for the  base-line tests voltage-current  characteristics
were taken on all  zones under  clean electrode and gas conditions at 17°C and
124 C with both static  and  flowing  air.

After overnight cooling fly ash was injected  for  3 hrs with voltage of 32 kV
applied to all zones  and with  all the rapping operating  as given in Table 1.
This allowed dust  to  build  up  on the electrodes  in a  realistic manner.  The
voltage was then raised to  the maximum (Test  1) and when stable conditions
were indicated by  the PPV mass sampling commenced.  Further tests at ambient
conditions were made  at lower  voltages.   The  plant was then shut down over-
night with the residual dust left on the electrodes.

The following morning the gas  temperature was raised  to  120 C over a period of
two hours with nominal  electrical settings, V-I characteristics were then
taken and dust injection commenced.   Further  characteristics were taken and
the information used  to adjust the   electrical settings  to give minimum
emission readings  on  the PPV before performance testing  began.  These pro-
cedures for temperature raising,  performance  optimisation, testing and rig
shut-down were followed on  a daily  basis.  Priority was  given to mass sampling
with impactor, charge and resistivity measurements  being made less frequently.
Some eight to nine hours running  were achieved daily.

Electrode changes  were  accomplished with the  minimum  electrode cleaning, but
dust samples were  taken from the  all-plate zone for resistivity measurements
at Denver.

Tests with each electrode arrangement commenced with measurements at ambient
conditions.

                            RESULTS  AND DISCUSSION

Electrical Characteristics

At ambient temperature, ^17°C,  the  voltage-current  characteristics of the
wire-plate zones were normal for  electrodes coated  with  a moderately
resistive dust and a  typical example  is  given in Figure  1.  The all-plate zone

                                      393
-------
shows virtually zero current until sparkover occurs.  The characteristics were
not dissimilar to those taken with clean electrodes.  At temperature  the
characteristics for wire-plate zones exhibited classic back-ionization
features of run-away current at low voltage, a depressed sparkover voltage and
a very marked hysteresis effect (Figure 2).  In contrast, the all-plate zone
showed zero current to sparkover and was very little changed from the clean
characteristic.  Sparkover in this zone triggered some flow of current which
was small by comparison with wire-plate zones.  The voltage at sparkover for
the all-plate zone was always well in excess of the value for other zones at
the same time, by as much as 20 kV in some instances.  The low moisture
content of the flue gas, 3.4% compared with 6% normally encountered, plus the
fact that there was no S02 may have contributed to the steepness of the V-I
curves, although this would have been offset somewhat by the 02 content which
was 10% compared with a usual 6% for power station precipitators.

Observation of the PPV trace indicated that minimum emission was obtained at
the onset of back-ionization and this coincided with the point of maximum
voltage as shown in Figure 2.  This point was selected from the V-I curves
taken prior to each test and used as the operational setting for that test.
There was considerable daily variation in the characteristics, but this was
only to be expected with the severe back-ionization experienced.  An attempt
was made to obtain more stable conditions by altering the gas temperature and,
thus, the dust resistivity.  However, the resistivity changed very little over
the range of interest and, hence, the effect was minimal.

Resistivity

New measurements at Denver showed that the resistivity of the feed dust had
risen from the previous value of 1 x 10" ohm cm to 1 x lO^ ohm cm at 150 C in
an atmosphere containing 5% moisture.  As the moisture content of the rig gas
was 3.4% a rather higher value would be expected in the rig.   Measurements on
ash taken from the electrodes of the all-plate zone gave a resistivity of
2 x lo^^ ohm cm.  The in-situ measurements taken with the SRI probe gave a
somewhat lower value of 1 x 10" ohm cm at 150 C.  This order of magnitude
difference is no doubt due to the different methods.  Never-the-less, the
measurements showed that the dust was highly resistive and,  thus, must be
responsible for the severe back-ionization at temperature.
The Denver resistivity measurements indicated that for temperatures around
17 C the resistivity would be below 10   ohm cm, this would account for th
fact that no back-ionization was experienced in low temperature tests.

Examination of Electrodes
Inspection of the electrodes after each test series showed that between 3 mm
and 10 mm of flakey dust had accumulated on the collecting plates of wire-
plate zones.  The all-plate zone had a fairly even coating <2 mm thick on both
electrodes when in the final zone and up to 5 mm when in zone three.  The
discharge wires were coated to <1 mm.  The dust thickness on shut-down
reflects the immediate rapping history and it is probable that variations in
deposit thickness could have caused some of the day-to-day variations in
electrical conditions.

                                     394
-------
The level of deposits observed  is  quite  normal  in power station precipitators
and this gives confidence  that  the test  conditions were closely representative
of full-scale operation.

Charge-to-Mass Ratio

The calculated and measured  charge-to-mass  ratio  plotted in Figure 3 shows
that the latter is less by a factor of ^3 than  predicted.  This is very probably
due to the presence of positive ions under  the  strongly back-ionizing
conditions encountered, which may  not only  have reduced the charge on the
particles but could have resulted  in positively charged particles being
collected in the probe.  This hypothesis could  not be checked  for non-back-
ionizing conditions at ambient  temperature  because of the problems with the
probe mentioned previously.   The experimental data in Figure 3 was obtained at
various positions along the  ESP and it is apparent that the charge is a
function of the voltage applied upstream of the measuring point  rather than
residence time.  Examination of the data showed that particles  entering the
precipitator had a small,  but significant charge  and the maximum charge was
attained in the first zone (Figure 3, Test  27).   It was also found that
particles escaping from the  all-plate zone  three  had virtually  the same
charge-to-mass ratio as when they  entered the zone (the two experimental
points for Test 24).

Performance

The performance, in terms  of efficiency  and emv,  is given for all the tests in
Tables 4 to 7, which are grouped into gas temperature bands.  It is apparent
from Table 4 that at low temperatures, with no  back-ionization problems, over-
all efficiencies are well  over  99% and overall  emv's were correspondingly high
at over 20 cm/s.  It is also apparent that  both overall and zonal performance
varied considerably from day-to-day under similar operating conditions.

There is a marked drop in  performance at temperature with back-ionization
present (Tables 5 to 7) and,  again,  there are considerable differences between
tests where temperature, rapping and electrical conditions were  reasonably
similar, (e.g. Tests 6 and 7, 21 and 22, Table  6).  There is a weak tendency
for the emv to increase with gas temperature over the range 107  to 157 C.

The effects of rapping are far  from consistent, e.g. test 10 with rapping has
a higher efficiency than tests  11  and 12 with no  rapping (Table  6).  This also
applies to test pairs 17-18  and 26-27 (Table 7),  where  the last  zone has
accelerated rapping.  In other  cases, tests 15-16 (Table 5)  for example, the
efficiency, as would be expected,  is greater without rapping.

In view of these anomalies,  which  can be partly but not wholly  explained by
the uncontrolled effects of  back-ionization, the  results for similar test
conditions have been averaged and  presented in  Tables 8 and 9  as "emv enhance-
ment factors".  This is defined as the emv  with plate-plate electrodes/the emv
with wire-plate electrodes.   Table 8 shows  the  very considerable increase in
emv for zones three and four when  the plate-plate arrangement  replaced the
wire-plate geometry.  The  greatest emv enhancement is for zone  three with no
back-ionization (a factor  of 13).   Zone  three,  however, shows  a greater

                                      395
-------
variability in this factor than zone four which has a value of about  three  at
all temperatures with rapping.  Its value without rapping is strangely,  less
at 1.2.  In terms of the overall emv the factor, apart from two cases, varies
from 1.2 to 1.6 irrespective of the position of the all-plate zone  (Table 9).
The exceptions are due to less efficient collection in the wire-plate upstream
zones, e.g. in test 29 the combined efficiency of zones one to three was
82.58% (wire-plate in zone four) whereas in test 17 (plate-plate in zone four)
it was only 75.29%.  Taking all the high temperature tests together gives
enhancement factors of 1.4 and 1.3 for zones three and four respectively.
Thus, there is without doubt, a clear advantage for the all-plate system.

The reason for the enhanced performance was originally attributed to a com-
bination of a higher and more uniform electric field in the absence of corona
and less gas turbulence with the simpler plate arrangement(l).  No turbulence
measurements were made during the present trials but a typical plot of emv
versus applied voltage (Figure 4) clearly shows the importance of the higher
operating point and, hence, the increased field with the all-plate zone.

Particle Size and Grade Efficiency

The average mass geometric mean diameter of the fly ash at the precipitator
inlet, as measured by the cascade impactor, was 6 ym.  This represents an
extremely fine dust and compares with a typical value for a CEGB power station
precipitator of 18 jam.  The size distribution for test 25 is given in Figure 5
for the inlets to zones one, three and four and the outlet of zone four.  The
dust becomes progressively finer with distance from the inlet, the mean  sizes
being 6.7, 5.2, 3.2 and 2.9 ym respectively.

The size data has been used to calculate grade efficiencies and Figure 6 shows
this information both for the complete precipitator and for the all-plate zone
on test 25.  It may be noted that there is a well defined minimum efficiency
for the complete precipitator as would be expected from changes in charging
mechanisms and this lies at 0.8 ym.  There is a less obvious minimum at 0.4 ym
for the all-plate zone alone.  These results are typical of the limited number
of cascade impactor tests but there is insufficient data to draw firm con-
clusions regarding the size-efficiency characteristics of the all-plate  zone
except to say that it appears better at collecting the fine particles.

                         IMPLICATIONS FOR FULL SCALE

The use of an all-plate zone is zeen initially as a method of up-grading an
existing precipitator by substitution in, say, the final wire-plate zone.   If
for example, a precipitator having a measured efficiency of 98.3%, a  specific
surface of 58.85 m^/m^/s and an emv of 6.9 cm/s has its emv enhanced by  a
factor of 1.2 the efficiency rises to 99.24% and an enhancement of 1.3 gives
a predicted efficiency of 99.5%.  This corresponds to a reduction in  emission
of 55% and 71% respectively.  A real case, would obviously have to be
investigated in depth rather than using this simple approach, before  any
decision on fitting an all-plate zone were made.  However, in anticipation  of
a retrofit, the CEGB have commissioned engineering rigs at the works  of  two
manufacturers with the object of investigating detail design, electrical and
rapping problems.  The rigs consist of five, full height plates  (three

                                     396
-------
earthed and two h.t.) and testing  is well  under  way.   The  designs are
proprietary, so no details can  be  given  here.

                                 CONCLUSIONS

Trials of an 'All-Plate' collector zone  in the four  zone EPA pilot electro-
static precipitator under conditions realistic of  power station plant have
amply confirmed earlier  tests at CERL  under atmospheric conditions using
artificial dusts.  The results  have shown  that the substitution of an all-
plate arrangement for the normal wire-plate configuration  enhances the zonal
emv by an average factor of  2.9 when used  in  the final zone and by 3.1 when
in the penultimate position.  The  overall  emv was  increased by a factor of at
least 1.3.

                              ACKNOWLEDGEMENTS

The author wishes to thank Mr J.H.  Abbott  and Dr L.E.  Sparks of the U.S.
Environmental Protection Agency for making this  investigation possible.  The
help of G.H. Ramsey, B.E. Daniel,  R. Valentine and R.  Ogan with the
experimental work is gratefully acknoweldged.  The paper is published by
permission of the Central Electricity  Generating Board.

                                   ENDNOTES

1.  Dalmon, J.  The Performance of an  Experimental Electrostatic
    Precipitator with a  Low  Turbulence Zone, Australian Inst. Fuel Symposium,
    Adelaide, Nov. 1974.

2.  Lawless, P.A., Daniel, B.E. and Ramsey, G.H.   Characterisation of the
    EPA/IERL-RTP Pilot-Scale Precipitator,  EPA-600/7-79-052, Feb. 1979.

3.  Nichols, G.B. and Banks, S.M.   Test  Methods  and Apparatus for Conducting
    Resistivity Measurements, SORI-EAS-75-090, Sept. 1977.

4.  Ensor, D.S., Sevan,  L.D. and Markowski, G.R.   Application of Nephelometry
    to the Monitoring of Air Pollution Sources,  Paper  74-110, June 1974,
    A.P.C.A. Meeting, Denver, June.
                                      397
-------
           TABLE 4.  PERFORMANCE IN TEMPERATURE RANGE  9°C  TO  24°C


Test      .             Efficiency  (%)                   Emv  (cm/s)
No.   RaPPin8   zi    Z2    Z3     Z4   Overall   Zl    Z2   Z3   Z4   Overall

      All Wire-Plate
 1       /       -               36.11  99.93     -    -    -   5.9   23.8
 2       /       -               55.33  99.81     -             10.6   20.6
 3       /       -               43.80  99.73     -    -    -   7.6   19.4

      Zone 4 Plate-Plate

 8       /     97.89 82.24 15.38 91.78  99-91   50.6 22.7  2.2  32.8   23.0
 9       /     99.22 71.54 28.38 63.64  99.95   63.7 16.5  4.4  13.3   24.9

      Zone 3 Plate-Plate
19       /     +•  99.61  -> 87.5    -      -     «-  35.2 ->  27.3
20       /     -e  99.49  -> 99.0  33.33  99.997  +-  34.6 -»•  60.4   5.3   34.2
                TABLE 5.  PERFORMANCE AT TEMPERATURE OF 107°C


Test      .            Efficiency (%)                   Emv (cm/s)
No.   KaPPin§   21    Z2    Z3    Z4   Overall   Zl   Z2   Z3   Z4  Overall

15
16
Zone 4 Plate-Plate
x - - - 53.85
/ - 18.09

85.90
73.94

- - - 10.2
2.6

6.4
4.4
         /  =  Normal rapping
         x  =  No rapping.
                                     398
-------
TABLE 6.  PERFORMANCE IN TEMPERATURE RANGE 112-125°C
Test
No.

4
5
6
7
30
31

10
11
12

21
22
Efficiency (%)
Rapping n Z2 Z3 ^
All Wire-Plate
/ _
/ -
/ 60.70 49.51 23.69
/ 22.02 33.26 27.75
+ f 73.51 + 35.71
x «- 72.97 + 16.29
Zone 4 Plate-Plate
/ -
x 57.71 68.39 58.75
x - - -
Zone 3 Plate-Plate
/ «- 34.48 + 36.45
/ •*- 52.92 + 48.58

5.87
Zero
4.30
34.35
14.45
43.24

40.62
60.53
30.99

27.86
27.70
Overall

78
73
85
81
85
87

99
93
91

69
82

.64
.08
.51
.48
.43
.16

.03
.23
.07

.96
.50
Zl

Emv (cm/s)
Z2 Z3 Z4




- 0.8
- Zero
12.3
3.3
«- 8
«- 8

_
10.7
-

«- 2
*• 4
10.0 3
9.0 3
.7 + 5
.6 + 2

_
15.1 11
-

.8 + 6
.9 + 8
TABLE 7. PERFORMANCE IN TEMPERATURE RANGE 142°
Test
No.

28
29

13
14
17
18

23
24
25
26
27

Efficiency (%)
Rapping zl Z2 Z3 Z4
All Zones Wire-Plate
+ «- 80.74 + 3.64
x *- 76.94 + 24.43
Zone 4 Plate-Plate
/ 69.97 33.30 35.44
/ 68.50 12.45 55.75
x — — -
+ - - -
Zone 3 Plate-Plate
/ •«- 67.83 + 61.19
/ «- 73.91 •* 61.37
/ x- 80.78 + 71.69
+ +- 61.48 + 76.71
x x- 79-47 + 56.91
/ = Normal rapping

35.04
38.89

70.54
83.57
43.23
41.27

54.07
7.07
47.26
57.41
47.35

Overall

87
89

94
98
85
88

94
90
97
96
95


.94
.35

.93
.09
.97
.70

.27
.64
.13
.18
.34

Zl

«- 10
«- 9

15.8
15.2
-
—

+- 7
-<- 8
4- 10
4- 6
«- 10

.6
.6
.8
.3

_
.6
-

.0
.7
0.
5.
2.
7.

6.
13.
4.

4.
4.
C-157°
6
5
1
4

8
2
9

3
3
C
Emv (cm/s)
Z2 Z3 Z4

.8 + 0
.6 + 3

5.3 5
1.7 10
-
—

.4 + 12
.8 + 12
.8 + 16
.3 + 19
.4 + 11


.5
.9

.7
.7
-
—

.4
.5
.6
.1
.1


5.
6.

16.
23.
7.
7.

10.
1.
8.
11.
8.


7
5

0
7
4
0

2
0
4
2
4

Overall

5
4
6
5
6
6

8
8
7

4
5


.1
.3
.3
.3
.3
.7

.7
.8
.9

.0
.7

Overall

6
7

9
13
6
7

9
7
11
10
10


.9
.3

.8
.0
.4
.2

.4
.8
.7
.7
.1

+ = Zone 4 rapped every 10 mins

x = No rapping.










                        399
-------
    TABLE 8.   EMV ENHANCEMENT FACTOR FOR ZONES FITTED WITH PLATE-PLATE
               ELECTRODES IN PLACE OF WIRE-PLATE ELECTRODES

Gas
Gas
Gas
Gas
Gas
Test
temp.
temp.
temp.
temp.
temp.
Conditions
17°
118
149
118
149
C
o
o
0
o
J
C
C
C
C
with
, with
, with
rapping
rapping
rapping
Zone
13.
1.
4.
3
3
3
9
, no rapping
, no rapping
3.
0
Zone 4
3.
3.
2.
1.
1.
1
1X
8
2
2
     TABLE 9.  OVERALL EMV ENHANCEMENT FACTOR FOR 4 ZONE PILOT E.S.P.
WIRE-PLATE ELECTRODES IN THREE ZONES,  PLATE-PLATE ELECTRODES IN ONE ZONE
                                         Zone 3           Zone 4
                                       Plate-Plate      Plate-Plate

     All tests in gas temp,  range         1  .               ,  _
       107°C to 157°C                     i-^              i-J
17 C with rapping
118 C with rapping
149 C with rapping
118 C no rapping
149 C no rapping
Emv F.nhanrpinpnl" ~Fartn
1.6
0.9
1.4
-
1.4
Emv with Plate-Plate
T" — 	 —...-.— . . . . .- 	 . 	 	 ,, — . — ___
1.2
1.6
1.4
1.2
0.9
Electrodes
                                Emv with Wire-Plate  Electrodes

     The results are averages of several tests.

     Gas temperatures are the averages  of the  bands in performance
     tables.

     x Ignores test 5 which has zero emv for zone 4.
                                    400
-------
cr
ce
o
31
O
co
5.6


4.8


4.0


3.2

2.4

1.6


0.8


0



 FIG
                ZONE 3, WIRE-PLATE
                 MAXIMUM  {56.2 KV
                          L 5.66 ma
 GAS TEMPERATURE  17°C
 GAS VELOCITY     1.6m/s
FLY ASH COATED ELECTRODES
                                        ZONE 4, PLATE-PLATE
             10
        20    30    40    50
          APPLIED VOLTAGE, kV
                                                     SPARK 0.16 ma
 60
         1  VOLTAGE CURRENT CHARACTERISTICS OF A WIRE-PLATE
            ZONE COMPARED WITH A PLATE-PLATE ZONE

\
LU
O*
OC
O
LU
O
cr
o
to
O
4.0
3.2
2.4

1.6
0.8

                      ZONE 3, WIRE-PLATE
         SPARK ^ 26 kV,
           3.06 ma
            10
                             GAS TEMPERATURE  125 C
                             GAS VELOCITY      1.6 m/s
                            FLY ASH COATED ELECTRODES

                       ONSET OF
                       BACK-IONIZATION
                             ZONE 4, PLATE-PLATE
                                    SPARK » 49 kV
                                      I      I     I

                               50
       20     30     40

          APPLIED VOLTAGE, kV
60
70
80
     FIG, 2 VOLTAGE CURRENT CHARACTERISTICS OF A WIRE-PLATE

           ZONE COMPARED WITH A PLATE-PLATE ZONE
                              401
-------
    M
    u
    3.
    HI
    u
    cc
o CALCULATED
• MEASURED (WITH
   TEST NO.)
           ESP INLET
         0
  10     15     20

APPLIED VOLTAGE, kV
                                         25
                              30
         FIG 3  CALCULATED AND MEASURED CHARGE:/MASS
               RATIO RELATED TO THE VOLTAGE UPSTREAM
               OF THE MEASURING PLANE.
o
25 r ZONE


20

15
10
5


3
4
WIRE-PLATE PLATE-PLATE
o
• D •
• O "
•
GAS TEMP. 149°C / /
TESTS
-
—
i
WITH RAPPING / /
mQ
-------
    99


    95
    90
o   85
E9   70
a:
t/o
O
 >.
 o
 o
 UJ
    50

    30

    15
    10
      5
            1.  INLETZ1
            2-  INLET Z3
            3.  INLET Z4
            4.  OUTLET Z4
      1
      0.1
         1

PARTICLE DIAMETER
                                         10
      FIG. 5  PARTICLE SIZE DISTRIBUTION TEST 25
              H.T.PLATE IN ZONE 3-
    100

     90

     80

     70

     60

     50

     40
                                     ZONE 3
                                     OVERALL
      0.1
                        1
                PARTICLE DIAMETER
                                         10
       FIG. 6 EFFICIENCY v PARTICLE DIAMETER
             H.T.PLATE IN ZONE 3. TEST 25
                     403
-------
     THE PHYSICS OF PULSE ENERGIZATION OF ELECTROSTATIC PRECIPITATORS
                    By:   Lionel Menegozzi,  P.L.  Feldman
                         Research-Cottrell
                         P.O.  Box 1500
                         Somerville, NJ 08876
                                 ABSTRACT

     Analytical/numerical computations for pulser operation are given for
a simple geometry.   The model describes and calculates the negative ion
density produced by corona discharge due to a DC voltage with superimposed
pulses.  The space charge is then used to estimate wire-quenching effects,
and compute the charging and collecting fields.   The computer calculations
also provide the amplitude and duration of the resultant current pulses.
This information is needed for analytical treatment of back-corona sup-
pression with pulse energization.  Finally, we obtain the precipitation
enhancement factors for both large and small particles.   The results indi-
cate that pulser application is a conceptually sound technology that would
improve the precipitator performance in power-limited situations such as
those arising with high resisitivity dusts.  The results also give a deeper
insight into the dynamics of pulse energization.
     NOTE:   Please contact the author for information regarding this paper.
                                  404
-------
           ADVANCED ELECTRODE DESIGN  FOR ELECTROSTATIC PRECIPITATORS

                  By:   S.  Bernstein, K. Ushimaru, E. W.  Geller
                       Flow Research Company
                       21414 68th Avenue South
                       Kent, Washington  98031

                                    ABSTRACT

    The  subject of this paper is a wavy electrode precipitator which employs a
new collecting  electrode.  The  wavy electrode precipitator  incorporates a con-
toured collecting electrode geometry which produces a nearly uniform electric
field along  the electrode.   This feature allows  operation with higher average
electric field  strengths  near the electrode than in conventional designs.  The
wavy electrode  provides separated flow zones within its  "valleys", which fluid
dynamically  shield particles from the main stream.   The  design also provides
mechanical rigidity to vertical bending wtihout  the sharp flanges normal to
the flow commonly found in conventional precipitators.   These features permit
increased performance for all particle sizes,  but especially help in the pre-
cipitation of fine particles which are normally  the most difficult to collect.
                                   INTRODUCTION

    This paper  describes  the wavy electrtode  precipitator and gives an analy-
sis of its performance.   This advanced  electrode design is based on a unique
geometrical  configuration of electrodes which is expected to provide signifi-
cantly improved  performance  as well  as  reduced cost over conventional preci-
pitators .

    Overall  reviews  of electrostatic precipitator  technology are given by
White (1,2), and Robinson (3). A state-of-the-art precipitator configuration
is illustrated  in  Figure  la.   The collecting  electrodes are typified by baf-
fles which provide plate  stiffness and  shield the collected particulate layer
from direct  scouring by the  mainstream,  particularly during rapping.  The ad-
vanced design presented in Figure Ib is  typified by a wavy collecting elec-
trode shape  which  provides the same  functions with certain advantages.  A
brief outline of fundamental  collection processes is given here to emphasize
the advantages  of  the latter  configuration.   These processes are ion genera-
tion, particle  charging,  particle capture, particle removal, particle re-
entrainment, and sparkover.

    Gas ions are obtained from a  corona discharge at the anode to provide a
source for particle  charging  by diffusion and by field charging.  Diffusion
charging dominates the charging process  for very small particles (submicron

                                      405
-------
size); the charge increases as the free ion density increases.  High ion
density in turn is obtained by high operating voltage and uniform current dis-
tribution in the interelectrode space.  Field charging occurs when a collision
of an ion and a particle takes place.  A particle continues to capture ions
until the repulsive electric field of the accumulated free ions on the parti-
cle balances the attracting electric field of the background.  The value of
the saturation charge increases linearly with the magnitude of the background
electric field, which is determined by the voltage applied to the electrodes
and by the electrode configuration.

    An electrically charged particle experiences a coulomb force directed
toward the collecting electrode and proportional to the charge on the particle
and the intensity of the electric field.  Since this force increases with
particle charge and with the driving electric field intensity, highly charged
particles and a high average field intensity along the collecting electrode
surface promote efficient collection.  The resulting migration accelerations
for fine particles are small compared to the accelerations imparted by the
turbulent fluctuations present in the flow through the precipitator.  However,
in regions near the collecting electrode where the flow is separated and rela-
tively protected from the turbulent eddies, the electric field forces are ef-
fective in moving the particles to the collector.  Thus, one design goal is to
increase the sparkover threshold (discussed below), which is the determining
factor for the maximum voltage that can be applied.  Another goal is to en-
hance the charging and collecting electric field and current distribution for
a given operating voltage which will also increase the charging effectiveness.

    It is desirable to maximize both the charging and the collecting field
operation at the highest possible voltage.  The upper limit for the operating
voltage is determined by the sparkover voltage of the given electrode.   Thus,
peaks in the distribution of electric field intensity along the collecting
electrode surface tend to promote early sparkover.  Conversely, a uniform dis-
tribution delays the onset of sparkover.

    Several other phenomena play significant roles in precipitator performance
(e.g*, chemical composition of the particulate, back corona generation, the
conveyance of particulates into the hopper, and particulate removal).  We do
not intend to minimize the significance of these process, although a full dis-
cussion of these topics is beyond the scope of the paper.

                         THE  WAVY ELECTRODE PRECIPITATOR

    The wavy electrode precipitator is illustrated in Figure Ib.  In terms of
electric field and gas flow characteristics, the fundamental improvements over
the standard precipitator configuration shown in Figure la are:  (1) A more
uniform electric field which gives a higher average electric field for a
specified operating voltage and which allows operation at a higher operating
voltage without sparkover.  (2) The ability to obtain this condition while
maintaining separated flow regions which, for the standard configuration, are
created by flanges that are generally detrimental to the electric field.  The
bulging of the flow channel at each of the wire electrodes provides the more
uniform electric field (i.e., voltage gradient) and also provides the pockets

                                     406
-------
of "dead" air.   (3)The wave  shape provides structural stiffness for the  col-
lector plates so that no  stiffeners  need be attached during manufacture.

    The disadvantages and unknowns associated with the wavy electrode  precipi-
tator are the increased cost of  manufacturing the wavy shape,  the problem of
how to design the hanging of plate so as not to promote sparkover,  and the
fraction of the collector plate  area that will be protected from scouring by
the mainstream as compared to a  standard configuration.  The theoretical in-
vestigation has provided  some quantitative measure of the improvement  in col-
lection efficiency  that can  be expected from the wavy electrode configuration.
This work is reported in  the next section.

                                 PERFORMANCE ANALYSIS

    In order to compare the  performance of the advanced wavy electrode with
that of the conventional  electrode,  a detailed analysis of  fluid dynamic flow
field, electrical characteristics and efficiency was performed.  The analysis
included the adaptation of a numerical model introduced by  Bernstein et al.,
(4,5) which consists of a system of  seven coupled nonlinear  partial differen-
tial equations:  a  three-equation fluid dynamics model which describes conser-
vation of mass and  momentum  for  the  fluid flow,  a two-equation  turbulence
model, and a two-equation electrostatic model.  In order  to  predict the effi-
ciency of the precipitator for each  particle size,  a differential form of the
classical Deutch efficiency equation is used.   The precipitator section is
divided into fine elements,  and  the  local electric field  is  used to compute
the charging field  and the collecting field.  The charging  field is assumed to
be the average field normal  to the electrode surface and  the collecting field
is the field near  the collecting electrode at  each point  consistent with the
Deutch model.

    It should be recalled that for the purpose of computing  the efficiency,
the restrictive assumptions  of the Deutch model  (see,  for example, White(l,2))
are still applicable.  These include the assumption that  the fluid  turbulence
is dominant and effective in dispersing the particles  to  obtain uniform parti-
cle concentration  in the  interelectrode space.  Electric  field  effects are
dominant near the collecting electrode.  No reentrainment of particulates is
allowed, and no  particle-particle interactions.

    In this section, we describe the basic structure of the  model,  the bound-
ary conditions,  and the techniques of solution which have been  employed.  The
solution in the numerical model  employs a finite difference  scheme.  Since the
wavy electrode is not easily represented by a  finite difference  scheme, a co-
ordinate transformation was  employed to map the  wavy electrode  into a  rectan-
gular grid.

Numerical Model
    In  the  present  study we have adapted  and modified a model  for the des-
cription of  time-averaged fluid  motion  (4).  The model includes fluid trans-
port equations  and  electrostatic equations.
                                      407
-------
Transport Equations

    For time-averaged flow, the continuity and momentum  equations  can be
written as
                                 3p U.
                                 _i = 0                                 (1)
                                  3x.

                       3p U  U      a     3t

                       -T5^--%-*f+P*Ei                     <2>
where p(x) is the time-averaged fluid density at position, U  is  the  averaged
velocity, P the pressure, and T-M the stress tensor.  The term p*  E± is  the
pressure gradient caused by the discharge electric field where p*  is the space
charge density, and E^ is the electric field intensity.  The  turbulence  kine-
tic energy equation is written as
                                       rvi       J
                                       O Kf      1                           / O \
                                      ~^~ ~ -%=- T^ ~ P£                  (3)
and  the  turbulence energy dissipation equation as
                                             9U
                                        C
            -8x—  3^7 — 3x      T   cl 3x   ij    £2

and furthermore
                                                      3U  x
                                                   ff ^JL\   S- •   •
                                                  eff 3x., /   ij

Equations  (3) and (4) describe the transport of the turbulent kinetic  energy,
k, and  its rate  of dissipation, e.   The  effective viscosity are  given by

                              yeff = Cu pk2/E     '                        (6)
The parameters Cy,  Cei, C£2> <^k» anc^ ae are empirical constants.

Electrostatic Equations

    The two dependent variables describing the electrostatic field are the
voltage and the  ion space charge.  These dependent variables may  be described
using Maxwell's  electrostatic equation and conservation of  current.

                                         32        o*
               Maxwell's Equation        — —  = - -2—                      (7)
                                         r\ *••       O
                                         3x.        o
                                         a
               Conservation of Current   -g — (p* b E.) = 0                 (8)
                                           i
where <|> is the electrostatic portential, eo is the dielectric constant of
the medium, p* is the space charge density as before, and b is  the mobility
assumed to be constant.

    These  equations model the interaction between the electrostatics and
fluid dynamics.   However, the model  presently excludes the  effects of  particle
dynamics on fluid field using the following assumptions:  (1) The response
time of any given particle to adjust itself to the surrounding  fluid environ-


                                      408
-------
ment is sufficiently small to ensure that the particles follow the small-scale
fluid motion.   (2)  In case of low particle loading,  the fluid  dynamics  field
exerts little  influence on the electric field and,  consequently,  the  correla-
tion between fluid  velocity and electric field is neglected.

                             COMPUTATIONAL MESH GENERATION

    The above  set of equations is numerically solved using  finite  differencing
techniques.  These  computational methods for solving physical  problems  in
complex-shaped regions require the generation of  an  appropriate grid  system in
the region.  A good grid system ensures the economy  and accuracy in the compu-
tational procedure.  This section will discuss the two  independent methods of
grid generation:  one for the conventional precipitator and the other for the
wavy electrode precipitator,

Grid System  for Conventional Precipitator

    A state-of-the-art conventional precipitator  is  illustrated in Figure la.
An appropriate grid system conforming to this configuration is the one  that
models the regions  of high gradient (in fluid transport and electrostatics)
in detail.   For this reason, denser grids were allocated to the regions near
the wires and  the baffles.  Additional constraint is imposed on the field in-
tensity near the discharge electrode.  Following  the discussion in Cobine (6),
a characteristic corona radius is used for the calculations of the field in-
tensity near the wire.  The grid system was constructed by  taking  the first
grid node at that characteristic distance away from  the wire and gradually
increasing the grid size to the center of the precipitator  cell.  To achieve a
better resolution of the velocity gradients near  the baffles, a similar method
was used  in  the lateral direction to allocate a denser  mesh in those  regions.
Thus, nonuniform mesh system of 59 x 30 grids was constructed.

Grid System  for Wavy Electrode Precipitator

    For the  wavy electrode precipitator,  a boundary-conforming, orthogonal,
curvilinear  coordinate system was generated.  A recent  series of papers by
Reid et al.  (7) and Mobley and Stewart (8) presented a  method with flexibility
in generating  grid  systems more appropriate to physical problems.  A  second
transformation is incorporated to stretch and pack the  grid as the problem
necessitates.

    A sample mesh system of 59 x 29 grids for a wavy electrode precipitator
was used.  The coordinates were packed towards the wire in  order to place the
first computational grid node at the characteristic  wire radius as before.
The grids were stretched gradually away from the  wire to optimize  the number
of grid nodes  for economical computation.

    To utilize the  general orthogonal coordinate  system,  the governing  equa-
tions were rewritten using the usual rules of coordinate transformation.
                                      409
-------
                             RESULTS AND DISCUSSION

    The results of numerical modeling for electrostatics are presented in
Figure 2.  The applied conditions were a wire voltage of -70 kV at a room
temperature of 15°C.  These conditions were selected for comparisons with
laboratory conditions.  Obviously, in a full-scale precipitator, the operating
temperature will be higher and the maximum allowable operating voltage, as
well as the current densities, will be much lower.  However, both the conven-
tional and wavy electrodes were considered under the same conditions for our
analysis.  The test wavy collector was a sine wave with height (from the
centerline to the peak of the wave) of 15 cm, wave amplitude (peak-to-valley
distance) of 3 cm, and wavelength of 22.5 cm.  The solution for the electric
potential takes into account the effect of ion space charge, and the boundary
conditions along the inlet and outlet are assumed periodic.  As can be ob-
served in Figure 2a, there is a strong gradient of potential near the baffles
in a conventional precipitator, indicating an imminent sparkover under the
given test conditions.  The wavy electrode shows, on the other hand, more uni-
form distribution of potential throughout the precipitator cell, as seen in
Figure 2b.  This favorable property of the wavy electrode may allow the
applied voltage to be increased without promoting a sparkover as discussed
earlier.

    Figures 3a and b show the fluid dynamic flow field at corresponding ap-
plied conditions.  The volume flow rate was set at 0.6 m^/s which translated
to the inlet velocities of 0.6 m/s for the conventional precipitator and
0.5 m/s for the wavy electrode precipitator because of varying inlet geome-
tries.  As illustrated, the numerical predictions show the dramatic effect of
ion motion and ionic wind along the electric field lines.

    The influence of the ionic flow is strong enough that just beyond the cell
entrance the calculations show a rapid adjustment of the flow to the electric
field.  The region between the discharge wires experiences a minimum in both
the electric field and space charge concentrations; this results in a locally
minimum electric pressure gradient p* Ej_ which allows the fluid a return
path to the wire.

    The wavy electrode exhibits a different pattern in separation region near
the collection plate.  The strong interaction between the ion migration and
the fluid motion along the wavy-shaped plate promotes a large recirculation
zone behind the inlet section.  In the conventional precipitator, the recir-
culation zone extends to about one-quarter of the cell behind the inlet
baffles.  On the other hand, nearly two-thirds of the collection plate in the
wavy precipitator is protected by the recirculation flows.

    A comparison of current flux distribution is given in Figure 4.  The ap-
plied voltage was at -70 kV and at room temperature.  The solid line corres-
ponds to a numerical prediction of current density of a wavy electrode, and
the broken line corresponds to that of a conventional precipitator electrode.
The conventional precipitator has a jpeak current at the plate exactly opposite
to the wire, whereas the wavy electrode shows nearly uniform current density
over the collection plate.


                                     410
-------
    This condition is  conducive  to  a  higher average migration velocity  for
the particles being collected.   Particularly for fine  particles  (submicron-
diameter particles), the wavy electrode shows favorable trends in  improving
the particle charging  process by two  ways:   (1)  The average  charging field
over the precipitator  cell  is higher  for the wavy electrode,  giving the higher
field charging,   (2) The average current density in interelectrode space is
higher in the wavy electrode, to promote the higher diffusion charging on the
fine particles.

    The results  of these theoretical  calculations have been  utilized to obtain
a qualitative measure  of the  improvements offered by the wavy design:  namely,
the increase in  particle collection efficiency per unit collector area.
Figure 5 shows the efficiency per cell versus particle size  using a modified
Deutch-Anderson  technique.   The  efficiency predicted in this  figure is a re-
lative value and should not be taken  as equivalent to  the total  performance of
a given precipitator system.   As expected,  the submicron particles are most
difficult to collect.  However,  the wavy collector consistently  outperforms
the conventional collector  counterpart.  Since the field intensity peaks are
not present in the wavy design,  sparkover may be delayed and  the operating
voltage may be increased as schematically illustrated  in Figure  5.

                                   CONCLUSIONS

    The advanced precipitator electrode with the wavy  geometry shows promis-
ing improvements over  the conventional precipitator electrode.   Based on the
present analysis, this improved  collector design will  offer  (1)  a higher
average electric field and  (2) a larger flow-shielding region without sharp
flanges.  A combination of  fluid dynamic advantages and improvements in elec-
trostatic properties may become  attractive for precipitator  optimization once
the improvement  provided by each technique is proven.   The design may also be
incorporated with other design innovations,  such as the pulse-charging,
rigid-frame electrode  and multistaging of precipitators.

                                    ENDNOTES

1.  White, H.J.   Industrial Electrostatic Precipitation.  Reading,
    Massachusetts, Addison-Wesley Publishing Co., Inc., 1963.

2.  White  H.J.   Electrostatic Precipitation of  Fly Ash.  Journal of the Air
    Pollution  Control  Association,  27(1), January 1977; 27(2), February 1977;
    27(3), March 1977; 27(4), April 1977.

3.  Robinson,  M.  Electrostatic  Precipitation in Air Pollution Control, Part
    I., Strauss, W.  (ed.).   New  York, John Wiley & Sons, Inc., 1971.

4.  Bernstein, S. and  C.T.  Crowe.  Interaction Between Electrostatics and
    Fluid Dynamics in  Electrostatic Precipitators.  (Presented at the 2nd
    Symposium  on the  Transfer and Utilization of Particulate Control
    Technology,  Denver, Colorado, July 23-27, 1979.)
                                       411
-------
5.  Butler, G. W., K. Ushimaru,  and S.  Bernstein.  An Investigation of Fluid
    Dynamics, Electrostatics,  and  Fine  Particle Interaction in Electrostatic
    Precipitators, Flow Research Note No. 187.  Kent, Washington, Flow
    Research Company, December 1980.

6.  Cobine, J.D.  Gaseous  Conductors.  New York, McGraw-Hill, 1941.

7.  Reid, R.O., A.C. Vastano,  R.E. Whitaker,  and J.J. Wanstrath.  Experiments
    in Storm Surge Simulation.  In:  The Sea, Vol. 6, Chap. 5, Goldberg, E.D.
    (ed.).  New York, John Wiley & Sons, 1977.

8.  Mobley, C.D., and R.J. Stewart.  On the Numerical Generation of
    Boundary-Fitted Orthogonal Curvilinear Coordinate Systems.  Seattle,
    Washington, NOAA/ERL Pacific Marine Environmental Laboratory, 1980.

                                 ACKNOWLEDGEMENT

    This effort was supported  by the U.S. Department of Energy (DOE) under
contract DE-AC03-8EV10506. We wish to  acknowledge the encouragement for this
effort provided by Dr.  Fred Witmer of the DOE.
                         Discharge
                         Electrode
    (a)  A State-of-the-Art Electrostatic
        Precipitator
(b)  The Wavy Electrode Electrostatic
    Precipitator
                   Figure 1.  Precipitator Electrode Geometry
                                       412
-------
                 Average Wire apontential = - 70 kV
                                                                  0.1Sm
                  (a)   A State-of-the-Art Electostatic Precipitator
                Average Wire Potential = -70kV
               (b)  The Wavy Electrode Electrostatic Precipitator
           Figure 2.   Computed Electrostatic Potential, -70kV
                   Average Wire Potential = - 70 kV
                                       -0.45m-
                     Scale: 1.0 mis
                                                                0.15m
                  (a)   A State-of-the-Art Electrostatic Precipitator
               (b)  The Wavy Electrode Electrostatic Precipitator
Figure 3.   Computed Velocity Field with Average Wire Potential of -70 kV
                                    413
-------
CM
 .O
 a.
 CO
 03
 a
o
    0.25
     0.20 -
    0.15 -
                                   Conv. ESP (-70 kV)
                                       Wavy ESP (-70 kV)
                              15      20      25
                                     X Station (cm)
           Figure 4.  Comparison of Computed Current Density, -70 kV
                                                               Wavy ESP (-80KV)
                                                              Wavy ESP (-70 KV)
                                                               Conv. ESP (-70KV)
                                   234
                                    Particle Diameter (10'6 m)
       Figure 5.   Particle Collection Efficiency as a Function of Particle Size
                                         414
-------
              PROBLEMS IN APPLYING AN ELECTROSTATIC  PRECIPITATOR
                        TO A SALVAGE FUEL-FIRED  BOILER

                            BY CHARLES R. THOMPSON
                      ATLANTIC  DIVISION,, NAVAL FACILITIES
                              ENGINEERING COMMAND
                           NORFOLK, VIRGINIA   23511

                                   ABSTRACT

    This  paper  addresses the  problems  encountered  in applying  electrostatic
precipitator  technology  to  a trash burning,  salvage fuel-fired boiler.  The
electrostatic  precipitators  were  designed   and  constructed   in  a   period
between  1972  through 1976.   Stack  emission  tests  were  undertaken  after
construction  was   completed   and  revealed  that  the  emissions  were  .86
grains/DSCF  at  12 percent C02.   The Virginia Air Emission Standard  was .14
grains/DSCF  at  12  percent  C02«   Outlined   are the   steps   undertaken  to
achieve  final  compliance.  Each  improvement   of  electrostatic   precipitators
such  as  correction  of inlet  gas flow  patterns,  plate and wire alignment,
wire  improvements,  increase  in the number  of collecting fields and increase
in  the  power to  fields,  is  described and  discussed.   Also, improvements to
the salvage  fuel-fired boiler  operating  modes such  as trash feed, over-fired
and under-fired  air  and  excess   air  are outlined.   Finally, the design and
actual   operating  parameters   for   the  electrostatic   precipitators  are
presented and compared.
                                 INTRODUCTION

    In the early  1960s,  the  U.S.  Navy in Norfolk, Virginia, foresaw a future
problem  in  the disposal  of  its refuse.  An  engineering analysis  showed the
feasibility of burning  refuse to  produce steam.  A salvage fuel-fired boiler
plant  was constructed  and  completed  in  May  1967.   The  constructed  plant
consists  of   two  180  tons-per-day water  wall  furnaces.   The  furnaces  are
capable  of producing  50,000  pounds-per-hour of steam each.  The facility was
the  first steam  generating  water  wall furnace  to be  built  in  the United
States for incinerating refuse. (1)

    The  plant operation  involves  the  dumping  of refuse  into  a storage pit
and  lifting  the  refuse,  by  crane,  into a  charging  hopper.   The  refuse is
burned in three  stages on incline  recipitating grates.   The  grates move the
refuse through the  combustion chamber with a vertical  drop off between each
grate stage to tumble the refuse.   The tumbling  improves refuse distribution
and  enhances  its combustion.   Under-fired and  over-fired air  are  provided
for  combustion.   Each   furnace  also  has   an  auxiliary  oil  burner  used to
assist   refuse  burning   and   to  supplement  any  equipment  breakdown   that
prevents  refuse burning.

    The   plant   was  originally  constructed  with  cyclone  separators  for
particulate control.   The collectors were  made  of 24-inch diameter  cyclones
with  12  cyclones  per   boiler.    The  cyclones'   operation  efficiency  was
approximately 30  percent. (2)  The passage of  the  Clean  Air Act in  1970 and

                                      415
-------
the  State  of  Virginia  Air  Pollution  Implementation   Plans  applied  more
stringent environmental  regulations  to this facility.  The  use  of these low
efficiency  cyclones  on the  incinerators  did not meet  the  new environmental
regulations.   These  regulations required  the  construction  of new  and  more
efficient pollution abatement controls on the incinerators.

    Design  of electrostatic precipitators  for  the  facility was  started  in
1972.  The  design  specifications  for  the  electrostatic  precipitators called
for  a  single  electrostatic  precipitator for each  incinerator.   The design
specifications  for the precipitators  were  a performance  specification.   The
only  items   specified  were  inlet  flue  gas  parameters,   maximum  flue  gas,
velocity  through  precipitator, a  minimum collection efficiency  and a final
outlet grain  loading.

    In February 1973,  a  contract  was awarded for the construction of the two
precipitators.  The  precipitators  were supplied by Buell  Emission Control
Division   of  Envirotech   Corporation;   each   included   two  transformer-
rectifiers  rated  at 45KVA  DC  and  550MA DC output,  two  collection fields,
electrode  cleaning by  vibration  and  collecting  plate cleaning  by rappers.
Table  1  lists   the   electrostatic   precipitator's   parameters   as  it   was
originally  constructed in 1976.

    The  salvage  fuel-fired  boiler  exhaust with the  completed electrostatic
precipitators  was  stack  tested in 1976  to  demonstrate  final compliance  with
Environmental  Protection Agency  new  source performance  standards  and  the
State  of  Virginia Air Pollution Regulations.  Test  results  showed emissions
to  be  above  the  state standard.   The emissions varied  from a  low of  .167
grains/DSCF  corrected  to 12 percent  C02  up to .6  grains/DSCF  corrected  to
12  percent  CC^.   During  this  operating  time  visible emissions  ranged  from
20  percent  opacity to  60  percent  opacity with a normal average  of around 40
percent.  After  the results of the  stack acceptance test  were  reviewed,  a
program was  undertaken to  correct  the electrostatic  precipitators1 operation
and  to meet  the State  Air Pollution  Standards.

                  ANALYSIS  OF THE  ELECTROSTATIC  PRECIPITATORS

Investigation

    The  Atlantic   Division,  Naval  Facilities Engineering Command,  the  Navy
Public  Works  Center,  Norfolk,  and  a  consultant  investigated   the  poor
electrostatic  precipitator  performance.   The  evaluation revealed  that  the
precipitator  had   performed  at a  collecting efficiency  between  89  and  96
percent.   But  the  current  operating efficiency  of  the  precipitator  had
decreased to approximately  88  percent and  had  not operated  near  96 percent
since  initial  completion  of  the precipitators.   This efficiency decrease was
consistent  with  the  power decrease  in  the  Corona  power  input.   It was  also
found  that  the original  equipment  design was among the least  conservative of
municipal incinerator  percipitators  already  constructed.  (3)

     Stack   test  data  taken during   the  acceptance  test   showed   low  C02
readings  which continually  dropped  to an  average  of  eight  percent  at the
stack.   The excess air  had increased  to 200 percent.   Temperatures of the

                                      416
-------
flue gas stream were between 600° and 700°F.

    The  assessment  of   the   electrical  performance  of   the   precipitator
revealed very  low voltage  and current with  heavy sparking, particularly  in
the  inlet  section of  the  unit.  This,  in  effect, caused 40  percent of  the
precipitator collection  fields not  to be operational.  A suspected cause  was
poor  combustion  in  the  furnace  which  caused large  carbon  flakes  to  be
entrained  in  the gas  stream  and enter  the  precipitator.   The hopper of  the
precipitator also had  a problem filling up  with  ash  and  possibly  shorting
out  the  bottom  of  the HT  frame and weights.   The  anti-sway insulators  and
top  support  insulators were evaluated  and  a  high  build up  of  ash was  found
on  them.   This  situation  fostered   tracking across  the  insulators   and
eventual cracking of  them.   The anti-sway  insulators  in particular  had a
continuing   problem   of   cracking   and   breaking   requiring   continuous
replacement.    All  the   air    load   electrical  data  taken   during   the
investigation  and  past  operating   performance showed  lower  than   expected
readings.  In  addition to  cracked insulators contributing  to poor electrical
readings,  poor electrode  alignment  evolved,  caused by  large  expansion from
high  flue   gas  temperatures   of  600 to  700°F.   Also,  at  that   temperature
range,  the  carbon  content  of  the  ash  would  yield  very  low  effective
resistivity.

    A  check  was  made  of the  resistivity of  the ash.  The  results  showed a
range  of  5   X   106   to  3  X  105  ohms  -   centimeters;  consistent  with
expectations and  with  results  from  other incinerators.   At this resistivity,
it would  take  a  high  Corona  power  input, low  average  gas  velocity  of three
to four  feet  per second and good  gas distribution  for  this  percipitator  to
achieve high performance.

    The refuse characteristics that were estimated  for  the  U.S.  Navy refuse
are  in  line  with the  expectations of refuse  in general.  It must be noted,
however,  that  observations of  the  refuse  by the  author  showed   a  large
quantity of  metals and  large   wood  pieces  entering  the furnace.  The test
results  showed that the  incinerator was operating  at excess  air far above
what was considered  good practice.    (Good  practice  would  dictate excess air
of approximately 70  to 80  percent.)  Nevertheless,  the  operators  at this
facility had always maintained a high flow of  under  fire  conbustion  air and
no over fire air. The non use of the over fire air was to keep any  low melt
point metals from melting down.  Any  melting metals have caused the grates
to freeze up and  the entire boiler to be  shut down.

    A  final  and  important  area investigated was  the  flow  distribution into
the  precipitator.  A  model study was  accomplished during  the precipitator
design  using  a   30  percent,  open   perforated   plate  at  the  inlet   to   the
precipitator.   The precipitator was  constructed using  vertical  channels  at
that  location.   The  vertical  channels  do  not offer  any  control   of flow
vectors in the up or down direction.

    These vectors in   the up or down direction  would  cause  leaking under or
over  the  treatment  zones  and  lead  to  poor  collecting  efficiency   of  the
precipitator.  The data  from the model  study showed gas distribution quality
in the  model  to be  only  marginally acceptable  with maximum  gas  flow  40

                                     417
-------
percent above  average velocities  in  several  areas and  a standard deviation
of 28  percent.   The  Industrial  Gas  Cleaning Institute  Standard mandates no
reading above  a maximum gas  flow  of 40 percent  and  a standard deviation of
25 percent.

Deficiencies

    The  investigation revealed  numerous  deficiencies in  the facility.  The
deficiencies  could be  categorized into  two  areas:   boiler  combustion and
electrostatic precipitator deficencies.

    The  boiler  deficiencies  found were type of  refuse,  classified material
and  combustion  air.   The  refuse  fed  to   the   boiler  was  a  mixture  of
industrial  wastes   with   some  municipal  refuse.   The  industrial  refuse
included  large  items of metal and wood,  such as  pallets,  sheets  of plywood
and  four  by four  boards.   These items caused  pockets of poor burning.  The
low melting point metals would melt  down  and  solidify on the grates.  It was
also  found that  the  trash feed included, many times during  the  day,   large
quantities  of   classified  material.    This  material   was   shredded  and
manifesting  a  tendency  to burn at  a  very  high  temperature  causing sudden
boiler operation  changes and  upsets.   Another deficiency found on the boiler
was  that  the  over-fired  air  was  not being  used to  assist  the combustion.
This was  due  to the fact  that in the past when over  fired  air was  used, the
grates tended to  freeze up and break.

    The  electrostatic precipitator had  numerous   deficiencies  in  its design
and  operation.   It  was  found that  the air  leakage   into  the precipitator,
both  from the  boiler and  the  expansion  joints,  was  highly  excessive.  The
gas distribution  into the  precipitator was found  to be very suspect due to a
90  degree  turn  immediately  in  front  of   the   precipitator  and  a   rapid
expansion  in  all  four  directions at the inlet   transition.   The  discharge
electrodes  were found to  be highly  corroded  and  a large build up of ash was
found  in many  locations.   Ash  build up  was also  found along  the support
insulators  at  the  top  of  the  precipitator  and  the  anti-sway insulators at
the  bottom of  the  precipitator.  A  check of  the plate and  wire  alignment
showed  some possible poor alignment but  nothing extreme.   From  the design
perspective,  it was  found  that the aspect ratio (the  length-to-height of the
precipitator) was poor, .536, and  that the   power applied  to each  field and
the amount of collection areas should have been increased.

    After  the above deficiencies had been identified, steps  were undertaken
to  correct each one.   The items were listed  in  order of  those which   would
merit  the  most  impact in  meeting  the  air pollution regulations in the  least
amount of time.  Essentially, this  led to a  first  attempt  in  bringing the
precipitator  into  compliance  with   Virginia  State   Air  Standards.    These
efforts   did   not  have   the  results  that   was  originally  intended  and
subsequently  compliance with  the  Virginia Air Emission  Regulations did not
meet  or  exceed  projected  goals.   A  second   attempt  was  undertaken  to
collectively  correct most  of the above   deficiencies  to assure  that   final
compliance  would  be  obtained  by  the  July 1979  deadline the U.S.  Navy had
agreed upon with  the  State of Virginia Air Pollution Control  Board.


                                     418
-------
                              CORRECTIVE  ACTIONS

First Corrective Implementation

    The   first   major  corrective   action  involved  moving  the  anti-sway
insulators  outside  the  precipitator  shell  and  blowing  hot  purge  air over
them.   It  was thought  that  this correction  would eliminate the low voltage
levels  and  numerous  electrical  trip-outs  that were occurring regularly.  The
top support insulators were also manifesting a problem of dust build up and
tracking across  them.   The  purge air system  to  these insulators was slightly
modified.   The  modification  involved  the way in which  the purge  air was
being directed across  the insulators.

    Stack  emissions  were tested after correction of the  above  items.   The
results  of  the  tests  revealed  that there was  essentially no change  in the
particulate   or   visible  emissions   from  the   facility.   The  particular
emissions  were  still  averaging  .86  grains/DSCF  corrected  to  12  percent
C02«   Visible emissions  averaged  greater  than  20  percent opacity  most  of
the  time.    The   performance  of   the  electrostatic  precipitator  had  not
improved.   The  number  of electrical  trip-outs  and  the  power level  to the
fields  were   the  same  as   before  the  corrected  items  were  attempted.
Consequently,  it became apparent that improving  the precipitator performance
with  piece-meal  corrective  actions would  not  bring  the  precipitator  into
compliance  with  State  of Virginia air  pollution  standard by July 1979.

Second  Corrective  Implementation

    The   second   phase  of   corrective  action  was  a  simultaneous  effort
attacking  all the other  deficiencies  previously  noted.   These  deficiencies
were  grouped  in  five  primary  areas:   the  insulator's  purge  air  system,
precipitator  alignment and  internals,  inlet  flow  distribution,  total  power
available to  the collection fields   and,  finally,  boiler operation problems.

    Continued investigation into the purge air  systems  and the dust build up
and  tracking across   the  insulators led  to measuring  the  volumetric  flow in
this  system.   The measurements  found  that the  purge air  flow was  very low
and that  many insulators  were obtaining no purge  air at all.  To correct the
situation  the total  volume  of   purge  air for both  the  top support  and the
anti-sway  insulators was doubled.   This  was  done  by adding additional fans
and separating duct  work to  the insulators.   This effort  led  to  a dramatic
decrease  in  the  amount  of  dust  being   deposited upon  the insulator  and
essentially eliminated  any tracking problems  across them.

    The   electrode  and  collecting   plate   alignment   were  rechecked  for
tolerances.   This check  found  a  few  minor  alignment  problems which  were
immediately corrected.   The biggest  deficiency found was  the  fact that the
existing  electrodes  were shroudless wires,   and  that some  necking  down was
taking  place where  the  wires   pass  the  top and  bottom  stiffeners  of the
plates.   To  correct   this  deficiency   new  electrodes  were  procured  and
installed  with  shrouds.   The  shrouds were  one-half  inch in  diameter and
extended approximately  eight  inches past  the  plate stiffeners.  It was  found
that   immediately   after  installating   the  new  electrodes   there  was   a

                                     419
-------
significant increase  in current to  the  collecting plates and less  sparking.
But heavy sparking was  still taking  place  in  the  inlet  field.

    The  third  area of  correction  was the  flow  distribution.   Flow  readings
were  taken  at  the  inlet flow  transition  approximately  three  to  four  feet
before  the  precipitator.  The  results  showed  high flows  in  the center  and
top  of  the  electrostatic precipitator  and  some  reverse  flow  on  one  side
(4).   It was  also expected  that much of  the  flow was  dividing into vectors
heading  to the  top and  bottom  of the precipitator.  These flow vectors  could
cause  the gas  steam  to channel  itself  to   the  top  and/or  bottom  of  the
precipitator.   To  correct this  possibility,  the  existing  vertical  channels
were  replaced  with  a  50   percent   open   perforated  plate.   On  the   inlet
transition side of the  plate,  eight  inch wide horizontal shelving was  placed
at  approximately 1.5  feet   intervals.   It was  expected that  these shelves
would  assist  in minimizing  the upward and  downward flow vectors.  Also,  the
existing flow  straighteners  at  the 90 degree  bend before the transition were
extended 15 inches.

    The  fourth  area  corrected  involved increasing the total power within  the
precipitator.   The  original  design  of  the  precipitator  called   for  the
construction  of two  collection fields.  But the precipitator was constructed
with  the ability  to  divide  into a  five-field precipitator,  each field  being
three  feet  in  length.   A new  third transformer-rectifier  set  was  added  to
each   precipitator   to  modify   the   precipitator   into   a   three-field
precipitator.   The new transformer-rectifier  was attached to the first  three
foot  field of the precipitator; with the  second  and  third  collecting  field
each  being  six foot  in  length using  the existing  transformer rectifiers.
This  set up  allowed   the  isolation  of any  continued  heavy   sparking  and
fostered low  power levels to only  the  first  three feet of  the precipitator.
Hence,  the  second  and third  fields could operate at  a much  higher   power
level  and improve  the  overall collection efficiency  of  the  precipitator.
After  energizing  the  precipitator  with  this  set up,  the above   intended
impact did occur.

    The   final   deficiency   area  corrected  was  boiler  operations.    This
involved resealing   all  possible  air   inleakage  spots  such  as  furnace
observation  and entrance door  seals and grate  access  door  seals.   Also  the
electrostatic  precipitator   extension joints  were  rebuilt.  As  previously
discussed the over-fired air had  been used very  rarely  due to  the  freezing
up  of   the  grates.    To  allow   greater   use  of  the  over-fired  air   the
under-fired   air   was  more   evenly  redistributed  throughout  the  furnace
combustion zone instead of  being  isolated  in one  primary  grate area.    Also
damper  controls were  installed on  over-fired air  fans.   An  evaluation  of
combustion  activity  in the modified furnace  revealed  an  increase  in  the
recirculation  of air  over   the  flame zone with much  better  combustion  and
less  large  unburned particulate matter  leaving the  furnace area.   When  all
of  the above work was  accomplished  there was a  decrease  in gas flow  volume
from  the precipitator of approximately 30 percent.

                                    RESULTS

    Completion  of  the corrective  work  on  the  furnace and  electrostatic

                                      420
-------
precipitators  brought  visible emissions  down  to 10 percent  opacity  or less.
A  stack  emission  test  was  then  run  for  particulate  emissions  with  the
results  showing a significant decrease  from the previous  testing.   However,
compliance with  the  State  standards was  still not met.   The  results  were for
particulate  matter   .244   grains/DSCF  corrected   to  12  percent   C02   and
visibile emission of approximately  three percent opacity.

    A  review of  the  situation revealed  that  the electrostatic  precipitator
was  performing  competently  with  the  Corona  power  level   as  expected.   A
review  supported that all  corrective  items  discussed  had  been completed.
The  stack emission  testing  crew reviewed  their testing  procedure  and  the
collected sample.   It  was   then discovered  that approximately six times  more
of  the  particulate  catch  was in  the wash  than on the  filter.   This meant
that a  substantial  amount   of  the particles  emitted from the stack were  very
large.   Consequently,  since large  particulate matter would  not   emerge
through  the  precipitator   the reason  for  failure of  the  stack  test  was
probably due to  reintrainment of  particulate matter  after the precipitator.

    A   maintenance   review   of   the  downstream  breaching  and  stack  was
undertaken.   This examination found  large  rust flakes  continually peeling
off breaching  walls and an  ash build up in the  bottom of  the  stack all the
way  up  to  the  point  where  the  breeching enters  the  stack.   A  thorough
cleaning  of  the duct work  and  stack   area  was undertaken and new  stack
emission  tests  run.   Results  of  the tests  showed that both  the particulate
and  visible  emissions  were in  compliance  with  Virginia  Air  Pollution
Regulations.   The particulate results were  .052  grains/DSCF  corrected  to 12
percent CC>2 and  opacity at seven  percent.

                       COMPARISON  OF OPERATING PARAMETERS

    A  comparison of the   final operating  parameters to  the  original  design
warranty  is shown in Table 2.  A  review  of  the  table shows  that  it  took a 50
percent  decrease in the flue  gas volume for particulate  emissions  to comply
with  the  required design  emission.  After  the  first  corrective  action,  the
flue   gas   volumetric   flow   and  temperature   met   design  specifications;
therefore,  the  other design  parameters  were met.   But as previously stated,
the  outlet  particulate concentration exceeds  the air  standards.   I believe
this again  documents the  fact that the  original precipitator design  is  not
conservative

    An  additional parameter  that has changed  but   can not  be shown  on  the
table  is  the  manner in which refuse is  fed  to  the boilers.   Before  the  last
stack  emission  tests  all  refuse was  presorted on  the tipping  floor.   This
sorting  eliminated   large  metal   and wood items.   The change  meant a more
complete  and consistent burn of  the refuse.   This presorting had  a  large
impact   on   lowering   the  gas   flow,   flue   gas  temperature   and  inlet
concentration.
                                     421
-------
                                   ENDNOTES

1.  Collins,  John P.,  Refuse:   The Urban  Ord.  U.S.  Navy.   (Presented  at
American  Defense  Preparedness  Association,  Energy/Environment  Conference,
Kansas City, Missouri, March 27-31 19771) p.3-4.

2.  Bibbens, Richard N., Air Pollution  Source  Emissions  Test of Salvage Fuel
Boiler (technical  report YF 38.534.003.01.001  for Navy Public  Works  Center,
Norfolk, Virginia).

3.   Hall,  H.   J.,   Electrostatic   Precipitator  Operations,   Problems  and
Solutions in Refuse  Incinerator  Applications  (Presented at  American  Society
Mechanical  Engineers,  Solid Waste Processing Division  Meeting,  New York,  New
York, January 16,  1980)

4.    Buhmann,   K.  A.,  Technical  Report  651:KAB  (Technical   Report  for
Department  of Navy, Atlantic Division,  Naval Facilities Engineering Command).
                                     422
-------
         TABLE 1  SUMMARY PARAMETERS FOR ELECTROSTATIC  PRECIPITATORS
                        QN SALVAGE FUEL-FIRED  BOILERS
                       NAVAL  STATION,  NORFOLK,  VIRGINIA
    Item

 1.  Particulate Perf.  Reqm't
      Federal Emission Reqm't
      State of VA.  emission reqm't
 2.  Fuel fired
 3.  H20, % vol in gas
 4.  A  P across PPTR,  Max "H20
 5.  Airtight seals
 6.  Design gas flow
 7.  Guar. Eff., % wgt
 8.  Inlet Cone., gr/ACFM
 9.  Gas vel., ft/sec
10.  Chambers, cells
11.  No. of Fields
12.  No. of Ducts
13.  Plate Height, ft.
14.  Coll. Plate Length, ft.
15.  Coll. Area, ft2
16. SCA, ft2/1000 ACFM
17. Design w, ft/sec
18. Aspect Ratio, 1/h
19. Area/rapper, ft2
20. Discharge wire
21. Effect wire length, ft
22. Ft. of wire/rapper
23. No. Rect. Sets, Si
24. Total Installed I
25. Avg J, ma/100 ft2
                     avj
MA
26. Automatic Control
27. Hopper heat
28. Disch. wire spacing, in
29. Gas Distribution
                            Parameter
0.08 gr/SCF @ 12% C02
0.14 gr/SCFD @ 12% CO2
Solid waste material & fuel  oil
13-15
0.5
Max drop 0.25% C02 inlet to  outlet
76,000 ACFM @ 690F
96
0.92
3.35
1 each
2 (91 + 6')
18, 9" width
28
15
15120     inlet  9072
          outlet 6048
199
0.27
0.536
3024    Impulse (5)
0.105" dia. Hi C steel - 25# wgt
10080
2016
2 - 35KVA ea. 45KVDC, 550 MA
1100
58     inlet  J = 48.5
       outlet J = 73
Ana Cotnp II
yes  one hopper/PPTR
8
Vertical channel + Perf. PL. on  inlet/
Perf. plate on outlet
                                     423
-------
          TABLE  2  PERFORMANCE DATA FOR ELECTROSTATIC  PRECIPITATORS
                        ON  SALVAGE FUEL-FIRED BOILERS
                       NAVAL  STATION. NORFOLK, VIRINIA

    Parameter                       Design                November 1979

1.  Gas 1000 ACFM inlet               76                       34.0
2.  Inlet gas temperature, °F      690-725                      573
3.  SCA, ft2/1000 ACFM               199                      444.7
4.  Eff. % wgt*                       96                       97.0
5.  w, ft/sec                       0.27                      .0079
6.  Avg v, ft/sec                   3.35                       1.49
7.  Outlet gr/SCFD @ 12% C02        0.08                      0.052
8.  Avg inlet C02%                                              7.2
9.  Avg outlet C02%                                             6.9
               EA %                                             161
10. Max drop in C02
    inlet to outlet                  .25%                       .3%
11. Inlet cone Ib/hr, avg            600                      12.04
12. Outlet cone Ib/hr, avg            24                       5.39
*Calculated using gr/SCFD measurements
                                    424
-------
         THE APPLICATION  OF  ELECTROSTATIC PRECIPITATORS

               TO BOILERS FIRING  MULTIPLE FUELS

            By:  Robert L. Bump,  Technical Manager
                      Industrial Precipitators
                      Research-Cottrell,  Inc.
                      Somerville, New Jersey

                             ABSTRACT

     Recent years have seen  a  swing to boiler designs which
afford versatility in the fuel fired.   This is, of course, a
result of the economics and  availability  of conventional fuels.
Many process industries are  using their own waste  products as
a principle fuel source.   The  varying conditions imposed on the
air pollution control device as a result  of this practice will
be discussed in this  paper.  Case histories will be presented
as well as a discussion of some of the operational considerations
which must be recognized.  Since  the  pulp and paper industry has
been a forerunner in  this activity, the majority of the data
presented will derive from this application.


                           INTRODUCTION

     The theoretical  aspects of electrostatic precipitator
application and behavior  are well known and need little commen-
tary.  The original Deutsch-Anderson  equation has  given way to
more sophisticated methods of  establishing  precipitator size for
a given duty which involve computer programs that  take into
account the many process  and fuel variables which  affect design.
Figure 1 indicates the various inputs  required to  derive the
proper specific collecting area.  The  division of  responsibility
for the input data between the engineer/user segment and the
supplier is also indicated.  The  complexity is caused primarily
by the fuel types and sources, ash analysis and their resultant
impacts on design.  Coupled  with  the  increasing stringency of
legislation and demand for a very high level of reliability we
have a ball game which if not  new is  certainly different.  To
make the problem more challenging, we  have  the added ingredient
of most industrial power  boilers  being designed for multiple
firing of coal, waste and oil.  This  is,  of course, a result
of the uncertainty of adequate fuel supplies, the  attendant
costs and the nebulous state of S02 legislation for industrial
power boilers.

Application Considerations

     How does this situation affect the engineer specifying air
pollution control equipment  and his counterpart applying it?

                               425
-------
There are fundamental data, usually furnished by the boiler manu-
facturer, which are basic to the application process.  Figure  2
outlines this information.  It is obvious that in the case of
multiple fuel usage this information is required for each oper-
ating mode.  Figures 3 to 8 are from a typical specification and
indicate fuel analysis and operating data for a boiler designed
to burn coal, wood waste and oil in various combinations.  The
most important conditions to be considered in sizing the preci-
pitator for the specified efficiency are the gas volume for each
operating mode, particulate loading, temperature and ash analy-
sis as affecting the precipitability (resistivity) for each fuel
or combination.  In the example cited, the gas volume and the
particulate loading for the coal-wood combination is greater than
the 100% coal or oil-wood combination.  The frontal area of the
precipitator is, then, set by the highest gas volume and the
maximum gas velocity permitted by the suppliers design or the
specification.  In this case the specification set 3.5 feet per
second as maximum.  The resistivity of the coal ash was found
to be approximately one order of magnitude greater than the
other fuel combinations necessitating the highest specific
collecting area for the specified efficiency.  In summary, we
have a situation where the coal-wood gas volume set the cross
sectional area and the coal ash resistivity set the length of
the precipitator.  It is not unusual for the "best to worst"
conditions on a multiple fuel application to involve as much as
50% spread in the sizing requirements.  Figure 9 demonstrates
this point.

Mechanical & Electrical Design Considerations

     Multiple firing conditions also require additional care in
several mechanical and electrical design areas.  Since power
density varies with changes in gas and particulate composition,
it is necessary for the designer to be aware of these ramifica-
tions in order to make proper selection of the transformer-
rectifiers.  In our example because the ash from wood waste is
of a lower resistivity and the moisture content of the gas about
20% higher than the coal fired mode it controlled the applied
power density.  Probably the most important area of design care
on multiple fuel applications is versatility in the collecting
surface rapping system.  It is understood that most losses  (of
particulate) in a precipitator occur as a result of reentrain-
ment during rapping.  It is also understood that the energy re-
quired to remove high resistivity ash from collecting surfaces
can be as much as six times that needed to dislodge a highly
carbonaceous, conductive material.  A rapping system must,
therefore, have a very high degree of intensity and cycle ad-
justment which may be readily made in order to cope with fuel
variations.  A microcomputer based control permits keyboard
adjustment of each rapper as well as the cycle time.  All oper-
ational parameters should be displayed at the controls via  digi-
tal readout devices.

                              426
-------
     The last area where design care  is of  utmost  importance  is
the dust removal system.  A large percentage  of  precipitator
operational and maintenance problems  stem from improper  ash
removal design or operation.  The topic is  too involved  to treat
here, however, suffice to say the system must be sized properly,
preferably of the continuous removal  type,  with  sufficient
auxiliaries (hopper heat, vibrators,  level  detectors, etc.) to
do the job.  This is particularly true with multiple fuels
since much of the particulate has a high combustible content
and the danger of fires exists if the ash is  allowed to  accumu-
late in the hoppers.  There has been  good experience with trough
type hoppers and continuous removal by the  use of  screw  convey-
ors.  Reference 1 treats this subject in considerable detail.

Operational Considerations

     The final consideration in applying precipitators to multi-
ple fuel operations is the variations in operating characteris-
tics.  Each firing mode and fuel results in gas  composition and
particulate chemistry which is different and  the precipitator
reacts operationally to these variations.   It is,  therefore,
essential that not only the rapping system  have  the versatility
described but that the precipitator power control  system design
affords fast, effective reactions to  the process changes.
Modern digital and analog controls accomplish this in milli-
seconds.  Figure 10 demonstrates the  operation of an effective
precipitator control circuit.  Other  control  areas which are
important have to do with the combustion process.  As an example,
take a boiler firing wood waste and coal.   The wood waste is
typically fired on a grate, requires  considerably more air to
burn than coal, and the resultant particulate may be as high  as
40-50% combustible.  Care must be taken to  minimize fire hazard.
Oxygen monitors interlocked with the  precipitator power supplies
are advisable.  Periodic checks to insure a "tight" system
should be made since 02 inleakage at  doors, expansion joints,
dust valves, etc. can cause hazardous conditions.  The importance
of a proper dust removal system and auxiliaries has already been
emphasized.  Thorough operator training on  the effects of fuel
variations not only at the firing end but in  the entire process
is a necessity.

Conclusion

     The technology for the application of  electrostatic preci-
pitators to multiple fuel fired boilers has been demonstrated.
Care and accuracy in specifying the conditions for the various
firing modes is essential from the user end.  The  supplier must
recognize the effects of the variables on equipment application
and design.
                               427
-------
Reference 1 - "Design and Operation of Reliable Central
               Station Flyash Hopper Evacuation Systems"
               by Joint Technical Committee of American
               Boiler Manufacturers Association and Indus-
               trial Gas Cleaning Institute, Inc.  Presented
               at American Power Conference, April 21-23,
               1980, Chicago, Illinois.
                               428
-------
                       PRECIPITATOR SIZING DATA
          ENGINEER/USER
                                    SUPPLIER
K5
     HV
     ULTIMATE
     ANALYSIS
     ASH CONTENT
     CODES, ETC.
FUEL FIRING RATE
          FIGURE 1
                     GAS VOLUME
                      EFFICIENCY'
ASH ANALYSIS
                                     RESISTIVITY
                               T
                            VELOCITY
                                                               POWER DENSITY
                                          REGRESSION EQUATION


                                                 SCA
                                               GAS VOLUME
                                                      COLLECTION AREA
-------
             Basic Specification Information



           Gas volume,  acfm,  normal and peak



           Temperature, normal and peak



           Pressure,  normal and design



           Dust loading, grains per acf or scfd



           Efficiency required



           Type of boiler (P.C. stoker, etc.)



           Type of fuel(s)



           Fuel analysis



           Particle size








                       Figure 2








                   Coal Analysis (.Identify State and Mine)



Ultimate (percent by weight)       Performance      Range



Carbon                                73.50     65.0-85.0



Hydrogen                               4.80      4.5-5-5



Sulfur                                 0.70      0.5-1.1



Nitrogen                               1.20      1.0-2.1



Oxygen                                 5.60      4.0-8.0



Moisture                               7.00      3.0-15.0



Ash                                    7.20      2.0-16.0
                        Figure 3
                           430
-------
   P2°5
           Ash Analysis  (Percent  by  weight)
                         29.79
   Fe2°3                 38.07
   A12°4                 20.13
   Ti02                   0.81
   CaO                    4.27
   MgO                    0.91
                          4.45
                          1.01
                          0.41
   Misc.                  1.31

                  Figure  4

    Analysis of Wood Waste and Bark (Percent by weight)

Type                      3/8" max, size
Carbon                    24.0
Hydrogen                   2 . 5
Sulfur                    ----
Nitrogen                  ----
Oxygen                    18.5
Moisture                  45.0 - 65.0
Ash                        5.0
Sand                      29.0 (of the ash as Si02)
HHV                       4300 BTU/lb.


                 Figure 5
                     431
-------
            Analysis  of  No.  6  Fuel  Oil  (Percent by weight)

           Sulfur                      0.70

           Hydrogen                  10.33

           Carbon                     88.33

           Nitrogen                   0.14 (0.3 max.)

           Oxygen                      0.50

           Heating Value              18,400 BTU/lb.




                           Figure 6
Size

% less than
 30 microns

% less than
 20 microns

% less than
 10 microns
Particulate Size Leaving Air Heater

   Coal      Coal/Wood     Oil     Oil/Wood

    83         83/50        53       53/50

    65         65/36        34       34/36

    44         44/27        27       20/27
                            Figure 7
                               432
-------
                Precipitator Design Criteria
Gas Flow, acfm
Temperature, °F
Heat Input to Boiler,
BTU/hr. x 106
Fly ash, inlet, Ibs./hr.
Fly ash, outlet, lbs./106 Btu
Ground Elevation, ft.
Hopper Storage, hrs.
Minimum Aspect Ratio
Maximum Plate Height, ft.
Maximum Pressure Loss, "Wg.
Maximum Gas velocity,
ft. /sec.
Efficiency, %
Opacity, %
100% Coal
335,000
273 +20
1055
13,950
0.075
14'-6"
12
1.0
32
0.5
3.5
99.5
20
50% Coal
50% Wood
435,000
335 +20
1055
19,500
0.075
14'-6"
12
1.0
32
0.5
3.5
99.5
20
50% Oil
50% Wood
348,000
318 +20
1055
17,500
0.075
14'-6"
12
1.0
32
0.5
3.5
99.5
20
Figure 8
Precipitator
Design Conditions
Gas volume, acfm
Inlet loading, gr/scfd
Outlet loading, gr/scfd
Efficiency, %
Note: Coal to be worst Eastern
Sizina Parameters
Velocity, fps
Treatment time, sees.
Tl -—• w» *-± f^ J— ii" •> 4" 1 /"\
Sizing Example
Wood
288,
6.
0.
98.
Waste
360
5
02
97
Bituminous
4.2
X
1.5 ,
Coal
246,040
1.47
0.02
98.57
3.64
1.4X
1.8 10
f^t-/ K' V»- V-" w. J- »-* v— —
Resistivity,  ohms/cm.
3.5 x 10
6.3 x ID'
                           Figure 9

                              433
-------
                           FIGURE 10
        TYPICAL POWERTRAC™ CONTROL OPERATION

          DURING UNSTABLE PROCESS CONDITIONS
UJ
-p-
    tu
    cc
    cc
    3
    o
    cc
    o
    o
    uu
    DC
           SPIT
           or
          SPARK
               I  RAMP  I


                 >8SEC
RAMP f  RAMp



<8 SEC I  >8 SEC
                          INTERRUPT ARC ONLY • PEDESTAL OUT
               1 CYCLE

               (16 mS)
                             TIME
-------
                            AUTHOR INDEX

AUTHOR NAME                                                   PAGE

Albrecht,  P.R.                                                  IV-116
Anderson, M.H.                                                  11-334
Arce-Medina, E.                                                   II-76
Ariman, T.                                                      III-290
Armstrong,  J.A.                                         IV-188, 1V-252
Bakke,  E.                                                        1-236
Balfour, W.D.                                                   III-119
Bamberger,  J.A.                                                 III-398
Bergmann, L.                                                    1-323
Berlant, MJ.                                                   H-218
Bernstein, S.                                                   H-405
Beutner, H.P.                                            HI-71, III-228
Bickelhaupt, R.E.                                                !-165
Boericke, R.R.                                                  III-353
Bohn, R,                                                       IV'344
Borenstein,  M.                                                  ni"90
Brookman, E.T.                                                 IV-125
Bump,  R.L.                                                     H-425
Bush, P.V.                                                      I"157
 Calvert, S.                                           ~'    ~>
 Games,  D.                                                     IV~135
 Carr,  R.C.
 Chamberlain, H.L.                                               IV-406
                                  435
-------
AUTHOR INDEX  (cont.)
AUTHOR NAME                                                    PAGE
Chambers,  R.                                                       1-45
Chiang, T.                                               III-250,  III-261
Chou,  K.H.                                                       IV-73
Cowen, S.J.                                                      IV-264
Crippen, L.K.                                                     1-148
Crowson, F.                                                      Ill-438
Crynack, R.R.                                                   11-242
Czuchra, P.A.                                                     IV-55
Dalmon, J.                                                        11-390
Demski, R.J.                                                      1-341
Dennis, R.                                                  1-1,  III-140
Dietz,  P.W.                                              III-449,  III-459
Donovan, R.P.                                                     1-11
Drehmel, D.C.                                            III-341,  IV-210
DuBard, J.L.                                                     IV-383
Durham, M.                                               11-54,  IV-285
Ensor, D.S.                                               1-176,  IV-242
Eskinazi, D.                                                      III-238
Faulkner, M.G.                                           11-199,  IV-395
Feldman, P.L.                                                      IV-3
Ferrigan III, JJ.                                                  I_197
Finney, W.C.                                                     11-358
Fjeld,  R.A.                                                      n_179
Fortune, O.F.                                                      TOO
                                  436
-------
AUTHOR INDEX (cont.)

AUTHOR NAME
Frazier,  W.F.                                                    III-171

Gardner, R.P.                                                   III-128

Gaunt, R.H.                                                       T"216

Gehri, D.C.                                                       I"333

Gentry, J.W.                                                    m-406

 Giles, W.B.                                                     HI-468

 TT  j-     TO                                                     ni-33
 Hardison, L.C.

          T^ T                                                     IV-317
 Harmon, D.L.
                                                                  III-221
 Hawks,  R.L.
                                                                    1-75
 Helfritch,  D.
                                                                  III-301
 Henry,  F.
                                                                   IV- 63
 Henry, R.F.
                                                                   IV-222
 Hesketh, H.E.
                                                                   III-382
 Hoenig, S.A.
                                                                     1-23
 Hovis,  L.S.
                                                                    1-129
  Hyde,  R.C.
                                                           III-181,  III-321
  lionya,  K.
                                                                    1-185
  Jaworowski, R.J.
                                                                    1-138
  Jensen, R.M.
                                                                    II-370
  Joergensen, H.J.
                                                                    1-352
  Johnson,  C.A.
                                                                   III-311
  Kalinowski, T.W.
                                                                   III-280
  Kanaoka, C.
                                                                   III-373
  Kirstein, B.E.                      437
-------
AUTHOR INDEX (cont.)
AUTHOR NAME                                                    PAGE
Kolnsberg, HJ.                                                  IV-179
Krishnamurthy,  N.                                                IV-232
Ladd, K.                                                      1-55,  1-65
Lagarias,  J.S.                                                     1-272
Landham,  Jr.,  E.G.                                                1-237
Langan, W.T.                                                    111-211
Lawless, P.A.                                         11-25, 11-35, 11-44
Leith, D.                                                         III-270
Leonard,  G.L.                                                     11-120
Maartmann, S.                                                     11-130
Mahoney,  D.F.                                                    1-206
Mappes, I.E.                                                    III-150
Martin,  D.                                                       IV-145
Masuda, S.                                                11-189,  11-380
Mathai,  C.V.                                                     IV-200
Mazumder, M.K.                                            11-160,  11-169
McCrillis.  R.C.                                                   IV-306
McElroy,  M.W.                                                       1-94
McLean, KJ.                                               1-265,  11-304
Menegozzi, L.                                                     11-404
Menoher,  C.                                                     Ill-Ill
Mitchner,  M.                                                       II-97
Moore, W.E.                                                     IV-105
Mormile, D.                                                      IV-363
                                  438
-------
AUTHOR INDEX (cont.)
AUTHOR NAME                                                    PAGE
Moslehi,  G.B.                                                    11-109
Mosley, R.B.                                                 II-l, 11-13
Musgrove,  J.G.                                          III-193,  III-201
Noonan,  F.M.                                                    IV-326
Oglesby, H.S.                                                    HI-80
Ostop, R.L.                                                       J'107
Parker,  R.                                                 111-51, IV-2
Parquet, D.                                                      III-363
Parsons, Jr., E.L.                                                I"303
Fatten, J.D.                                                     HI-160
Pearson, G.L.                                                     I~120
Pedersen,  G.C.                                                   m-60
Petersen, H.H.                                                   T'291
Piulle, W.                                                         I"253
Potokar, R.W.                                                    III-417
       ,                                                          IH-21
Prem, A.
         a u                                                     IV'26
Presser, A.M.
Pyle, B.E.                                                        l™
Raemhild,  G.A.                                                   U"349
          r- v                                                    III-102
Reardon, F.X.
                                                                  11-262
Rimberg, D.B.
Rinaldi, G.M.
      ,  _                                                 11-283, 11-295
Rinard,  G.
         r  TVT                                                      IV-83
Rubow,  L.N.                      439
-------
AUTHOR INDEX (cont.)
AUTHOR NAME                                                   PAGE
Rugg,  D.                                                       11-273
Samuel, E.A.                                                    H"149
Schliesser, S.P.                                                  11-252
Semrau,  K.T.                                                     m'43
Shilling, N.Z.                                                    11-230
Smith, W.B.                                                       !-96
Snaddon, R.W.L.                                                  IV-74
Sparks,  L.E.                                              H-314,  11-326
Spawn, P.D.                                                    IV-335
Starke, J.                                                      IH-428
Stevens, NJ.                                                    1-313
Sullivan,  K.M.                                                  H-141
Tatsch,  C.E.                                                    IV-353
Teller, AJ.                                                     HI-393
Thompson, C.R.                                                  11-415
Urone, P.                                                      IV-275
VanOsdell, D.W.                                                   1-35
Viner,  A.S.                                                     IV-168
Wakabayashi,  A.                                                 III-332
Wang, H.H.                                                      IV-36
Wang, J.C.F.                                                   IV-373
Wegrzyn, J.                                                      IV-46
Weyers, L.L.                                                     1-226
Wilks, W.H.                                                      IV-15
                                  440
-------
AUTHOR INDEX (cont.)


AUTHOR NAME                                                  PAGE


Williamson, A.D.                                                IV-297


Yamamoto, T.                                                    n"87


Yung, S.                                                 IV-1( IV-155


Zarfoss, J.R.                                                   n-208
                                    441
                                                  6USGPO: 1982 — 559-092/0430
-------