-------
-103-
In the size region between about 0. 2 and 0. 5jU, both diffusion and field
charging are significant, and hence the migration velocity would be that
associated with the combined charge resulting from the two charging regimes.
The migration velocity related to field charging is useful for most
types of dusts. However, one should realize the limitations under which the
equations were derived, and that they do not apply to very fine dusts and
fumes.
Another modification to the previous migration velocity equations is
necessary for very small particles. The equation as stated applies to
particles where Stokes1 law governs the equation of motion. For smaller
particles, the Cunningham correction factor is applied to the equation for
migration velocity. Under these conditions, the expression for migration
velocity as given in Equations 4.4, 4. 5, and 4. 6 must be multiplied by a
factor (1 + A — ), where
a
A = mean free path of gas molecules,
a = radius of the dust particles, and
A = a numerical factor.
Note that the Cunningham correction factor is significant for small particles
in the size region where diffusion charging is important.
4. 2 PARTICLE COLLECTION WITH LAMINAR GAS FLOW
Although laminar flow is only obtained in some laboratory models,
and approached in only a small percentage of the commercial two-stage
precipitators, it is informative to begin the discussion of the theory of
particle collection with the laminar flow model. In this model, the maxi-
mum velocity point is generally associated with the center of the pipe,
and reduces with increasing radius. If we neglect the variation of velocity
with radius, and assume a "plug" type flow, we can assign a velocity, Vg,
to both the gas stream and the dust through the pipe (see Figure 4. 3 for a
sketch of this flow system).
Charged dust particles in an electric field will attain a velocity
(w) towards the collection electrode. Since there are only two velocity
components related to the dust motion, the net velocity is the vector sum
of these two components. Since all dust particles will flow in this same
resultant direction, a dust-free zone will develop in the center of the pipe
-------
-104-
R
vr - Resultant Velocity
Figure 4.3. Length of Precipitator Required for 100%
Collection for Idealized Laminar Gas Flow
in a Pipe Precipitator of Radius, R.
SOUTHERN RESEARCH INSTITUTE
-------
-105-
and grow to fill the entire pipe as the gas flows through a distance (L) that
is determined by the pipe radius R, gas velocity v and migration velocity w.
The equation relating these quantities to the length required for 100 percent
collection efficiency is:
L = -32- (4.7)
w
This assumes that there is no erosion or other reentrainment losses
within the precipitator. Since migration velocity is proportional to
particle radius (field charging), it becomes necessary to size the precipitator
for the smallest particles expected in the dust.
This idealized case occurs so infrequently that it is almost of academic
interest alone.
4. 3 PARTICLE COLLECTION WITH TURBULENT GAS FLOW
In all practical precipitators, the gas flow is turbulent and the motion
of a dust particle will be primarily the result of the turbulent gas flow, gas
motion caused by collision between neutral gas molecules and the ions
driven by the electric field (electric wind) as well as the migration velocity.
Therefore, the specific trajectory for any one particle cannot be completely
predicted for this case.
The equation for the collection efficiency of an electrostatic precipi-
tator was discovered experimentally based on field tests by Anderson in 1919.
Deutsch derived a similar equation based on theoretical considerations in
1922. In the early 1950's, White2 derived an identical expression based on
the probability of collection for a single particle. Both derivations lead to
collection efficiency equations of the form
rj = 1 - exp (- — w)
Vg
where
r? = collection efficiency,
A/v = specific collection electrode area,
w = particle migration velocity.
*Refer to the bibliography for this chapter.
-------
-106-
The exponential collection equation has also been verified by laboratory
experiment, and is generally accepted as describing precipitator performance.
Since Deutsch is credited with the initial development of the theory of
collection, equations of the above form which relate to precipitator operation
are known as Deutsch-Anderson efficiency equations.
The development of the original Deutsch-Anderson equation was based
on several simplifying assumptions. These were:
(1) The particles are considered to be fully charged immediately on
introduction into the collection system.
(2) Turbulent and diffusive forces cause the particles to be distributed
uniformly in any cross section. (This assumption is more restrictive than
necessary in actual practice.)
(3) The velocity of the gas stream does not affect the migration
velocity of the particle.
(4) Particle motion is governed by viscous drag where Stokes' law
applies.
(5) The particle always moves at its electrical terminal velocity.
(6) Dust particles are sufficiently separated so that their mutual
repulsion can be neglected.
(7) The effect of collision between ions and neutral gas molecules can
be neglected.
I
(8) There are no disturbing effects such as erosion, reentrainment,
uneven gas flow distribution or back corona present.
As a review of the original derivation, consider a single charged
particle in a wire and pipe precipitator suspended in a gas stream under
the influence of an electric field. The particle will acquire a velocity (w)
in a direction toward the collection electrode as described by Equation 4. 2a.
In the central turbulent region of a precipitator, the electrical migration
velocity is small in comparison to that component due to the gas flow. But,
in a boundary zone of thickness (6 ), the gas flow will be laminar because
of the friction between the gas stream and the wall. An example of the
SOUTHERN RESEARCH INSTITUTE
-------
-107-
velocity distribution as a function of radius is shown in Figure 4. 4. In this
boundary layer, since the gas flow is laminar, the resultant velocity for a
dust particle will be the vector sum of the gas velocity and the drift velocity.
Therefore, in an increment of time At, all the dust particles present within
the boundary layer (6 ) will be driven to the collection electrode if
6 = wAt
During this same interval of time, the gas stream will have moved through
the precipitator a distance
AL = vAt
where v is the average gas velocity.
From these conditions, the incremental equation for particle removal can
be established
AN _ S6 _ SwAt SwAL wAAP
N " Ad Ad Adv = " Adv (4.8)
where
AN = particle concentration removed,
N = particle concentration in the gas stream,
AAC = incremental collection area = SAL
S = circumference of pipe
vg = volume flow rate of gas = Adv
Ad = cross sectional area
v = gas velocity
Passing to the limit with Equation 4.8, we find
(4.9)
N Adv Vg
Integration of Equation 4.9 yields an expression for the particle concentration
at the precipitator outlet.
N = N0exp ( -
-------
-108-
Position
g
Gas Velocity
vgxQ
g
vgxQ
actual gas velocity.distribution,
average gas velocity outside boundary
layer x quality factor, and
boundary layer.
Figure 4. 4. Possible Gas Distribution in a Precipitator
Duct.
SOUTHERN RESEARCH INSTITUTE
-------
-109-
where N0 is the particle concentration at the inlet. The collection efficiency
can be stated as the ratio of particles removed to the inlet concentration.
r, = Nn - N = 1 _ (__^cw_) (4.10)
1\ V
0 vg
White derived an identical equation based on the probability of a
particle being collected in a circular wire and pipe precipitator during an
interval of time, At. If the particle is within a distance, wAt, of the collection
electrode, it will be trapped by the electric field and will be collected. Then
the probability of collection within a given interval of time is the probability
of the particle being located within the outer annulus. Since the particle has
equal probability of being located at any point within the cross section, then
the probability of collection, Pc, is the ratio of the area of the annulus to
the cross-sectional area of the precipitator.
The probability of escape within each section AL, is
P' = 1 - P - 1 2wAt
es c R •
and the probability of escape through the complete precipitator is the proba-
bility of escape in one section raised to the power of the number of sections
in the unit
(4. 11)
"- *• 4TV11
where
At = total time divided by the number of sections considered on the
precipitator.
Algebraic manipulation yields an equation identical to that previously derived
in Equation 4. 10.
A somewhat more general expression of the same form will result if
the original assumptions are somewhat modified.
(1) The dust particle concentration, Ng, in the collection layer, 6, is
proportional (but not necessarily equal) to the average particle
concentration, N0, in any cross section.
-------
-110-
(2) The effective gas flow through the precipitator can be described
as the product of the average gas velocity, vg, times a quality
factor, Q.
(3) The collection layer thickness, 6, is defined as the distance away
from the collection electrode to the point where the migration
velocity, w, is equal to the linear velocity of the gas stream down
the precipitator in the laminar flow region.
In the central region of the precipitator where the gas velocity and turbulent
velocity components are large compared with the migration velocity, the dust
trajectory will be random. In this region the effect of the migration velocity
will be negligible. But, in the region near the wall, in the collection zone,
the longitudinal velocity is limited by the frictional forces of the wall. The
translational velocities are effectively reduced, while the migration velocity
component towards the collection electrode remains constant. In this col-
lection layer, 6, the migration velocity dominates and all particles within 6
will be removed within an incremental time, At, which is defined by the
relation
At = 6/w
In the collection layer, the migration velocity causes all particles
within the distance, 6, from the wall to be collected within a time, At.
During this time the gas moves a distance AL = v QAt. The number of par-
ticles in a particular size range removed will be the product of the number
density of particles in the boundary layer and the volume of this region
cleared, or
AN = -NcSwAtAL (4.12)
o
The number of particles within the volume of the precipitator defined by the
length, AL, and the cross section of the pipe is
N = N^AL (4.13)
which yields an equation for the increment of particles removed
AN _ NgSwAtAL _ N6 Sw = N§ Sw AL
N -' N0AdAL =~ NQAd ^ Ad vQ
SOUTHERN RESEARCH INSTITUTE
-------
-111-
where
Ng = particle concentration in the boundary layer,
v = average gas velocity, and
Q = quality factor for the gas flow.
Passing to the limit we have
^. = -^..|_._^_dL (4.15)
N N0 Ad vQ
Integration of Equation 4. 10 yields an expression for the particles col-
lected or removed
N, Sw
N = N0exp( -- 2- -- L). (4.16)
N0 AdvQ
The efficiency can be determined from
Nn - N N5 SL
» 1 - exp (- — • -
.
w)
which can be expressed as
o
The derivation for the collection efficiency given above relates the
collection efficiency of an installation to the migration velocity of the dust
particles. Since the migration velocity of dust particles is dependent upon
the particle size as well as other factors, the equation as stated applies
specifically to a particular particle size for which the migration velocity
is given. Thus in the strict sense, the Deutsch-Anderson equation applies
only to a group of particles for which the migration velocity is known. If
the performance for a typical dust is desired, it is necessary to separate
the specific particle size distribution into several discrete size ranges,
compute the migration velocity and collection efficiency for each size range,
and then determine the overall efficiency. This is discussed in more detail
in Chapter 8, Systems Analysis.
-------
-112-
Equation 4. 17 differs from the previously described Deutsch-Anderson
equation in two respects: the particle concentration in any cross section,
and a quality factor, Q, can be used to modify the average gas flow rate
through the precipitator.
The quality factor, Q, is included to provide a means to modify the
efficiency equation for installations with a highly uneven gas flow. The
Deutsch-Anderson equation utilizes an average gas velocity, v. In the above
derivation, the average gas velocity is multiplied by Q to yield an effective
gas velocity. The factor, Q, could be different for each precipitator instal-
lation and must be determined individually. For installations with a rela-
tively even gas flow the numerical value of the factor would approach unity,
while in some extreme cases, it could approach a value of 2. A numerical
example will show some insight into the manner in which the uneven gas
distribution could cause a variation in the overall efficiency of an installation.
For this example, consider a simplified precipitator where a uniform
gas flow exists in each of four identical areas of the unit, as depicted in
Figure 4. 5. The average gas velocity for this example is determined by
summing the products of the velocity and the area, which yields an average
gas velocity of 2. 5 v where v is an arbitrary value. If the expected col-
lection efficiency for this unit with a uniform gas flow of 2. 5v is 90%, the
Deutsch equation can be used to determine the value of the exponent,
rj = 99% = 1 - exp ( - g A™ ),
from which
,j.
Aw/2. 5v = In 1QO_r? = In 100 = 4. 6.
The example can be analyzed as if the simple precipitator in Figure
4.5 were really four individual precipitators with different gas flow rates.
The percent of material not removed in each precipitator could be expressed
as Aj, A2, A3, A4, as follows:
At = exp ( - -^-) = exp ( - 11.75) = 0.000007
O
A2 = A3 = exp ( - Aw ) = exp (- 5. 875) = 0. 0029
SOUTHERN RESEARCH INSTITUTE
-------
-113-
g
2V
g
Figure 4. 5. Assumed Gas Flow Distribution for
Quality Factor Example. Average
Velocity Proportional to Average
Volume Flow Rate = 2. 5 v .
5
-------
-114-
\ = exp (- ^2L.) = exp (-2.35) = 0.0954
5 vg
The total material escaping is then the weighted sum of that lost in each
section, or
ATotal = °-4887
for constant loading per cu ft, From this, the overall efficiency can be
expressed as
q .»-°-«" ,.8.8. 94.4».
The efficiency for the actual case is about 4. 6% less than expected for
one with the average flow.
The numerical value for the quality factor can be determined from
the relation
O - i 100 4. n 100 _ In 100 _ 4.6
V ~ 100-Th ' 100-T)a " lnl8 " 2-9
Q = 1.6
Therefore for this example, the effective velocity is 1.6 times the average
velocity.
4. 4 FACTORS MODIFYING THE PARTICLE
COLLECTION IN A PRECIPITATOR
Several factors can lead to a significant change in the collection
efficiency of the idealized precipitator as described in the section on par-
ticle collection theory. These factors include particle agglomeration, back
corona, uneven gas flow, and erosion and rapping reentrainment, among
others. Most of these factors tend to decrease collection efficiency.
However, agglomeration would result in a larger effective particle size and
would tend to increase collection efficiency.
The migration velocity of a particle is proportional to the radius of
the dust, as seen from Equation 4.3. As the particles agglomerate, the
ratio of electrostatic force to aerodynamic drag increases, with a resultant
SOUTHERN RESEARCH INSTITUTE
-------
-115-
increase in migration velocity. Thus as the particles agglomerate, the col-
lection efficiency of the precipitator tends to increase.
Electrical sparkover and back corona are two related phenomena that
tend to decrease the collection efficiency of electrostatic precipitators.
Electrical sparkover results when a localized gas breakdown occurs across
the interelectrode space.
The phenomena of spark initiation have been studied extensively by
Loeb3, Penney? and others. In terms of electrostatic precipitator per-
formance, the following factors are significant.
(1) Spark initiation occurs at the anode.
(2) Spark initiation is influenced by the magnitude of the electric field
near the anode.
(3) Propagation of a spark depends on the magnitude of the applied
voltage.
For a negative corona, the magnitude of the electric field near the
collection electrode is related to the resistivity of the dust and the current
density according to the equation
E = jp
where
p = electrical resistivity, ohm-cm.
The influence of the electric field on sparking can best be explained
in terms of the voltage-current curve, Figure 4. 6. The voltage impressed
across the electrode of a precipitator will result in a current flow, the
magnitude of which will depend upon the gas composition, temperature,
pressure, and precipitator electrode dimensions. If the electrodes are
clean, the voltage can be increased until sparking occurs. If the collection
electrode has a dust layer with a resistivity of around 10 ohm-cm, a
current density of 10"6 amp/cm2 will result in a field in the deposit of
around 10 kV/cm (indicated by Point A in Figure 4. 6).
-------
-116-
A-p
B-p
C-p
.10
10 " ohm-cin
5xlOl° ohm-cm
10U ohm-cm
e
o
"a
S
rt
0]
£
e
0)
3
U
13.80
40
10.35 <30
£ 6.90 120
3,45 U10
20 40 60 80 100
Applied Voltage. kV
Figure 4.
-------
-117-
A further increase in the resistivity to 5 x 1010 ohm- cm causes the
electric field in the deposit to reach 10 kV/cm at a current density of about
7 x 10~7 amp/cm2 corresponding to Point B on Figure 4.6. Since spark
initiation will occur at a lower voltage due to the higher field at the anode,
the precipitator will operate at Point B. If the dust resistivity is further
increased to 1011 ohm- cm, the same field in the dust deposit will result from
a current density of only 10"8 amp/cm2. Consequently, the voltage and
current at which the precipitator operates will be further reduced and oper-
ate at Point C on the curve.
As resistivity of the dust layer is increased, a point will be reached
at which the applied voltage will be inadequate to propagate a spark across
the interelectrode space. Under these conditions, the current and resistivity
will be such that breakdown of the gas in the interstitial regions of the dust
layer will occur. This condition results in the formation of a corona in the
dust layer. The resulting positive ions formed travel across the interelectrode
space toward the discharge electrode. These ions will impact upon dust
particles in the interelectrode region and neutralize or reduce the charge
present on the dust due to the negative ions flowing from the region of the
discharge electrode.
The effects of sparking and back corona influence precipitator oper-
ation in several ways. Within the range of resistivities where sparking
occurs, operating voltage and current are reduced over that of a clean
electrode precipitator. The electric field is reduced by virtue of the
reduced voltage and current, and the charging time is increased by virtue
of the reduced current. In a sparking mode, the precipitator is momen-
tarily deenergized during the interval of the spark. Consequently, the
efficiency is influenced in proportion to the total deenergization time.
The effects of uneven gas flow, erosion, and rapping reentrainment
were discussed previously and need not be repeated here.
4. 5 PRACTICAL ASPECTS OF PARTICLE COLLECTION
The Deutsch-Anderson equation, rj = 1 - exp (- *$-. w), is widely used
to describe the performance of precipitators. Because of this wide ac-
ceptance, a general term, effective migration velocity (EMV) or precipi-
tation rate parameter, is customarily associated with the collection
efficiency of installed units. The value of the EMV can be determined when
values are known for the collection area (A), volume flow rate (vg), and
efficiency rj .
-------
-118-
This w is not the velocity with which particles move toward the collection
electrode as given in the previous section, rather it is the net average
migration velocity for an average particle which would result in the
measured efficiency for a particular installation, and therefore is, in effect,
a performance parameter.
Several factors contribute to the fact that the precipitation rate
parameter is different from the theoretical migration velocity. These
factors include reentrainment, uneven gas flow, particle size distribution,
and back corona, and have been discussed individually in other sections of
the report. Collectively, the result is a net flow of particles from the wall
back into the precipitator such that the effective flow toward the collection
electrode is reduced.
Thus, while the precipitation rate parameter is useful for describing
the actual performance of a precipitator, care should be exercised in order
to insure that the value of this number is understood to be related only to
the performance of a precipitator, and not to any real measurable velocity
of particles toward the collection electrode.
4.6 RECENT MODIFICATIONS TO THE DEUTSCH EQUATION
It is generally recognized that the basic Deutsch-Anderson equation
has a number of limitations principally associated with the assumptions
made in the derivation. Several investigators have attempted to modify the
basic Deutsch equation so that it would more nearly describe precipitator
performance.
Cooperman3 states that the Deutsch equation neglects the role of
diffusional forces in the precipitator which can account for the variation
between observed and theoretical migration velocities and the apparent
increase in the precipitation rate parameter with increasing gas velocity.
Cooperman postulates that the difference in particle concentration along
the precipitator length gives rise to a diffusional force which results in
a particle velocity through the precipitator that is greater than the gas
velocity. At low gas velocities, the effect is pronounced, but is masked
at higher velocities.
SOUTHERN RESEARCH INSTITUTE
-------
-119-
In addition to the linear diffusion, Copperman suggests that a
difference in particle concentration would result in a back diffusion if
the particle concentration at the collection surface were higher than
that within the central sector, and hence result in a lower migration
velocity.
Cooperman suggests an equation of the form
r, = 1 - exp [.- 0/a (1 - ) -- ] (4.20)
where
b = half width of the precipitator duct (comparable to radius),
L = precipitator length,
f = ratio of particle transport away from the collecting electrode to
the migration velocity towards it,
w = electrical migration velocity (theoretical),
Dx = particle diffusion coefficient in the direction of the gas flow, and
v = mean gas transport velocity.
If we perform a few algebraic manipulations the exponential term can be
made to appear more like the Deutsch terms. This is desirable because
of the widespread use of the Deutsch equations.
- =- (4.2D
Now let us add a factor (h) = height of the collection electrode from which
L. x JL = __£_ (4.22)
b h Ad
where
Ac = collection electrode area, and
AJ = cross-sectional area of the duct.
-------
-120-
Thus the product of #/<* and —- becomes
b
0/a-x ±- = U-f) 2L . h. (4.23)
b vg Ad
and vx AJ = the cross-sectional area of the duct times the mean gas
velocity which is the volume flow rate (y±) of the preeipitator. /Therefore,
Equation 4. 24 becomes
(l-f)w
vg
Equation 4. 20 can be rewritten with more familiar terras so that
i? = 1 - exp-[-(1-f) (1 - -~-) A w ] (4.24)
.3
-------
-121-
but (we- wr) is merely the net particle velocity towards the collection
electrode (wn) so that Equation 4. 24a becomes
n - l-expt-U-V^M-11^-)] (4.24c)
Therefore, the new theory developed by Cooperman is closely related
to the original theory developed by Deutsch, and takes into account two
factors which can significantly modify the electrical and convective
parameters used in the original derivation. These factors can bring theory
and practice into better agreement.
No experimental verification of this theory was included.
Robinson derived an equation for collection efficiency* which also
offers an explanation for the wide disparity between the precipitation rate
parameter and the theoretical migration velocity. In this analysis, two
erosion terms are introduced. The first erosion term is proportional to
the density of dust particles in the boundary layer. Erosion of this type is
based on the concept that the greater number of impacts between dust particles
and the deposit layer, the greater the quantity of dust that will be reentrained.
The second erosion term is a constant term such as one would expect to find
with a constant velocity gas stream flowing near a collected layer of dust,
which for a fixed velocity should give a constant erosion rate. These two
factors are combined to yield an efficiency equation which is a modified
Oeutsch equation.
„ = l - (l - _£__ )exp[- (1 - a) X w A/Vg- -*L-] (4.26)
1 - or s 1 - a
where
a = coefficient of the variable erosion,
/3 = coefficient of the constant erosion,
X = ratio of dust concentration near the wall to the average
concentration in the cross section, and
w, A, v = standard precipitation parameters.
o
-------
-122-
Equation 4. 26 can be algebraically manipulated to become
rj = (1 - -y-M 0 - exp [ - (1 - a)X w A/vg ]} (4. 26a)
which corresponds to modifying the Deutsch equation in three respects:
(1) The entire equation is multiplied by an erosion sensitive
term, 1 - j3 (1 - a).
(2) The particle concentration near the wall can be different
from the average particle concentration (X).
(3) The migration velocity is modified by the variable erosion
effect, w-eff = (1 - a)w.
The two papers discussed above represent the more significant
attempts to bring the theoretical and experimental aspects of precipitation
phenomena into closer agreement. Experimental verification of one of
the above models appears to be necessary to resolve the differences, if in
fact differences exist.
SOUTHERN RESEARCH INSTITUTE
-------
-123-
CHAPTER 4
BIBLIOGRAPHY
1. Deutsch, W., Ann, der Physik 68, p 335 (1922).
2. White, H. J., jndustrial Electrostatic Precipitation, Addison-
Wesley (1963).
3. Cooperman, P., "A New Theory of Precipitator Efficiency, " Paper
No. 69-4. APCA Meeting, New York City (1969).
4. Robinson, Myron, "A Modified Deutsch Efficiency Equation for
Electrostatic Precipitation, " Atmospheric Environ 1, No. 3, pp
193-204 (May 1967).
-------
-124-
CHAPTER 5
GAS FLOW
5. 1 INTRODUCTION
Gas flow is important in the precipitation process for a variety of
reasons. First, the gas volumes handled can be large so that the cost
of equipment can be high if high pressure drops are involved. The power
costs also will be excessively high if large losses occur in the duct. An
even more important consideration, however, is the uniformity of the gas
flow entering the precipitator. Severe nonuniformity causes erosion of
the collected dust from hoppers and collecting electrode and causes
variable gas treatment times in the precipitator. The combined effects
can cause marked reduction in precipitator performance, often as much
as 20-30 percent. Poor gas flow can also result in excessive deposits of
dust on ducts, turning vanes, elbows, and distribution plates. These
deposits can further alter the flow pattern and result in further nonuniformity
of flow.
The fundamentals of fluid flow are covered in many standard texts
on fluid mechanics. Basic types of flow are defined depending upon the
gas velocity, duct size, and gas properties. One of the most important
dimensionless parameters in fluid flow is the Reynolds number which is
defined as
«. -
where
p = gas density,
v = velocity of gas,
M = viscosity, and
D = equivalent diameter based on wetted perimeter
(D = 4 area
perimeter
If the Reynolds number is less than critical (about 2000 to 26fOO
in a duct), the flow will be laminar and the velocity profile across the
duct will be parabolic as illustrated below. In this type flow, the
maximum velocity is twice the average for a cylindrical pipe.
SOUTHERN RESEARCH' INSTITUTE
-------
-125-
When the Reynolds number is greater than critical, any small
disturbance will not die out but will continue to grow until the gas flow
has fully developed turbulence. The flow profile will have a boundary
layer that increases in velocity from zero at the wall up to the mean
velocity of the core of the flow. With reasonably constant average
velocity and turbulent flow, the maximum velocity will be about 110%
of the average velocity. The velocity profile will be nearly parabolic
in the boundary layer and nearly flat, but fluctuating, in the core:
3
With an uneven velocity distribution, the maximum velocity can
be much higher than the average velocity and the profile can be almost
any shape.
All large size precipitators operate well into the turbulent region.
Typical values for Reynolds numbers for electric precipitators are
10, 000 and higher. The gas flow in all practical sizes of duct work is
also in the turbulent region. Gas flow in duct work is subject to wide
variations in velocity distribution.
When a duct changes cross section, the velocity profile changes.
In a converging transformation, the velocity vectors must increase to
compensate for the smaller area of cross section. If the profile is
reasonably constant, the magnitude of the vectors change with little
-------
-126-
variation between maximum and average velocity. However, with a
diverging transformation, the entering air tends to maintain its velocity
distribution because of inertia. The flow separates from the wall, and
eddies form which tend to grow to the full size of the duct because of
the large Reynolds number (see sketch below). The stream tends to
spread out slightly, but separation will occur for diverging angles greater
than about 7° included angle.
Separation will also occur on one side of a turn, as illustrated
below.
Reverse flow ,in the separation zone as indicated above is common
in flue systems unless special precautions are taken to prevent it.
As illustration of the effect of a sudden transformation in cross
section of a duct, let us consider the kinetic energy of the moving gas:
At a point (1) upstream of a transformation: (KE)j = 1/2 mvx2
At a point (2) downstream of the transformation: (KE)2 = 1/2 mv22.
SOUTHERN RESEARCH INSTITUTE
-------
-127-
The continuity equation for a square duct of side S would be:
v p m = Av = S2v;
o
then
. Vl _ , _§2_
and
\T
V2
(KE)i _ 1/2 mvt2 = , vl 2 , S2 *
(KE)2 " 1/2 mv22 v2 ) " S, '•
or the kinetic energies vary inversely as the fourth power of the side
of the duct; thus doubling the side of a square duct reduces the average
kinetic energy far downstream to 1/16 of what it was upstream. The
other 15/16 of the energy must be dissipated in turbulence and viscous
losses.
The relationship of velocity to pressure, or "head" is based on
kinetic energy. The velocity head equals the kinetic energy:
1 2 1 P 2
h = -5- mv = -s- — v
e> & g
or
v =
In English units, we have
p = pressure in inches water gage,
w = pounds per cubic foot, and
v = feet per minute.
w
-------
-128-
For standard air, 70°F at 29. 92 inches Hg barometer,
v = 4005A/p"~~.
Now, a loss due to friction in any part of duct system is a loss
in energy. Therefore, the loss can be expressed as a function of the
"Velocity Pressure":
p = w
Losses in elbows, wyes, transformations, and so on, are often
expressed as "percent of velocity pressure". Some texts convert this
to "equivalent feet of duct". Either kind of data can be used in the
calculation of friction losses in ductwork.
There is also friction in the boundary layer. This "wall friction"
depends on the Reynolds number and on the roughness parameter of the
wall construction. The roughness parameter is the variation in the wall
divided by the duct equivalent diameter, D = (4 x - : - - — ). In the
J perimeter
large sizes of ducts used in precipitator work, the roughness would
generally be considered "smooth", unless there were structural
protuberances, rough welds, extreme, corrosion, or severe dust deposits.
Numerical values of friction factors can be found in textbooks and
handbooks such as the ASHRAE Handbook (see Figure 5. 1).
The method of calculating frictional losses in gas handling systems is
straightforward and well developed. The most common form of expressing
frictional losses in straight runs of duct is that described by the Panning
equation:
A fLpv2
Ap=
where
Ap = static pressure loss:,
f = friction of factor,
*Refer to the bibliography for this chapter.
SOUTHERN RESEARCH- INSTITUTE
-------
-129-
i.o
r TIT rii
Wholly rough zone :
£ roughness = .0333
d
0.1
1000
Figure 5.1 Air Friction Vs. Reynolds Number and Roughness.
-------
-130-
L = length of duct,
p = density of gas,
v = velocity,
g = gravity, and
M = hydraulic radius.
Let
v = Av = volume rate of gas flow, and
&
M = -TJ- = area/perimeter;
Combining the constants:
Ap • k -?•
p
or pressure loss for odd-shaped ducts varies with —- .
A3
P A3
If -£3 or — is held constant, an approximate equivalent diameter
JTl Jt
can be calculated for any odd-shaped duct. It will be approximate
because the velocity used in the derivation is an average velocity
calculated from volume flow rate and cross-section area, twit the real
friction will be the sum of the friction due to the actual velocity
distribution, which is never known exactly. Therefore, any form of
equivalent diameter calculation is approximate. For example, l«t us
compare the above with another form of equivalent diameter:
^ 2ab
SOUTHERN RESEARCH INSTITUTE
-------
-131-
where a and b are dimensions of a rectangular duct.
bb)'
2 (a + b)
a3b3
64 2 (a + b)
Let b = Ka in both formulas
F+T
^5 16
D2 = -F
tR V*
n - s / lb K
Uz ' 7T2 K + 1 '
2Ka2 2K
and
Di = a (K + 1) = a K + 1
when
K = 2, Dx = 1. 33 a
D2 = 1. 34 a.
So the two formulae are almost the same if the aspect ratio, K,
of the duct cross section is not too high:
when
K = 10, Dx = 1.82 a
D2 = 2. 12 a.
-------
-132-
The friction loss with such an extreme aspect ratio should be checked by a
model if possible.
Friction, however, is only part of the problem. If sharp bends,
sudden transformations and rough walls in the system generate severe
turbulence and nonuniform velocity profiles, there will be trouble in
any equipment sensitive to air flow. Because fans, dust collectors,
and similar equipment are designed on the basis of some average
velocity, any extreme variation from that average may degrade the
operation. For example, the standard fan test code calls for a straight
inlet duct 10 diameters long with a straightener 6 diameters from the fan.
If in the field, a sharp bend is located at the fan inlet so that most of the
air is thrown to one side by inertia, then part of the fan is operating unloaded
and the volume handled will be less than design by 10 to 15%.
The best operating condition for an electrostatic precipitator will
occur when the velocity distribution is uniform. Because of the logarithmic
nature of the efficiency formula, an increased velocity in one plate section
will decrease the efficiency of that section more than the decreased velocity
in a parallel section would increase the efficiency of that section. As a
consequence, for a precipitator with nonuniform gas velocity, the total
plate area required to achieve a given efficiency will be greater than that
for a precipitator with uniform gas velocity.
Figure 5. 2 illustrates the adverse effect of uneven gas flow on
overall precipitator efficiency. Efficiencies are plotted for each unit of
a 3-unit precipitator with 1/2, 1, and 1-1/2 meters per second gas flow
in the three units respectively. Although the collection efficiency for the
average velocity is 97.3%, the average collection efficiency for the three
units is only 94. 5%. In practice, for dry collection the actual loss in
efficiency would ve'ry likely be far worse because of reentrainment of
collected particles in the high velocity zones.
In order to describe nonuniformity of gas flow, some easily
understood means must be employed, either qualitative, or quantitative.
The readings of each row or column of a velocity traverse can be
plotted to arrive at a series of velocity profiles such as shown in
Figure 5. 3. Alternatively, an isopleth, or contour plot, of the lines
of constant velocity can be constructed as shown in Figure 5.4. These
methods of plotting are useful in visualizing the velocity distribution
SOUTHERN RESEARCH INStltlJIF
-------
-133-
100
95
o
o
c
0)
85
80
99.
91.0
Weighted
Average
Q4. 5
0
1 r^i
Velocity, meters/sec
Figure 5.2. Collection Efficiency as a Function
of Flow Velocity.
-------
c
X
PI
z
JO
8
5
|
•1-4
o
o
i-H
CO
O
CO
i£t
i
8 9 10 11
Traverse Points
13
14
15
16 17 18 19 20
Figure 5.3. Gas Velocity Profiles.
-------
CO
CJ1
Figure 5.4. Gas Velocity Contour Plot.
-------
-136-
and in determining the location and type of corrective measures to be
taken. Note that the velocity at the walls in both these plots is zero.
In order to reduce the nonuniform ity of the data on gas velocity
to a single quantitative measure, statistical calculations can be made,
or the ratio of the maximum velocity to the average velocity can be
determined. Because the average velocity enters the electrostatic
precipitation formula, the ratio of maximum velocity-to-average-velocity
will give some measure of the degradation of efficiency due to nonuniform
air flow. Alternatively, the ratio of the root-mean-square of the readings
to the arithmetic average can be calculated. Both ratios would be 1. 00 for
perfectly uniform flow, but the maximum velocity-to-average-velocity ratio
would be the larger of the two when the velocities are nonuniform.
The frequency of occurrence of velocities falling into a series of
intervals can also be plotted, and thus a histogram showing the statistical
velocity distribution can be constructed as shown in Figure 5. 5. This
curve and the data for the profiles and contour plots were taken from a
particularly poor velocity distribution cited by White in Industrial
Electrostatic Precipitation.4 The two peaks indicate extremely bad
distribution. A good distribution should approximate a Gaussian or
normal distribution in the core of the gas flow, with a tendency to skew
somewhat toward zero because of the wall friction. The narrower the
spread of the distribution, the better the uniformity. Statistically, the
spread is measured by the standard deviation, the root-mean-square of the
sum of the deviations from the average. The standard deviation of the
example shown is about 50% of the average velocity. No references were
found giving examples of the standard deviation to be expected for good
distribution, but the Industrial Gas Cleaning Institute recommends 85% of
local velocities within ±25% of the mean, and no single reading more than
±40% from the mean as an acceptable criteria. This set of conditions
corresponds to a standard deviation of about 23% of the average velocity.
Note that it is impossible to achieve a completely uniform velocity
distribution in a practical duct system. There will be zero velocity
at the wall of the ducts and there will be random variations due to
turbulence in the core of the flow. A practical goal, then, must be to
achieve a nearly uniform gas flow using the best duct design procedures
in the present state-of-the-art plus field corrective measures as required.
Present practice is to use models constructed of plastic or hardboard
from 1/4 to 1/6 scale of the full-size plant to determine a practical
design. Field corrections are used as a last resort.
SOUTHERN RESEARCH INSTITUTE
-------
Maximum * 14. 0 ft/ sec .. __
Average* 7. 8 ft/ sec }Ratio « 1.80
Root- mean-square » 8. 4 ft/ sec }Ratl° * 1- 08
25
^ 20
!
I
•S 15
rt
W
_c
too
e
1 10
0)
K
5
M*
-
) 2
_ ^ '
f .
/
x
y
/
/
/
i
'
1
1
1
i
1
1
1
1
1
1
/
1
1
8
. Hypothetical curve with
. good distribution
\
\
\
\
\
I
1
i
\
\
\
i
\
\
\
\
\
1
\
^.
Sample histogram of
bad distribution
-
l — I
b 12 1*4 '
ft/sec
Figure 5. 5. Histogram Showing Statistical Velocity Distribution.
CO
-3
-------
-138-
To achieve uniform flow in a duct according to ASME test procedures,
one should have at least ten diameters of duct before and after any disturbance
such as the elbow, expansion or contraction. In practice, such conditions
cannot always be economically realized because of space limitations and the
high cost of ducts in the sizes involved. However, given a reasonable space,
it is possible to approach an acceptable quality of flow at the precipitator
inlet by the use of straighteners, splitters, vanes, and diffusion plates.
5.2 STRAIGHTENERS
Partitions in a straight section of duct for the purpose of eliminating
swirl are called straighteners. They may be "egg crate" dividers or
nested tubes as shown below:
A straightener will reduce the angle of a helical flow path to some
angle less than that defined by arc tan =
length
spacing
The nonuniformity
in the velocity in the axial direction will not be reduced. The scale of
turbulence, or eddy diameter, will be temporarily reduced to the same
size as the spacing of the straighteners, but because the Reynolds number
is usually well above critical, the eddies will not die out, but will grow
until they again reach the order of magnitude of the full duct. It is theo-
retically possible to make the spacing of straighteners small enough to
obtain a Reynolds number less than critical and to obtain laminar flow
through the straighteners, and to obtain nearly absolute uniformity.
Unfortunately, such spacing would only be about the size of soda straws
and the straightener would be expected to plug up with dust almost immedi-
ately.
A recommended straightener, according to AMCA Bulletin 210,
is an egg crate with a spacing of 7-1/2 to 15% of the diameter of a round
duct or the average side of a rectangular d.uct, and with a length equal
SOUTHERN RESEARCH INSTITUTE
-------
-139-
to three times the spacing. This reduces the swirl angle to arc tan
1/3 = 18-1/2°. An alternate straightener is a simple criss-cross at least
one and a half diameters long. The loss in these straighteners is equal to
the loss in four plain duct diameters. If it is necessary to reduce the
swirl angle to smaller values, the ratio of length to spacing must be
increased, and the resulting friction will be higher.
5.3 SPLITTERS
A duct section that changes size or direction may be divided into
smaller ducts over the full length of the change by partitions called
splitters. Splitters may be used in elbows where direction is changed,
or in transformations where velocity is changed. Splitters add wall
friction, but can reduce total friction by optimizing velocity pressure
losses.
Losses in elbows depend in part on how sharp the bend is. A
square, or mitered, elbow will have a loss of 1. 25 times the velocity
pressure, but an elbow of optimum configuration could have a loss of
only 0.11 to 0. 14 velocity pressure. As shown in Figure 5. 6, the optimum
configuration is an elbow with a ratio of inside radius to outside radius
of about 0. 66 for a square duct, or 0. 7 for a round duct.
To design a splitter elbow, therefore, it is only necessary to divide
the given elbow into segments all having a radius ratio of about 2/3.
5. 4 TRANSFORMATION SPLITTERS
Splitters may also be used in a diverging duct transformation to
divide the flow into nearly equal parts and then distribute the flow to the
larger* section:
-------
-140-
100
n
n
£
"8
S
o
09
09
O
10
Ratio inside radius
outside radius
Figure 5. 6. Elbow Loss as a Function of Radius Ratio.
SOUTHERN RESEARCH INSTITUTE
-------
-141-
The gas flow will not be uniform within each segment of the trans-
formation section, but the volume through every segment can be made
equal to that in every other segment if the splitters are manufactured to
be field adjustable.
A sharp angle in the transformation causes the gas to separate from
the walls of the duct, introducing turbulence and nonuniform flow. The
maximum angle of divergence for no separation is about 7° included angle.
Therefore, splitters in a transformation should be selected to have 7° to
10° included angle between successive splitters. For example, a trans-
formation with 60° included angle could be split into 8 channels with
7-1/2° spread, or 6 channels with 10° spread:
Note that a transformation in one direction is the simpler.
Transforming in two directions would require pyramidal splitters.
5.5 VANES
Another kind of deflector for redirecting gas flow is the turning
vane. Turning vanes are flat, bent, or curved plates which are short
relative to the duct section in which they are installed, as opposed to
splitters which extend the full length. A plain flat plate vane used to
deflect the air stream is partly effective, but it tends to increase
turbulence as shown in the sketch:
-------
-142-
The low pressure area behind the plate also tends to pull the gas
flow back toward its original path.
Curved turning vanes in an elbow can be quite effective if they are
spaced to give about a 6:1 aspect ratio and a 2/3 radius ratio, and are
streamlined to give constant cross section through the turn.
The streamlined turning vanes shown in Figure 5.7 will preserve the
flow pattern and will have a loss of about 10% of the velocity pressure. A set
of single thickness turning vanes will also preserve the flow pattern, but
will have about 35% velocity pressure loss and may introduce some turbu-
lence because of the unequal cross sections between them. Single thickness
vanes should have a straight extension downstream with length about twice
the spacing. In practice, single thickness turning vanes are generally used
because of cost considerations.
For rectangular elbows, one parameter is the aspect ratio, or the
ratio of the depth of the elbow measured parallel to the axis of the bend
to the width of the elbow measured in the plane of the bend:
"Hard" Bend
w
"Easy" Bend
It is intuitively obvious from the sketches that a low aspect ratio
elbow is a "hard" bend with high pressure loss which has very nonuniform
flow caused by inertial forces. Any aspect ratio greater than unity will
make a fair elbow, but aspect ratios from 4 to 6 are recommended.
Turning vanes are also used in transformation elbows; that is,
elbows that change cross section between inlet and outlet. Although
the combination of elbow and transformation is relatively poor design
practice, severe space limitations may force it upon the designer. If
a transformation elbow must be used, then turning vanes are essential.
SOUTHERN RESEARCH INSTITUTE
-------
-143-
Figure 5. 7. Streamlined Turning Vane Elbow.
-------
-144-
and they must be closely spaced to about the same spacing as the
precipitator plates. They must also be followed by additional flow
rectification means such as diffusion plates.
5. 6 DIFFUSION PLATES
Diffusion plates, or screens, are simply perforated plates or wire
screens which improve the uniformity of air flow by a combination of
effects. First, they reduce the scale of turbulence from the order of
magnitude of the duct to the order of magnitude of the holes. Of course,
the kinetic energy that existed in the large scale eddies will reappear in
the small scale eddies, but the large differences in velocity will be
reduced. Second, there is a pressure drop across the screen and a
reduction in area. The pressure drop will partly reappear as a velocity
vector perpendicular to the plate. This vector added to the original
velocity vector will give a resultant velocity always more nearly
perpendicular to the plate: w'
Thus, it might be possible to design a diffuser plate to turn the gas
stream through a precise small angle. However, in practice, it is
usually simpler and less costly to use two or more diffusion screens in
series to achieve a fair degree of uniformity.
Perforated plate screens break the gas stream up into a multiplicity
of small jets with high turbulent intensity and small scale of turbulence.
These jets eventually coalesce downstream. The turbulent intensity
reaches a peak at 2 to 3 mesh lengths (center to center of holes) down-
stream and declines exponentially thereafter. The scale of turbulence
is of the order of the hole size at the screen and increases until it
reaches the size of the duct. There is a critical parameter of 50%
SOUTHERN RESEARCH INSTITUTE
-------
-145-
open area for diffusion screens. When the percentage of open area is less
than 50%, the jets seem to be too far apart to coalesce uniformly and
the screen introduces nonuniformity. When the percentage of open area is
between 50% and 85%, the jets appear to coalesce within 5 to 10 mesh
lengths (center to center of holes) with improvement in uniformity.6
Screens may be used in series to provide greater uniformity at
the cost of larger pressure drops. Dryden and Schubauer7 developed
the following relationship for the reduction in turbulent intensity:
r = (14-k)-2 ,
where
r = reduction factor
k = pressure drop coefficient = ^r-
PVO/Z
n = number of screens in series
Ap = pressure drop
p = density
vo = average velocity.
All of the preceding ductwork design techniques are available to
the designer. None of the criteria are rigid, so there is considerable
freedom in design. It is the designer's choice as to whether to use
splitters, turning vanes, or screens to control the air distribution.
On the inlet side of a precipitator, there may be a heavy dust loading
of particle sizes large enough to settle out. Horizontal splitters or
vanes form convenient shelves for the deposition of disastrous quantities
of dust. Therefore, horizontal splitters and vanes are generally used
only when the velocity is higher than the erosion velocity of deposited
dust. Dust will collect by impaction on the diffusion screens, so some
means of cleaning then is required, such as regular rapping or soot
blowing.
-------
-146-
The pressure drop through a screen at large Reynolds numbers
depends on the coefficient of contraction of the jets formed by the gas
flow through the openings:
Cc = coefficient of contraction
s = solidity ratio
(1 - s) 100 = percent open area.
The coefficient of contraction depends on whether the holes are
sharp edged, or round edged, whether the screen is made of non-
coplanar bars, on the thickness of the plate, and on the accuracy of
manufacture. The best procedure is to measure the pressure drop
for the actual screen to be used. Alternatively, use values of the
coefficient of contraction as reported in the literature:
For sharp edges, Cc is about 0.62.
For round edges, Cc equals unity.
For thick plates (more than 2-1/2 times hole diameter), Cc is unity.
Other values lie between 0. 6 and 1. 0.
The design criteria for diffusion screens are summarized below:
(1) Open percent area should be between 50% and 65%.
(2) Screens in series must not be closer than 5 to 10 mesh lengths
apart.
(3) Hole size must be large enough to be unlikely to become plugged
with dust, but small enough to reduce the scale of turbulence as much as
possible. Sizes from 1 to 2 inches in diameter are reported.
(4) Screens should be oriented to refract the gas stream toward
the desired direction.
5.7 MODELLING
Because of the difficulties in the design of ducts for large size
precipitation installations, the use of models for air flow studies is
SOUTHERN RESEARCH INSTITUTE
-------
-147-
almost universally practiced. The models usually consist of sheet acrylic
plastic or pressed hardwood constructed to be geometrically similar to the
full scale system. The models are used to obtain a visual pattern of the
gas flow by the injection of smoke streamers or by the placement of yarn
tufts in the air stream. Quantitative measurements are made with pitot
tube traverses or thermoanemometers. A visual evaluation of the particulate
drop-out characteristics of the system can be made by the introduction of
dust into the gas stream. An evaluation of the erosion of deposited dust can
be made by placing a layer of ground cork on the bottom of the duct before
starting the blower.
Information from model studies can be presented in the form of
photographs of smoke streamers, yarn tuft deflections, and cork erosion
patterns. Data on nonuniformity of gas flow can be presented in the form
of plots of velocity profiles or velocity contour diagrams.
Modelling has been practiced for many years in designing air flow
systems. Early work was done at scales of 1/4 or 3/8 size with the Reynolds
number held substantially constant. Because of the increasing size of modern
installations, the trend has been toward the use of models constructed to
1/16 scale. Whereas models can give a considerable amount of valuable
information on velocity distribution, dust fallout patterns, and static pressure
loss, they do not necessarily accurately reflect the actual velocity distribution
in a full size unit, and a small scale model may be less accurate than a larger
scale model. The results of models in part depend on such factors as the
fidelity of the model and the skill of the investigator.
In the theory of modelling fluid systems, certain dimensionless
parameters should be made equal in the full size system and in the model.
There are as many dimensionless numbers as there are fundamental
dimensions in the physical system. Therefore, if some of the physical
parameters are arbitrarily chosen or fixed by the nature of the fluid, it
becomes impossible to hold all the dimensionless numbers constant. The
dimensionless numbers in fluid flow depend on the ratio between the
important physical parameters as follows:
(1) Inertial force = pvL = Reynolds Number
Viscous force
(2) Inertial force = _vf_ = (Froude's Number)
Gravity force Lg
-------
-148-
(3) Inertial force . _v^_ . (^c^g Number)2
Pressure force vc2
(4) Acceleration force = _Ag__ (Euler,8 Number)2
Tnertial force 2 va
Note that within the electric field of the precipitator an "electrical"
Froude's Number can be defined:8
Inertial force _ PV
Body force 6?TNa/iwL
Symbols in the above are as follows, in consistent units;
p = density
v = velocity
L = linear characteristic dimension
V- = viscosity
g = gravity
VG = velocity of sound in the fluid
N = number of dust particles per unit volume
a - radius of dust particles
w = migration velocity
Ap = pressure gradient.
In equipment the size of industrial electrostatic precipitators, the
flow nearly always has a high Reynolds number and is in the turbulent
regime. Theoretically, for testing in a model, the Reynolds number in the
model should be exactly the same as in the full scale unit. Unfortunately,
this is usually impractical. For example, to make a -—%• scale model of
16
SOUTHERN RESEARCH INSTITUTE
-------
-149-
a system having air at 300°F in a duct with a characteristic dimension of
16 feet, and using room air for the model fluid, the Reynolds number for the
system must be equated to the Reynolds number for the model:
PQ (
460 + 70 v
460 + 300 r
fi if"
M0 (1.33)
R = po Vo Lo
e
Note that the density and viscosity at 70°F have been replaced by a
temperature correction factor times the values.
Simplify
v = 530 xv. _ v,
0 760 + 1.33 x 16 ' 30.515
Solve:
vt = 30. 5 v0 .
Therefore, to maintain the Reynolds number constant, the velocity
in the model should be 30. 5 times as high as in the system. If the
system were to be run at 10 ft/sec, then 305 ft/sec would be required in
the model, or about 27% of the speed of sound. Thus, the model would
have a Mach number of 0. 27 where the system would have Mach number
about a tenth as much and the model would not be dynamically similar.
Similarly, if the actual system velocity were 50 ft/sec, the model flow
would be supersonic. If the scale ratio were to be 1/4 instead of 1/16, the
situation would be better by a factor of 4, even if still theoretically incor-
rect.
Because the main problem in ductwork design for electrostatic
precipitators is to achieve uniform gas distribution, unrealistically
high velocities in the model are not recommended. Also, the actual
velocity distribution from the upstream equipment cannot be fully known
or duplicated because of the effects of temperature and density variations
and of the effects of the presence of dust. Dust changes the density and
viscosity of the gas slightly; dust tends to deposit on the walls, splitters,
and vanes by settling or impaction and changes the roughness parameter
or even the size and shape of the ductwork. The only system that perfectly
models a given system is the system itself. This is the one reason why
-------
-150-
field engineering may be required to rectify the gas flow distribution if the
system was modelled.
This is not to say that models should not, or need not, be used. In
fact, models are essential for difficult duct configurations. A more im-
portant reason to conduct model studies is to investigate the gas flow pattern
expected in the commercial installation. Poor gas flow conditions may
cause severe dust reentrainment that can result in the precipitator perw
formance falling far short of design expectations.
The calculation of the pressure losses of a complicated system with
nonuniform gas flow may be impossible. A model will give a conservative
estimate of the pressure loss provided the roughness parameter and the
Reynolds number are properly duplicated. Note on Figure 5.1, which plots
friction against Reynolds number for various roughness numbers, that in
the "wholly rough zone", the friction is essentially constant. Therefore, the
friction will be correct if the Reynolds number is high enough, even though
it is not correct for theoretical similarity.
In those cases where severe space limitations force the designer
to violate all the criteria for good air flow, it would be highly desirable
to use a model to make the air flow distribution as good as possible under
the circumstances. Note, however, that the scale of turbulence in any
duct is of the order of magnitude of the scale of the duct, so that the size
of eddies in the model will be much smaller than the size of eddies in the
full size system. It can be expected, and it has been shown,8 that the
gas flow distribution in the full size system will be worse than in the model,
and the smaller the scale of the model, the greater the deviation.
In the cited reference, an air flow distribution with the ratio of
maximum velocity to average velocity of about 1. 20 was obtained in the
model, but the full scale unit had the maximum velocity about 1.40 times
the average. Experience built up over a long period of time is required
before one would have any real confidence in scaling up from models having
a very small scale factor such as * to 1 and smaller. An example of
successful modelling with this scale factor is reported by Burton et. al. 9
In general, .the recommended plan would be to design good duct work
with field adjustable vanes and splitters whenever possible. When poor
duct work is required because of space limitations, a model study is abso-
lutely essential, with a safety factor included as determined by experience.
Careful field adjustments by competent people will then provide a reasonable
probability of acceptable operation.
SOUTHERN RESEARCH INSTITUTE
-------
-151-
CHAPTER 5
BIBLIOGRAPHY
1. ASHRAE Handbook of Fundamentals, American Society of Heating,
Refrigerating, and Air Conditioning Engineers, Inc. (1967).
2. Fan Engineering, Sixth Edition (1961) Buffalo Forge Company.
3. Remmers, Karl and Bingel, Rolf, "The Importance of Gas
Distribution in Electrostatic Precipitators, " Staub 19, 422
(Dec. 1959).
4. White, H. J., Industrial Electrostatic Precipitation, Addison-Wesley
Publ. Co., Reading, Massachusetts (1963).
5. A MCA Bulletin 210, "Standard Test Code for Air Moving Devices, "
Air Moving and Conditioning Association, Detroit, Michigan (1960).
6. Baines, W. D. and Peterson, E. G., "An Investigation of Flow
Through Screens, " ASME Trans. (July 1961).
7. Dryden and Schubauer, "The Use of Damping Screens for the
Reduction of Turbulence," J. of Aero. Science 14, No. 4 (1947).
8. Qpfell, J. B. and Sproull, W. T., "Limitations of Model Studies
in Predicting Gas Velocity Distribution in Cottrell Precipitators, "
I and EC Proc. Des. and Dev. 4, No. 2 (April 1965).
9. Burton, C. L., et al., "Application of Model Studies to Industrial
Gas Flow Systems, "Am. Soc. Mech. Engrs. Ann. Mtg.,
Atlantic City, New Jersey, Nov.-Dec. 1959, Paper 59-A-280,
9P (1959).
-------
-152-
CHAPTER 6
RAPPING AND REENTRAINMENT
6.1 RAPPING
After collection of the dust on the collection electrode, it must be
removed from the system. This is accomplished by washing or flushing the
plates in the case of a wet collection electrode, or by impacting or rapping
in the case of a dry collection system.
The successful removal of dust from a precipitator by rapping depends
upon the formation of a coherent dust so that on removal, it falls in sheets
and large agglomerates into the hoppers. The adhesive and cohesive forces
acting on the dust layer are electrical, mechanical, and molecular in nature,
and influence the difficulty of dislodging the dust from the electrode, as well
as reentrainment of the dust during the rapping process.
Within the collected dust layer, the forces acting on the particles are
the cohesive forces due to adsorbed gases on-the surface, mechanical forces
due to interlocking of the particles and interparticle friction, Van der Waal's
forces due to proximity of adjacent particles, and electrical forces.
One component of the electrical force is generated by the ion current
flow to the dust layer. The magnitude of the net electrical force is given by
Lowe and Lucas1 as:
Pel.« d2 f kiEjp - Ef]
32
where
P j = electrical adhesive force,
d = particle diameter,
E = electrical field "strength,
j = current density,
p = specific electrical resistance,
kj = constant,
kjEd jp = proportion of Coulomb force due to ion current, and
d2E2 = share of Coulomb repelling force caused by influence
32 effects.
iRefer to the bibliography for this chapter.
SOUTHERN RESEARCH INSTITUTE
-------
-153-
If the dust resistivity is high, as in the case of fly ash, cement dust,
etc., the net force will tend to cause the dust layer to adhere more tena-
ciously to the collection electrode, thereby increasing the rapping intensity
required to dislodge it. Highly conductive dust, such as carbon black, tends
to reduce the net electrical force.
A second component of the electrical retentive force is electrostatic
in nature. If the particles are collected under the influence of an electric
field as is the case here, the dipole moments of the individual particles
will tend to align in such a manner that adjacent particles will be attracted
by a coulombic force.
Fundamental studies of the adhesive forces on dust particles have been
made byDavies, Penney,3 and others. In terms of practical precipitator
operation, the forces will vary with the type, size, and condition of the dust,
as well as its electrical properties.
Temperature of the gas and gas composition also influence the adhesive
forces. When operating at temperatures below the dew point, the dust layer
can become moist and extremely difficult to dislodge.
In practical precipitators, rapping is usually required on both the
collection and discharge electrodes, although the problems of reentrainment
are of no consequence in the case of the discharge electrode, due to the small
quantities of dust involved. The considerations for dust removal are the
same as those for the collecting electrode, since deposits can form on the
discharge wire. However, the current density through the dust layer is
larger than that through the dust layer on the collecting electrode by the
ratio of the collecting surface area to the area of the discharge wire. Con-
sequently, the electrical forces would tend to hold the dust more tenaciously
to the discharge electrode than to the collection plate. These deposits can
reach the point of serious impairment to precipitator efficiency if the dis-
charge electrode rapping is inadequate.
Rapping systems for both collection and discharge electrodes can be
of the impact or vibrator type. Impact rappers are normally actuated pneu-
matically, electrically, or by gravity, and can vary from a single blow to
a rapid succession of impacts. The rapping cycle can be varied over a
wide range depending upon the requirements of the precipitator. Vibrator
types are generally motor-driven vibrators that shake the electrode sup-
port structure, and are often used on the discharge electrode in conjunction
-------
-154-
with an impact rapper on the collection plate.
The principal consideration in any rapping system is the magnitude of
the forces required to dislodge the dust without mechanical damage to the
electrodes or support structure, or causing excessive reentrainment. It is
a general rule to rap just hard enough to dislodge the dust and keep the col-
lected dust layer at an acceptable thickness. It is impractical to maintain
the collecting electrode completely clean, since successful rapping is accom-
plished by removal of the dust in clumps or sheets. Dust thicknesses of \ to
1 inch prior to dislodging are usual. Figure 6.1 shows the results of several
tests reported by the British Central Electricity Generating Board relating
the acceleration required to dislodge dusts remaining on the collection elec-
trode to a given thickness.
Some investigations have been made of the magnitude of the rapping
blow required to achieve the desired rapping efficiency. Brandt and Heinrich
attribute the dust removal from the electrodes to the peak value of accelera-
tion normal to the plane of the surface. Sproull5 reports that the determining
factor in dust removal is the acceleration imparted to the electrodes and
that it varies with type of dust, whether the electrode is rapped in the plane
of the electrode (shear) or perpendicular to the plane (normal). Figures
6. 2 and 6. 3 show the accelerations required for shear and normal rapping
for several dusts and various temperatures. As shown on the curves, accel-
eration of the order of lOOg's may be required for 90% removal of the col-
lecting electrode dust for rapping normal to the plate, and 200g's for rapping
in the plane of the electrode for diff icult-to-remove dust. For fly ash and
easier-to-remove dusts, the accelerations required are considerably less
of the order of 10-30 g1 s.
The efficiency of rapping, measured in terms of residual dust, is in
general improved if the power is removed during the rap as indicated in
Figure 6. 2. This is due to the removal of the corona current and the re-
duction of the force holding the dust layer to the plate. Since this force
is dependent on resistivity, its magnitude will vary for different types
of dust and the requirement for power-off rapping will depend on the dust
characteristics. Power-off rapping is usually resorted to in the event
normal rapping practice is inadequate. It is to be recognized that the por-
tion of the precipitator that is in operation during rapping will not be col-
lecting material and stack puffs may result.
SOUTHERN RESEARCH INSTITUTE
-------
-155-
4
OU
f-,
9
"53
a
•iH
^ 1
•!-< «J
§
M
of
CO
S
o
2 2
-4-J
M
Q
1
i
o
0 o Single Measurement
O Average of Five
Measurements
\
0 \ 0 95% certain that the
\+^J average dust thick-
L ness for a given
\ acceleration lies
\ O below this
0 \ line.
o o \
o x
0 \
\
o O O ^ ^ n
X\°
8 ^ -a.
°° °o ° ""
o
o O
o 0
o
o o ^
w wo
L L L
10 20 30 4
Acceleration in Vertical Direction,
Units of g
Figure 6.1. Acceleration to Remove Dust Down to
a Given Remaining Thickness for Fly
Ash in British Installations.
-------
-156-
100
0 20 40 60 80 100 120 140
Max Shear Acceleration of Collecting Electrode Plate Produced by Shear Rap ("g")
Figure 6. 2.
Shear (parallel) rapping efficiency for various precipitated dust layers
having about 0. 2 gram of dust per square inch, as a function of maxi-
mum acceleration in multiples of "g. " Curve 1, fly ash, 70 to 300°F,
power off. Curve 2, fly ash, 300°F, power on. Curve 3, fly ash,
200°F, power on. Curve 4, cement kiln feed, 70°F, power off.
Curve 5, same as 4, except at 300°F. Curve 6, same as 4 and 5,
except at 200°F. Curve 7, cement dust, 300°F, power off. Curve 8,
same as 7, except at 200°F. Curve 9, cement kiln feed, 200 or 300°F,
power on. Curve 10, cement dust, 300°F, power on. Curve 11, same
as 10, except at 200°F. Curve 12, fly ash, 70°F, power on. Curve 13,
cement kiln feed, 70°F, power on.
SOUTHERN RESEARCH INSTITUTE
-------
-157-
100
0 20 40 60 80 100 120 140
Max Normal Acceleration of Collecting Electrode Plate Produced by Normal Rap ("g")
Figure 6.3. Normal (perpendicular) rapping efficiency for various precipitated
dust layers having about 0. 2 gram of dust per square inch, as a
function of maximum acceleration in multiples of "g." Curve 1,
fly ash, 200 or 300°F, power off. Curve 2, fly ash, 70°F, power
off; also 200 or 300°F, power on. Curve 3, fly ash, 70°F, power
on. Curve 4, cement kiln feed, 300°F, power off. Curve 5,
cement dust, 300°F, power off. Curve 6, same as 5, except
power on. Curve 7, cement kiln feed, 300°F, power on. Curve
8, cement dust, 200°F, power off. Curve 9, same as 8, except
power on. Curve 10, cement kiln feed, 200°F, power off. Curve
11, same as 10, except at 70°F. Curve 12, cement kiln feed,
200°F, power on. Curve 13, cement kiln feed, 70°F, power on.
-------
-158-
One problem encountered in dry dust removal is reentrainment of the
dust during rapping. Because of the height of the plates is often 30 ft or
more, it is apparent that some dust will be reentrained as it falls .through
the passage to the hoppers. This gives rise to a rapping "puff, " which is
often visible in the plume during the rap. To minimize this effect, rapping
is done in stages with only a portion of the plates rapped at one time. Var-
ious collecting electrode shapes are also used in an effort to shield the fal-
ling dust from the gas stream, thereby minimizing reentrainment.
There are many variations possible in rapping practice depending on
the particular installation and process. If parallel units are used, the gas
flow can be directed by dampers during rapping so that there is %no gas flow
through the unit. Where the dust load is intermittent, as in the case of a
basic oxygen furnace, it is possible to eliminate rapping of the plates. Fol-
lowing the oxygen blow, when the dust load is reduced, the draft fan can be
shut off and the plates rapped.
The accelerations required to maintain a given rapping efficiency are
those at the plate itself and the design of the rapping system should include
consideration of the rigidity of the supporting members and the electrode,
and their ability to transmit the impact to all areas of the electrode. Recent
studies of operating precipitators in Europe have shown that accelerations
measured on the collection electrode are as low as 5 g's, whereas minor
changes in the method of applying the blow could yield accelerations of 30-
50 g's. The latter range appears adequate for most fly ash precipitators.
Although it has been recognized that rapping is more than just a
minor adjunct to precipitator operation, there is a conspicuous lack of
experimental .data that defines the conditions required to effectively re-
move the types of dusts encountered. There are reports of inability to
dislodge high resistivity dust by rapping, and there is no experimental ver-
ification of the rapping requirements as a function of dust resistivity. Also,
there should be some measure of the reentrainment tendency for various
dusts as related to particle size, specific gravity, particle configuration,
and quality of gas flow.
6.2 REENTRAINMENT
\
Particles on a dust surface at the boundary of a gas stream are trans-
ported by suspension, by saltation (repeated rebounding), or by "creep. "
SOUTHERN RESEARCH INSTITUTE
-------
-159-
A compact layer of settled dust will not move until the least secure surface
particle is dislodged by the drag force of the gas velocity, by impaction by
another particle, or by the explosive impact of electric sparkover, which
also results in momentary disruption of power. When this particle again
strikes the surface of the dust, it dislodges more particles and initiates a
cascade of saltating particles. If an impacted particle is not ejected into the
gas stream, it may be displaced a small distance downstream which is the
motion described by Bagnold6 for dust held by gravitational forces. If the
force due to gravity is replaced by the electrical force, the equation can be
written as:
U* = kt Sp - S a a
S
where
U* = threshold drag velocity,
kx = experimental coefficient,
Sp'" = density of particle,
S = density of gas,
a - acceleration on deposit (electrical force divided by mass), and
a = particle size.
Bagnold found the coefficient Mk1" to be constant for particles larger
than 200 microns. Chepil7 obtained the curve shown in Figure 6. 4 for thres-
hold drag velocity versus particle size. Note that the minimum drag velo-
city occurs around 60-100 microns and that the threshold is very high for
very small particles such as might be found in electrostatic precipitator in-
stallations.
The initiation of erosion is strongly related to the adhesion of the dust
layer. Adhesion depends in part on the history of the accumulation of the
dust. If the dust layer accumulates essentially one particle at a time with
sufficient freedom of motion and sufficient time for each particle to orient
itself to the randomly situated positive and negatively charged areas on the
dust layer and on itself, then the dust layer will have an adhesive strength
an order of magnitude higher than the same dust piled in bulk.8 The require-
ment for the formation of a layer thus held together by Coulomb forces is
met by electrostatic precipitation and by sedimentation in low velocity zones
in the ductwork. On the collection plate, however, additional forces due to
-------
-i
H
PI
OS
o
10
100 200
Particle Size,
300
400
500
600 700 800 900 1000
(Square Root Scale)
Figure 6.4. Threshold Shear Velocity Vs. Particle Size.
-------
-161-
the ion current act on the dust layer. Loosely piled dust that results from
rapping of plates, electrodes or diffusion screens does not meet the require-
ments for an adhesive layer, so a much lower velocity will erode such piles
in the hoppers.
With particles larger than 100 microns, initiation of erosion can occur
when the drag on one particle is enough to start a saltation cascade. A ran-
dom gust of turbulence can initiate a cascade, so it is apparent that a uniform
gas velocity with the least amount of turbulent intensity is highly desirable.
With particles smaller than 100 microns, erosion will be initiated when the
velocity exerts enough drag force on the edge of a discontinuity. Such dis-
continuities will exist as the results of slabs of dust falling off the plates
through the action of rapping and gravity. As the slabs break up, agglom-
erates larger than 100 microns may occur, which can permit saltation to
begin.
Erosion may also be initiated by the impact of large particles on the
collected dust layer. The large particles have greater electrical migra-
tion velocities and therefore impact the deposit with greater momentum.
The impact energy may be sufficient to initiate erosion.
Erosion can occur at any place that dust accumulates. Dust accumula-
tion in the inlet ductwork does not necessarily result in carryover of dust
from the precipitator, but it does result in alteration of the gas flow pattern,
usually in the direction of greater nonuniformity of velocity and lower col-
lection efficiency. Therefore, it is desirable to have a velocity in the duct-
work high enough to prevent settling or high enough to erode away dust
deposits. Dust accumulation in the hoppers is a required function of pre-
cipitators, so gas velocity in the hopper area must be below some critical
velocity for loose unconsolidated dust. It is also important to prevent in-
leaks through the collection hopper discharge gates because the erosion
velocity in this case is equal to the terminal settling velocity of the parti-
cles. Air in-leak through conveyors should be less than the critical velo-
city for loosely piled dust.
Erosion in the precipitator itself is important because it results in
a direct loss of collected dust, and a direct reduction in collection efficien-
cy. Sharp edges on the dust layer always exist because of rapping and grav-
ity, and erosion can always occur if the local velocity is high enough. Con-
trol of uniformity of velocity and of the intensity of turbulence can hardly be
overemphasized.
-------
-162-
When the dust slab falls off the collection plate, some of the slab
breaks up, and the dust becomes reentrained. It is a usual practice to limit
the local gas velocity in the area of the falling dust by the use of pockets in
the plate or baffles on the plate, perpendicular to the gas flow.
In addition to erosion due to wind velocity, there is some reentrain-
ment of particles in a precipitator which can be called "electrostatic re-
entrainment". Particles which have a low resistivity can discharge their
acquired electrical charge when they reach the collector plate. With no
electrical force holding the particle to the plate, the particle may simply
fall off, or be eroded by the wind or saltating particles. The particles may
also acquire an induced positive charge and migrate toward the negative
discharge wire. Reeritrainment can also occur during sparking when a
precipitator section is momentarily out of service.
The opposite problem to reentrainment can exist in some cases
where the adhesion of the dust is extremely high. The dust cakes on the
collector plates or the corona wires and even severe rapping may not
dislodge all the dust.
Table 6.1 summarizes the different erosion situations that can exist
in a system using electrostatic precipitation.
Note that there are many critical velocities for erosion:
(1) Threshold velocity for particles greater than 100 microns.
(2) Threshold velocity for sharp edged discontinuity on a compact
dust layer.
(3) Threshold velocity for undisturbed compact dust layer.
(4) Threshold velocity for mechanically compacted dust.
(5) Threshold velocity for loose dust.
Furthermore, each of the above velocities will be different for every
possible particle size distribution, as well as for different adhesion strengths
induced by chemical, mechanical, and thermal conditions. Thus, no simple
analytical method has been developed to predict erosion and reentrainment.
The cascade effect of saltating particles cannot increase without limit.
Eventually, the number of particles ejected by impaction will equal the num-
ber of particles that become reembedded in the dust layer and do not bounce
SOUTHERN RESEARCH INSTITUTE
-------
-163-
Table 6.1
Erosion Situations
Location Adhesion
Precipitated dust on corona wires High
Precipitated dust on collector plates High
Settled dust in ductwork High
Dislodged dust in hoppers Low
Dislodged dust in ductwork Low
Dust in conveyors and air locks Low
Desired Velocity
More than critical
for compact dust
Less than critical
for compact dust
Greater than critical
for compact dust
Less than critical
for loose dust
Greater than critical
for loose dust
Less than critical
for loose dust
Falling dislodged dust
Zero
Minimum practical
-------
-164-
out again. The rate of erosion will therefore eventually reach a saturation
value. The saturation point will be reached in a precipitator at some point
downstream from the inlet. At the inlet, the erosion rate is necessarily
zero, because there is no material collected at zero length.
Eroded material from one area of the collector will normally be
recaptured downstream. The last section of the precipitator has no down-
stream collector, so that a constant erosion rate will result in a net
loss of dust and a reduction in efficiency of collection. Robinson9 has
derived a modification of the usual efficiency equation
TJ = 1 - exp (- — w)
vg
to include the effect of erosion
= 1- (l-j~) exp[-(l-a)AXW.-1l_
where
a = mass of dust eroded per unit mass precipitated
for normal dust,
j3 = mass of dust eroded per unit mass precipitated
for "problem" dust (i.e., low resistivity dust that
is repeatedly precipitated and reentrained with no
change in concentration in the length of the pre-
cipitator),
^ = concentration uniformity factor, the ratio of the
dust concentration near the wall to the average
dust concentration over the cross section between
precipitator plates,
A = collection area,
v- = volume gas flow rate, and
O
w = migration velocity.
When X = 1 for uniform distribution and a - 0, and j3 = 0, the
modified equation reduces to the usual form. If j8 = 0, X = 0; and
a - 0, the equation reduces to an exponential form which could be
expressed using an equivalent migration velocity.
SOUTHERN RESEARCH^ INSTITUTE
-------
-165-
CHAPTER 6
BIBLIOGRAPHY
1. Lowe, H. J. and Lucas, D. H., "The Physics of Electrostatic
Precipitation, " Brit. J. Appl. Phys. 24, No. 2, pp 40-47 (1953).
2. Davies,C. N., Aerosol Science,Academic Press, London-New York
(1966).
3. Penney, G. W. and Klingler, E. H., "Contact Potentials and the
Adhesion of Dust, " Communications and Electronics (July 1962).
4. Heinrich, D. O., "The Cleaning of Collection Electrodes in
Electrostatic Precipitators, " Staub 22, No. 9, pp 360-364 (1962).
5. Sproull, W. T., "Fundamentals of Electrode Rapping in Industrial
Electrical Precipitators," JAPCA 15, No. 2, pp 50-55 (Feb. 1965).
6. Bagnold, R. A., The Physics of Blown Sand and Desert Dunes,
Wm. Morrow and Co., New York City, pp 322-334 (1943).
7. Chepil, W. S., 'Dynamics of Wind Erosion: II - Initiation of Soil
Movement, " Soil Science 60. pp 397-441 (1945).
8. Niedra, J. M. and Penney, G. W., "Orientation and Adhesion of
Particles," IEEE Trans. IECI 12, No. 2, p 46 (1965).
9. Robinson, Myron, "A Modified Deutsch Efficiency Equation for
Electrostatic Precipitation, " Atmospheric Environment, Pergamon
Press I, pp 193-204 (1967).
-------
-166-
CHAPTER 7
RESISTIVITY AND CONDITIONING
Resistivity of the collected particulate layer has a dominant influence
on the operation of electrostatic precipitators. If the resistivity is greater
than about 2 x 10l° ohm-cm, the electric field in the dust layer resulting
from normal current densities will exceed the breakdown strength in local-
ized areas, and excessive sparking and back corona may occur. Conse-
quently, the high tension voltage and resulting corona current must be
reduced, adversely affecting precipitator performance. If the resistivity
is less than about 107 ohm-cm, the electrical force holding the dust deposit
will be low, and excessive reentrainment can occur for certain types of
dust.
There are two conduction mechanisms which determine the resistiv-
ity of the collected dust layer in an electrostatic precipitator. These
mechanisms are termed volume conduction, which is related to the bulk
composition of the material,, and surface conduction, which depends on an
adsorbed film on the particle surface and which is related to the composi-
tion of the gas and dust surface.
7. 1 VOLUME CONDUCTION
In volume conduction, electric charge is transferred through the bulk
material that comprises the collected dust layer. In most materials of
interest in electrostatic precipitator applications, volume conduction occurs
by means of electron carriers within the materials and is dependent upon
the thermal excitation of the electrons in the molecular structure of the
materials. It has been shown that for most industrial dusts and fumes,
the relationship of volume conductivity and volume resistivity to tempera-
ture may be approximated by
E
p =
SOUTHERN RESEARCH INSTITUTE
-------
-167-
where
p = resistivity,
C = constant,
k = Boltzmann's constant,
T = temperature, and
E = Activation energy.
cl
White1 has shown that resistivities of dusts as measured in the labora-
tory follow the theoretical relationship between volume conduction and acti-
vation energies above about 250°F. Figure 7. 1 shows the experimental re-
lationship between resistivity and the factor _L_ for a variety of types of
kT
materials. Figure 7. 2 shows the quantitative relationship between resisti-
vity and electron activation energies.
The influence of temperature on resistivity of collected dusts is shown
in Figure 7.3. In the absence of moisture, the resistivity would continue to
increase with decreasing temperatures due to the decreased electron excita-
tion. Conversely, increases in temperature increase the thermal electron
excitation resulting in decreased resistivity.
This decrease in resistivity at elevated temperatures is significant
in many electrostatic precipitator applications which operate in the tempera-
ture range in which volume conduction is predominant.
7. 2 SURFACE CONDUCTION
At gas temperatures below 300-400°F, surface conduction becomes
the increasingly important mode provided ample moisture and other con-
ditioning materials are present in the gas. Surface conduction depends on
the presence of a conductive film of adsorbed material on the surface of the
particulate which provides for transfer of the electric charge along the sur-
face. If ample moisture is present in the flue gas, it will be adsorbed on
the surface of the particles following the basic laws of adsorption. It has
been calculated that the existence of a film 5 to 10 molecules thick on the
*Refer to the bibliography for this chapter.
-------
-168-
lO1
1.3 1.4 1.5 1.6 1.7 l.BxlO13
Figure 7.1 Plots of Resistivity Vs. 1/kT.
SOUTHERN RESEARCH INSTITUTE
-------
-169-
Resistivity, ohm- cm
J.U ~
10"
10io
109
10s
io7
ioe
m5
\* Talc
A1203 \
CaO\
MgO\
Leached fly ash >?
Si02\
0.1 0.2 0.3 0.4 0.5 0.6 0.7
Energy, electron-volts
Figure 7. 2 Experimental Values of Electron
Activation Energy. (Reference 1)
-------
-170-
10
15
10
14
10
13
s 10*
u
E
to
I
00
11
09
0)
io
10"
10
10"
Reagent Grade
' CaO
Cement Kiln
Dust
i
\
100
200 300 400 500
Temperature, °F
600
700
"Figure 7.3 Resistivity of Bone-Dry Dusts at Elevated Temperatures.
SOUTHERN RESEARCH INSTITUTE
-------
-171-
surface is sufficient to explain the observed decrease in apparent resistivity
of collected layers of particulate with increased moisture content. The
effect of adsorbed moisture on resistivity is illustrated in Figure 7.4. If
moisture is present in the gas, the resistivity departs from the bone dry
resistivity curve at temperatures where significant adsorption occurs.
In processes where sufficient moisture is present in the gas, either
from the process itself or as a result of evaporative water cooling, the
resistivity of the dust is usually sufficiently low to be readily accommodated
by electrostatic precipitators.
Many processes that emit dust do not have sufficient moisture in the
flue gas for adequate conditioning in the absence of other conditioning agents.
If such substances are present, however, as they are in many combustion
processes, they may interact to alter the surface conduction properties and
reduce the resistivities beyond the values indicated by the moisture content
alone.
Because of the dominant role of moisture in the conditioning process,
it is termed the primary conditioning agent. Other materials, either
naturally occurring or added to the flue gas, are termed secondary condi-
tioning agents.
7. 3 MECHANISMS OF SURFACE CONDITIONING
The role of secondary conditioning agents has received considerable
study to determine how naturally occurring constituents of the gases, as
well as chemical additions to the gases, act to alter the dust resistivity.
The extent of the use of electrostatic precipitators in the collection of fly
ash from low sulfur coals has resulted in more work on the resistivity of
fly ash than on any other type of dust.
Research carried out by Chittum and others2 in the early 1940's on
the fundamentals of conditioning led to the advancement of a theory of con-
ditioning by alteration of the moisture adsorption properties of dust sur-
faces. Chittum proposed that an intermediate chemical adsorption film,
which was strongly bound to the particle and which in turn strongly adsorbed
water, would be an effective conditioner. For example, he postulated and
later showed qualitatively, that a weak basic particle such as ZnO would
-------
-172-
10"
Temperature, °F
Figure 7.4 Effect of Humidity on Particle Resistivity.
SOUTHERN
INSTITUTE
-------
-173-
exhibit improved adsorption capacity with prior additions of HC1 or H2SO4.
Similarly, weak acidic particles such as A12O3, when exposed to ammonia,
showed improved water adsorption characteristics.
In fly ash precipitators, the influence of sulfur content of the fuel has
been recognized as a highly significant factor in precipitator performance.
The sulfur appears in the flue gas principally as SO2. However, approxi-
mately 1 to 2% of the sulfur is oxidized to SO3, and it is sulfur in this form
that is active in the conditioning process. The quantity of SO3 present de-
pends upon the sulfur content of the fuel, the operating conditions in the
boiler, and the composition of the fly ash.
At the temperatures that exist downstream of the air heaters, the SO3
probably combines with moisture to form adsorbed H2SO4 on the fly ash. The
HgSC^ is highly hygroscopic and attracts free moisture to further enhance
the conduction.
The rate at which H2SO4 is adsorbed is dependent on the composition
of the fly ash surface. If the particulate surface is acidic or neutral, ad-
sorption occurs only by van der Waal's forces or physical adsorption, which
involves nonspecific, relatively weak attractive forces between the solid
particle surface and the gas molecule! On the other hand, if the particu-
late surface is basic, strong chemical attractive forces of the same nature
and magnitude as those involved in chemical bonds are present, resulting
in much higher adsorption rates.
This mechanism of the formation of the conductive film appears to be
substantiated by experience with specific fly ash which indicates that ash
high in basic constituents is conditioned much more readily by SO3 than is
ash high in acidic constituents. Conversely, acidic particles appear to be
more readily conditioned by a basic conditioning agent.
Darby and Heinrich4 reported apparent physical differences in fly ash
which has been conditioned by the addition of SO3. Electron microscope •
photographs of nonconditioned dust showed completely smooth surfaces,
while conditioned dust exhibited a surface film. They postulated that as
a result of this film or layer, a more conductive path was provided. Figure
7. 5 is a photograph provided by the Central Electricity Generating Board
illustrating a "conditioned" fly ash. It is proposed that SO3 tended to be
-------
-174-
-.« ~
% a
Figure 7. 5. Conditioned Fly Ash Particle - CERL, England (20, OOOX)
-------
-175-
adsorbed on the particle surface at temperatures up to 400°C, and upon
cooling, the adsorbed SO3 brought down water vapors to form liquid/semi-
plastic nodules as shown. These nodules, in turn, resulted in lower re-
sistivity and stronger mechanical bonds between adjacent particles. Diam-
eter of the particles shown is in the range of from 0. 3-3 microns. No
photograph of nonconditioned particles was available for comparison.
A sample of fly ash from a power plant in Alabama burning low sulfur
(approximately 0, 5%) coal was obtained for comparison by Southern Research
Institute. According to observations mentioned previously, the fly ash
should have a high resistivity and be difficult to precipitate.
Figure 7. 6 is an electron microscope photograph taken at Southern
Research Institute of one of the typical particles enlarged approximately
80, 000 X. Clearly visible on the surface are aggregates similar to the
"nodules" shown on the conditioned fly ash. This observation leaves some
question as to the significance of these surface characteristics with respect
to resistivity.
A review of much of the literature available on the fundamentals of
electrostatic precipitation reveals that very little recent experimental
work, and no recent theoretical work, has been reported in the field of
gas conditioning and resistivity of dusts. This is particularly surprising
in view of the relatively important role of resistivity in precipitation.
A fundamental investigation with modern instrumental methods now
available should provide a much needed quantitative model of conductivity
and conditioning mechanisms.
There is some evidence that the resistivity of collected dust may not
be a static property. Cohen3 reports that the resistivity of fly ash collected
in a cyclone separator varies by two orders of magnitude in a period of 15
to 20 minutes. It is postulated that the increase in resistivity is due to a
loss of the absorbed film due either to physical or chemical interaction with
the bulk of the dust particle. If similar changes occur in the collected dust
layer, a resistivity gradient could occur with the freshly collected dust
having a lower resistivity than that deposited earlier.
SOUTHERN RESEARCH INSTITUTE
-------
»»*
OS
I
0.5
Figure 7.6. Unconditioned Fly Ash Particle - Low Sulfur Coal (80, OOOX)
-------
-177-
7. 4 FIELD CONDITIONING STUDIES
In the precipitation of high resistivity dusts, the addition of condition-
ing agents to the gas stream is an option in improving precipitator perform-
ance. The choice of whether to meet the requirements by additional precipi-
tator capacity or by the use of conditioning agents is largely one of economics.
The types of conditioning agents used in dust conditioning are moisture,
sulfur trioxide, and ammonia. Other agents, such as sulfuric acid, ammo-
nium sulfate, and ammonium bisulfite, have been tried experimentally, but
no operating plants utilize these agents on a continuous basis.
7. 5" MOISTURE CONDITIONING
Moisture can be used as a conditioning agent in processes in which
gases exit at temperatures in the range of 1500-2500° F. Normal practice
is to cool the gases to temperatures that can be handled by the precipitator
by the injection and subsequent evaporation of water. The process of cool-
ing raises the humidity of the gases entering the precipitator and can result
in resistivities in a suitable range for precipitation.
The effect of gas cooling by water spray can be seen from the high
temperature psychrometric chart as shown in Figure 7.7. If gases from
the process enter the cooling chamber at, say, 1300° F with a moisture
content (wt. moisture) of 3% and are cooled by evaporation of water to
(wt. dry air)
300° F, the resulting moisture content will be approximately 24%. Under
these conditions, the resistivity of the collected dust should be reasonably
low. On the other hand, if the starting temperature is only 700° F, cooling
to 300°F would result in a moisture content of only about 11%.
On processes in which heat recovery is used, the gas temperature
is normally too low to evaporate the amount of water necessary to provide
adequate conditioning. An example is the flue gas from steam electric
generating plants which leaves the air preheater at temperatures of from
250-400° F.
Field tests on water conditioning were carried out at the Pyrmont
Station of the Electricity Commission of New South Wales using steam from
SOUTHERN RESEARCH INSTITUTE
-------
0.30
o
I
100 200 300 400 500 600 700 800 900 1000 1100 1200 1300
Temperature, °F
Figure 7.7. High-Temperature Psychrometric Chart.
-------
-179-
the soot blowers, water sprays in the combustion chamber, and water sprays
in the flue following the air heater. It was found that spray water injection
in the duct following the air heater did not result in complete evaporation of
the water, and carryover to the precipitator caused excessive dust buildup
on the electrodes.
As steam, injections of 12, 000 Ib/hr were found to be required on the
plant at Pyrmont, which was considered uneconomical as compared with
other conditioning agents.
Moisture conditioning in the form of steam injection is used on inter-
mittent processes, such as EOF furnaces, to provide moisture during the
interval between the start of oxygen lancing and the beginning of water
spray cooling. Water sprays in the cooling tower are normally thermo-
statically controlled to start at a preset gas temperature to ensure com-
plete evaporation of the water. In the absence of the steam conditioning,
high stack emissions known as lance puffs will occur, until such time as
the temperature reaches the point that resistivity is in the proper range.
7.6 CHEMICAL CONDITIONING
Extensive pilot and full trials with sulfur trioxide conditioning have
been carried out in this country, the United Kingdom, and Australia. An
operating plant at the Kincardine, Scotland station of the South of Scotland
Electricity Generating Board utilizes SO3 injection as a regular operating
practice. Tests on the plant show improvements from collection efficien-
cies from 60-65% to somewhat over 90% with additions of around 10 ppm SO3.
Results at Kincardine have led to the design of SO3 conditioning plants for
the Central Electricity Generating Board plant at Rugeley and another in
South Africa in which sulfur is to be burned on site and the SO2 catalytically
oxidized to SO3.
The SO3 injection facilities are generally rather simple in concept.
The sulfur trioxide used is a stabilized form to prevent polymerization,
and is supplied to the plant from a temperature controlled railroad tank
car or truck. It is generally stored in a heated insulated tank with the tem-
perature being 20-30°C (68-86°F).
SOUTHERN RESEARCH INSTITUTE
-------
-180-
From the storage tank, the liquid SO3 is piped to a metering flow con-
troller and than to an evaporator to convert the liquid SO3 to a gas. The SO3
gas can be mixed with dry air for transport through heated pipes to the dis-
tribution manifold. From the manifold it is distributed in the flue gas duct
through a series of pipes fitted with teflon nozzles.
The major problems with SO3 additions are the cost of process, the
requirements for temperature and moisture control, and the safety aspects
associated with handling of SO3. Temperature control is required in the
system because liquid SO3 solidifies at temperatures below around 20°C,
and boils at around 44°C. Both extremes must be avoided during storage
and transport. Moisture must also be excluded from storage, metering,
and piping to prevent the formation of sulfuric acid and subsequent corro-
sive attack on the system.
There is some reluctance on the part of plant personnel to the use of
SO3 conditioning due to the safety problems. In either liquid or gaseous
form, SO3 is an effective oxidizer and strongly attacks tissue. Leaks from
the system form dense acid fumes as SO3 reacts with atmospheric moisture,
and safety precautions must be considered. The problems, however, are
not unlike those associated with many chemical plants and difficulties can
be avoided by proper design of the system, especially selection of valves,
seals, etc., to prevent leakage.
Figure 7.8 is a schematic diagram of an SO3 conditioning plant used
in pilot scale studies in tests at the Creil station of Electricite de France.
Experience with SO3 conditioning of fly ash has been varied, ranging
from substantial improvements with additions of 10-20 ppm to requirements
of over 100 ppm without striking improvement. The difference in effective-
ness appears to be in the nature of the fly ash. Tests conducted on the
Tallawarra plant of the Electricity Commission of New South Wales showed
that SO3 was effective in reducing the resistivity, but quantities were about
600% greater than estimated from prior experience with low sulfur coals.
Subsequent tests showed the ash to be acid (pH 3-5. 5) and perhaps account
for the relative ineffectiveness of the SO3 additions. Baxter5 reports sim-
ilar experience with SO3 conditioning of acidic fly ash.
-------
-181-
uu
(1)
(10)
(12)
(U)
1. Heating Filaments
2. SO3 Evaporator
3. Flow-Controller
4. Air Reheater
5. Condensation Pots
6. Air Tanks
7. Air Dryer
8. Pressure Regulator
9. Air
10. Duct, No. 1
11. Duct, No. 2
12. Injection Manifold
Figure 7.8 Schematic Diagram of SO3 Conditioning Plant.
SOUTHERN RESEARCH INSTITUTE
-------
-182-
Ammonia has been used extensively as a conditioning agent to reduce
the normally high resistivity of aluminum silicate catalyst dust from petro-
leum cracking processes used in the production of high-octane gasoline.
The dust resistivity from the cracking operation is normally high (about
5 x 10 ohm-cm). It was found that additions of about 20 ppm of ammonia
reduced the resistivity to around ltf° ohm-cm and minimized sparking and
back corona problems associated with the higher resistivity material. Ammo-
nia additions were utilized as a standard procedure on such plants, since
the cost of conditioning proved to be low in relation to improved precipitator
performance and recovery of the valuable catalyst.
Conditioning of fly ash by ammonia addition to the flue gas is not so
straightforward as SO3 addition. Baxter reports that ammonia addition to
the flue gas of power plants must be made ahead of the air heater at tem-
peratures of around 700° F. Injections downstream of the air heater at
lower temperatures showed no conditioning effect.5
The formation of ammonium bisulfate in the temperature range of
570-670°F
SO3 + NH3 + H2O = NH4 HSO4
(gas) (gas) (gas) (liquid)
was suggested as a possible reaction.
Consideration of this product as a conditioning agent is consistent
with the experience and theory advanced by Dalmon and Tidy6 who investi-
gated the use of sulfamic acid and ammonium sulfate conditioning of fly ash
in a laboratory precipitator. Both compounds were believed to be converted
in this temperature range to a mist of ammonium bisulfate droplets which
were subsequently attached to fly ash particles with a resultant improve-
ment in precipitability.
Baxter reports associated problems with ammonia conditioning due
to the fouling of heat exchanger surfaces by solid ammonium bisulfate at
temperatures below 300°F.
Full scale experience with ammonia conditioning has been somewhat
varied. Tests on the acidic ash of the Tallawarra station of the Electricity
-------
-183-
Commission of NSW showed an increase in efficiency from 85-99% from
additions of 15-20 ppm of ammonia. This is contrasted to the require-
ments of SO3 in excess of 100 ppm with the same ash. Figure 7.9 shows
the effect of ammonia additions on the electrical readings. The increase
in voltage and decrease in current are indicative of the decrease in resis-
tivity accompanying ammonia additions.
The major reason for addition of conditioning agents is to reduce the
resistivity of a normally high resistivity ash. However, it has been reported
that difficulties arising from the operation of precipitators below the acid
dew point can be corrected by conditioning additives. Tests on full scale
plants at the Tennessee Valley Authority's Widow's Creek Plant have shown
increases in efficiency of from 50% to 95% with additions of around 15 ppm
of ammonia.
Reese and Greco7 have shown that increases in efficiency of the same
order can be achieved by increasing the precipitator inlet gas temperature,
which would tend to result in an increase rather than a decrease in resis-
tivity. This evidence would indicate that increases in efficiency with am-
monia injection are associated with an increase in resistivity by effectively
removing some of the SO3 or H2SO4. This would be in contrast to the
usual procedure of adding conditioning agents to reduce resistivity.
Full scale SO3 injection operations have been carried out in Australia,
England, Scotland, France, and South Africa.
Darby and Heinrich8 present some economic considerations for a full
scale SO3 injection system similar to the installation at Kincardine,
Scotland. The equipment for a 300 MW plant consisting primarily of
SO3 storage tanks, an evaporator, and an injector manifold, was estima-
ted to cost approximately $63, 000 (1966). At an injection rate of 10 ppm
SO3 for 8000 hrs, annual operating costs were estimated at $25,000 based
on the 1966 price of SO3 in England, and $5000 for heating costs.
Coutaller and Richard9 describe a similar installation in France and
estimate investment costs for a 125 MW to 250 MW plant at $60, 000 to
$80, 000 (1967). Operating costs for SO3 at about 17 ppm (v/v) were esti-
mated at about $ 11, 000 to $ 12, 000 per year (125 MW).
SOUTHERN RESEARCH INSTITUTE
-------
-184-
Start with Start with
clean electrodes^ NHa injection
20r 30
1 10 (-5 20
OL 10
10j- 40 -- -
1 5 (£30
OL 20
4r 50
1 11£40
OL 30
0 20 40 60 80 100 120 140
Time, min
Figure 7.9. Effect of Ammonia Conditioning on Southern
N.S.W. Fly Ash. Specific Collecting Area,
300 ft2/1000 ft3 per min; Mean Gas Temperature,
280° F.
-------
-185-
Alternatives for the use of stabilized liquid SO3, a source of chemi-
cal conditioning, may affect the economics. For example, sulfuric acid
injections have been tried with some success, although quantitative data
on cost and effectiveness as compared with SO3 are not substantiated.
SOUTHERN RESEARCH INSTITUTE
-------
-186-
CHAPTER 7
BIBLIOGRAPHY
1. White, H. J., Industrial Electrostatic Precipitation, Addison-Wesley
(1963).
2. Schmidt, W. A., "Electrical Precipitation and Mechanical Dust
Collection". I. &E. C. Proc., 41, No. 11, 2428 (1949).
3. Cohen, L. and Dickinson, R. W., "The Measurement of the Resistivity
of Power Station Flue Dust", J. Sci. Instrum., _40, 72 (1963).
4. Darby, K. H. and Heinrich, D. O., 'Conditioning of Boiler Flue Gases
for Improving Efficiency of Electrofilters", Staub 26. No. 11, 12(1966).
5. Baxter, W. A., "Recent Experience with Ammonia Conditioning of
Power Boiler Flue Gases", JAPCA 18. No. 12, 817 (Dec. 1968)
6. Dalmon, J. and Tidy, D., "Effect of Ammonium Sulphate and Related
Compounds on the Performance of an Experimental Electrostatic
Precipitator", CERL Laboratory Note, Central Electricity Generating
Board, England, (1968).
7. Reese, J. T. and Greco, J., "Experience with Electrostatic Fly-Ash
Collection Equipment Serving Steam-Electric Generating Plants",
JAPCA 18, No. 8, 523 (1968).
8. Darby, K. and Heinrich, D. O., "Conditioning of Boiler Flue Gases
for Improving Efficiency of Electrofilters", Staub, 26, No. 11, 12
(Nov. 1966), Engl. Transl.
9. Coutaller, J. and Richard, C., "Improvement of Electrostatic Dust-
Filtering by SO3 Injection", Pollution Atmospherique (Paris) 9, No. 33,
(Mar. 1967), Fr.
-------
-187-
CHAPTER 8
ELECTROSTATIC PRECIPITATOR SYSTEMS ANALYSIS
Systems analysis is the term applied to the analytical study of a com-
plete process. This technique is particularly useful for analyzing the behav-
ior of processes that are characterized by strong interdependencies between
the system subfunctions. The significant subsystem functions of an electro-
static precipitator were described in the preceding chapters of this document.
These fundamental relations have been available either qualitatively or quan-
titatively for many years; however, the comprehensive behavior of the entire
system has not been presented in a form that is useful for either the design
or analysis of complete installations. The purpose of this chapter is to re-
port on the progress in the development of the application of this systems
analysis technique to the electrostatic precipitator.
The benefits of a complete systems analysis of electrostatic preci-
pitation are threefold. First, it permits a systematic review of the param-
eters that influence precipitator performance. A review of this nature points
out the significance of the relationships that are known and also pinpoints
those areas where quantitative relationships are not available. Second, it
may permit the design of precipitators from theoretical relationships rather
than empirical ones. Finally, utilizing this approach, there is a possibility
for optimizing the performance of a precipitator for a given set of dust and
gas properties.
Present design methodology utilizes empirical data accumulated from
past experience. The relationships that are used do in fact contain informa-
tion that could be derived from theoretical considerations. There is a pos-
sibility of improving the design techniques by replacing the empirical rela-
tionships with theoretical ones. Experience has shown that the performance
of fly ash collectors is, in general, related to the particle size distribution
of the dust. The overall systems analysis should be able to provide informa-
tion to go one step further and explain how and why this is so.
The electrostatic precipitator is actually a single subsystem in the
overall operation of an installation. The input parameters to this subsystem
are dictated by the overall behavior of the entire installation. For example,
if the dust collector is installed in a coal-fired steam electric generating
plant, the input conditions are established by the boiler load and other oper-
ating parameters of the installation. The primary input parameters of the
SOUTHERN RESEARCH INSTITUTE
-------
-188-
installation are:
dust load,
particle size,
resistivity,
gas volume flow rate,
temperature, and
composition.
These input parameters may be adjusted by the plant operator on the basis
of overall system optimization, but in general are beyond the control of the
precipitator manufacturer. These parameters do dictate the operational
constraints on the electrostatic precipitator unit.
A second constraint on the system is the participate emission limit
established for the output gas conditions. These constraints determine the
required behavior of the precipitator. For the systems analysis approach,
the above factors constitute the independent input and output constraints
for the precipitator system.
8.1 GENERALIZED PRECIPITATOR MODEL
The unit subfunctions required for electrostatic precipitation have
been discussed elsewhere in this report. These subfunctions, namely,
source of unipolar ions,
formation of an electric field,
particle charging,
particle collection, and
dust removal,
together with their interdependent relationships can be described in a
functional block diagram. One possible system block diagram is shown
in Figure 8.1.
The behavior of each unit subfunction has been discussed in previous
chapters. However, in the fundamental theory section, the functional be-
havior of the subsystem was stressed. In systems analysis, a description of
-------
I Resistivity I
in
0
c
x
PI
a
Wire Rad.
Collector Rad.
Wire Roughness
Sec. Er-.is.
Aval. Coeff.
Ionizing Rad. ^^
van der Waals, Molecular,
and Mechanical
Electroneg. Gas
Gas Velocity
Velocity Dist.
Collection Area
Volume Flow
Applied Voltage
Gas Dnnsity
Ion Mobility
Dust Thickness
S e ctionalization
Collection Area
Wire Rad.
Collector-Had.
Partif!-- Size
Dielect. Const.
Time
Temperature
Dust Load
Gas and Dust
Ion Velocity
00
CO
I
Gas and Uncollected Dust
Collected Dust
Dust Load
Plate Design
Hopper Design
Gas Velocity
Gas Dist.
Rapping Force
Happing Interval
Particle Size
Dust Prop.
Temperature
H
Figure 8.1. Electrostatic Precipitator System Model.
-------
-190-
the strong Lnterdependencies among these subsystems is necessary to pro-
vide a description of the entire system function. The complexity evident
from Figure 8.1 points to the necessity for a high speed computer to allow
definitive computations. Without the computer, the system flow diagram
will only serve to point out the strong interrelationships in the system.
There are several system subfunctions where the detailed quantitative
relationships are not yet available. The qualitative effects are known, but
further research is required to complete the definition of the system behav-
ior.
Examples of these factors include;
(1) specific relationship between the electric wind and particle
collection,,
(2) specific particle reentrainment factors,
(3) relationship between resistivity, current density, and
sparking conditions as a function of position within the
precipitator, and
(4) detailed relationship between particulate and ionic space
charge as a function of position within the precipitator.
These factors are known to modify the behavior of an electrostatic
precipitator, but the specific relationships necessary for detailed calcu-
lations must be determined.
8. 2 SIMPLIFIED SYSTEM MODEL
A simplified system model was developed as the first step in the
detailed model. In this model, the unknown factors listed above were
neglected. A second simplification was possible because the model was
used for evaluation of operating installations rather than for the design
of new installations. Voltage and current measurements were available,
thus eliminating the need for the inclusion of the corona generation sub-
system. The simplified model block diagram is shown in Figure 8. 2.
-------
I Resistivity)
c
X
PI
a
z
n
z
H
H
n
Wire Had.
Collector Rad.
Wire Roughness
Sec. Erms.
Aval. Coeff.
Ionizing Rad.
Electroneg. Gas
Gas Velocity
Velocity Dist.
Collection Area
v'ol'jme Flow
Applied Voltage
Gas Density
Ion Mobility
Dust Thickne:
Soctionalization
Collection Area
Wire Rad.
Collector Rad.
Size
Const.
Par tic!'.
Dielect.
Time
Temperature
Dust Load
Gas and Dust
Ion Velocity
van der Waals, Molecular
i
t—i
CD
t-L
I
Dust Load
Plate Design
Hopper Design
Gas Velocity
Gas Dist.
Rapping Force
Rapping Interval
. Gas and Uncollected Dust
Collected Dust
Particle Size
Dust Prop.
Temperature
Figure 8.2,
Electrostatic Precipitator System Model - Parts Simulated in Computer Program in
Heavy Line.
-------
-192-
The inputs to the model are listed in Tables 8.1 and 8. 2
Table 8.1
Input Parameters from Boiler
Dust Load, D
Gas Volume Flow, v
Particle Size Distribution
Dust Resistivity, p
Gas Temperature, T
Particle Dielectric Constant, c
grains/ft
ft3/ min
ohm-cm
°p
F/meter
Table 8. 2
Precipitator Parameters
Collection Electrode Area, A
Corona Electrode Length, Lw
Corona Electrode Radius, a
Collection Electrode Radius, b
Corona Current, I
Applied Voltage, V^
Corona Wire Roughness Factor,
Relative Air Density, d
Ion Mobility, ju
ft2
ft
in.
in.
amps
kilovolts
cm2/ESU (volt) sec
The present model is used to compute the incremental collection
efficiency of the installation as a function of particle size. The collection
efficiency per unit distance through the precipitator is determined from
the particle charge and collection electric field. Calculations are based
on wire and pipe equations utilizing the electrostatic system of units. The
computation begins with a determination of the electric field within the
first foot of the unit, and proceeds to the collection efficiency for each
particle size. The inlet particle size distribution is modified by the amount
of material removed in each size range. This computation is iteratively
carried through the dust collector until the outlet dust concentration is
determined. A list of the derived quantities is given in Table 8.3. A
-------
-193-
system flow chart with the equations associated with each function is shown
in Figure 8.3
Table 8 . 3
Derived Quantities
Breakdown Field Strength, E0
Breakdown Voltage, V0
Corona Current per Unit Length,
Dust Surface Area per cm3, S
Current Density at
Collection Electrode, j -,
Average Electric Field, E
Saturation and Actual Charge, qs,
Migration Velocity for Each Size,
Gas Velocity (average)
ESU/cm
ESU (volts)
ESU (amp/cm)
cm2/cm3
ESU (amp/cm2)
ESU/cm
ESU (Coulomb)
cm/sec
cm/sec
8.3 EXAMPLE UTILIZING EXISTING SIMPLIFIED SYSTEM MODEL
Electric field. The equation used for the electric field in this particu
lar model is given below* x
ii \3Sr n S2y2 / r (Q 1 \
I |_ ^ * ' M-"-' ~ _| U- \«>->-1 P >-> L ' t \O.I^
This equation is valid where the contribution to the space charge is
primarily from the charged dust.
E0 is the breakdown field strength as shown in Chapter 1. The equa-
tion relating E0 to applied voltage and geometry is:
•e
E0 =30 f d (1 + 0. 3,Vd7a)
x 3.3 ESU/cm
(8.2)
Current required to charge the dust. The total amount of current re-
quired to charge the dust can be computed if the quantity of dust in each
size range per unit time and the saturation charge for each particle is
known. The equation for saturation charge due to field charging developed
* See list of symbols on Figure 8. 3.
SOUTHERN RESEARCH INSTITUTE
-------
I
t
i
i
• 1
f
t
i A
1
E 2
tt(-r> *tiv]exp6Sr'
\
a
L
>
If
I ! • * 0
Pi * H
i 2
"JT 3Sr
1
1
t
S
^
+.
1
S»r
Ec
E
j«
i
i
r»>
c
rt
/
S
1 ' |
k
E(r)
ai-
b!
AI
Average
Value
t ,
Ni * Ai
I Materii
!
- ZANj
I
J Lp
i ^
'v
7
i
1 v~*
ri
n
L_
P
q
~$
1 - e
i
(t)
1TT)
q
'
Ec
ri
j
v
ft)
\ Wi
g Lp
I
EC
|
I
i
I
ZS /ftN
i
i >
"
AN
~N~
a
b
f
d
E0
V0
I
Lw
i
ji
S
A.
j
Ec
t
wj
Nj
Lp
v
ti
A
Wire Radius
Collector Radius
Roughness Factor
Relative Air Density
Breakdown Field Strength
Breakdown Voltage
Total Current
Corona Wire Length
Current /Length
Ion Mobility ,
Surface Area of Dust per cm of Gas
Area of Plate
Current Density of Plate
Average Electric Field
- Saturation and Present Charge
* Dielectric Strength Relative
* Radius of Particulate
* Volume Flow Rate
» Time Lapsed in Collector
" Migration Velocity
• Number of Particles
* Precipitator Length
• Gas Velocity
"- Efficiency
* Increment
* Summation
Figure 8.3. Computer System Flow Diagram.
-------
-195-
in Chapter 3 converted to ESU is:
r» Ec (8.3)
where EC is the charging field. Therefore, the only remaining require-
ment is determination of the number of particle charges in each particle
size range that is introduced into the precipitator per unit time.
Particle size distributions are generally available for dusts that are
to be collected. The particle size continuum given for a given installation
can be approximated by an equivalent set of discrete particle sizes that
approaches the specific distribution, as shown in Figure 8.4. The con-
tinuum was replaced by a set of particle sizes where all the material in
each size range interval can be considered to be made up of identical
spheres.
The total weight of dust introduced per unit time is obtained from the
product of the dust concentration (D) and the gas volume flow rate
W = vg/\ x D ,g£|n_\ x/J£§m_\ (8.4)
&\sec/ Vft / \grains /
where the total weight of dust per second (W) is given in units of grams/ sec.
The total weight of dust in each size range is given by the product of the
total weight of dust per second and the discrete mass size distribution.
The total number of particles in each size range is determined by dividing
the total weight of material in each size range by the weight (volume x
density) of one particle. This yields a number distribution of particles
per unit time into the precipitator.
The total current required to charge the particles (C ) is obtained by
summing the products of the saturation charge per particle (qig) and the
number of particles per second in each size range (Nj).
n
Cp = E qis Ni (8.5)
1=1
The percentage of the total current required to charge the dust to saturation
is
SOUTHERN RESEARCH INSTITUTE
-------
-196-
30
(0
-. 20
I
§ 10
b
Q>
cu
0
.35 .70 1.0 5.0 10.0 20.0 45 80 150 300 700
Particle Size, \i
Figure 8.4. Input Particle Size Distribution and Discrete Approximation.
-------
-197-
x 100 (8.6)
The surface area of the dust per unit volume of gas is determined from
the particle size distribution and the dust load. The number of particles in
each size range times the surface area of one particle (A^) summed over all
size intervals yields this quantity
n
Si = S Nj A. (8.7)
i=0
Particle charge as a function of position. The charge on each particle as
a function of time is related to the free ion density in the vicinity of the par-
ticle. The charging rate is independent of particle size for field charging.
Thus a general expression relating charge to saturation charge as derived
in Chapter 3 is
q(0 _ I"
qs " 1 +T /t (8.8)
where r - charging time constant = 1 = —
N0 e ju 7T j 7T
The free ion charge density (N0 e) is related to the current density (j) in the
interelectrode space. The current density can be determined from either
the known voltage-current.characteristics of the unit, or from measured
current when working with operating installations. In this model, measured
currents were used.
Since only average values of current were known for individual instal-
lations, estimates of current density as a function of position within the
precipitator were used. The variation of current with position can probably
be determined by an iterative calculation of charge vs. time (distance) in
the computer, but an approximation to this was used in this demonstration
model. For lack of a more definitive value, the assumption made for this
current distribution was that the amount of current that went to apply charge
in each foot of the precipitator was that amount of charge required to charge
to saturation the quantity of material expected to be removed within that
foot of the unit. This is a conservative estimate of the amount of charge
bound to particulate in the inlet section. The actual bound charge is ex-
pected to be somewhat greater in practice.
SOUTHERN RESEARCH INSTITUTE
-------
-198-
The amount of material removed per foot, used for a gross estimate
of bound particulate charge, is calculated from the Deutsch-Anderson equa-
tion using the expected efficiency.
(8.9)
The free ion density is then estimated by multiplying the average free ion
density in that section of the precipitator by the ratio of the free ion charge
to total space charge within each foot of the precipitator. This free ion
density is used to compute the charging time constants within that increment
of the precipitator. The charge on each particle is computed from Equation
8.3. The time used is the distance through the precipitator divided.by the
velocity
t = ^- and (8.10)
E
r- ... c_ from which (8. II)1
3 if
1
(8.12)
Collection efficiency. At this point, values are available for the electric
field, saturation charge for each size range, and percent, of saturation charge.
Thus, the collection efficiency for each size range within this increment can
be computed. The equation for this efficiency for each size range is
Values are computed for each size range and stored.
The amount of material removed within each size range is determined
by multiplying the efficiency per foot by the particle count. This quantity
of material is subtracted from the incoming particles to yield a new parti-
cle size distribution as the input to the second foot of the precipitator. The
process is, repeated for each foot of the collector until an output particle
size and overall efficiency is computed.
-------
-199-
8.4 LIMITATIONS OF EXISTING MODEL
As indicated previously, the simplified precipitator model neglects
several factors that significantly influence precipitator performance. These
factors were omitted primarily because data were not available or because
analytical relationships had not been developed.
A review of the factors that were omitted shows that some of them
tend to increase the predicted efficiency, whereas others would tend to de-
crease the efficiency. The equations for electric field strength used in the
model include the electrostatic component and the contribution of charged
particles. The contribution of the free ions to the electric field is consid-
erable. However, equations relating field strength to both free ion density
and charged particle density were not available. Consequently, the field
predicted by the simplified computer model will be low. The particle charge
and collecting field will be greater than predicted by the model. The effect
of neglecting the ion density contribution to the electric field would be es-
pecially significant in installations preceded by a mechanical collector.
The computer model also does not include diffusion charging. For
most fly ash applications, the particle size distribution is sufficiently large
that diffusion charging is not too significant. However, for small size dusts,
such as metallurgical fumes, diffusion is significant and the effect of neg-
lecting it will result in lower predicted efficiencies.
The effect of particle reentrainment reduces overall collection effi-
ciency. The amount of dust lost from each section during rapping must be
small, otherwise overall collection efficiencies would be low. Since quan-
titative data relating dust loss to rapping conditions were not available, they
were not included in the model.
Since the total corona current was used as the basis for determining
electric field, resistivity of the dust layer did not influence the predicted
efficiency. High dust resistivity would serve to limit corona current and
operating voltage. For arriving at a precipitator design, effect of dust
resistivity in determining current-voltage characteristics must be included.
This necessitates a knowledge of the current-voltage characteristics of
the gas and the sparking limitations as determined by precipitator geometry
and dust resistivity.
SOUTHERN RESEARCH INSTITUTE
-------
-200-
8. 5 MODEL VERIFICATION
The simplified model was used to predict the collection efficiency of
a few installations for which test data were available. The purpose of the
verification was to determine how closely the simplified theory could come
to predicting efficiency, and to indicate the influence of some of the operat-
ing variables.
Within the constraints of the factors that were omitted in the simplified
model, predicted efficiencies agreed reasonably well with measured effi-
ciencies, except where excessive reentrainment was suspected.
Table 8. 4 lists the input data to the model and the predicted and
measured efficiencies. In Cases 1, 2, and 3, predicted and measured values
were quite close. These installations have reasonably normal conditions
and the neglect of rapping losses is probably balanced by the neglect of field
contribution to ion current and diffusion charging.
Figure 8. 5 shows the predicted collection efficiencies for various par-
ticle sizes for the five cases. The effect of the neglect of ion contribution
to the field is apparent in Case 4, which has a mechanical collector preceding
the precipitator. Consequently, the space charge due to charged particles
is low and the space charge due to ions would be proportionately higher for
Case 4 than for the other installations.
Cases 4 and 5 are operated under conditions where excessive reentrain-
ment is suspected. Also, gas flow quality has not been determined. In view
of the neglect of these factors in the model, it is not surprising that varia-
tions between measured and predicted'efficiencies occur in the two cases.
-------
-201-
Table 8.4
Comparison Between the Measured and Computed
Collection Efficiency for Five Fly Ash Precipitators
Unit No. Computed Efficiency, % Measured Efficiency,
1 97.8 99.1
2 98.9 96.5
3 98.5 98.4
4 96.5 84.1
5 96.0 55.0
SOUTHERN RESEARCH INSTITUTE
-------
-202-
8
0)
i
u
99.9
.8
.7
.6
.5
.4
99! 0
8
7
6
5
4
90
80
70
60
50
40
38
'8
m = measured
c = computed
o
i-«-
§
90
.5 .7 1.0 2.5 5.0 10.0 20.0 40 80 160 320 overall
Particle Size, n (diameter)
Figure 8. 5. Computed Efficiency Vs. Particle Size for Five Installations,
Fly Ash from Electric Utilities. Overall Computed and
Measured Efficiencies Are Compared on Right Margin.
-------
-203-
CHAPTER 9
DESIGN METHODOLOGY
The design of an electrostatic precipitator for a particular installation
involves many parameters that can influence both cost and performance. The
more significant variables involved in the design are:
(1) Area and type of collection electrodes,
(2) Dimensions of the precipitator shell,
(3) Size, spacing, and type of discharge electrodes,
(4) Size and type of power supply units,
(5) Degree of sectionalization,
(6) Layout of the precipitator in accordance with physical
space limitations,
(7) Design of the gas handling system,
(8) Size and shape of hoppers,
(9) Type and number of electrode rappers, and
(10) Type of dust removal equipment.
There are several methods used for the selection of suitable values for
these variables and each manufacturer may utilize slightly different methods
in arriving at a particular design.
The following discussions are intended to point out the methods that
can be used in the design or the assessment of the adequacy of a design rather
than in presenting a detailed design handbook.
Two approaches to the selection of precipitator size will be presented.
One approach, Method I, is based on the conventional Deutsch-Anderson effi-
ciency equation and the other, Method II, approaches the design from "the
standpoint of the electrical requirements. These methods must obviously
give compatible results. They differ mainly in the fundamental way in which
design is approached.
9.1 DESIGN METHOD I
A common approach to the selection of the area of collecting plate
required is to utilize the Deutsch-Anderson equation
SOUTHERN RESEARCH INSTITUTE
-------
-204-
A
rj = 1 - exp ( w)
where
A = area of collecting surface,
V- = gas flow rate,
w = precipitation rate parameter,
17 = efficiency, %, and
exp = base of natural logarithms.
The critical parameter in this equation is the precipitation rate w.
This parameter varies with each installation depending upon resistivity
and particle size of the dust, quality of gas flow, reentrainment losses,
and sectionalization, among other factors. The values of w are selected
by the equipment manufacturers on the basis of past experience with a par-
ticular dust, or from the composition of the dust that can be related to past
experience. Each precipitator manufacturer therefore has a file of experi-
ence from which a precipitation rate parameter can be selected, and this
file of information is kept as proprietary data.
The values of the precipitation rate parameter w vary with the applica-
tion as a result of variations in dust properties. Variations also occur
within each application area. Table 9. 1 lists the average values of preci-
pitation rate parameters for various applications, and the range of values
that might be expected within each application. From this table, it is ap-
parent that the spread in the values of the precipitation rate parameter is
large in some instances such as fly ash precipitators, and within a reason-
ably narrow range within others. For the pulp and paper industry, a preci-
pitator designed for recovery boilers would have an uncertainty of around
15-20% in precipitation rate parameter. For a precipitator designed for a
98% collection efficiency, the measured efficiency would range from 97 to
99.4% based on the range of design precipitation rate parameters.
For fly ash precipitators, on the other hand, the variation in preci-
pitation rate parameters is quite large, so that a precipitator designed on
the basis of a w of 0.43 for 98% efficiency would give an efficiency of only
around 75% if the precipitation rate parameter were 0. 13 ft/sec.
The major problem in the design of precipitators based on this
approach is in the selection of the precipitation rate parameter for the
specific application. Several techniques can be used to narrow
-------
-205-
Table 9.1
Representative Precipitation Rates for Various Applications
Precipitation Rate w ft/sec
Application Average Range
Utility Fly Ash 0.43 0.13-0.67
Pulp & Paper 0.25 0.21-0.31
Sulfuric Acid 0.24 0.20-0.28
Cement (wet) 0.35 0.30-0.40
Smelter 0.06
Open Hearth 0.16
Cupola 0.10
Blast Furnace 0.36 0.20-0.46
SOUTHERN RESEARCH INSTITUTE
-------
-206-
the uncertainty of the value of w to be used. If the in-situ resistivity of the
dust is known, the precipitation rate parameter can be determined for some
applications. Figure 9. 1 shows the variation in w with resistivity for fly
ash precipitators. If the precipitator being designed is a replacement for,
or an addition to an existing unit, resistivity can be measured, and the
uncertainty in the value of w can be reduced. Alternatively, if a similar
installation burning the same fuel is available, measurements of resistivity
can be made and the value of w selected with some confidence.
The data from Figure 9. 1 only apply to fly ash or to a dust with simi-
lar properties. If the particle size differs significantly, the absolute values
of w will change, although the general character of the curves would be sim-
ilar.
In-situ resistivity data have not been determined to the same degree
in applications other than fly ash, so that statistically reliable data relating
w and resistivity are not generally available.
If it is impractical to select w on the basis of resistivity, other factors
can often be used. In fly ash precipitators, resistivity is influenced by the
sulfur content of the fuel, and relationships have been developed between
precipitation rate parameter w and percent sulfur. Figure 9. 2 shows a curve
developed by Ramsdell for a group of fly ash precipitators burning coals of
varying sulfur contents. On a statistical basis, the precipitation rate can be
predicted within reasonable accuracy. However, on an individual installation,
the variations are too great to predict w with acceptable precision based upon
sulfur content alone. In many instances, the only information available is
the sulfur content of the coal, and designs are sometimes based solely upon
this parameter.
Particle size of the dust is a very important consideration in determin-
ing the value of w for design purposes. Referring to Table 9. 1, the variations
in w between the various application areas are due largely to particle size
variations. In cement kilns, the alkali content of the raw material alters the
size distribution of the dust. Metallurgical operations characteristically
produce smaller size dusts from high temperature melting operations. Size
of dusts from recovery boilers in pulp and paper mills can change with tem-
perature. These factors result in variations in precipitation rate param-
eters between the various applications, and within the same application area.
1Refer to the bibliography for this chapter.
-------
0.656
u ° 492
id v
0« »
at
K
U
••-I
o
0.328
0.164
20 1 1—» I MM
10
u
a>
w
"B
o
I I I I 8 I I
-207-
T 1—I I I >M
l i i i
T 1—I I I I I
10"
10 10
Resistivity, ohm-cm
11
10
Figure 9.1. Relationship between Precipitation Rate Parameter and
Resistivity. (Reference 2)
SOUTHERN RESEARCH INSTITUTE
-------
-208-
0.656
E
cd
(4
cd
P*
0)
td
O
D.
•«-i
O
0.492
u
0)
(0
0.328
0.164
r-20
Temp - 300° F
- 5
01 2
Sulfur Content of the Fuel,
Figure 9.2.
Relationship between Precipitation Rate Parameter
and Sulfur Content for Electric Utility Installations
at a Temperature of 300°F (Reference 1). '
-------
-209-
The sections covering the specific application areas in Part II de-
scribe the emissions from each process in greater detail, and also present
data showing the range of design and test precipitation rate parameters
where they are available. Reference should be made to each section for a
more detailed analysis of the range of precipitation rate parameters.
When the selection of precipitation rate parameter has been made,
the area of collecting surface required to achieve a given efficiency when
handling a given gas volume can be determined.
The power required for a particular application is determined on an
empirical basis. The power requirements for a given application are rela-
ted to the efficiency and the gas volume handled. Figure 9. 3 is a curve
showing the power requirements per unit of gas volume for two applications;
fly ash and pulp and paper mill recovery boilers. Similar curves can be
developed for other applications. The second step in design, therefore, is
to determine the total power requirements based on efficiency and gas flow.
Note that the recovery boiler precipitators require greater power per unit
of gas flow to achieve the same efficiency as a fly ash precipitator. This
is primarily due to differences in particle size of the dust, and is related
to the precipitation rate parameter.
Figure 9. 4 shows the variation in collection efficiency with the number
of independently powered bus sections. The number of sections required
to reach a given efficiency can be determined from curves for the specific
application. The curves shown in Figure 9.4 are for fly ash precipitators,
and were developed from empirical relationships from a large number of
tests.
The above procedure will provide a rational basis for arriving at
plate area, total power, and degree of sectionalization required. It should
be recognized that the selection of the value of w and the curves relating
power and sectionalization requirements are all interrelated. If inadequate
sectionalization is used, a lower value of w would result, the precipitator
could not be operated at the required power level, and the efficiency would
be reduced. Consequently, curves relating the design parameter should be
internally consistent.
The type and number of rappers for the collecting and discharge elec-
trodes depend upon the properties of the dust, gas properties, current
SOUTHERN RESEARCH INSTITUTE
-------
-210-
c
0»
c
o
3
o
V
99
0
rFly Ash
Prepipitator
Recovery Boiler
Precipitator
0
100 200 300 400
Power Rate, watts per 1000 cfm
500
600
Figure 9.3. Collection Efficiency Vs. Power Rate.
-------
-211-
c
4)
w
u
0)
1—I
•3
u
99.9,
99.0
90.0
80.0
70.0
60.0
50.0
/ 1.8% Sulfur,
300°F
012345
Number of Bus Sections per 100, 000 cfm
Figure 9.4. Relationship between Collection Efficiency and
Sectionalization (Reference 2).
SOUTHERN RESEARCH INSTITUTE
-------
-212-
densities, and the configuration of the electrodes and electrode support
structures. High resistivity dust is usually harder to remove because of
the increased force holding it to the plate. Low temperature operation also
tends to give a moist dust that is more difficult to remove. Table 9. 2 shows
the number of rappers per unit area of collecting electrode and the number
per unit length of discharge wire for a group of installations.
The following example will serve to illustrate this design approach.
Design an electrostatic precipitator for a pulverized fuel boiler with the
following given conditions:
(1) Dust resistivity 7 x 1010 ohm- cm
(2) Gas temperature 300°F
(3) Gas volume 750, 000 acfm
(4) Sulfur content 1.8%
Example 1
(a) Select a precipitation rate parameter.
From Figure 9. 1, the precipitation rate parameter corresponding
to a dust resistivity of 7 x 10 ohm-cm is 0.35 ft/sec or 21 ft /mitt.
(b) From the Deutsch-Anderson equation
i /> A v
r? - 1 - exp ( -— • w)
g
or rearranging
100
- w 100 - r?
750,000 100
21 m 100 - 99
= 35; 600 In 100*
= 164, 000 sq ft
(c) Compute total power requirements.
From Figure 9.3, power required for 99% efficiency is 140 watts/
1000 cfm. For 750, 000 cfm, total power = 750 x 140 = 105, 000 watts.
(d) Determine number of bus sections.
From Figure 9.4, number of bus sections required is 3. 5 per
1000 cfm. 7. 5 x 3. 5 = about 26 bus sections.
-------
-213-
Table 9. 2
Rapping Practices
Application
Utilities
Pulp and Paper
Metals
Cement
Collection Electrode
Rappers/I OOP ft2
0.25 - 0.90
0.25 - 0.99
0.11 - 0.82
0.33 - 0.52
Corona Electrode
Rappers/1000 ft
0.09 - 0.66
0.21 - 0.32
0. 28 - 0. 50
0.19 - 0.33
SOUTHERN RESEARCH INSTITUTE
-------
-214-
9. 2 DESIGN METHOD II
A second approach to the sizing of precipitators is to determine the elec-
trical characteristics first and develop relationships that relate efficiency and
power requirements to the required collection surface area. The concept of
the electrical approach is based on the theoretical factors that influence the
precipitation rate parameter, that is, the current required to charge the par-
ticles and provide the space charge field, and the voltage required to estab-
lish the electrostatic collection field.
In this design method, the curves relating efficiency and corona power
density are developed on an empirical basis. The curves are the same as
that for Design Method I, and are given for fly ash and recovery boiler preci-
pitators in Figure 9.3. Figure 9.5 is an empirically developed curve show-
ing relationship between precipitation rate parameter w and corona power den-
sity. Since the total power has been determined previously, the collecting
surface area can be computed from the data read from Figure 9. 5. The num-
ber of independent bus sections can be determined in the same manner as in
method I. The following example will serve to illustrate the design proce-
dure based on this method.
Example 2
(a) Select a precipitation rate parameter.
From experience with this coal, the precipitation rate parameter is
selected to be 0.35 ft/sec.
(b) Compute corona power.
From Figure 9.3, power per unit volume corresponding to a col-
lection efficiency of 99% is 141 watts/1000 cfm. The total power
is then computed:
141 watts x 750>0oOcfm = 105, 000 watts
1000 cfm
(c) Determine collection electrode area
From Figure 9.4, for w = 0.35 ft/sec, power density =0.64
watts/ft, area = power total power density = 105, 000
watts x .1 ft = 164, 000 ft2
0. 64 watt
-------
-215-
20 g.
18
16
14
12
10
8
6
4
2
'0.53
0,40
m
5
0.27
0.14
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Power Density, watts/ft2
1.0 1.1 1.2
Figure 9. 5. Linear Relationship between Precipitation Rate Parameter
and Power Density for Fly Ash Collectors.
SOUTHERN RESEARCH INSTITUTE
-------
-216-
(d) Number of electrical sections
From Figure 9. 5 the number of sections per 100, 000 cfm at 99%
efficiency = 3.5
_n_
Number of sections 2. 8 x > ™ — = 26
It should be noted that the curves relating the various factors in
method II must also be mutually compatible. The major difference between
these two design methods is that the empirical data are used to arrive at
the collecting surface area as opposed to the use of the Deutsch-Anderson
equation. In the examples shown, the precipitation rate parameter w was
used in both instances. However, it is possible to develop other relation-
ships that would eliminate the necessity for arriving at a value of w alto-
gether. Such relationships would be empirical and be subject to the same
degree of uncertainty as the selection of the precipitation rate parameter.
A more fundamental approach to precipitator design would be based
upon theoretical factors such as particle size, gas composition, dust resis-
tivity, precipitator dimensions, and other input conditions. This method is
discussed in the chapter on systems analysis. However, at this time,
techniques for design based only on theoretical relationships have not been
developed to the extent that they can be used on a commercial basis.
Consideration in the design of the gas handling equipment and ash
removal systems is covered in other sections.
-------
-217-
CHAPTER 9
DESIGN METHODOLOGY
1. Ramsdell, R. G., "Design Criteria for Precipitators for Modern
Control Station Power Plants, " Proc. Am. Power Conf. 30,
(April, 1968) Abstr. Combustion 39.
2. White, H. J., Industrial Electrostatic Precipitation, Addison-Wesley,
Reading, Massachusetts (1963). "~~
SOUTHERN RESEARCH INSTITUTE
-------
-218-
CHAPTER 10
MECHANICAL COMPONENTS
The mechanical construction of an electrostatic precipitator varies
according to the particular process to which it is applied and according
to the design philosophy of the precipitator manufacturer. Broadly clas-
sified, precipitators are of the tubular or plate type depending on the col-
lection electrode geometry. Precipitators are further classified as hori-
zontal or vertical-flow types depending on the direction of gas flow through
the precipitator.
10. 1 TUBULAR TYPE PRECIPITATORS
Tubular type precipitators are composed of cylindrical collection
electrodes with discharge electrodes passing through the axis of the cylin-
ders. A typical arrangement of a commerical tubular precipitator is shown
in Figure 10.1. Gas to be cleaned flows around the outside of the cylindri-
cal electrodes and up through the inside of the cylinders where precipita-
tion takes place. The collected dust or liquid is removed from the bottom
of the chamber.
Hexagonal tubes are also used in tubular type precipitators to increase
the amount of collecting surface available in a given volume since closer
packing is possible with a hexagonal configuration.
Tubular type precipitators are frequently used where the gas flow is
low, where mists or fogs are being precipitated, or where water flushing
is used to remove collected material. When water flushing is used, the
tubes do not have to withstand the rapping forces and can be made of lighter-
weight materials. Tube diameters vary from 6 to 12 inches, and lengths
of the tubes are typically between 6 and 15 feet.
10. 2 PLATE TYPE PRECIPITATORS
The great majority of electrostatic precipitators in service are of
the plate type. Collection plates are spaced 8 to 12 inches apart with a
series of discharge electrodes spaced along the centerline of adja-
cent plates. A typical arrangement is shown in Figure 10. 2. The gas to
SOUTHERN RESEARCH INSTITUTE
-------
-219-
.High Voltage Insulator
Compartment
Support Insulator
Steam Collector
High Tension
Support Frame
Collecting
Electrode Pipes
Shell
High Tension
Electrode
Electrode
Weight —
Gas Deflector
Cone
Collected^
Dust Out
Figure 10.1. A Single-Stage Vertical Wire and Pipe Unit.
-------
-220-
Figure 10.2. Parallel Plate Precipitator.
SOUTHERN RESEARCH INSTITUTE
-------
-221-
be cleaned passes horizontally between the plates (horizontal flow type) or
vertically up through the plates (vertical flow type). Material collected by
the precipitators is generally deposited in bins or hoppers at the base of the
precipitator in the dry bottom type, and in a liquid in the case of the wet
bottom precipitators. The collected material is then removed by appropri-
ate dust handling equipment.
The purpose of the shell is to confine the gas flow for proper exposure
to the electrodes, to avoid excess heat loss, and provide structural support
for the electrodes and rapping equipment. The shell is normally rectangular,
where plate electrodes are used^or cylindrical if tube electrodes are used.
Cylindrical shells may also be used with plate type electrode precipitators
where relatively high or low gas pressures are encountered.1 Shell mate-
rial is usually steel; but because of particular corrosion problems, it may
be made or lined with tile, brick, concrete, or special corrosion-resistant
steels. Insulation is usually required to maintain the shell at a temperature
above the dew point if the gases contain corrosive materials.2 Access doors
and stairways and safety provisions are provided as auxiliary equipment.
Gas diffuser plates can be provided as a part of the shell to improve
gas flow. These plates are usually of the order of ^ inch thick with holes
to equalize gas flow. Design of diffuser plates is covered in more detail
in the section on gas flow. Roof and wall baffles are used to minimize the
amount of gas which may by-pass the electrodes.
10.3 DISCHARGE ELECTRODES
Discharge electrodes can be of a wide variety of types. Practice
differs between American and European manufacturers in the method of
supporting the discharge electrodes. Typical European practice is to pro-
vide a frame or tubular support for the electrodes, whereas most American
manufacturers suspend the electrodes from a support and maintain them in
position by weights and guides at the bottom.
The shape and size of the discharge electrodes are governed by the
corona current and mechanical requirements of the system. Where high con-
centrations of fine dusts are encountered, space charge limits the current
flow, especially in the inlet sections. In such cases, special electrodes
*Refer to the bibliography for this chapter.
-------
-222-
that give higher currents may be used to achieve a high power density within
the inlet sections. Variation in the current flow and electric field within
limits is possible by controlling the type and size of discharge electrode.
Weighted wire. Weighted wire discharge electrodes consist of vertically
hung wires spanning the full height of the collecting electrodes as illustra-
ted in Figure 10. 3. They are typically from 0.100 to 0.150 inch in diam-
eter. In practice, they are suspended from a support frame at the top and
held taut by a weight at the bottom. The wires are kept in place by a steady-
ing frame, equipped with guides at the bottom.
There is considerable variation among manufacturers in the method
of supporting the discharge wire from the support frame. The principal
concern with the support is to minimize wire breakage due to mechanical
fatigue. The discharge wires move under the influence of both aerodynamic
and electrical forces, and under severe conditions, mechanical fatigue
failure can occur. Various methods of allowing some movement of the
support have been attempted to minimize the fatigue problem.
Wires are also subjected to localized sparking in regions of high
field strength and shrouds are sometimes used to give a larger diameter,
and hence low field strength in critical regions near the ends of the elec-
trodes.
Weighted twisted squares. Weighted bar discharge electrodes consist
of vertically hung square bars spanning the full height of the collecting
electrodes. They are usually ^inch or -5 inch square and are twisted
longitudinally. The twisting helps straighten the rods and increases the
length of sharp edge, which increases the corona current. In practice,
they are suspended from a support frame at the top and held taut by a
weight at the bottom. The wires are kept in place by a steadying frame
equipped with guides at the bottom.
Spiral wires. Spiral wires are formed as a spring and then pulled out.
The ends are attached to a formed frame. The spring tension developed
keeps the wires taut. A complete frame containing a multiplicity of wires
is then installed between the collecting plates as a discharge electrode.
The wire used is on the order of No. 12 gauge.
SOUTHERN RESEARCH INSTITUTE
-------
-223-
Figure 10.3. Weighted Wire Corona Electrodes.
-------
-224-
Barbed wire. Barbed wire is also employed for discharge electrode ser-
vice. These wires are hung vertically and weighted at the bottom to hold
them taut. The wires span the full height of the collecting plates. Commer-
cial grade barbed wire has been used for this application. The wires are
supported at the top and are guided and kept in place at the bottom by a
steadying frame.
Stiff electrodes. Stiff electrodes consist of formed sheet or structural
members rigidly fastened to a support frame at the top. The length can
span the height of the collecting plates or be supported from frames. For
the longer spans, the electrodes are kept in place at the bottom by an align-
ment frame. The thickness of the base material is on the order of ^j inch.
In order to enhance corona generation, the sheets are formed with sharp
edges or with prongs pierced out of the base metal. Figure 10. 4 shows
typical shapes of stiff discharge electrodes.
Supported wires. As indicated previously, European practice generally
is to utilize a stiff structural member for the support of the discharge elec-
trodes. The support members (usually tubes) are fastened rigidly to a top
support frame. The electrodes themselves can be of a variety of shapes
ranging from circular, square twisted, or formed sheets to other special
shapes to give the desired cur rent-voltage relations. The electrodes are
strung through crossarms fastened to the central support member or from
sections of a rigid frame. The distance between supports is around 2 to
4 feet.
The supported wire electrode system has the advantage of minimiz-
ing the wire breakage problem since the electrodes are supported by rigid
members and remain in position and energized even if breakage of the
electrode occurs. Figure 10. 5 illustrates two types of support frames.
Horizontal rods. Horizontal rod type discharge electrodes are used in
vertical flow applications. Small diameter rods (order of ^ inch) or
twisted square rods (^ to ^ inch) are placed horizontally in a metal frame.
The rods are fastened at one end only. The remaining end is allowed to
float freely in the frame to prevent rod bowing due to expansion. The
rods are extended beyond the frame and capped with a contoured knob to
reduce end arcing. The frame assembly is installed between the collect-
ing plates and suspended from a support frame above.
SOUTHERN RESEARCH INSTITUTE
-------
-225-
Figure 10.4. Typical Shapes of Stiff Discharge Electrodes (Reference 3).
-------
-226-
Frame
Support Bolt
Corona Wires
\
-J. Central Mast
\
Central Mast Type
Frame Type
. ^
Cross Member
Figure 10. 5. Two Types of Support Structures for Corona Wires.
SOUTHERN RESEARCH INSTITUTE
-------
-227-
10.4 DISCHARGE ELECTRODE SUPPORT
The main functions of the discharge electrode support are to provide
the necessary high voltage electrical insulation and to give mechanical
support to the discharge electrode frame.
There are several types of support systems currently used in precipi-
tator design. One type is a support insulator with bushings as illustrated in
Figure 10.6. In this design, the high voltage insulators are located on the
roof of the precipitator. A bus beam is mounted on top of the insulators
and the discharge electrode assembly is suspended from the bus beam by
hanger rods. Porcelain pin-type insulators support the mechanical load
of the internal framework and are located in a relatively low temperature
zone comparatively free of contaminants. The entrance bushings, commonly
called flower pots, are made of a clay refractory and cemented in place
with a refractory cement. These bushings are not gas tight, and it is com-
mon practice to provide a flow of air into the insulator compartment to pre-
vent entrance of dust laden air from the precipitator. This type of arrange-
ment has several inherent disadvantages.
1. The refractory bushing is porous and therefore adsorbs
moisture during low temperature operation.
2. It is made in separate pieces and must be cemented in place.
Heat is required to dry the joints before it becomes an accep-
table insulator, which results in a slow startup procedure.
3. If a positive pressure exists in the precipitator, the insulator
compartment must be pressurized to maintain a positive
pressure in excess of the precipitator pressure at all times.
This means that there is a flow of air through the open space
between the hanger support rod and the bushing.
4. In many instances the introduction of-air into the treating
zone of the precipitator can cause disturbance of the
electrical field with the result that sparking will exist in
the colder areas.
5. The surfaces of the refractory are generally rough and
difficult to clean. In many instances, fly ash precipitators
-------
H. T* Conductor
From Rectifier
Insulator
Compartment
Metal
H. T.
Pptr.
L«
i —
^
'nclosed
Bus Due
Roof
-
t
Support Conventional Pin Type
Beam Porcelain Insulator
; ^^ i ; * ' ' : '• ; 1
Air ! I • ' | F^f^ i I
Flow ^t~\ >-T-k ! !
/.. 1- — -i ••.-• i / x / \ i ' \' '-
| . 1 / ..'. v / \ 1 < \ i
, :'! \ ^ ^ / ;!i \
• 1 — , \ ^--fe riai | 1
L^^ T T / •: i
Refractory
Type Entra
Bushing
O
I
Hanger
Support
High Temp.- Extreme
Environmental Zone
to
to
co
J H. T. Discharge
-L Electrode Frame
r r
Figure 10.6. High Temperature Support Bushings, Style 1.
-i
n
-------
-229-
are subject to fume from oil firing. This fume condenses
on the inside of the bushing with moisture and sulfur com-
pounds, most of which are relatively conductive and practi-
cally impossible to remove.
6. Another disadvantage of the arrangement in Figure 10. 6
is that in many instances the internal operating tempera-
ture in the insulator compartment is above the safe
operating temperature of conventional porcelain insulators.
This condition is, as a rule, relieved by introducing addi-
tional air or by adding thermal insulation to the floor of
the compartment.
7. In many processes there are periods during startup when
a large quantity of moisture exists in the gas. During these
phases, the refractory bushing is wet and becomes conduc-
tive. To alleviate this condition, electric or steam heaters
are installed in the compartment to keep the temperature of
the bushings above the dew point of the gases.
Figure 10. 7 illustrates a support bushing arrangement in which the discharge
electrode assembly is suspended by hanger rods which are supported directly
by bushings. In this case, the bushings are constructed of alumina or
Pyroceram and have higher mechanical strength and better thermal shock
resistance, permitting a much simpler electrode support design.
The low porosity of the insulation materials and better gas seal pro-
vided by the gasket minimize the gas inleakage to the insulator compart-
ment. However, for some applications the bushings are continuously
purged with air, either induced when the precipitator is under suction or
forced by blowers. The bushings are housed in either individual roof tun-
nels or in a common housing on top of the precipitator. There are other
types of electrode support structures in use, and each manufacturer will
use a different type based upon experience. Most of the types will be some
variation of these basic concepts.
10. 5 COLLECTING ELECTRODES
Various types of collecting electrode structures have been used in
the plate type precipitator. The desirable properties are: (1) that the plate
act as a shield to provide protection of the collected dust and minimize
-------
H. T. Conductor
From Rectifier
Insulator
Compartment
Metal Enclosed
H. T. Bus Duct
Pptr. Roof
Metal
Cover
Hanger
Support
f Pyroceram
Bushing
High Temp. - Extreme
Environmental Zone
i
CO
u
H. T. Discharge
Electrode Frame
IT
Figure 10.7. High Temperature Support Bushings, Style 2.
-------
-231-
reentrainment by scouring from the gas stream, (2) should be free of points
or sharp edges that may cause sparking and reduction in the operating volt-
age, and (3) that the design provide mechanical rigidity. The latter is also
important from the standpoint of transmitting the rapping accelerations to
all parts of the collecting plate.
Various types of collecting electrodes are illustrated in Figure 10.8.
This list is by no means exhaustive. The patent literature contains numer-
ous other electrode types which have been prepared to shield the collected
dust and minimize reentrainment. Many of these are unacceptable because
of excessive weight or cost.
Shielded flat plates. This type of collecting electrode is the most popu-
lar in present day use in this country. The electrode is used chiefly in
horizontal flow, duct type precipitators. The plate proper consists of
either solid sheet, expanded metal,or perforated sheet. In order to shield
the precipitated dust from the gas passing across the plate, baffles are
mounted along the plate. The baffles are fabricated as formed shapes,
and welded to the ends and surfaces of the collecting plate. Baffle shapes
vary from flat strips perpendicular to the collecting surface to aerodynamic
designs to minimize gas turbulence. The size of the collecting electrodes
ranges from 10 feet to 40 feet in height and from 3 feet to 12 feet in the
direction of gas flow.
Offset plates. These collecting plates are made by bending a flat sheet
into a square or angular zig-zag or a corrugated pattern. The dust preci-
pitated in the troughs is shielded from the main gas stream, minimizing
reentrainment. The plates are usually from 10 feet to 30 feet in height,
and from 3 feet to 9 feet in the direction of gas flow.
Pocket plates. As the name implies, these plates are made up by fabri-
cating pockets or louvers out of the base sheet. The plates are arranged
so that the pockets face into the gas stream. Two plates, spaced about
inches back-to-back, are employed to make up a collecting electrode
assembly. The hollow space between the plates is used as a chute where-
in the dust collected on the plate can fall down to the hopper out of contact
with the gas stream. Plates are from 18 feet to 24 feet high and 6 feet in
the direction of the gas flow.
-------
-232-
«
-------
-233-
Vee plates. A vee plate is a composite assembly of metal strips bent in
the shape of a vee or chevron. The vees are spaced about 1^- inch apart for
the full length of the plate. The assembled plate is about 2 inches thick.
The plates are hung so that the points of the vees face upstream. The spaces
between the vee members act as quiescent zones in which the dust is preci-
pitated with minimum reentrainment. The collecting plates in use today
range from about 18 feet to 36 feet in height. An individual plate is 3 feet
in the direction of gas flow. It is customary to fasten two plates together
in order to make up a 6 foot section.
Rod curtain. Rod curtain collecting electrodes are solid metal rods
which are hung vertically in a straight line, relatively close together. Re-
latively large collection areas are thus formed that are effectively shielded
from the gas stream. These are very poor aerodynamically and are sub-
ject to high reentrainment losses. Their main advantage is the ability to
withstand high temperatures (700° to 800° F) without much warping or dis-
tortion.
10.6 ELECTRODE CLEANING EQUIPMENT
Dust accumulates on both the collection and discharge electrodes and
must be periodically removed for proper operation of the precipitators. The
requirements for rapping are discussed in detail in the chapter on rapping
and reentrainment. Methods of dust removal include impact, vibration, or
liquid flushing. For most precipitators, mechanical means are used to
remove the collected dust in the dry state.
Single impact rapper - (electromagnetic solenoid). Electrostatic sole-
noid rappers consist of a plunger which is lifted by energizing the solenoid.
On release of the plunger by deenergizing the coil, it falls against an anvil
which transmits the rap through a rod to the electrodes to be cleaned.
Figure 10.9 illustrates the construction. Solenoid type rappers are used
for both discharge electrode and collecting electrode cleaning. They are
mainly used in connection with horizontal flow, dry type plate precipita-
tors. When used for plate or discharge electrode rapping, they are located
on top of the precipitator.
Solenoid rappers can also be spring actuated as well as gravity actu-
ated, and in such instances, can be located in other than upright position
-------
-234-
Conduit Box
Cover and Gaske"
Plunger Guide
Coil Cover
Coil Assembly
Plunger
Casing Gaskets
Flange Bolts and Nuts
Lower Casing
Adjusting Nuts
4ff* Adjusting Bolt
Adapter or Mounting
Rapper Rod
Figure 10.9. Typical Electromagnetic Rapper Assembly.
SOUTHERN RESEARCH INSTITUTE
-------
-235-
on the side or end for plate rapping. Control consists of varying the elec-
trical energy, which changes the magnitude of the impulse, or the frequency
of rapping. Final selection of magnitude and frequency is usually made
after trials in the actual plant under operating conditions. The accelera-
tion of the rap can be as low as 5 g's, but raps from 30 g's to 50 g's are
required for most fly ash precipitators.
Vibrators (electromagnetic). Electromagnetic vibrators consist of a bal-
anced, spring loaded armature suspended between two synchronized electro-
magnetic coils. When energized, the armature vibrates at a high frequency.
This vibrating energy is transmitted through a rapper rod to the electrodes
and also collecting electrodes. When employed for discharge electrode
cleaning, the rapper rod is provided with an electrical insulating section
in order to isolate the high voltage electrode charge from ground. This
device is used mainly in the vertical position for discharge electrode clean-
ing and is suitable for both horizontal and vertical flow dry type precipita-
tors. Control consists of varying the electrical energy input, which changes
the amplitude of vibrations, the operation time duration, and the frequency
of vibration. The armature must be adjusted for the operating position
required. Figure 10.10 is a typical electromagnetic vibrator installation.
Vibrators (eccentrically unbalanced motors). Mechanical vibrators con-
sist of an electric motor equipped with adjustable cam weights mounted on
a single shaft or on both shafts of a double ended motor. When operated,
the eccentrically positioned cam weights set the entire assembly into vibra-
tion. The motor is mounted directly on the rapper shaft which transmits
the generated vibration to the electrodes to be cleaned.
Vibrators are used for both discharge electrode and collecting elec-
trode cleaning. They are used mainly in connection with horizontal-flow, dry-
type plate precipitators. This type of rapper is used for either top, side,
or end operation. Control consists of varying the degree of eccentricity
by cam weight adjustment, the length of time operated, and the frequency
of operation.
Single impact (motor-driven cams). This rapping system is usually
used to clean dry-type horizontal flow precipitators. The mechanism
consists of a motor-driven shaft running horizontally across the precipi-
tator. Cams are located along the shaft which raise small hammers by
their handles. When the rotating cam reaches the end of its lobe, the
-------
-236-
Enclosure
Quick Opening
Clamp
Ground Connection
Clamps
(Rapper Rods to
Ceramic Shaft)
Precipitator R oof
Dust Laden Vibratiorf :
Gas Area
To
Discharge
Wire
J
Enclosure
Vibrator
Mounting Plate
Stuffing Box and Guide
Flexible Conduit
Conduit Fitting
Housing
Ceramic Insulating Shaft
Closure Plate
High Voltage Bushing
Rapper Rod Assembly,
Must Be Plumb
Anvil
High Tension Frame
Discharge Wires
Figure 10.10. Typical Electromagnetic Vibrator Assembly.
SOUTHERN RESEARCH INSTITUTE
-------
-237-
hammer swings downward, striking an anvil located on the end of a single
collecting electrode. The cam shaft can be located at any height along the
plate ends. Rapping control is limited to adjustment of operating time
and shaft speed.
Single impact (motor-driven swing hammers). This rapping system is
limited to collecting electrode cleaning of dry type horizontal flow preci-
pitators. The mechanism consists of a shaft running horizontally across
the precipitator between banks of collecting plates. The shaft is oscillated
by a motor-driven mechanical linkage. Hammer heads are connected to
the shaft by spring leaf arms. The hammers strike against anvils attached
to the ends of all collecting plates. The striking anvils are located near
the bottom of the plates. The rapping blow can be varied by adjusting the
arc of the hammer swing. Further control is obtained by operating time.
Single impact mechanical rappers. The mechanical arrangement for a
mechanically actuated impact-type rapper is illustrated in Figure 10. 11.
The system consists of a cam shaft running across the precipitator. The
cams located along the shaft engage the cam disc and raise the hammer
rods. The hammer rods are released when the cams reach the end of
their lobes and fall under gravitational forces, striking an anvil which
is attached to the support structure of the electrodes being rapped. The
system of rapping can be used on both discharge and collection electrodes.
Rapping control is limited to operating time and shaft speed.
«
A variation of the mechanical rapper system is the use of swing
hammers to provide the impact. These can be located on the sides or
ends of precipitator as well as on the top.
Vibrators (air). The major components of a typical vibrator consist
of a reciprocating piston in a sleeve type cylinder. The valve is of the
centrally located pin type. The vibrator assembly is fastened directly
to the end of a rapper rod which transmits the rapping energy to the
electrodes to be cleaned. This device is used for both discharge elec-
trode (employing an insulating section of rod) and collecting electrode
cleaning. This type of rapper is used for both horizontal and vertical
flow, dry type precipitators. For collecting electrodes, the vibrators
can be used for either top, side, or end rapping since they operate in any
position. For discharge electrode cleaning they are usually used as top
rappers. Control consists of varying the air pressure, the duration of
-------
-238-
Roller Lever Kicks Free
When it Passes This Line
Roller
Lever - Loose
On Shaft
Driver- Keyed
To Shaft
To Collection Plates
Figure 10.11.
Mechanical Single Impact Rapper. Plates Are
Raised and Dropped by the Action of the Rapper.
Impact of the Plates in the Guide Supplies
Rapping Force.
SOUTHERN RESEARCH INSTITUTE
-------
-239-
the rapping period, and the time elapsed between cleaning.
Water flush and liquid removal. Some electrostatic precipitators are used
to precipitate mists from the gas stream. Most of these applications are
found in chemical, petroleum, and paper plants. The mist droplets which
collect on electrodes run together to form larger droplets, and then run down
the electrode surface where they drain away. In some cases, water or other
liquid is introduced at the top of the electrode to improve mist removal.l
Tube electrodes are more adaptable than plates or rods for this application.2
Because these electrodes do not require rapping, they can be made thinner,
but may need to be fabricated from corrosion resistant material.
10.7 HOPPER SYSTEMS
Hoppers are used to collect and store dry precipitate which is removed
from the electrodes. The physical properties and quantity of the dust must
be considered when designing hoppers. Heaters are sometimes required to
prevent moisture condensation and caking of the dust. The outlet and slope
of the sides must allow the dust to be removed adequately. If hoppers are
allowed to overflow, collected dust will be reentrained in the gas thereby
greatly reducing precipitator efficiency. Baffles are frequently placed in
hoppers to minimize undesirable gas flows which may reentrain dusts.
If the precipitator system is operated with internal pressures less
than ambient atmospheric, then air inleakage through the hopper can cause
a reentrainment of the dust from the hoppers. Care must be taken to pro-
vide good seals around hopper doors and dust removal connections for
systems operating under partial vacuum conditions.
10. 8 DUST REMOVAL SYSTEMS
Container removal. This system is used on small installations collect-
ing dry material in a. hopper. The hoppers are usually of the conical or
pyramidal type. The system consists of placing a transportable container
below the hopper. The collected material stored in the hopper is transferred
to the container through a simple manual valve or slide gate. When filled,
the container is removed for emptying. In some instances the container is
embodied as part of a truck.
-------
-240-
Dry vacuum systems. In this system, dry bulk material is transferred
from a precipitator hopper to a transport pipe system which is under vacuum.
The material is metered from the hopper to the transport system through
automatic rotary feeder valves or dump valves. The system vacuum is
developed by an air pump. In order to maintain system fluidity, ambient
air or hopper gas is induced as a carrier. The pump discharges the dust
into a silo for storage.
Wet vacuum systems. In this system, dry dust is removed from a pre-
cipitator hopper into a transport pipe system which is maintained under
vacuum by a water aspirator. The collected dust or ash is metered from
the hopper into the transport system through automatic feeder valves or
dump valves. In order to keep the dust suspended in the gas carrier, am-
bient air or additional hopper gas is induced into the transport line. The
dry material being transported mixes with the water used for aspiration
and forms a slurry. From this point the water-dust mixture is run to
waste.
Screw conveyors. A screw conveyor system usually starts with an open
screw in the bottom of a trough type hopper which moves the dry dust to the
outside. At the turns in the system each screw run passes the dust on to
each successive screw by a gravity drop. The dust is moved on to a system
silo or directly to some mobile conveyance. A screw conveyor system is
also applicable with a conical or pyramidal type hopper. A rotary valve is
required when the system is operating under vacuum.
Scraper bottom. The precipitator hopper is a flat bottom pan. An end-
less belt type scraper moves the collected dust to one end where a screw
conveyor is located. The screw moves the dust out of the hopper. Once
outside, the dust is conveyed to some remote point by any form of system
such as container removal, vacuum, or screw.
SOUTHERN RESEARCH INSTITUTE
-------
-241-
CHAPTER 10
BIBLIOGRAPHY
1. Strauss, W., Industrial Gas Cleaning, Per gam on Press, Long Island
City, New York (1966).
2. White, H. J., Industrial Electrostatic Precipitation. Addison-Wesley,
Reading, Massachusetts (1963).
3. Rose, H. E. and Wood, A. J., An Introduction to Electrostatic.
Precipitation in Theory and Practice, Constable and Company,
London (1966).
-------
-242-
. CHAPTER 11
POWER SUPPLIES AND CONTROLS
The fundamental driving force that removes the dust from a gas
stream in electrostatic precipitators is that between an electric field and
a charged dust particle. Therefore, the electric field in the vicinity of
the collection electrode should be maintained as high as practical in order
to provide the maximum possible gas cleaning efficiency. The function of
the power supply and related control system (power pack) is to provide
optimum voltage and current conditions.
11.1 THE POWER SYSTEM
The power supply system consists of four components as shown in
Figure 11.1: a step-up transformer, a high voltage rectifier, a control
element, and a driving function or sensor for the control system. The
step-up transformer is required because the voltage levels for modern
precipitators can range from about 30 to 100 kV.
The high voltage rectifier is required to convert the high voltage
ac power to varying direct current to be compatible with the electrostatic
precipitator requirements.
One function of the control system is to vary the amplitude of the
dc voltage that is to be applied to the electrode system. This control
can be applied to either the primary or the secondary circuit in the
power supply, but it is customarily utilized in the primary or low voltage
side. The control system can be operated either manually or in one of
several automatic modes. Automatic control systems are typically in-
stalled in all modern commercial installations. A well-designed auto-
matic control system serves to maintain the voltage level at the optimum
value, even when the dust characteristics and concentration exhibit
dynamic behavior.
High voltage generation. Techniques for generating high voltage include
the use of frictional types of high voltage machines, such as the Van de
Graff generator, high frequency transformer, and the standard iron core
voltage transformer. Both the frictional type generator and the high fre-
quency transformers are too limited in current capacity for practical
SOUTHERN RESEARCH INSTITUTE
-------
LOW-VOLTAGE
POWER IN -»
CONTROL
ELEMENT
t '
MANUAL
CONTROL
STEP- UP
TRANSFORMER
AUTOMATIC
CONTROL
HI-VOLTAGE
RECTIFIER
SENSOR FOR
AUTOMATIC
CONTROL
ELECTRO-
ASTATIC l
JPRECIPITATOR ]
i
to
•*»•
CO
Figure 11.1. The Power Supply System for Modern Precipitators.
-------
-244-
commercial precipitators, hence, the conventional high voltage power trans-
former is universally used as a means for converting the normal power line
voltage to the values required for precipitator operation.
High voltage rectifiers. The high voltage rectifying equipment has passed
through several stages of evolution to reach its present stage of development.
The invention of the synchronous-mechanical rectifier in 1904 by Lemp1 and
utilized by Cottrell in 1907 was the first means for providing sufficient power
at high voltage so that electrostatic precipitation became feasible. In this
system, a high voltage alternating current is fed to a rotating spark gap
that is driven in synchronism with the applied ac power. The phasing-of the
spark timer is such that ac is converted to pulsating dc. A schematic dia-
gram of the mechanical rectifier, included primarily for historical purposes,
is shown in Figure 11. 2.
Point A lines up with Point 1 when Point 1 is at a positive voltage. As
the sine wave changes to where Point 1 becomes negative, the rotor turns to
align Point B with Point 1 so that the output always maintains a voltage of
the same polarity. It is imperative that the rotational speed of the spark gap
maintain synchronism with the driving voltage.
A
Shortly after the development of the synchronous rectifier, Lodge in
England patented the use of mercury vapor rectifiers. However, high volt-
age rectifier tubes were not used commercially until 1920.
The first nonrotating (static) power supply utilizing rectifier tubes was
installed in the early 1920's. Copper oxide rectifiers3 were utilized in about
1928. Selenium rectifiers were introduced about 1939. These devices, with
significantly smaller electrical losses, allowed the development of static
power supplies that were economically competitive with the mechanical unit.
A diagram of the static rectifying system is shown in Figure 11. 3 where
either vacuum tubes or diodes act as the rectifying element.
More recent rectifier developments include high-voltage, high-efficiency
vacuum tube and single-crystal silicon rectifiers. The silicon rectifiers are
currently the most widely used for new installations since they provide high
conversion efficiency and high reliability.
Voltage control. In the normal operating range, the transformer is linear,
so that the output voltage is directly proportional to the input voltage. The
output voltage can therefore be controlled by changing the voltage input to the
1 Refer to the bibliography for this chapter.
SOUTHERN RESEARCH INSTITUTE
-------
-245-
O
Low Voltage
ac Input
High Voltage
Transformer
Synchronous
Motor - 4 Pole
C\ \ _ Precipitato
i jl>-»- PVV./V O
High Voltag
dc _O utp ut
Note: Motor and transformer connected
to same electrical source.
Figure 11.2. Full-Wave Mechanical Rectifier H-V Power Supply.
-------
-246-
AC INPUT POWER
DC OUTPUT
Figure 11.3. Vacuum Tube Rectifier.
SOUTHERN RESEARCH INSTITUTE
-------
-247-
transformer. Some common techniques for voltage control include tapped
series dropping resistors, series rheostats, tapped transformer primaries,
variable inductances or solid state control devices. Some installations use
combinations of control elements where gross voltage control is accom-
plished by tapped devices and continuously variable devices such as rheo-
stats or inductances provide the fine control. Some of the more modern
control devices include saturable reactors (magnetic devices), silicon con-
trolled rectifiers (SCR), and thyristors (four-element, solid-state devices).
The reactive control elements in general lead to a more efficient power con-
version subsystem than do the resistive ones.
The current limiting function is necessary to avoid catastrophic fail-
ure of the power supply under arcing conditions. When an arc exists be-
tween the corona wire and the collection electrode, the ionized path con-
stitutes essentially a short circuit across the power supply. Without a
means for current limiting, excessive current will flow which will lead to
a catastrophic failure of the power supply. This protection can be provided
by a variety of devices including resistors, various combinations of reactive
elements or active devices in the control subsystem.
The signal to the control device determines the operational charac-
teristics of the power supply so long as the design of the overall system
is sufficient to provide the needed capability. This control signal can range
from a simple manual setting of a passive device (tapped element) to a dy-
namic control signal related to current, voltage, spark rate, or combina-
tion of these. The shortcomings of manual control became evident early in
the period of commercial application to precipitators. The volt age-current
characteristics of the precipitator are related to the dust concentration and
composition. Extremely heavy dust loads tend to quench the corona current
in the input section of the unit and high resistivity dusts lead to increased
voltage drops in the deposit layer, with the result that excessive sparking,
back corona, or simply poor collection efficiency occurs. Therefore, the
need for dynamic control becomes obvious.
Various combinations of the above parameters (voltage-current-
spark rate) have been utilized in power supply automatic controls, but the
predominant one currently used in this country is spark rate together with
a current limit. The spark rate control establishes the applied voltage at
a point where a fixed number of sparks occur per minute (typically 50-150
per corona section). The spark rate is a function of the applied voltage
for a given set of precipitator conditions. The higher the voltage, the
-------
-248-
greater the spark rate. As the spark rate increases, a greater percentage
of the input power is wasted in the spark current, and consequently less
useful power is applied to dust collection. Thus, the spark rate control
should be designed to maintain the precipitator at or near the optimum oper-
ating point. The optimum conditions correspond to the highest time average
voltage on the precipitator.
As the dust conditions and concentration change, sparking conditions
change. Specifically, a highly resistive dust can cause sparking to occur
at a lower than normal applied voltage (this is discussed in detail in the
section on particle collection). The automatic spark rate control adjusts
the applied voltage downward to bring the collector into a more efficient
operating range if high dust resistivity is encountered.
The spark rate control senses voltage or current surges caused by
the sparks. A schematic diagram of one type of spark rate control is
shown in Figure 11.4. The operation of the circuit is as follows: the cur-
rent surge in the primary circuit, associated with a spark in the secondary,
is sensed by the detector. This signal fires the thyraton V1 which discharges
capacitor Cj. When V^ is turned off, capacitor Cj charges through Rx apply-
ing a voltage across the series combination of R2 and C2- Thus, with con-
tinual sparking, a voltage is developed across C2 which is proportional to
the spark rate of the unit. C2 tends to discharge through resistor R3 gen-
erating a signal Vs on the plate of diode V2.
A reference signal V^ that opposes Vs is developed by voltage divider
R4 across battery B that is connected in series opposition to Vs. These
two voltages are compared at the diode where an increasing voltage signal
is developed across R5 until Vs just exceeds V]-, where the circuit estab-
lishes equilibrium. A detailed description of this circuit is given by Van
Hoe sen. 4
In the case where the dust conditions are such that no sparking
occurs, the voltage or current will increase up to the maximum capability
of the power supply. In this case it may be necessary to include a maxi-
mum voltage or current limiter to avoid damage to the power supply.
Sectionalization. It has been experimentally determined that the over-
all collection efficiency of a spark rate limited installation can be improved
by increasing the number of independently powered electrical sections
SOUTHERN RESEARCH INSTITUTE
-------
FROM
HI-VOLTAGE -
TRANSFORMER
SPARK
DETECTOR
B+
-wVVW.
T Y-
_L_L_-x' +
TO CONTROLLER
CO
Figure 11.4. Spark Rate Control Circuit from Van Hoesen, et al.
-------
-250-
(corona) in the precipitator.5
The spark rate in an electrostatic precipitator as given by White6
follows an exponential relationship as shown below.
S = expb (Vp - Vi) (11.1)
where
S = spark rate sparks/min,
b = empirical constant,
Vp = peak voltage, and
Vj = initiation voltage (voltage for one spark per minute).
Thus, as the applied voltage increases above V1; the spark rate
increases accordingly. This increase in voltage leads to two conflicting
effects. First, the increased voltage increases the charge and electric
field in the precipitator, which tends to increase the collection efficiency.
But, in contrast to this, a spark tends to short circuit the power supply
and reduce the collection efficiency for the duration of the spark. The
overall effect of these two factors is shown in Figure 11. 5, where the
maximum efficiency occurs at a spark rate of about 125 per min. This
discussion specifically applies to a precipitator operating from a single
power supply where one spark short circuits the entire collector unit.
Next, consider a similar situation where two independent power
supplies are used, each supplying only one half of the collection area.
The sparking condition for the entire precipitator will still follow the
same spark rate equation above, but a spark will now disable only one-
half of the precipitator. In this case, two sparks will be required to be
as detrimental as one was in the original example, which leads to an
effective spark rate of about 250 per minute. A further increase to ten
electrical sections will similarly raise the overall effective spark rate
to 1250. This increased spark rate will occur at a significantly higher
applied voltage than the original single section case.
A numerical example will serve to show the changes in applied
voltage to be expected. Consider the case where spark initiation occurs
at an applied voltage of 33 kV and a spark rate of 125 per min occurs
at an average voltage of 42 kV. Substitution into Equation 11. 1 yields
SOUTHERN RESEARCH INSTITUTE
-------
-251-
bO
cti
50
40
(8 >
30
20
Optimum Point
i 1 i i
0 100 200 300
No. of Sparks/min
Figure 11. 5. Optimum Operating Voltage for Single Section
of a Precipitator (Reference 7).
Req
-L Ceq
I
Figure 11. 6. Equivalent Circuit of the Precipitator
Collection Electrode System.
-------
-252-
S = expb (Vp - V;) (11.2)
125 = exp[D(42 - 33)] = exp (9b)
b = g- In 125 = 0.536
The voltage required for a spark rate of 250 per min yields a value for the
peak voltage of
VT ln s + vi
which for S = 250 yields
V =10.3 + 33 = 43.3 kV
and similarly for S = 1250
Vp = 13.3 +33 = 46.3 kV
Thus, the effect of the increased sectionalization is seen to lead to a signi-
ficant increase in the applied voltage, which generally leads to increased
collection efficiency.
In addition to the sparking effect, increased sectionalization tends to
isolate portions of the precipitator with mechanical defects such as broken
or misaligned wires, and misaligned collection electrodes or other factors
that tend to cause a reduction in the operating voltage of localized sections.
A further advantage of sectionalization of the precipitator power sup-
ply is the higher internal impedance associated with the smaller capacity
power supplies. Higher impedance tends to limit the current and minimize
the tendency for power arcs to develop.
The precipitator constitutes a capacitive load and the energy stored
in the precipitator depends upon the capacitance of the section and the volt-
age to which it is charged.
If a spark occurs within a precipitator, the energy stored in the pre-
cipitator due to its capacitance tends to sustain the spark for a given time.
However, the current drain during a spark is heavy relative to normal
SOUTHERN RESEARCH INSTITUTE
-------
-253-
current flow from the power supply, and the spark would tend to be quenched
when the energy from the precipitator has been dissipated in the spark.
However, if the power supply impedance is low, the power supply volt-
age can supply a sustaining value of current while the arc is maintained.
Consequently, the precipitator capacitance may not be completely discharged
during the interval that the power supply voltage reaches its minimum value.
In such case, the spark would not be extinguished and a condition known as
a power arc would develop. The current limit on the power supply normally
quenches a power arc; however, considerable disruption of the precipitator
occurs during the period of the arc. Low impedance power supplies are not
as effective in spark quenching and power arcs can develop more frequently
than in precipitators powered by high impedance supplies.
Waveform selection. Electrostatic precipitation theory shows that the
collection efficiency is fundamentally related to two factors: the charge on
the dust particles and the average value of the electric field in the collection
zone. The charge on the particles is related to the peak value of the elec-
tric field in the charging region (field charging). The basis for the selection
of the waveform should take these factors into consideration.
There are several choices of waveform including: pure dc, half-wave
and full-wave rectification and pulsed supplies. Pure dc is seldom, if ever,
used, primarily because of its poor spark quenching characteristics.
In general, half-wave power supplies provide a greater degree of sec-
tionalization, although the sections are not completely independent, since con-
trol is normally associated with the transformer primary. Full-wave recti-
fication is used where higher average currents are required, as for example,
where large dust loading or extremely fine particles lead to a large space
charge which limits current flow.
The effect of the voltage waveform on precipitator operation can be
seen from the equivalent circuit, Figure 11.6.
The dust layer constitutes a distributed capacitance, the magnitude of
which is given by Qj = e e0 where e = relative dielectric constant and e0 =
permittivity of free space and a resistance, given by the dust resistivity p .
The effect of this combination is to provide a filtering action to the buildup
of voltage which results in a delay in the voltage appearing across the dust
-------
-254-
layer. The higher the dust resistivity and the greater the dust capacitance,
the greater the delay in charging. Thus, for a time varying applied voltage,
the voltage appearing across the dust layer will in effect be filtered and will
not follow the swing of the applied voltage.
The parallel capacitance Ceq constitutes the distributed capacitance
of the precipitator due to the electrodes. The large area represented by
the collection surface and the discharge wires can lead to an appreciable
capacitance. If the wires are assumed to be a plate, the capacitance due
to the geometry can be calculated from
r - e°A
e(l " d
where £c = permittivity of free space, A = plate area, d = distance between
electrodes. For a 10 cm plate-to-wire spacing and 1500 m2 of plate area,
the capacitance would be
Ceq= 8. 85x10^x1500 = 0.126uF
In practice, the capacitance must be around half this value for the given
dimensions due to the effect of the discharge wires.
The equivalent resistance of the precipitator can be computed from
Req = T
For a precipitator of 1500 rn2 plate area, a current of 400 mA at an applied
voltage of 40 kV would be typical. For these conditions
R _40_kV_ = 100xl03ohms
eq 400 mA
This equivalent resistance is not a constant since the voltage-current rela-
tionships in the corona systems are not linear. However, as a final approx-
imation, the assumption of a constant resistance will serve to illustrate
the operating conditions.
The time required for the charging and discharging of the capacitance
represented by the equivalent circuit is determined by the value of the capa-
citance and resistance. The time constant can be computed from T = RC =
SOUTHERN RESEARCH INSTITUTE
-------
-255-
(0.126 x 10'6 x 100 x 103) = 12. 6 x 10'3 sec, which means that about 12. 6
milliseconds would be required for the charge in the capacitance to reach
63% of its final value if a steady dc voltage is applied.
Since the normal voltage source is a time varying one, the effective
voltage applied to the precipitator electrodes would be a function of the
applied voltage waveform and the precipitator impedance. Figure 11.7
shows the precipitator voltage resulting from a 60 Hz half-wave and full-
wave supply and a 500 Hz pulsed voltage supply. The waveforms were
computed on the basis of the equivalent circuit of Figure 11.6 and the values
of capacitance and resistance given above.
As indicated ip Figure 11.7, the capacitance of the precipitator main-
tains the applied voltage at some minimum value even though the applied
voltage would go to zero for a half cycle in the absence of the precipitator
capacitance.
The 500 Hz pulse power supply indicates the effect of the higher fre-
quency. Since charging current is applied more frequently, the resultant
precipitator voltage approaches that of a dc voltage.
Studies of the effect of waveform have shown that higher peak voltage
can be applied before sparkover if the voltage rise rate is rapid. It can be
shown from the equivalent circuit that the integrating effect of the dust
layer prevents the voltage across it from rising as rapidly as the applied
voltage. Consequently, since sparking is related to the electric field at
the dust surface, a higher peak voltage can be applied before sparking occurs.
In general, the value of the peak voltage determines the maximum
charge on a particle in the field charging mode. For optimum collection,
the product of the average and maximum field should be as high as possible.
This condition would be met by a rapid voltage rise rate.
However, the particle charge rate is also a very significant variable
in terms of precipitator performance. Pulse type power supplies must
provide sufficient pulse width for the particles to acquire their charge and
this requirement tends to limit the effectiveness of pulse power supplies.
The practical high voltage power supply. The practical aspects of power
supply requirements place additional requirements on the designer. The
-------
100
O
I
i
m
1
-i
PI
80
flS
u
0)
60
0}
40
20
0
500 pps
500 fi sec
Corona
* Quench
\ x \ Full Wave
•100 pps - \
\ SOOjusec .
\
4
1
I
14
16
2 4 0 3 U
Time, milliseconds
Figure 11.7. Time Response of the Precipitator Equivalent Circuit to Various Voltage Waveforms
With the Same Peak Amplitude.
to
m
OS
I
-------
-257-
various alternative techniques for fulfilling these requirements provide
some insight into the problems facing the designer of a precipitator power
supply.
H-V transformers
H-V transformers for precipitation rectifiers are conservatively
designed and ruggedly constructed to withstand high average ambient tem-
peratures and continuously occurring electrical transients due to precipi-
tator sparking.
The high voltage and power level requirements are best satisfied by
liquid-immersed, self-cooled designs. The insulating and cooling liquid
is usually transformer grade mineral oil. In some instances fire insur-
ance regulations dictate use of nonflammable oil (Askarel). The latter
material is expensive and requires special insulation system designs be-
cause of its electrical dielectric and solvent properties. Transformers
50 kVA and larger require extended cooling surfaces, in the form of con-
vectors, to maintain low temperature rise.
The shell-type, single-phase transformer core and coil arrangement
is normally used. Stacked cut core designs are common and in some in-
stances tape-wound (cut "c") cores are used. The L-V primary winding
is placed adjacent to the core and the H-V secondary winding is placed
over the primary winding. Though a single-layer-type secondary wind-
ing can be used, it is advantageous to use two coils, which can be arranged
to minimize electrical stress on the primary-secondary insulation.
Numerous oil ducts are included in both primary and secondary windings
to minimize thermal gradients and assure thorough impregnation. Surge
screens are used at both ends of the secondary winding for protection
against transient voltages. A grounded electrostatic shield between pri-
mary and secondary windings is useful for minimizing radio interference
caused by precipitator corona and sparking. The core and coil structures
also require adequate structural bracing to withstand mechanical stresses
and shocks induced by precipitator short circuits and sparkover.
Mechanical rectifiers
Figures 11.2 and 11.8 illustrate typical FW and HW rectifier arrange-
ments. The rotors are driven by an 1800 RPM synchronous motor and,
-------
Low Voltage
ac Input
in
O
c
1*1
2
a
High Voltage
Transformer
Precipitator
Synchronous Motor
4 Pole
High Voltage
dc Output
i
(S3
Ol
00
I
Note: Motor and transformer connected
to same electrical source.
Figure 11,8. Half-Wave Mechanical Rectifier H.-V Power Supply.
n
I
H
H
PI
-------
-259-
in effect, alternately switch H-V transformer connections to the precipitator
once each half-cycle. In the HW rectifier, the second rotor essentially dis-
tributes alternate pulses to two precipitator sections. Because of high tip
velocities, a small gap (~£") is needed between the rotor tips and arc shoes.
These gaps (4 in series for FW and 6 in series for HW) introduce substantial
voltage and power losses in the system, particularly when clearance increases
to -|" or -g" due to tip burning. The intense electrical discharge in these gaps
generates toxic gases, heat, and radio interference radiation. The former
are dissipated by special ventilating systems, and the latter by grounded en-
closures and radio interference chokes and filters. Quite often resistors (4K
to 10K) are placed in the H-V rectifier leads to reduce severity of precipitator
sparking transients.
These rectifiers are often driven with 3-phase reluctance-type syn-
chronous motors of approximately 1 HP. These motors are essentially
squirrel cage induction motors with salient poles milled in the rotor. Since
these rotors do not have fixed magnetic polarity, special polarity sensing
and transformer polarity reversing contractors are needed to insure nega-
tive H-V output. In the more modern designs permanent magnets were in-
corporated in the rotor structure. This change eliminated the special polar-
ity controls and improved motor performance.
Electron tube rectifiers
Figure 11. 9 illustrates the single-phase bridge commonly used for
FW and HW energization. It will be appreciated that electron tube rectifiers
offer considerable practical improvement over the mechanical rectifier in
regard to electrical efficiency, noise, toxic gases, lack of moving parts,
fixed polarity, and size.
The principal disadvantages are tube replacement costs, need for
filament power, and some X-ray radiation. X-ray radiation is seldom a
problem because it is soft and easily shielded by the steel rectifier en-
closure provided for weather and H-V hazard protection. The filament
power is provided by one or more small filament transformers which are
oil immersed with the H-V transformer.
Tungsten filament tubes provide tube life of from 10, 000 to 20, 000
hours, by operating at a filament voltage just sufficient to satisfy the peak
current requirements. Inherent current limitations of these tubes preclude
-------
High Voltage
Transformer
Low Voltage
ac Input
High Voltage
dc Output
01
o
X
m
Filament Circuit
ac Input
Filament Transformer
Precipitator
O
CO
o»
o
X
n
x
1
Figure 11.9. Tube Rectifier H-V Power Supply.
-------
-261-
their use in rectifiers larger than 750 mA if reasonably long life is to be
obtained. In most energization systems means are included to insure a one
minute tube filament warmup before application of H-V and quite often volt-
age stabilizers are provided to closely regulate filament voltage.
Silicon rectifiers
In 1956, application of silicon rectifier elements to precipitation rec-
tifiers quickly followed general commercial availability of the basic silicon
devices. Rectifier failures were common at first, but have been essentially
eliminated by appropriate design criteria and uniformity of rectifier junction
manufacture.
Series strings of rectifier junctions are used as arms in single-phase
bridge rectifier connection. Depending upon the characteristics of the rec-
tifier junctions, transient voltage distribution is accomplished along the
string by either resistance, capacitor, or res is tor-capacitor networks. With
development of the "avalanche diode, " it is sometimes possible to obtain
adequate transient voltage distribution solely through avalanche voltage limit-
ing without resorting to a compensating network. The rectifier legs are com-
posed of a sufficient number of 500 to 1200-V rectifier cells, each rated for
approximately 1A average current. These cells are mounted on insulating
boards with provision for simple connection in the circuit. It is general
practice to immerse the silicon rectifier assemblies with the H-V trans-
former. This provides a compact package and an ideal ambient for the H-V
elements. The insulating fluid can be either transformer oil or Askarel. In
case of Askarel, a rectifier failure generally produces sufficient contamina-
tion to also ruin the H-V transformer. In most cases, air core inductors are
included in the dc output leads of silicon rectifiers. These chokes serve to
reduce severity of the precipitator sparking transients on the rectifier assem-
bly by slowing voltage and current wavefronts.
H-V silicon rectifier assemblies have not been specified and typed
like the H-V rectifier tubes. This is largely due to differences in recti-
fier junction manufacture and specialty of the precipitation rectifier ap-
plication. As previously mentioned, several H-V rectifier modules are
available and are used to replace mechanical and tube rectifiers. In gen-
eral, these assemblies are for lower current sets, 500 mA or less.
-------
-262-
Description of L-V control components
Manually operated rheostats were used as a variable series impedance
means to vary energization levels in early electrical sets. In general, the
control range was quite limited and also directly related to precipitator cur-
rent requirements. This control situation was helped somewhat by use of
primary taps on the H"-V transformer and a tap switch which essentially
changed the transformation ratio.
Power rheostats similar to field rheostats with cast iron grids were
common. In some cases more modern strip or ribbon resistance elements
were used in conjunction with power tap switches to vary resistance. Suit-
able resistance values ranged from 1 to 8 ohms depending upon current levels
and power losses could run as high as 15 to 18 kW. The rheostats were con-
structed so that 20 to 40% of the total resistance was always in the circuit
as ballast. This ballast performed the essential function of spark quenching.
Tapped input autotransformer
Use of tapped input autotransformers provides true voltage control at
moderate costs. These arrangements often have a control range of ^50%
and permit relatively close adjustment by means of coarse and fine voltage
tap switches. Operation of the tap switches momentarily de-energizes the
system—generating transient electrical surges and often puffs from the pre-
cipitator. Ballast resistors are also incorporated for spark quenching.
Variable autotransformer
Variable autotransformers either manually or motor operated provide
true voltage control without the circuit interruption of tap switching. Motor
operated units also provide means to automatically raise and lower voltage
in accordance with relay controls. These assemblies are prone to commu-
tation difficulties, particularly when the brush remains stationary at one
point on the winding for long periods of sparking operation. The usual volt-
age control range is = 50% for precipitation rectifier control.
Induction regulator
The induction voltage regulator is essentially a variable transformer
wherein secondary voltage is varied by changing magnetic flux linkages
SOUTHERN RESEARCH INSTITUTE
-------
-263-
between primary and secondary windings. Its construction is similar to a
wound rotor induction motor with a blocked rotor whose angular position
can be varied to change magnetic coupling. It is usually connected in auto-
transformer fashion with the stator winding in shunt with the load and the
rotor winding in series with the load. These regulators are usually motor
operated, and hence adaptable to automatic control systems. Though
rugged, and the most expensive of the three voltage control means, these
regulators are also vulnerable to precipitator sparking. In this case,
sparking transients cause rapid wear of the gear drive train resulting in
noisy operation and jamming. The usual voltage control range is =* 50%.
Saturable reactor
The saturable core reactor provides a good means of control. Load
power is controlled in accord with the dc current supplied to the control
winding. Since control is accomplished by electrical signal, the saturable
reactor is well-adapted for continuously controlled automatic systems. It
is a static device and can be designed to provide long life under all sorts
of electrical abuse.
Both full-range and partial-range reactors are used for power control
purposes. The partial range reactor arrangements usually require fixed
resistance in the primary circuit to limit short-circuit currents. With high-
gain, full-range reactor designs, it is possible to combine the power control,
ballast, and automatic fault current limiting functions in a single device.
Saturable reactors are not well-suited for use with mechanical recti-
fiers because phase shift effects seriously affect the rectifier phasing.
They are also not well-suited for use in conjunction with monocyclic net-
works. High powered systems, 700 mA and larger, with saturable reactor
controls are sometimes adversely affected by H-V sparking under light
load. Basically, the spark transient unbalances core fluxes causing unsym-
metrical H-V waveforms, which induce additional sparks. In bad situations,
many cycles may be required to return to normal symmetrical operating
conditions.
Thyristor
Phase control with back-to-back thyristors (silicon controlled recti-
fiers) is an excellent means to control precipitator energization levels.
-------
-264-
Fast, full-range control is easily accomplished with negligible control
power and the system is easily adapted to automatic control. Control is
achieved by synchronous switching techniques and the switching devices
(SCR's) have low losses in the "on" and "off" states. Successful appli-
cation of thryristor phase control was realized in 1967. It is highly
contingent upon selection of reliable, adequate devices, proper firing
circuitry, and due consideration of electrical conditions in precipitator
energization. Use of thyristors in conjunction with silicon rectifiers
now permits electrical set conversion efficiencies (KWa x 100) of
76% at 1500 mA, 50 kV power levels. The theoretical maximum conver-
sion efficiency for a full-wave, single-phase rectifier is 81%, assuming
lossless components.
Ballast elements
The circuit ballast function is simply satisfied by the fixed resistors
of suitable rating. Resistors with ribbon or expanded metal grids prove
reliable and operate at reasonably low temperatures without hot spots.
The monocyclic network originally proposed by Steinmetz is an
excellent ballast element because it is an energy storage device. It is
simply a resonant bridge circuit which draws power from the source only
when load impedance is high. As such, it inherently limits power under
short circuit precipitator conditions. Unfortunately, monocyclic networks
are expensive to construct and they have the ability to severely overload
components in the event of an open-circuit condition.
The ac reactor is another good ballast element which will limit
fault currents quite efficiently. This is an inductive element, and some
care is needed in its application to make certain it does not introduce
high voltage transients into the system.
SOUTHERN RESEARCH INSTITUTE
-------
-265-
CHAPTER 11
BIBLIOGRAPHY
1. Lemp, H., U.S. Patent 774, 090 (1904).
2. Lodge, Sir Oliver, U.S. Patent 803, 180.
3. Stevens, S. A., etal., "High Voltage Supplies for Electrostatic
Precipitators, 1I Institute of Electrical Engineers, Colloquium on
Electrostatic Precipitators, London (1965).
4. Van Hoesen, H. E., "Automatic Control of Electrical Precipitators, "
Paper 58-229, AIEE Winter General Meeting (Feb. 2-7, 1958).
5. Ramsdell, Roger, 'iDesign Criteria for Precipitators, " American
Power Conference (April 23-25, 1968).
6. White, H. J., Industrial Electrostatic Precipitation, Addison-Wesley,
Reading, Mass. (1963).
-------
-266-
CHAPTER 12
MEASUREMENTS
12. 1 MEASUREMENT OF GAS FLOW AND EFFICIENCY
Standardized procedures have been established by various organi-
zations for the determination of the dust content of gases and the effi-
ciency of dust collection equipment. Many of the standards are either
based upon, or are similar to the ASME Performance Test Code PTC-
27, 1957. Table 12. 1 compares some specifications of six different
measurement standards.
The usual procedure in dust sampling is to determine the gas velocity
at a number of points in a cross section of the ductwork as in a standard
pitot tube traverse. The individual readings are used to determine the
required velocity for isokinetic sampling of the dust laden air. Alterna-
tively, a null type pitot tube may be used to obtain isokinetic sampling.
The mean velocity is used to calculate the total volume of the air per unit
time.
-------
Table 12.1
Comparison of Six Measurement Standards
Specification
Number of test
points
Location of test
points
Array of test points
(rectangular)
Array of test points
(circular)
Velocity measurement
Oust sampling
nozzle
Tubing
Filters
Sampling pump
Dust calculation
Other
Organization and Publication
A. S. M. E. »
PTC-27
12-20 if velocity
range less than 2: 1
and area more than
25ft1.
Double or treble if
In a straight run with
uniform velocity
Centroids of equal
areas
Centroids of quad-
rants of equal
area anuli
Average of pilot
tube traverse
Thin-edged isokinetic.
1/4-in. diameter
minimum
Smooth, cleanable,
short, airtight.
above dew point
99% collection
efficiency, tare
less than 20% of
dust weight
Adequate capacity
Sum over all
traverse points
Preliminary runs to
check instruments
and personnel
I. O. C. I. *
Pub No 1
V sq it area
Same as A. S. M. E.
Same as A. S. M. E.
Same as A. S. M. E.
Same as A. S. M. E.
Isokinetic if particles
larger than 5 microns.
Taper less than 15*
Same as A. S. M. E.
99% collection
efficiency
Same as A. S. M. E.
Same as A. S. M. E.
Same as A. S. M. E.
V. D. I. *
2086
Not specified
In a straight run at
least ±3 hydraulic
diameters long
Same as A. S. M. E.
Same as A. S. M. E.
Same as A. S. M. E.
Thin-walled isokinetic.
aim within ±5" of flow
Short, airtight,
corrosion resistant,
and no electric
charge
"complete capture"
Same as A. S. M. E.
Same as A. S. M. E.
Include estimate of
error
National Council for
Paper Industries
Same as A. S. M. E.
In a stable flow
pattern 8 to 10
diameters upstream
'and 3 to 5 diameters
downstream to
nearest disturbance
Same as A. S. M. E.
Same as A. S. M. E.
Same as A. S. M. E.
Isokinetic if particles
larger than S microns.
Aim within ±5° of
How.
Not specified
Paper, alundum,
membrane, glass
wool, wet scrubber,
or electrof liter
Not specified
Same as A. S. M. E.
5-10 mini ft
50-60 min total
time
Manufacturer No 1
Same as A. S. M. E.
la a straight run
with 10 times largest
dimension upstream
to nearest distur-
bance and with
s. p. il-in. w. e.
Same as A. S. M. E.
Same as A. S. M. E.
Same as A.S.M.E.
Isokinetic
Not specified
Paper, alundum.
asbestos, or tar
camera
Aspirator or numo
Same as A. S. M. E.
Sample at least one-
hour
Manufacturer No. 2
4 if less than 2 ft*.
12 if between 2 and
25 If.
20 if more than 25 ft .
In a straight run
with uniform velocity
and 8 diameters
upstream to the
nearest disturbance
Same as A. S. M. E.
Same as A. S. M. E.
Same as A. S. M. E.
Isokinetic
1/4-in. diameter
minimum
Smooth, corrosion
resistant
Paper, alundum,
glass cloth, membrane.
Impinger, or tar
camera
Same as A. S. M. E.
Maximum velocity
less than 1. 5 times
minimum
to
* A. S. M. E.: American Society of Mechanical Engineers.
I. O. C. I.: Industrial Gas Cleaning Institute.
V. D. I. Verein Deutscher Ingenieure (German Engineer Association).
-------
208
quadrants as shown in Figure 12. 1. For a rectangular cross section, the
sampling points are taken at the centroids of equaJ area rectangular sub-
divisions of the duct. The velocity measurements are made by measuring
the velocity pressure with a standard pitot tube. Special pitot tubes for
extreme dust conditions may be used provided proper calibration factors
are available.
It is important to select the location of the traverse carefully.
Locations which have a high degree of turbulence or nonuniformity are
undesirable, as are locations with very low velocities which would be
difficult to measure accurately. Two location are required, one before
the inlet to the dust collector and one after the outlet. The second
location is often chosen at the inlet to the induced draft fan where the
turbulence and nonuniformity are low and the velocity is high enough to
be measured with a reasonable degree of accuracy. The selection of a
point before the inlet to the dust collector is usually difficult, because the
gas velocity right at the inlet is rather low for accurate measurement,
and there is often a diffuser plate in the way. Any other spot upstream
from the dust collector inlet is often either turbulent, nonuniform, inaccess-
ible, or hot.
If it is necessary to run a pitot traverse close to the precipitator inlet,
the traverse location should be downstream from a diffuser plate at least
5-10 hole diameters or mesh lengths, and far enough upstream from the
electrostatic precipitator to prevent sparkover to the sampling probe.
(
12. 3 DETERMINATION OF DUST CONCENTRATION
Samples of dust laden air are taken isokinetically at measured time
intervals at the traverse points in the duct. The sample probe should be
inserted to the location of the first traverse point with the velocity pre-
adjusted to approximately the isokinetic rate. The probe should be pointed
downstream, and the flow cut off until ready to sample. At the start of
sampling, the probe is rotated into the flow and the sample flow started.
The sampling rate is then checked for isokinetic flow and adjusted if neces-
sary. Gas temperatures and pressures are recorded during the sampling
at each point. The probe is held at the sampling point for a predetermined
period of time, calculated to give a total dust collection sufficient for accu-
rate weighing, but not less than 10 minutes at each traverse point. A mini-
mum of about 100 milligrams of dust should be collected for accurate
weighing. At the proper time, the probe is moved quickly to the next
SOUTHERN RESEARCH INSTITUTE
-------
-269-
•/
2R2
.316R
r, « . 548R
N
r a radius to point p
N * no. points (20 in example)
p = index (1 to 5 in example)
R = radius of duct
Figure 12.1. Sampling Points in a Round Duct.
-------
-270-
traverse point and the process is repeated. The total run should cover the
circuit of all traverse points twice. The weight of dust collected on the
filter medium divided by the total sampling volume is then the mean weight
per unit volume, or the mean concentration of dust in the gas stream. This
mean concentration times the mean gas volume per unit time gives the total
weight per unit time in the gas stream. Test runs made in the dust collector
inlet and in the outlet give the required data to calculate the collection ef-
ficiency:
W, - W,
T, = - i - ?_
where
Wj = weight per unit time- dust collected at inlet, and
W2 = weight per unit time- dust collected at outlet.
The sampling line includes the sampling probe, the filter holder,
possibly a condenser if the gas is humid or a cooler if the gas is hot, a
gas meter, a pump, and the necessary instrumentation to measure
temperature, humidity, and pressure for the calculation of gas density.
An Orsat apparatus or other gas analysis instrumentation may be used if
information about the constituents of the gas is required, such as CO and
CO2 for carbon balance calculations or SO2 and SO3 for information on the
effect of sulfur on resistivity.
The sample probe may be of a null pressure type so that isokinetic
sampling can be achieved without reference to a previous velocity traverse.
The filter medium may be one of several types, according to the
requirements of the particular situation:
1. Paper filter for dry gas at temperatures below 300°F.
2. Glass cloth filter for coarse dusts and temperatures to 750°F.
3. Alundum thimble for high temperature and chemical resistance.
4. Membrane filter for fine dusts to sizes below 1 micron.
5. Impinger trains for particle size distribution of insoluble dusts.
6. Electrostatic electrofilter for high efficiency at low sampling rates.
7. Tar camera (colorimetric filter for nonaqueous mists).
The filter selected should resist the conditions of temperature, pressure,
and chemical attack to which it will be exposed, and should have a collec-
tion efficiency of greater than 99%. The filter must be protected against
SOUTHERN RESEARCH INSTITUTE
-------
-271-
condensation during the test run. If the filter assembly is part of the
sample probe, it should be held inactive in the gas stream long enough for
it to warm up to the gas temperature so that condensation will not occur.
The following items are usually measured or calculated in testing
dust collection equipment:
1. Gas volume
a. Main stream - pitot traverse or stoichiometric calculation.
b. Sample stream - gas meter, flowmeter or calibrated orifice.
c. Corrections calculated for density.
2. Humidity - psychrometric dry bulb and wet bulb temperatures or
cooled condensate.
3. Gas composition - Orsat or other gas analysis instrumentation.
4. Gas density - calculated from composition, temperature, humidity,
and pressure.
5. Gas pressure - manometer and barometer.
6. Dust concentration - weight of sample collected by isokinetically
sampled traverse.
7. Collection efficiency - calculated from concentrations at inlet and
outlet of collector.
8. Dust properties - determined by analysis of dust sample (see ASME
PTC-28, 1965).
a. Size distribution - calculate from Stokes' law and terminal
velocity distribution (ASME PTC-28, 1965).
b. Specific gravity - ASTM C188-44: change in volume of kero-
sene or naptha with addition of dry sample.
c. Moisture content - weight loss in drying oven.
d. Water soluble content - weight loss after dissolving in water.
e. Water soluble sulfate content - chemical analysis of solute.
f. Resistivity - resistance of 5-mm layer of dust in a 3-in.
diameter cup through a 1-in. electrode weighted to
10 gm/cm .
-------
g-
-272-
Specific surface - ASTM C204-55: calculate from air per-
meability.
Figure 12. 2 shows one schematic arrangement of dust sampling
apparatus for testing dust collectors. Other arrangements are possible,
depending on the instrumentation available. Two setups are required,
one upstream from the dust collector and one downstream. It would be ,.
desirable to run both upstream and downstream traverses simultaneously
to insure exactly the same operating conditions for both runs. However,
standards usually require only that the operating conditions be maintained
constant.
12.4 COMPUTATION
The computation of results of dust collector tests requires careful
attention to all corrections for temperature, humidity, density, and instru-
ment calibration.
The gas velocity may be calculated from the pitot tube readings with
necessary corrections. A standard pitot tube reads the impact, or velocity,
pressure directly with no correction for calibration. All other types of
velocity pressure tubes must be calibrated and the data corrected. The pitot
tube velocity at each traverse point is calculated by the formula:
V = 1096.5
where
P
d
Density =
ft/min
the velocity pressure in inches water gage, and
the density of the gas calculated from the composition,
temperature, and humidity of the gas.
144p.
R (T, + 460) 2. 036
pounds per cubic foot,
where
PS =
T, =
static pressure, inches of mercury,
dry bulb temperature °F, and
SOUTHERN RESEARCH INSTITUTE
-------
-273-
Wall of Flue
-+. To
Exhaust
or Or sat
Apparatus
(1) Sample probe and pitot tube
(2) Thermocouple
(3) Filter holder
(4) Temperature readout
(5) Inclined manometer
(6) Flow meter
(7) Humidity measurement
(8) Pump or blower
(9) Flow control valve
(10) Cooler or condensate trap if required
Figure 12. 2. Schematic Arrangement of Dust Sampling Apparatus.
-------
-274-
R = gas constant calculated for composition of gas:
R = 53. 3 for standard air, and
R = 52.4 for 12% CO2 flue gas.
Note that turbulent fluctuations in the gas flow show up as variations
in the velocity pressure and that the velocity varies as the square root of
the pressure. If there are wide variations in velocity pressure readings,
averaging the velocity pressures will not represent the average velocity.
Therefore, pitot traverses should be made at locations where the turbu-
lence is relatively small. The pitot tube manometer becomes unreliable
when the velocity is below about 10 ft/sec (600 ft/min). Therefore, pitot
traverses should be made at locations where the velocity is around 30-50
ft/sec (1800-3000 ft/min).
The total gas flow may be calculated either from the integration of
the pitot tube traverse or from the stoichiometric relationships of com-
bustion. The standard calculation for the mean velocity from a pitot tube
traverse is simply the arithmetic mean of all the traverse points:
and the total gas flow is determined from the continuity equation:
Q = Avm
where A is the actual measured cross-section area at the test section
including the effect of errors in construction, warping of plates, and dust
deposits. (Refer to the section on sources of error for comments on the
use of the arithmetic mean velocity).
To determine the total gas flow from stoichiometric considerations,
it is desirable to have the ultimate analysis of the fuel being burned, the
operating conditions of the furnace, and an analysis of the flue gas com-
position. The total gas flow in cfm is given by
WCT
Q = __ Cfm
SOUTHERN RESEARCH INSTITUTE
-------
-275-
where
Wc = the weight of fuel burned per hour, and
T = the volume of gas produced per pound of fuel.
The weight of fuel per hour may be obtained by weighing or cal-
culating from
Q ^T7l _ IT* \
o \& - n, )
W = — V—
where
S = the rate of steam generation,
E = the enthalpy of steam,
S
EW = the enthalpy of feed water,
n = the boiler efficiency, and
H = the heating value of fuel.
To determine T, the volume of gas produced per pound of fuel, a
material balance may be made by following a known and detectable chemi-
cal element throughout the system. Common examples of such elements
are carbon and sulfur. The amounts of carbon and sulfur in the fuel are
known from the chemical analysis of the fuel and the amounts of CO2 and
SO2 in the exhaust gases are known from the tests on the gas with the Orsat
apparatus or other analytical procedures. Trace elements can also be
used.
The sample gas flow rate may be calculated from standard orifice
meter formulae if orifices are used to meter the sample flow, or may be
calculated from the calibration curves of whatever flow meter is used.
Because immediate adjustment of sample flow to isokinetic sampling is
required, curves or tables are required in the field for making these cal-
culations rapidly.
i
The dust concentration by weight is determined as the ratio of the
weight of dust to the weight of gas:
wr
m =
Wg
-------
-276-
where
n
wr = weight of dust = S w^ = sum of traverse dust samples, and
i
... - , , total volume of gas sampled .
w = weight of gas sampled = - —
g density
The collection efficiency can be calculated from the dust concen-
tration in the inlet and outlet to the collector:
n -
where
m = mass concentration, outlet, and
m^ = mass concentration, inlet.
A similar calculation using dust concentration by volume may be made,
provided that the volumes are corrected to the same basis, that is, at the
same temperature and pressure. Because this would represent equivalent
densities, the calculation would be identical. Also, a similar calculation
may be made using dust collected per unit time, provided the installation
is operating under steady-state conditions. Because steady-state conditions
imply the same weight and volume of gas per unit time, the calculation
would again give an identical result.
12. 5 SOURCES OF ERROR
Errors may occur in measurement, or in calculation, and also may be
intrinsic due to random variations. Table 12.2 lists the sources of measure-
ment errors. Some are unavoidable and irreducible, but some can be
avoided or reduced to a small value.
Isokinetic sampling is essential to accurate sampling in order to
obtain representative samples of all particle sizes. If the velocity into the
sampling nozzle is greater than the velocity of the gas stream, the larger
particles will be under-represented because they do not accelerate with the
gas. If the velocity into the sampling nozzle is lower than the velocity of the
gas stream, the larger particles will be over-represented because they ram
SOUTHERN RESEARCH INSTITUTE
-------
-277-
Table 12.2
Sources of Measurement Errors
1. Errors due to probe
a. Anisokinetic sampling velocity
b. Probe not parallel to gas flow
c. Dust collected on walls or probe and sample line
d. Condensation in probe or sample line
2. Errors due to sample collector
a. Efficiency less than 100%
b. Condensation in collector
3. Errors in sample flow
a. Meter inaccuracy
b. Leaks
4. Errors in weighing
a. Hygroscopic loss or regain of sample or filter media
b. Loss of sample in handling
c. Scale accuracy
5. Errors in integrating
a. Variations over time or space of dust load or velocity
b. Integration error
c. Too few traverse points
6. Errors due to personnel
a. Reading and recording errors
b. Failure to comply with requirements
c. Calculation errors
-------
-278-
into the nozzles where smaller particles detour around the streamlines.
Figure 12.3 shows the flow lines in the vicinity of a sampling probe. It can
be seen that a representative sample will be taken only when the sample flow
is parallel to and equal in magnitude to the main gas flow—that is, isokinetic.
It is difficult to obtain precise isokinetic sampling in the case of velocity
variations, so some unavoidable error may occur.
A relationship between the accuracy of anisokinetic sampling and the
ratio of sampling velocity to approach velocity has been developed by Ruping.*
Figure 12.4 shows his curves of 1% and 5% error plotted on a semilog grid
of sampling velocity ratio versus the dimensionless parameter:
k =
ws wo
where
R - the radius of sampling probe,
wg = the settling velocity,
w = the approach velocity, and
g = the acceleration of gravity.
For all possible values of the dimensionless parameter, it is necessary
to hold the sampling velocity within about ±2% of the isokinetic velocity for a
1% error, or within about ±6% of isokinetic velocity for a 5% error.
For a particular situation, the required range need not be quite as
restricted. For example, for a 1/4-in. I. D. probe, an approach velocity of
8 ft/sec, and a 10-micron particulate matter of unit specific gravity,
k = 0. 13; and the allowable range of sampling velocity would be about +25%
to -12% of isokinetic velocity for 5% accuracy. However, in a practical
situation, a wide range of particle sizes will usually exist, hence a wide
range of values of the parameter will result, and a narrow range of sampling
velocities about isokinetic velocity will be required.
Isokinetic sampling also requires that the probe be parallel to the gas
flow at all times. In a turbulent flow, this may not be possible. Standard
methods usually call for aiming to within ±5° of parallel to the main gas
flow. Note that the cosine of 5° is 0.996, so that the reduction in cross-
section of the nozzle facing the gas stream will be only 0. 4% at that aiming
angle.
1Refer to the bibliography for this chapter.
SOUTHERN RESEARCH INSTITUTE
-------
I
to
-3
(£>
i
Local velocity greater
than sampling velocity
Sampling velocity' greater
than local velocity
Isokinetic
sampling
Figure 12. 3. Flow Lines in the Vicinity of a Sampling Probe.
-------
1.4
01
o
c
z
m
71
n
x
z
3
H
n
i
to
oo
o
0.2
0.5
10
Dimensionless Parameter 8&
wswo
Figure 12. 4. Accuracy of Anisokinetic Sampling.
-------
-281-
Although the pitot tube itself is usually considered to be error free and
to have a calibration factor of unity, the manometer used to read the velocity
pressure is not free of error, especially if the gas flow is turbulent so that
the manometer reading fluctuates. A slant tube manometer with draft gage
oil of specific gravity 0. 826 may have divisions down to 0. 01 or 0. 02 inch
of water. Thus, the accuracy of reading the manometer will be about +0.01
inch water gage with constant gas velocity. Figure 12. 5 shows the range
of apparent velocities for manometer accuracies of 0.01-in. w. g. and
0. 1-in. w. g. Note that the limits diverge rapidly at low velocities. At
densities different from that for standard air, the curves shown will shift
slightly, but the shape will be similar. Also note that turbulent variations
in velocity pressures on the order of +0. 1 in. w. g. can introduce appreci-
able error.
Errors due to the sample collector depend on the efficiency of col-
lection for the size particles in the gas stream. Because most of the mass
occurs in the larger particles, a small loss of mass in the smaller particles
will not necessarily be important. Standards call for 99% efficiency of the
sample collector.
Errors in measurement of the gas volume sampled depend upon the
accuracy of the meter and its calibration.
Errors in temperature readings depend upon the calibration of the
thermocouples, resistance thermometers or glass thermometers used,
random errors due to temperature variations, and heat losses due to heat
conduction through the thermometer stem or lead-in wires. In thermo-
couples, there is the added possibility of a deviation in the temperature of
the cold junction. Table 12. 3 shows the error limits for mercury thermom-
eters for various scale divisions and temperature ranges.2
Errors in weighing depend upon the scale accuracy and the size of
dust sample collected with respect to the tare weight of the filter. Care
must be taken to avoid or compensate for hygroscopic loss or gain. Care
must also be taken to include with the dust sample any dust deposited in
the sampling nozzle and sampling tube. A weight and collection efficiency
change can result from the paper thimble becoming charred if the gas
temperature exceeds 300°F.
The existence of an error in the calculation of the mean value of
either velocity or dust concentration is attributed in the literature to the
-------
-282-
3000
1000
•S 800
B
£ 500
o
£ 400
300
200
100
—a = 0,01 in.
X
/
a=0.
a = -0. 01 in. —
1 in
/
z
JL
a= -0. 1 in.
•y-
0 .01
0. 02 0. 03 0. 04 0. 05 0. 06 0. 08 0.10 0. 2 0. 3 O. 4 0. 5 0. 6
Manometer Reading, inches water gage
Figure 12. 5. Accuracy of Velocity Readings for Manometer Accuracy of
+ a Inches Water Gage and Standard Air Density.
SOUTHERN RESEARCH INSTITUTE
-------
-283-
Table 12.3
Permissible Error Limits for Glass/Mercury Thermometers
Range. *C
From
-58
0
50
100
200
300
400
515
To
0
50
100
200
300
400
515
700
More than 700
Error Limit ± "C for Scale Divisions of:
.01
-
.02
-
-
-
-
-
_
-
.02
-
.04
.04
-
-
-
-
-
-
.05
-
Oil
0.15
-
-
-
_
-
-
0.1
0.3
0.15
0.25
-
-
-
-
-
-
0.2
0.4
0.2
0.3
0.5
-
-
-
-
-
0.5
0.7
0.5
0.5
1
1.5
-
-
-
-
1
1
0.7
1
1.5
2
2.5
3
-
-
2
2
1
1.5
2
3
4
5
6
-
5 or 10
3
2.5
3
4
5
7
10
10
10
-------
-284-
ranclom variations in time and space of the variable being measured.
However, there is an additional systematic error due to the fact that the
value of the velocity is zero at the wall and the present calculation tech-
niques do not take this into account.
If the variable to be measured were perfectly random, a simple
average of all the points would be a satisfactory estimate of the mean
value. However, dust concentration and velocity are not perfectly random.
The velocity in the core of the gas flow will be some average gas flow plus
and minus turbulence, but near the walls, the turbulence will be damped out
and viscous friction will diminish the flow velocity all the way to zero at
the walls . The dust concentration in the core of the flow will be rapidly
mixed by turbulence to some mean concentration plus and minus both time
and space variation, but near the walls, the mixing effect will diminish
with decreasing turbulence so that dust concentration can be expected to be
lower near the walls, possibly close to zero. With large dust particles in
the gas stream, the settling velocities may overcome the mixing effect of
turbulence and a concentration stratification may occur with a heavy load
of larger particles near the bottom of the duct.
The mathematical definition of a mean value of a variable over an
area is as follows:
where
y = variable,
A = area, and
b, a = limits of A.
The problem of determining the mean velocity or the mean dust con-
centration is, therefore, a problem in numerical integration from a limited
number of traverse points. If the function "y" were a simple linear function,
the mean would be an arithmetic mean, or a simple average:
n
A.M. = JL S yi
ni = 1
SOUTHERN RESEARCH INSTITUTE
-------
-285-
If the variable were completely random, the arithmetic mean would
be satisfactory. Because the gas flow is zero at the walls, the velocity
readings at all traverse points located adjacent to a wall will be higher than
the local mean and an average velocity calculated from the arithmetic mean
will be 5% to 15% high. For example, suppose a traverse were made in a
square duct with 25 traverse points and every point had a velocity of 100.
Also suppose, for simplicity of calculation, that the boundary layer is para-
bolic and has a thickness equal to half the traverse spacing. An average
velocity using the arithmetic mean would obviously be simply 100. However,
by integration, a true mean velocity can be calculated.
/•
/
i t
100 1
1
1
1
1
100 1
r
i
i
i
100 '
1
1
1
100 1
1
1
k\
Velocity Profile
By elementary calculus, the mean ordinate of a parabola is 2/3 of the
maximum. Therefore, the edge points of the traverse would have a true
value as follows:
66.6
I
o
100
100 + 66.66
= 83.33
Also by elementary calculus, it can be shown that the intersecting
parabolas at the corners have a mean of 1/2 the maximum. Then the
corner points would have a true value as follows:
50 I 66.6
I
66.6 | 100
I
50 + 66.66 + 66.66 + 100 = 1? x i00 = 70. 83
4 24
-------
-286-
Then the corrected traverse would be as follows:
70.83
83.3
83.3
83.3
70.83
83.3
100
100
100
83.3
83.3
100
100
100
83.3
83.3
100
100
100
83.3
70.83
83.3
83.3
83.3
70.83
Calculated Mean Values
Then the overall average would equal
4 x 70. 83 + 12 x 83.33 + 9 x 100
25
2183.28
25
= 87.33
or 12. 66% less than the arithmetic mean. Also, duct, dimensions may not
be known with required accuracy because of construction errors and dis-
tortions of flues due to warping, temperature effects, dust buildup, etc.
These may be common sources of error.
Obviously, the result would be different if the boundary layer thickness
were different, but it is apparent that the cook-book use of the arithmetic
mean in calculating an average value can lead to a large systematic error in
the calculation of the average velocity in a rectangular duct. Research may
be required to determine the relationship of a number of traverse points,
boundary layer thickness, and accuracy. Where a high degree of accuracy
is required, it is recommended that as many traverse points as is economi-
cally possible be used and that the boundary layer thickness be measured
for calculation of the mean value as shown above. The foregoing remarks
do not strictly apply to a pitot traverse in a round duct, because only four
points will be in error in a standard equal annular area traverse, instead
of the 16 erroneous points in the above example.
The distribution of dust in the duct work at the measuring point may
be very nonuniform because the distribution can be distorted by transforms,
elbows, turning vanes, and simple settling. Therefore, the required num-
ber of traverse points will depend on the degree of uniformity. However, a
relatively inaccurate measure of dust concentration at the inlet to the pre-
cipitator will not greatly decrease the accuracy of the efficiency calculation.
SOUTHERN RESEARCH INSTITUTE
-------
-287-
Suppose the concentration at the inlet is about 100 times the concentration at
the outlet, both measured with ±10% accuracy. Then the minimum and
maximum efficiency calculations would be as follows:
loo- 10
= 0-989
^ = 1 - 100 + 10 = * " TTO" = l ~°-009 = °'991
Thus r? = 0. 99 ± 0.001.
Therefore, in efficiency calculations, the greatest accuracy is '
required in the determination of the outlet dust loading. Fortunately, the
outlet can usually be measured at a point with fairly uniform air flow,
fairly uniform dust distribution, and with velocities high enough for
accurate readings of isokinetic sampling velocities.
It is theoretically impossible to obtain a truly representative sample
of any aerosol with a wide range of particle sizes because of the differences
in the trajectories of the different sizes. Large particles tend to settle
out, but small particles tend to follow the gas flow. For example, the
terminal velocities of spherical particles with unit specific gravity are
approximately as follows:
Size, microns _ Terminal Velocity _
0. 1 8 x 10" cm /sec or 2-1/2 inches /day
1 3 x 10~3cm/sec or 4-1/2 inches/hour
10 0.3 cm /sec or 6-1/2 inches /min
100 25 cm /sec or 10 inches /sec
1000 400 cm /sec or 150 inches /sec
Thus, a 1- micron particle in a 250 cm /sec wind (8. 2 ft/sec or 5. 6
mph) is moving almost horizontally: arc tan 3 x 10"3 cm/sec = 0° 0' 2. 5 in.
250 cm/sec
arc tan 0. 000012, but a 100- micron particle in the same wind has an appreci-
able trajectory angle with respect to horizontal: arc tan 25 cm/sec _
250 cm /sec
5°40'. Therefore, from a theoretical point of view, it is not possible to
-------
-288-
aim a sampling probe at the correct angle for all particle sizes. However,
from a practical point of view, an aiming error of only 5° is inconsequential.
At extremely low gas velocities, however, the trajectory angle for large
particles gets disastrously large.
No comment on errors due to personnel needs to be made other than
to recommend the use of trained, experienced, reliable personnel for
testing.
The total effect of all errors combined may be serious when the col-
lection efficiency is low. Although there are many sources of error, and
care must be taken to minimize their effect, the efficiency calculation is
not very sensitive to these errors when the efficiency is high. If we assume
the error in measurement is ±a for both inlet and outlet, we can derive the
following relationship between true efficiency versus calculated efficiency.
"calc = 1 + l (H - 1)
Figure 12. 6 shows this relation plotted on an inverted log (1 - x) by
log (1 - y) scale for measurement errors of ±10%, ±20%, and ±50%. Note
that even a 20% measurement error at a calculated efficiency of 99. 5%
only represents a range of true efficiencies from 99. 25 to 99. 67%.
However, if the emission from a stack is specified, then the impor-
tant factor is the accuracy with which one can measure the dust concentration
at the outlet and the measurement errors apply directly.
12. 6 MEASUREMENT OF RESISTIVITY
Since resistivity plays such an important role in governing precipitator
performance, it is important that it be measured with sufficient accuracy so
that its influence on precipitator performance can be determined.
There are a variety of methods utilized for determining resistivity.
They are classified as laboratory measurements if the dust is extracted from
the duct and measurement subsequently made in the laboratory, and in- situ
if the resistivity measurement is made in the duct or in the presence of the
gaseous atmosphere of the duct.
SOUTHERN RESEARCH INSTITUTE
-------
-289
99.9
99.8
U
S
w
s-,
H
Calculated Efficiency
Figure 12.6. Relationship between True Efficiency Vs. Calculated Efficiency.
-------
-290-
Laboratory techniques vary in the manner in which the resistivity is
determined and in which the dust layer is deposited. The simplest method
of resistivity measurement is to utilize a pair of plates which serve as the
electrodes and measure the current and voltage while maintaining the cell at
the temperature corresponding to that in the duct.
Figure 12. 7 shows a high voltage conductivity cell that is used for
laboratory measurement. The cell is normally mounted in an oven and its
temperature controlled over the range of interest. Dust can be placed on
the plate manually, or it can be electrostatically deposited by increasing the
voltage on the point electrode so that a corona is formed. Once the deposit
is formed, a plate is lowered onto the dust surface and the resistivity
determined from the current-voltage readings.
Because of the dependence of resistivity on humidity as well as temp-
erature, other types of laboratory measurement equipment are arranged to
introduce moisture and perhaps other conditioning agents in a recirculating
system. Dust is introduced into the system and, after sufficient time for
equilibrium, is precipitated onto the plate of a high voltage conductivity cell.
Resistivity is then measured by lowering a disc onto the dust surface, and
the voltage and current measured for resistivity calculations.
Laboratory measurements have the inherent disadvantage that the
atmosphere to which the dust is exposed during measurement is different
from that in operation; also, the dust may have undergone chemical and
physical changes after removal from the original source. Consequently, the
resistivities determined by laboratory techniques can be in error by several
orders of magnitude in the temperature range where surface conduction pre-
dominates. Figure 12.8 shows resistivity data taken on the same dust by
laboratory and in-situ techniques. The values agree well in the temperature
range where volume conduction predominates, but the laboratory data are
considerably higher in the low temperature region. The disagreement between
laboratory and in-situ data depends upon the nature of the dust and gas.
Agreement can be good under some conditions, but very poor in others.
In-situ resistivity measurements are made under conditions in which
the only atmosphere to which the dust is exposed is that present in the pro-
cess. This eliminates uncertainties in attempting to duplicate the gaseous
conditions and generally gives more reliable and reproducible results than
those obtained by laboratory techniques.
SOUTHERN RESEARCH INSTITUTE
-------
-291-
Scale, in.
Li; i^u^^jk^^yaczz
> I!
11 ji
"'"•gEJf
^B
1!
liMl
III)
Figure 12.7. High Voltage Cell for Measurement of Dust
Resistivity.
-------
-292-
In-situ Tests
Laboratory
Tests
F . 1667
J «1.56-10
\
Bayswater Specimens
New Vale Specimens
= 1667 -*- / \
cm
Temperature, °C
150
200 250 300 350
Temperature, °F
400
450
500
Figure 12.8. Resistivity-Temperature Relations
(Reference 3).
SOUTHERN RESEARCH INSTITUTE
-------
-293-
Figure 12.9 is a schematic diagram of a typical in-situ resistivity probe.
The apparatus consists of a point-plane cell at the end of the probe, which is
inserted into the duct through openings of from 2-1/2" to 4" diameter. When
a voltage sufficient to generate a corona is applied to the cell, dust is pre-
cipitated onto the plate. After sufficient time has elapsed to collect a dust
layer, the point-plane disc is lowered onto the dust layer and the current
and voltage are measured. The probe is then removed and the dust thickness
measured. Resistivity can then be calculated from the dimensions of the
probe, thickness of the dust layer, and the electrical data. Thermocouples
on the probe are used to measure temperature.
Figure 12. 10 is another type of in-situ resistivity probe. It differs from
the first type discussed in that the dust and gas are drawn from the duct
into a small cyclone separator where the dust is separated and deposited
into a resistivity cell. The cell is composed of a central electrode and an
outer cylindrical electrode on the inside of the cyclone dust hopper. The
collected dust forms an annular layer between the two electrodes, and the
resistivity is determined from resistance measurements.
The two types of resistivity measuring instruments differ in the size
particle collected, and in the manner in which the deposit is formed. No
data have been found which show comparative resistivity values measured by
the two methods.
Cohen4 reports that for some types of dust, the resistivity varies with
aging time after collection. Figure 12.11 shows that for a particular dust the
resistivity measured 10 ohm-cm immediately on collection and rose to 10
after 20 minutes. Laboratory resistivity values are shown on the same curve,
and the data indicate that the in-situ resistivities approach the laboratory
values after about 30 minutes. It is postulated that the change of resistivity
with time is due to a change of the conducting film by chemical and/or physical
interaction with the bulk of the dust particles. If so, it is interesting to note
that the resistivity of the collected dust would also change with time, so that
resistivity would be a dynamic rather than a static property.
Since the techniques for resistivity measurements are not standardized,
the data should be interpreted cautiously, and the method used in the measure-
ments identified. It should be especially noted that critical resistivity values
in terms of precipitator operation should be identified with the method of mea-
surement, since the values can vary by as much as two orders of magnitude.
-------
Thermocouple Mater
High Voltage
Connection (
n ^
fTf\ (
"^ ^
•
Connection
I ^
V
Point
^
vn
Disc
ww\ ^n
-Tl
Plate
s
Hxxxxxx>coooo^
Point- Plane Cell
Microammeter
Connection
Li
Wvv
O _ __» _, 1 C .,
t
k>3.n)piirig
Hole to
CD
1
F/v/w
1 '
Adapter
Flange
Fiberglas String
8
2
m
a
z
a
E
i
n
X
z
5
-
'^
'
Flue
Wall
Figure 12.9. Resistivity Probe.
-t
-------
-295-
F: Exhaust from Cyclone
V
E: Thermometer
G: Inlet from
Sampling^
Probe
Hr^Connection to
Megohmeter
A: Cyclone
B: Resistivity Cell
C: Heater
D: Rapper
Figure 12.10. Diagram of Apparatus.
-------
-296-
10'
02 4 6 8 10 12
Time, min
Figure 12.11. Resistivity Changes During Site
Measurements on Four Different
Power Stations. (Reference 4)
SOUTHERN RESEARCH INSTITUTE
-------
-297-
CHAPTER 12
BIBLIOGRAPHY
1. Verein Deutscher Ingenieure, VDI-2066 - Standards. "Performance and
Measurements at Dust Collectors" (May 1966).
2. Darby, K. H. and Heinrich, D. O., "Conditioning of Boiler Flue
Gases for Improving Efficiency of Electrofilters," Staub 26, No. 11,
pp 12-17 (1966).
3. Tassicker, O. J., Herceg, A. and McLean, K., "The Electrical
Resistivity of Fly-Ash from Bayswater and Newvale," Bulletin No. 11,
University of New South Wales (Dec. 1966).
4. Cohen, L. and Dickinson, R., "The Measurement of the Resistivity
of Power Station Dust," J. Sci. Inst. 40, p 72 (1963).
-------
-298-
CHAPTER 13
TROUBLESHOOTING AND MAINTENANCE
When an electrostatic precipitator fails to achieve its design efficiency,
it is necessary to examine the causes for its poor performance and take cor-
rective action where possible. The number of causes for poor precipitator
performance is so large that it is impractical to establish a rigid procedure
for determining the exact problem. Rather it is a task of reviewing all
aspects of the precipitator operation and arriving at possible trouble based
on operating data and physical examination. Diagnosis of the problems and
corrective action requires a good understanding of precipitation theory, as
well as practical experience in precipitator operation.
Precipitator problems can arise either when the precipitator is brought
on line, or after an extended period of operation. In the latter case, the
difficulties may be associated with improper maintenance, changes in gas
flow due to dust accumulation, misalignment or warping of the electrodes, or
changes in the dust properties due to changes in the fuel or process param-
eters. In the former case problems are more likely to result from poor
gas distribution, inadequate area of collecting surface, or inadequate ener-
gization.
Causes for precipitator problems can be classified generally as follows:
1) Electrical
2) Gas Flow
3) Rapping
4) Mechanical
Since an electrostatic precipitator operates on the basis of electric
field and electric charge, it follows that the electrical energization must be
adequate to provide for particle charging, maintenance of the electric-field,
and for holding the collected dust to the collection plates. Field tests show
a close correlation between efficiency and useful corona power. Figure 13.1
is a plot of efficiency of fly ash precipitators vs. corona power. This
curve is useful as a guide to determining whether existing electrical
energization is sufficient to achieve the desired efficiency.
SOUTHERN RESEARCH INSTITUTE
-------
-299-
99
S
0)
W
o
o
o
U
98
97
96
95
94
93
92
91
90
80
70
60
50
40
30
20 _
10 L
0
7
0
25 50 75 100
Corona Power, watts/1000 cfm
125
150
Figure 13.1 Relationship between Corona Power and Efficiency for
Fly Ash Precipitators.
-------
-300-
There are several possible causes for inability to achieve the required
level of power input. The most common ones are
1) High dust resistivity
2) Excessive dust accumulations on the electrodes
3) Unusually fine particle size
4) Inadequate power supply range
5) Inadequate sectionalization
6) Improper rectifier and control operation
7) Misalignment of the electrodes
Normal power supply operation is indicated by voltages in the proper
range as determined by the precipitator electrode spacing. Typical values
for these currents range from 10 to 30 mA per 1000 sq ft of collecting
surface. For most applications, the power supply output is controlled by
the spark rate and is adjusted to give around 10 to 100 sparks per min per
section. The setting of the spark rate should be such as to give maximum
average high tension voltage, usually resulting in spark rates as shown
above.
If a precipitator is operating in a spark rate limited condition but with
low current and voltage, the problem commonly can be traced to high resis-
tivity dust, electrode misalignment, or uneven corona due to buildup on the
discharge electrode.
Because of the importance of resistivity in the precipitation process,
in-situ resistivity measurements should be made as one of the first steps in
troubleshooting. If the resistivity is found to be high (in excess of 1010 ohm-
cm) most of the difficulty may be due to this cause. If this is not the case,
other potential causes of abnormally low currents should be investigated.
High dust resistivity affects precipitator efficiency principally by
limiting the voltage and current at which the precipitator operates. If the
precipitator electrodes are clean, the high tension voltage can be increased
until a sparking condition is reached. The maximum voltage is determined
principally by the gas composition and precipitator dimensions.
If dust is deposited on the collection electrode, the voltage at which
sparking occurs is decreased due to the increased electric field at the dust
surfade; If the resistivity of the dust layer is increased, the voltage at
which sparking occurs will be further reduced. Finally, at very high
SOUTHERN RESEARCH INSTITUTE
-------
-301-
values of dust resistivity, the voltage will be reduced sufficiently so that
sparks will not propagate across the interelectrode space. Under these
conditions, the gas in the interstitial regions of the dust layer will break
down, resulting in a back corona. The positive ions, resulting from this
corona,How toward the discharge electrode and neutralize the negative
charge previously applied to the dust particles.
Back corona results in an increase in current at low voltage and is
manifested visibly as a diffuse glow at the surface of the dust layer. Visual
verification of the presence of back corona is usually very difficult; how-
ever, it can be observed under very dark conditions.
Corrective procedures for precipitators that are limited by high
resistivity include gas conditioning or alterations in temperature to reduce
the resistivity of the dust. Some improvements can also be made by
increased sectionalization.
If a precipitator is operating in a spark-limited mode with abnor-
mally low voltage on dust with resistivities less than 1010 ohm-cm, the
problems are likely to be associated with misalignment of the electrodes,
uneven deposits on the discharge wire or broken corona wires.
Occasionally, precipitators are encountered that operate at the maxi-
mum voltage or current settings on the power supply with no sparking. This
condition is likely to be associated with the collection of low resistivity dusts,
where the electric field in the deposit is insufficient to initiate sparking.
These installations are referred to as "power hogs. " The fact that the
precipitator is not sparking does not necessarily mean that the unit is under-
powered. These installations may have sufficient power to provide adequate
charging and collection electric fields without sparking. If low efficiencies
are encountered and tests show that sufficient power is provided, then other
disruptive conditions should be sought.
Other electrical problems encountered with precipitators are shorting
of the high tension frame by dust accumulation in the hoppers, broken wires,
insulator or "flower pot" bushing leakage, and broken down or leaky cables.
13.1 GAS FLOW
Uneven gas velocity distribution in a precipitator can reduce the efficiency
several ways. First, because of the exponential relationship between gas
-------
-302-
velocity and efficiency, an uneven gas distribution flow results in a lower
efficiency than would be predicted from the average. Second, uneven gas
How can cause scouring or erosion of the dust from the collection plates,
thereby reducing efficiency. Third, poor gas flow patterns can result in
dust fallout and accumulation on turning vanes and in the duct work. Such
accumulations can further interfere with gas flow, and cause decreased per-
formance. Plugging of distribution plates is also a commonly encountered
problem which disrupts gas flow.
Evidence of poor gas flow quality (uniformity of flow) is obtained from
a gas velocity distribution profile at the precipitator inlet. The IGCI recom-
mends a gas quality such that 85% of the local velocities is within 25% of the
mean with no single reading more than±40% from the mean. This is the
minimum requirement for good precipitation.
Other types of gas flow problems commonly experienced are air
inleakage into hoppers from the dust conveyor system and "gas sneakage"
(see below) through the upper portion of the precipitator around the high
tension frame. Air inleakage through the conveyor system results in
reentrainment of collected dust which is carried back into the precipitator.
Air inleakage can also occur through the precipitator shell or inlet flanges,
if the precipitator operates at less than atmospheric pressure. Often suf-
ficient air is bled into the system to cause intense sparking.
"Gas sneakage" is the term used to describe gas flow that consti-
tutes a by-pass around the effective precipitator section. It can occur
through dead passages of the precipitator above the collector plates, around
the high tension frame, or through the hoppers. The gas sneakage problem
usually accounts for only a few percent drop in efficiency unless it is
exceptionally severe. The problem can be identified by measurement of
gas flow in the suspected areas in a nonoperating or cold test. Corrective
measures usually involve baffling to direct the gas into the active precipi-
tator section.
Problems of reentrainment of dust from the hoppers by virtue of
air inleakage or gas sneakage can often be identified by an increase in dust
concentration at the bottom of the exit to the precipitator.
13.2 RAPPING
Rapping requirements for various types of dusts are discussed in the
section on rapping and reentrainment. Unfortunately, there are inadequate
data from which to determine rapping requirements based on variations in
SOUTHERN RESEARCH INSTITUTE
-------
-303-
dust resistivity, current density and other dust conditions.
Rapping is needed on both the discharge and collection electrodes.
The rapping forces, can be either too mild or too severe and result in poor
performance. Removal of the collected dust is best achieved when the
intensity and frequency of rapping is such that the dust falls from the col-
lection plates in sheets;! hence too intense and frequent rapping can result
in a clean plate with the collected dust reentrained rather than falling into
the hopper. On the other hand, inadequate rapping can result in an exces-
sive thickness of deposit on both collection and discharge electrodes, which
can lead to other disruptive effects.
Inadequate rapping of the discharge electrodes can result in heavy
dust build up with a localization of the corona, low corona current,and ex-
cessive sparking.
The presence of a deposit on the discharge electrodes can be a result
of several factors, including poor gas flow and the conditions of the dust.
Often deposits of up to two inches diameter can form. These deposits or
donuts are generally composed of the finer dust particles and often cling
tenaciously to the discharge wire. Deposits on the discharge wires do not
necessarily result in poor performance, although depending on resistivity,
power supply range, and uniformity of the deposit, it can constitute a source
of reduced efficiency.
Rapping of both collection and discharge electrodes is only effective
if the force is transmitted to the respective electrodes. Variations in the
design of the support structure and in the electrodes themselves can also
result in inadequate rapping. Recent investigations of rapping acceleration
in fly ash precipitators have shown measured accelerations of approxi-
mately 5g's on the plates when accelerations of SOg's may be required.
If practical, measurement of the rapping acceleration with accelerometer
mounted on the plates can be used to check rapping adequacy. A common
method of adjusting rappers is with the use of an optical dust measuring
instrument in the precipitator exit gas stream.
The discharge electrodes should be kept as clean as practical. The
rapping intensity in this case is limited only by mechanical damage to the
electrodes and support structure.
-------
-304-
13. 3 HOPPERS AND ASH REMOVAL
Hoppers and ash removal systems often constitute problems in pre-
cipitator operation. If the hoppers become full, the precipitator may
become short circuited by the collected dust. Often the power through
the dust is sufficient to fuse the dust, and a large clinker-type structure
called a "hornet's nest" is formed. This further interferes with ash removal
and must be removed for proper operation. Most problems associated
with the hoppers are in providing for proper flow of the dust. Improper
adjustment of the hopper vibrators or failure of the conveyor system are
the usual causes of the failure to empty the hoppers. In some instances it
may be necessary to provide heat and/or thermal insulation for the hoppers
to avoid moisture condensation and a related cementing of the collected dust.
Air leakage and gas sneakage as related to hoppers have been pre-
viously discussed.
13.4 TROUBLESHOOTING TECHNIQUES
The procedures used in locating precipitator troubles are principally
those of measurement of those quantities that can be readily determined
and a visual inspection and observation of the precipitator itself. Depending
upon the severity of the problem, the difficulty will be more or less obvious.
For example, if a precipitator is designed for 90-95% efficiency and is
operating at 50-60%, the difficulties are gross ones. In contrast, the
problems in increasing efficiency from 90-95% may involve subtle changes
or improvements.
Gross difficulties with precipitators are usually those associated
with inadequate electrical energization or excessive reentrainment. The
following techniques are generally useful in pinpointing the causes of
gross precipitator problems.
1) Measurement of high tension voltage, current, and
spark rate.
2) Measurement of gas flow distribution.
3) Observation of the collecting plates for evidence of back
corona.
SOUTHERN RESEARCH INSTITUTE
-------
-305-
4) Oscilloscope record of the high tension voltage to determine
the duration of the corona current.
5) Direct observation of the collection plate for evidence of
scouring or excessive reentrainment. (This requires
construction of a glass plate and wiper for an access port
and a means for illumination of the interelectrode space.)
6) Measurement of the dust distribution at the precipitator
exit.
7) Careful examination of the general alignment and the con-
ditions of the hoppers, insulators, and other components.
8) Measurement of the rapping accelerations.
9) Observation of the corona wire and the uniformity of the
corona tufts.
13.5 MAINTENANCE
Maintenance of precipitators falls into the categories of preventive
maintenance and maintenance to correct failures of some part of the system.
Preventive maintenance schedules should be established to conform to the
requirements for the particular installation. A typical maintenance schedule
is given in Table 13.1 for a fly ash precipitator. The items covered will give
an idea of the types of maintenance that need to be performed on a weekly to
monthly basis.
Problems with failure of precipitator components are usually asso-
ciated with power supply, wire breakage, failure of structural parts due to
corrosion, fatigue, or rapper failure.
Power supply failures should be handled by only those personnel com-
petent to service high voltage equipment. The types of failure that may be
encountered are: (1) malfunction of the control circuit, (2) rectifier failure,
or (3) transformer failure. Conventional high voltage servicing procedures
should be used to identify the particular power supply problem.
The majority of the precipitators now in service utilize vacuum tube
or solid-state rectifiers. Normal vacuum tube life ranges from 12, 000 to
-------
-306-
Table 13. 1
Typical Maintenance Schedule
A. Annual Inspection
1. Internal inspection
a. Observe dust deposits on collecting plates and wire before
cleaning - a j" deposit is normal. If metal plates are clean,
there is a possibility that a section is shorting out. If more
than ^-" of dust is on plates, rappers are not cleaning.
b. Observe dust buildup on wires.
c. Interior corrosion - corrosion could indicate air leak through
shell, or could indicate moisture carryover from air heater
washer.
d. Plate corrosion adjacent to door or near bottom of plate could
indicate inleakage through doors.
e. Check plates for alignment and equal spacing between plates.
f- Measure to see that discharge wires hang midway between
plates.
g. Check for and replace broken wires.
2. Hopper inspection
a. Check for dust buildup in upper corners of hoppers.
b. Check anti-sway insulators to see that they are cleaned and
not cracked.
c. Check high tension weights - if one has dropped 3", this
indicates broken wire.
d. Check hopper bottom and valve for debris.
SOUTHERN RESEARCH INSTITUTE
-------
-307-
Table 13.1 - Maintenance Schedule (Continued)
3. Penthouse inspection
a. Check for corrosion due to condensation and/or leakage of
flue gas into housing.
b. Excessive dust in penthouse indicates air sealing pressure too
low.
c. Clean all high tension insulators.
d. Check that all high tension connections are secure.
e. Check that collars on high tension vibrator insulators are
secure.
4. Transformer inspection
a. Check liquid level.
b. Clean high tension line, insulators, bushings, and terminals.
c. Check surge arresters, spark gap should be ^5".
5. Control cabinet inspection
a. Clean and dress relay contacts.
B. Rappers and Vibrators Checked Quarterly
1. Rappers
a. Check distributor switch contacts for wear and lubricate.
b. Clean dust, dirt, and moisture from cabinet.
c. Check rapper assembly for binding at plunger or misalignment.
-------
-308-
Table 13.1 - Maintenance Schedule (Continued)
2. Vibrators
a. Check contacts on load cams to see that they are clean.
b. Clean dirt, dust, and moisture from cabinet.
c. Check vibrators to see that they operate at proper intervals.
C. Checks to Be Made Each Shift
1. Electrical reading for each control unit should be recorded and
checked for abnormal readings.
2. Rapper controls should be checked to see that they operate.
3. Vibrator controls should be checked.
SOUTHERN RESEARCH INSTITUTE
-------
.309-
20, 000 hours, and servicing usually involves only replacement of defective
rectifier tubes. Most solid state rectifiers are relatively trouble free and
maintenance is not normally required. Transformer oil should be periodically
inspected to determine its condition, however transformers generally last
the life of the precipitator.
Insulators and flower pot bushings which hold the high tension frame
must be kept clean and dry to prevent arcing. This is normally done by
providing a source of clean, dry gas to the insulator compartment. Should
the dusty, moist gas from the precipitator get into the insulator compartment,
arcing would occur and ultimately result in the failure of the insulator.
One of the most common difficulties with electrostatic precipitators
of the suspended wire electrode type is wire failure. Depending on the
location of the break, the wire can fall in such a way as to short circuit
an entire section and it must be removed before the section can be reenergized.
Wire failures can be caused by corrosion, excessive localized
sparking or mechanical fatigue. It is often difficult to determine the exact
cause of failure; the best procedure is to keep account of the location of
the failed wire in the precipitator and the position of the break along the
wire to determine if the failures are random or associated with a particular
location in the unit. Corrosion failures generally occur around cooler
areas where condensation can take place. Excessive air bled in around
insulators can cause condensation leading to failure of the wires near
the insulators. Electrical sparking in localized areas is also a common
cause of failure. Normal sparking is a statistical phenomenon and the
location of the spark can be at any position along the wire and in any wire.
However, localized areas of high field strength can confine sparking to a
particular area, and this will ultimately lead to failure at that position.
Common faults leading to this type failure are misalignment of the elec-
trodes or improper shielding around edges of collection plates. These
conditions call for realignment of the plates, use of shrouds to reduce the
field at critical locations, or removal of sources that might cause localized
high fields.
Mechanical failure can occur because of movement of the wire under
the influence of aerodynamic and electrical forces. If a wire is moved
from the center of the field by an uneven gas flow, an electrical force
will develop causing the wire to move further in the same direction. This
movement could result in fatigue failure of the wires.
-------
-310-
When it is recognized that a typical precipitator contains approxi-
mately 10, 000-30, 000 feet of wire, it is reasonable to expect some wire
failure.
There have been some attempts to establish a normal failure rate
in terms of the number of wire failures per year per 1, 000 ft of wire;
however, there is a wide variation in experience depending on application,
and no standard or normal failure rate has been established.
Most industrial gases are corrosive under proper temperature
conditions, and corrosion must be considered as a problem in the design
and maintenance of electrostatic precipitators. This can be handled by
the use of corrosion-resistant materials or by maintaining the tempera-
ture of the critical components above the dew point. Use of corrosion-
resistant materials is practiced for precipitators used in the collection
of sulfuric acid fume, in cement kiln precipitators operating at low tempera-
tures, in some precipitators collecting dust from black liquor recovery
boilers, and in some fly ash precipitators operating with high sulfur coal
at low temperatures.
Because of the trend to lower exit gas temperatures in electric power
generating stations, problems of corrosion have become more acute in
that application. One of the major corrosion problems is that of the pre-
cipitator shell. In order to reduce corrosion to a minimum, it is custom-
ary to provide good insulation on the shell to keep its temperature as high
as practical. This technique has been adequate within the range of tempera-
tures and sulfur contents used.
In addition to the shell, corrosion can also occur at other locations.
Air inleakage at joints can cause severe localized corrosion. The major
consideration as far as maintenance is concerned is to maintain all
corrosion-susceptible surfaces above the dew point and to prevent air
inleakage into vulnerable areas of the precipitator.
SOUTHERN RESEARCH INSTITUTE
-------
-311-
CHAPTEM 14
ELECTROSTATIC AUGMENTATION AND UNUSUAL
PRECIPITATION DESIGNS
Augmentation of particulate control devices by superposition of an elec-
tric field or by charging of the dusts has been studied to varying degrees.
Also, the use of methods of establishing an electric field by space charge
and alternative methods of particle charging have been subjects of continuing
interest.
Electrostatic augmentation has been applied to fabric filters, packed
bed filters, wet scrubbers, and mechanical collectors, using both conventional
and alternative methods of establishing an electric field and particle charging.
The following discussion is limited to an examination of the concepts
involved in electrostatically augmented particulate control devices and in
unusual precipitator designs, as opposed to a critique of the implementation.
14.1 WET SCRUBBERS
The efficiency of wet particulate scrubbers is generally limited by the
ability to capture the dust particles by the water droplets. Conventional high
efficiency scrubbers are operated with relatively large pressure drops to
achieve a high relative velocity between the dust particle and the water drop-
let. This high velocity is necessary to propel the dust particle across the
boundary layer and to overcome the surface tension of the water drop so that
the particle will be captured.. Fine dusts require a very large pressure drop,
and hence high energy requirements, because the mechanism that transports
the dust particle through the boundary layer is primarily inertial,
One concept of electrical augmentation of wet scrubbers is to charge
the water and dust with an opposite polarity in an attempt to utilize the
attractive electrical forces to bring the dust and water droplets together.1
An analysis of the attractive forces involved between a 0. 5ju diameter
dust particle and lOjj, water droplet exhibits a variation as shown in
iRefer to the bibliography for this chapter.
SOUTHERN RESEARCH INSTITUTE
-------
312
Figure 14.1. The particles and water droplets are considered to be oppo-
sitely charged to saturation in a relatively high electric field of 6 kV/cm.
The force equation from electrostatics is
F = q • q2/47re0r2 Newtons
The saturation charge for the dust and water droplet is 78 and 31, 000 elec-
tronic charges, respectively. The attractive force as a function of separa-
tion distance is
F = 5- 65 * I0"" Newtons
r2
when the separation distance is measured in meters. The graph shows the
attractive force for separation distances from 10 to lOOju. At lOjm, the
dust particle will be within the boundary layer of the air around the water
droplet and they will contact. At distances greater than 100^, the force
becomes negligible.
A reference force for the gravitational attractive force acting on a
50ju water droplet is included which causes this water droplet to fall toward
the earth with a settling velocity of about 10 cm/sec. This suggests that
the effect of the electrostatic attractive force is rather small, except for
very close approaches.
The electrical force causes an increase in the relative velocity of
the dust particle and the water droplet of only about 3. 0 cm/sec for the
example considered. While this is a rather small change in the relative
velocity, the fact that the attractive force always exists will tend to increase
the collection efficiency of the very small particles somewhat. Since this
effect is rather small, it seems that it would be more desirable to install
an electrostatic precipitator initially than to include both positive and nega-
tive charging fields in front of the high energy wet scrubber.
Another technique for augmentation of wet scrubbers is through the
use of an electrostatic precipitator as an agglomerator to modify the par-
ticle size distribution of the dust to the scrubber.2 Since the collection
efficiency for scrubbers is greater for large particles, the reentrained
black liquor fume from a paper mill recovery system is more easily col-
lected after agglomeration. The tests reported were from a pilot scale
test.
-------
-313-
10"
10
01
u
o
n
•1-4
I 10-12
•a
t*
0)
1—I
V
o
• o
TJ
C
a)
to
C
o
+•»
•s
(a
0)
o
fc
(1)
u
rt
Gravitational Attraction for a 50jU Water Droplet
•3.0 —
10
-13
102i
-0.3 —
10
-14
10g
0.03
0
20
40 60 80
Separation Distance, fj.
100
H
m
i
3
I-'-
D
P)
o
o
i-*.
r+
<
o
o
03
r+
»-*•
O
3
i
o
1^
CO
o
Figure 14.1.
The Electrostatic Force between a 10u Water Droplet and
a 0. 5jU Dust Particle Charged to Saturation in an Electric
Field of 6 kV/cm.
SOUTHERN RESEARCH INSTITUTE
-------
-314-
14. 2 ELECTROSTATICALLY AUGMENTED PACKED BED FILTERS
A packed bed filter is a particulate emission control device that is
comprised of a column of loosely packed fibers, gravel or other filter ma-
terial. The dust laden gases flow through this filter column to provide the
opportunity for the dust particles to be collected on the filter medium. These
packed bed filters characteristically provide rather poor efficiency for small
particles.
The collection efficiency for this type of collector can be increased by
the use of electrostatic augmentation. If the filtration region is immersed
in an electrostatic field, the dust particles will be driven in a direction that
tends to increase the probability of impact between the particles and the fil-
ter medium.
If the dust particles are initially uncharged, the electric field will in-
duce an electrical dipole in each particle which tends to align each particle
with the electric field.3'4 This tends to orient the particle in such a way
that the negative end of the dipole contacts the positive surface charge of
the filter medium, and similarly, the positive ends tend to contact the nega-
tive surface charge of the medium (see Figure 14.1). The result of this is
that the collection efficiency is increased somewhat by the attractive force
between the dust particles and the filter medium.
The filtration efficiency of packed bed filters can be further enhanced
if the dust particles are electrically charged prior to introduction in the col-
lector. If charged particles are introduced into the filter medium with a
superimposed electric field, the dust particles are driven in the direction
of the electric field. Thus, the probability of collection and the retentive
forces for the dust particles are both increased by the addition of the elec-
trostatic augmentation.
Particle collection theory for turbulent flow electrostatic precipita-
tion shows that particles are captured only in the boundary layer of the
gas stream adjacent to the collection electrodes. In conventional electro-
static precipitators, the interelectrode space is largely inactive for par-
ticle collection. This problem is significantly reduced in the augmented
packed filter bed collector since the entire surface of the filter medium
acts as the collection surface. This results in a significant increase in
the collection area for a given volume of collector in comparison to that for
a conventional precipitator.
-------
-315-
The advantage of the loose packed filter bed is the low pressure drop
required to maintain the gas flow. It seems probable from the above consid-
erations that a reasonably high collection efficiency, even for the small par-
ticles, can be attained with electrostatic augmentation.
An obvious disadvantage to the electrostatically augmented bed filter
is the inability of the system to be rapped for dust removal. The dust col-
lector would require back flushing or filter media recycling for cleaning.
14.3 FABRIC FILTER
The collection efficiency and pressure drop in a fabric filter are
related to the open area in the fabric and dust layer. As dust builds up
in the fabric, the open area decreases leading to an increased collection
efficiency and pressure drop. The efficiency of a fabric filter is essen-
tially constant for a given dust and the resultant pressure drop is depen-
dent on the gas flow rate.
Electrostatic augmentation of a fabric filter can take the form of
charging the dust particles or establishing an electrostatic field across
the dust layer, or both. The action of the electrostatic field is to provide
a force between the dust particle and the dust layer such that openings
appear to be smaller. Particles that would otherwise pass through the
dust layer are therefore trapped, resulting in a somewhat higher efficien-
cy and pressure drop.
Only limited studies have been made of electrostatic augmentation of
fabric filters; the principal advantage noted has been a decrease in the time
required to establish a dust layer and hence less time required to reach
ultimate efficiency. Other studies have largely been inconclusive.
14.4 MECHANICAL COLLECTION
Consideration has been given to electrostatic augmentation of mechan-
ical collection; however, since the mechanical collectors are basically of
high gas flow type, there would appear to be little advantage to the addition
of electrostatic augmentation since:
SOUTHERN RESEARCH INSTITUTE
-------
-316-
(1) the area of collecting surface would he small relative l.o
the gas flow, hence in the standard efficiency equation,
the contribution of the electrostatic field to the efficiency
would be small; and
(2) the tendency toward reentrainment would be large.
14. 5 SPACE CHARGE PRECIPITATION
Conventional electrostatic precipitator theory shows that the migration
velocity of charged particles suspended in an electric field is proportional
to the product of their charge and the value of the electric field as given by
w = qEp/6na7] (14.1)
where q is the charge on the particle in Coulombs, E is the value of the
electric field in the collection zone in volts per meter, a is the radius of
the particle in meters, and TJ is the viscosity of the gas in kg-meter-second.
This basic equation is valid for the particle velocity independent of the me-
chanisms used for either providing the charge or establishing the electric
field.
The space charge precipitator5'6'7 utilizes the distributed space charge
of the charged particles, sometimes augmented by additional space charges
provided by the introduction of charged water droplets as the source of the
collection electric field. Conventional electrostatic precipitators of either
the single-stage or two-stage type utilize collection electrodes connected to
high voltage power supplies in addition to this space charge to provide the
collection field.
The motivation for the development of space charge precipitators is
to simplify the electrical requirements in the collection zone of the instal-
lation. Space charge precipitation is not a new concept in precipitation.
The space charge has always contributed to the electric field in the collec-
tion zone; but the removal of that component of the collection field caused
by the high voltage power supply is a new embodiment. However, it should
be pointed out that the value of the electric field in the collection zone will
always be smaller for the space charge collector than would result from a
conventional precipitator of the same dimensions if all other factors remain
-------
-31T-
the same. The removal of the electrical contributions from the power sup-
ply will result in a reduction in the collection fiel-d. Since the migration ve-
locity is proportional to the collection field, this results in a decrease in the
migration velocity. Therefore, it must be understood that the purpose for
pursuing the development of space charge precipitators is to simplify the de-
sign of precipitators, not to improve their performance.
Two factors can be used to promote the development of space charge
precipitators. The first is the absence of back corona in the collection zone
and the second is the reduction of reentrainment when charged water droplets
are utilized as part of the space charge. But it must be pointed out that these
factors are not peculiar to space charge collectors. It is common knowledge
that wet wall precipitators reduce the reentrainment of collected particles
and that back corona is related to the current density and resistivity of the
dust particles on the collection electrodes. Therefore, there are methods
other than space charge precipitation available for the solution of these
specialized problems.
The electric field in the collection zone of a wire and pipe precipitator
as determined from conventional theory is
J* (14. 2)
where E(r) is the electric field value as a function of the radius r; r0 is the
radius of the corona glow region; i is the current per unit length of the
corona wire; p and pj are the respective space charges, total and ionic;
jLij is the mobility of the free ions; g0 is the permittivity of free space, and
E0 is the breakdown field strength of the gas in the precipitator. This equa-
tion shows that the electric field is caused by both the electrostatic term and
the space charge term.
In the absence of an applied voltage, the electric field is caused ex-
clusively by the distributed space charge. Gauss' law can be used to eval-
uate the value of the field for this special case. Thus
Jf0Eds = q. (14.3)
For the wire and pipe case, where the electric field is everywhere normal
to the surface of the pipe and uniform, Equation 14. 3 reduces to
SOUTHERN RESEARCH INSTITUTE
-------
-318-
where E is the value of the electric field at the pipe; q is the total space
charge enclosed within the surface which is equal to the product of the space
charge density, p, and the volume irr2h; A is the surface area of the cylinder;
and €0 is previously defined.
A similar derivation for the parallel plate configuration yields
E (parallel) = -&• (14.5)
2e0
where h is the distance between the plates.
Equations 14.4 and 14.5 show that the value of the electric field is
linearly related to the plate separation for a uniform charge density situation.
Thus, theoretically, the collection efficiency for a space charge precipitator
is independent of plate spacing. As the plate spacing increases, the electric
field increases, which results in a linear increase in migration velocity, thus
maintaining a constant collection efficiency. However, this collection effi-
ciency is always less than that which theory predicts for a conventional pre-
cipitator since the electric field is increased by the applied voltage field.
Equations 14. 4 and 14. 5 show that the collection electric field is directly
proportional to the space charge density for space charge precipitators. As
the particles that provide the space charge are collected, the space charge
density is reduced. The result of this is that the collection electric field is
reduced by the amount of precipitate collected as the gas proceeds through
the collector. The migration velocity and incremental collection efficiency
approach zero as the total material collected approaches 100%.
This problem is overcome in some instances by the injection of charged
water droplets at various positions within the collector. This technique re-
duces the problem of a reduced space charge, but does not eliminate it.
Some problems are encountered in the formation of these charged
droplets. Spray nozzles typically form droplets of too great a diameter,
while grown condensation nuclei require considerable equipment as well as
additional corona sections.
-------
-319-
Some merit is attributable to space charge precipitators in that the
construction of the collection electrode system is simplified. This tech-
nique may find an application for the collection of particulate from small
gas volume installations where extremely high collection efficiencies are
not required.
A special case of the space charge precipitator that has been investi-
gated utilizes laminar flow in the collection zone.7 Since the collection effi-
ciency of the space charge precipitator is independent of plate spacing, it
is theoretically possible to build a collector with laminar flow rather than
with the turbulent flow conditions existing in conventional installations. The
motivation for this development is that the collection efficiency for laminar
flow precipitation is linearly related to the collection electrode area to volume
flow ratio rather than the exponential ratio associated with the Deutsch-
Anderson equation for turbulent flow. Whereas the Deutsch-Anderson equa-
tion relates as follows:
T, = 1 - exp(-^-w) (14.6)
s
the laminar flow equation can be stated as
„ _ A
Vg
w.
Thus it can be seen that the laminar flow equation predicts a 100% collection
efficiency when — becomes equal to the reciprocal of the migration velocity.
o
Initially, this approach appears to be extremely desirable. However,
when viewed in the light of attainable migration velocities, some difficulties
with the approach become apparent. Again, calling on conventional theory,
the migration velocity is
w = qEp/67ra7]
From Equation 14. 5, the electric field for the parallel plate configuration is
Plate separations from laminar flow must be maintained at about 0.1 in. ^
2 5 x 10~3 meter. The space charge density can be estimated for a reason-
able dust loading of about 10 grains/it3 = 2.3 x 10"5 gm/cm3 = 0. 23 gm/m3.
If this dust loading is brought about by a lju diameter particle with a density
SOUTHERN RESEARCH INSTITUTE
-------
-320-
of 3 gm/cm3, the number of particles per unit volume is determined by divid
ing the weight per unit volume of the gas by the weight of one particle. The
weight and volume of unit density particles differ by a factor of 3.
= jir (5 x lO'5 cm)3 = - x ID'" = 5. 22 x 10'13 cm3
o
The number of particles per cm is
• 4. 4, I* ,»„.«.../.»•
= 4.4 x 1013 particles/ m3
The charge per unit volume is estimated from the product of the charge
per particle and the particle concentration. Assume a reasonable value of a
charging field on the order of 5 x 105 V/m. The saturation charge for Iji
particles is about 260 electrons /particle = 4.2 x 10~17 Coulombs/particle.
Thus,
p = space charge density = 1. 4 x 107 part/ cms x 4. 2 x 10~17 Cou/part
» 5.90x 10'10Cou/cm3 = 5.90x 10"4 Cou/m3
Substituting into Equation 14. 5
E - 5.90xl(PCou/nfx2.5xlO-3meter = 0 83 x 10s V/m
2 x 8.85x 10-12 F/m
= 8.3 x 104 V/m = 830 V/cm
Customary values for the collection electric field range from 2-6 kV/cm.
The migration velocity for a particle collected in a field of 800 V/cm
will be only one-fifth of that for one collected in a field of 4 kV/cm. Thus,
the collection electrode area for a laminar flow space charge collector
would be approximately the same as for a conventional turbulent flow col-
lector for a collection efficiency of 99%.
If the laminar flow collector were contemplated for collecting larger
particles, the problem becomes more acute. The space charge from large
particles is reduced if the same dust loading is used. For example, if
-------
-321-
particles are used in theprevious example, the volume of one particle
is increased by a factor of 1(T, while the charge per particle is increased
by a factor of I0a for a net reduction in space charge of a factor of 10 to
83 V/cm.
Electrostatic precipitation theory shows that the migration velocity of
a charged particle suspended in an electric field is proportional to the pro-
duct of the charge on the particle and the value of the electric field. In this
example, the charge on the particle is of the same order of magnitude that
would be encountered in conventional precipitators. The value of the collec-
tion electric field is about one-tenth that in conventional collectors. There-
fore, the migration velocity of the particles would be only one-tenth that for
conventional precipitators.
The idea of approaching laminar flow in the collection zone does have
some merit, but the practical matter of constructing commercial sizes of
these units with plate spacings of 0.1 in. places an extremely tight manu-
facturing tolerance on the installation.
SOUTHERN RESEARCH INSTITUTE
-------
-322-
CHAPTER 14
BIBLIOGRAPHY
1. Joubert, Jacques, "Contribution to the Problem of Determination
of the Trajectory of a Conducting Particle Placed in a
Hydrodynamic Field and an Electric Ionizing Field, " University
of Lyon, France.
2. Walker, A. B., "Enhanced Scrubbing of Black Liquor Boiler
Fume by Electrostatic Pre-Agglomeration, "JAPCA 1.3, No. 12,
p 622 (December 1963).
3. Zebel, G., "Deposition of Aerosol Flowing Past a Cylindrical
Fiber in a Uniform Electric Field, " J. Colloid Sci. 20, pp 522-
543 (1965).
4. Mazumder, M. K. and Thomas, K. T., "Improvement of the
Efficiency of Particulate Filler by Superimposed Electrostatic
Forces, " Filtration and Separation, p 25 (Jan-Feb 1967).
5. Faith, L., etal., "Particle Precipitation by Space Charge in
Tubular Flow, " Ind. Eng. Chem. Fund. 6, p 519 (1957).
6. Webber, M. E., "Experimental Studies on Space-Charge
Precipitation, " thesis submitted to the University of California
at Berkeley (September 5, 1969).
7. Hanson, D. N. and Wilke, C. R., "Electrostatic Precipitator
Analysis, " Department of Chemical Engineering, University of
California at Berkeley.
-------