Removal of Respirable Particulate Matter from Flue Gases

BACKGROUND OF THE INVENTION

One of the principal means for controlling emission of

small particles from power plant exhausts is electrostatic

precipitation. This is a well-established method, which has

been in use since the early years of this century. However,

the effectiveness of conventional electrostatic precipitation

decreases with a decrease in particle size, and for particles

of lμm diameter and smaller, the effectiveness of conventional

electrostatic precipitation has decreased to the point that it

is not adequate to meet anticipated standards. The range of

particle size (approximately) from .2μ to lμ is a particular

concern because: (1) electrostatic precipitation is not

effective in this range; (2) exhausts from power plants - and

other plants - often have significant emissions in this range;

and (3) particles of this size are not effectively filtered by

the human respiratory system and consequently these particles

- when present as suspensions in the air - pass readily into

the throat and lungs.

Large particles are more easily removed from the air (or

other gases) than small particles. Particles having diameters

of 50μm, or greater, will settle quite rapidly simply under

the action of gravity. In addition, the effectiveness of

electrostatic precipitation decreases for particles of smaller

diameter. These well known facts have stimulated various

proposals to agglomerate particles, and thereby assist in the

removal of particles from the air. Some experiments have been

performed to accomplish agglomeration by acoustic means (see,

for example Report to the EPA, Midwestern Research Institute,

1975, and T. Watanabe, J ^" . Institute of Electri c Discharge of

Japan, , U 133 (1991) ) , presented a quadrupole field as a

means for assisting the agglomeration of charged particles.

However, most proposals involve bipolar agglomeration, which

was studied by N.A. Fuchs, Zeitschπft fur Physik m 1934.

This subject was taken up again by Eliasson, Egli and Hirth

(B. Eliasson, W. Egli and M. Hirth, Helvetica Phsyikalische

Acta ϋj), 1035 (1986)) and Hughes and Richardson (J.F. Hughes &

R.B. Richardson, Poce . 3rd ICESP, p. 337 (1987)) , beginning in

the late 1980' s.

A fundamental weakness of those bipolar schemes which

simply split the exhaust stream into two equal channels - one

to be charged to one polarity and the other to the opposite -

has been discussed by Eliasson, et al . (B. Eliasson, W. Egli,

J.R. Ferguson and H. Jodeit, Journal Aerosol Sci . , _18_, 869

(1987)) . The problem arises because the exhaust stream,

handled in this way, clearly will yield charged particles

whose size distributions are more or less identical for both

polarities (what has come to be called "symmetrical" bipolar

agglomeration) . In this case, as demonstrated by Eliasson and

co-workers, agglomeration occurs more readily for the largest

charged particles, so the agglomeration process shifts the

particle-size distribution toward finer particles. It is shown

that in this case the concentration of fine particles may be

reduced by only a factor of two or three, even after multiple

stages of the process. To overcome this poor performance of

the "symmetrical" bipolar process, Eliasson and Egli (B.

Eliasson & W. Egli, Journal Aerosol Sex . , 22, 429 (1991))

proposed that the exhaust first be processed in a cyclone type

particle separator, which could be arranged to separate the

particles into two distinct size distributions; one containing

particles which are mainly of diameter less than a certain

value dc, and one containing particles which are mainly of

diameter greater than dc. The two, separated size

distributions would then pass through separate channels in the

precipitator - one stream being charged to one polarity and

the other stream to the opposite polarity. Recombined in an

agglomeration chamber, one should find a process in which

large particles are combining predominantly with smaller

particles. (Eliasson and Egli call this "asymmetric" bipolar

agglomeration.) Koizumi, et al . (Y. Koizumi, et al . , Proc .

ESA-IEJ Joint Sy p . Electrosta tics , p. 66 (1994)) made

calculations based on a computerized model for small-particle

motion in a gas. They modeled both the agglomeration of

uncharged particles by random diffusion and the agglomeration

of a mass of charged particles, half of which were charged to

one polarity and half to the opposite polarity. A homogeneous

inter-mixture of these particles is assumed and the

calculations follow the agglomeration process from this

assumed initial state. Koizumi et al . present calculations of

the "symmetrical" and the "asymmetrical" bipolar processes.

Their work confirms that of Eliasson and co-workers, but shows

significant agglomeration in very short time; that is, times of the order of 0.01 seconds. Koizumi et al . do not present

any experimental results, as was done by Eliasson et al .

Many processes in industrial plants require substantial

input of air. After partaking in the process, the air flow

must be returned to the atmosphere. The quintessential example

of this is the power plant burning fossil fuel. The input air,

having supplied oxygen for combustion, must be returned to the

atmosphere, along with the "products of combustion." The

latter are predominantly C0 ₂ gas, but particulate matter may be

entrained in the exhaust gas; and especially so if solid fuel

(principally coal) is being burned. Plant exhausts can contain

a very wide spectrum of particle sizes. However, particles of

50μm to lOOμm diameter, and larger, are generally too heavy to

be suspended in the exhaust atmosphere and tend to precipitate

rapidly. This can be enhanced by baffles, filters and circular

flow patterns in the exhaust pipework, or flue. It is

particles of diameter less than 50μ which tend to be carried

into the atmosphere in untreated exhausts; and especially

particles of diameter a few microns - or less - which can

remain suspended indefinitely in gases at normal temperatures

and pressures.

Electrostatic precipitation is a widely used conventional

means of removing small particles from plant exhausts. The

electrostatic precipitation process electrically charges the

particles by exposing them to a stream of ions in the presence

of a substantial electric field - all within the exhaust

stream. The electric field causes ions in the gas to drift

with the field lines, in the course of which they are

intercepted by particles in the gas and in this way the

particles become charged. (This process is called

"Field-charging".) Then, by means of transverse electric

fields, the charged particles are swept across the exhaust

duct and out of the gas stream, where they are captured on

collection plates of the precipitator. The transverse drift or

migration velocity, w, experienced by the particle in the

transverse collecting field, is a parameter of fundamental

importance to the process. Throughout the duct space, between

the collection plates, charged particles drift toward the

plates, each with their particular value of w - depending upon

their charge and their average diameter. The particles have a

distribution in size and they are spatially distributed over

the width of the duct. Particles with higher migration

velocities clearly have a much higher probability of being

collected. For a precipitator operating at given potentials,

it can be shown (see, for example, Section 6.1 in Harry J.

White, Industrial Electrostatic Precipi ta tion, Addison-Wesley,

1963) that, when the field charging process is operative, w is

proportional to the average diameter d of the particle. In

other words, larger particles will have higher drift

velocities, w, and will more likely be collected.

There are two complexities to this situation which should

be mentioned at this point: (1) For particles having d » .2μm,

and less, charging in an electrostatic precipitator (ESP)

occurs primarily by diffusion of ions to the particles, rather

than by directed streams of ions which constitute the

field-charging process. Diffusion charging occurs at a rate

which increases only approximately linearly with the particle

diameter d in contrast to field charging where the rate of

charging is proportional to the square of d. Further, in

diffusion charging, there is no well-defined saturation value

as in field charging. Charging will continue so long as the

particle remains suspended in a gas containing ions. However,

in an actual ESP, charging time is limited by the size of the

precipitator and by the velocity of the exhaust and, in

practice, diffusion charging is significantly less effective

than field charging. (2) Beginning with d < .2μm, the motion

of the particle through the gas becomes less that of a

particle moving through a viscous fluid (as described by

Stokes' Law) and begins to have the aspects of a particle

moving amongst molecules (as described by Kinetic Theory) . The

consequence of this is that for d < .2μm, the migration

velocity w, begins to increase with decreasing d, as shown in

the theoretical curves of Figure 1.

The problem may be summarized by reference to Figure 1.

The effectiveness of conventional electrostatic precipitators

declines with decreasing particle size, reaching a minimum in

the range of particle diameters between .2μm and lμm. In fact,

throughout this range, the migration velocity w is too low to

permit effective particle collection in conventional ESP

equipment, as illustrated in Figure 2. Figure 2 presents the

theoretical collection efficiency as function of the particle

size for three different ratios of collection area versus gas

velocity. Figure 2 clearly shows, that the uncollected or

emitted fraction might be on the order of 30% for fine

particles while under the same conditions less than 0.1% are

uncollected for particles with lOμm diameter.

Increasing the operating electric field, E, or increasing

the size of the precipitator plates and reducing the gas

velocity will permit collection of smaller particles. However,

there exist limitations to all three options. The amplitude of

the electric field is limited by the breakdown strength of the

gas between corona wire and collection plate and cost imposes

boundaries on the size of the collection area and minimum gas

velocity.

What is needed is a new concept which would circumvent

the practical limitations on size and operating voltage in

conventional ESP's. Changing the size distribution of the

entrained particles towards larger particles could constitute

such a new concept .

If very small particles in the exhaust stream could be

caused to agglomerate and thus form larger particles, the

small-particle performance of electrostatic precipitators

could be improved to exactly the extent that agglomeration

could be achieved. Such a concept has been proposed and

experimental and analytical results have been published (N.A.

Fuchs, Zeitschrift fur Physik; B. Eliasson, W. Egli and M.

Hirth, Helvetica Physikalishce Acta jϋ, 1035 (1986) ; J.F.

Hughes & R.B. Richardson, Proc . 3rd ICESP, p. 337 (1987) ; B.

Eliasson, W. Egli, J.R. Ferguson and H. Jodeit, Journal

Aerosol Sci . , 18., 869 (1987) ; B. Eliasson & W. Egli, Journal

Aerosol Sci , 22., 429 (1991) ; T. Watanabe, J. Institute of

Electric Discharge of Japan , J 3, 133 (1991) ; Kanazawa et al . ,

Proc . Inst . of Electrosta tics Japan, 16, 323 (1992) ; Y.

Koizumi et al . , Proc . ESA-IEJ Joint Symp . Electrosta tics , p.66

(1994) ) . Calculations by various of these workers suggest that

agglomeration by simple diffusing together of the smaller

particles would require times of the order of ten seconds to

several minutes, depending on the assumptions made. In any

case, this mechanism is evidently too slow to be applicable in

electrostatic precipitation. The average residence time of an

entrained particle in an electrostatic precipitator is on the order of 2 to 10 seconds.

The agglomeration process could be greatly accelerated if

the small particles to be agglomerated were of oppositely

charged streams of equal particle size distribution (B.

Eliasson, W. Egli and M. Hirth, Helvetica Physikalishce Acta

59. 1035 (1986) ; J.F. Hughes & R.B. Richardson, Proc . 3rd

ICESP, p. 337 (1987) ; B. Eliasson, W. Egli, J.R. Ferguson and

H. Jodeit, Journal Aerosol Sci . , 18, 869 (1987) ; Kanazawa et

al . , Proc . Inst . of Electros tatics Japan , 16., 323 (1992)) . The

conditions for this so-called "symmetric bipolar agglomeration" could be achieved - for example - by dividing

the exhaust stream into two channels, in each of which corona

charging of respectively opposite polarities took place. The

two channels would then merge into an "agglomeration chamber. "

This approach achieves agglomeration in much shorter times

than uncharged agglomeration, but a fundamental difficulty

arises from the fact that the larger particles from the two

streams quickly merge and form inactive, electrically neutral

particles. As a consequence, symmetric bipolar charging

reduces the concentration of the fine particles by only a

factor of two or three, even after a passage through a

multistage, bipolar precipitator (B. Eliasson, W. Egli, J.R.

Ferguson and H. Jodeit, Journal Aerosol Sci . , 18 _. , 869 (1987)) .

An alternative method would be to divide the smaller and

larger particles within the flue gas stream into two separate

gas streams, i.e. all particles below a critical diameter dc

would be in one stream and all particles above dc would be m

the other stream. The particles within the two gas streams

would then be charged with opposite polarities and afterwards

allowed to merge and agglomerate. This so-called "asymmetric

bipolar agglomeration" was analyzed by Eliasson and Egli (B

Eliasson & W. Egli, Journal Aerosol Sci , 22, 429 (1991)) for a

typical size distribution of particulates in a flue gas

stream. This distribution is shown m Figure 3.

The results were encouraging if the critical diameter dc

was on the order of lμ and minimum overlap in the two size

distributions existed. For dc larger than 8μm, the asymmetric

bipolar agglomeration was inferior to the symmetrical case.

The shortcomings of the reported asymmetric bipolar

agglomeration were principally due to the fact that the

process was based on the size distribution existing at the

inlet to a typical ESP.

SUMMARY OF THE INVENTION

The method disclosed herein overcomes the limitations of

previous agglomeration schemes by utilizing the following six

features:

1. Introduction of large particles of approximately 10-50μm

in diameter from external sources.

2. Charging these large particles with the opposite polarity

to the charge on the entrained particles.

3. Introducing these large particles at a point in the

precipitator where most of the entrained particles have

already been collected and primarily small particles are left

in the stream.

4. The large particles are charged just prior to their entry

into the ESP where they attract the fine particles already

entrained in the gas stream and form agglomerated particles.

5. The resultant exhaust stream containing the agglomerated

particles would pass through a final stage of conventional ESP

or baghouse .

6. The scavenger particles collected in the hopper of the

outlet section are separated from the fines and prepared for

reuse .

Thus, the present invention relates to an apparatus and

method for agglomerating fine particles in a gas stream.

Larger particles are introduced into the apparatus and

imparted with an electrical charge opposite in polarity to the

charge imparted to the fine particles to be agglomerated. The

difference in charge polarity causes the fine particles to

agglomerate to the larger particles, and the resulting

agglomerated particles are then removed from the gas stream by

precipitation.

BRIEF DESCRIPTION OF THE DRAWINGS

Figure 1 is a graph showing the theoretical migration

velocity as a function of particle diameter for field and

diffusion charging;

Figure 2 is a graph of collection efficiency for various

particle size fractions;

Figure 3 is a graph of density, surface and mass

distribution of fly ash;

Figure 4 is a graph of expected particle size

distribution for three hoppers of an ESP in conventional

operation;

Figure 5 is a block diagram of an agglomeration system in

accordance with the present invention;

Figure 6 is a schematic diagram of an agglomeration

system in accordance with the present invention; and

Figure 7 is a cross-sectional view of means for achieving

a more uniform electric field in accordance with the present

invention.

DETAILED DESCRIPTION OF THE INVENTION

Large charged particles in the gas stream quickly

scavenge smaller particles which had previously been charged

to the opposite polarity by passing the exhaust through the

inlet sections of a conventional ESP. Because of their high

charge and large surface area, a single large scavenging

particle should be capable of accumulating a very large number

of small particles. For example, the saturation charge of a

50μm size particle is 50 = 2,500 times larger than the charge

on a lμm size particle.

The large scavenger particles can be charged positive by

field charging while the input stream can be charged

negatively, as is conventional m ESP' s. This choice

facilitates the adoption of the invention to conventional

wire-plate precipitators.

The method of deliberately introducing large particles to

scavenge the very fine particles is applied at a section of a

conventional precipitator where only a small fraction of the

original dust loading are present and the remaining particles

are predominantly fine particles which the conventional ESP

was unable to collect, as shown in Figure 4. Thus the large

scavenger particles introduced do not become attached to the

relatively large and medium size particles of opposite charge

in the exhaust stream. They remain available to agglomerate

large numbers of fine particles. Introducing the large

scavenger particles at this location keeps the consumption of

the scavenger particles to a minimum.

Assuming, for example, an inlet dust loading of lOgm/N

and that 1% of the inlet loading remains at the point of

scavenger particle injection, one would deal with

approximately lOOmgm/Nm of fine particles. To neutralize this

number of fine particles, - assuming that the scavengers are

solid (not hollow) particles and of the same specific gravity

as the fines - would require a number that is larger by the

ratio of the radii between scavengers and fines. For a ratio

of 50 one would then arrive at a required scavenger mass of

5gm/Nm ³ . If one further uses the approximate relationship that

lMWel corresponds roughly to 1 Nm /sec, one would then arrive

at a scavenger demand of 18kg/(hr,MW) . While this would be a

large amount to bring to the site on a daily basis, it should

certainly be feasible to reuse the scavenger particles which

will have collected primarily in the last hopper. Since the

relative size of the scavengers and fines is so large, it

should be easy to separate the two fractions from the ash.

Using smaller size or hollow scavengers would reduce the

weight, but might make it more difficult to reuse the

scavengers .

Scavenger particles might be obtained from sources

outside the power station, but in many instances it will be

possible to use particles from the fly-ash collected in the

various sections of the ESP. Figure 5 shows a system where

fly-ash particles are collected from the hoppers of section

ESP No. I and ESP No. II and then sorted according to size by

a classifier. The large particles are then utilized as

scavenger particles and the small ones discarded with the ash

flow.

ESP No. I is not necessarily limited to the first ESP

section, but represents a section of the ESP where most of the

particles (and especially the large ones) have been collected

and primarily small ones are left in the stream. Conversely,

ESP No. II can be more than one ESP section. Utilizing

collected fly-ash not only from ESP No. I but also from ESP

No. II allows for a very large variation m particulate

loading.

Scavenging particles also could be made of carbon or

carbon composition. In this embodiment they perform a dual

service. On one hand they attract fine particles and on the

other hand they also absorb mercury or mercuric chloride from

the flue gas stream.

To merge the scavenger particles with the stream of fine

particles, one skilled in the art could utilize various

arrangements. One possible arrangement is shown in Figure 6.

Fines entrained in a gas stream flow through the various

precipitator sections until they arrive at the agglomeration

zone. Large scavenger particles are charged inside the

injector ducts and emerge from the nozzles with a speed

matching the velocity of the stream of fine particles. In the

agglomeration zone the fines attach themselves to the large

scavenger particles and proceed as agglomerated particles

towards the outlet section or sections of the ESP where they

become recharged and collected.

As the rate of agglomeration is directly proportional to

the charge on the small and large particles, a substantial

benefit could be obtained if the particles could be more

highly charged. Consequently, it is of utmost importance to

maximize the electric charge on the fine and scavenger

particles. Maximum electric charge can be obtained by

generating the highest possible electric field over the entire

particle charging region. One possible way to achieve this is

by rendering the electric field as uniform as possible in the

charging zone. This, however, tends to limit generation or

corona current. To overcome this shortcoming, both pulsed and

dc electric fields are applied. Figure 7 shows one possible

geometry in which particle charging is accomplished by causing

the gas stream to pass through a space of concentric-

cylindrical cross-section; this concentric cylindrical space

consists of (a) an outer cylindrical metal wall; (b) an inner

cylindrical wall made of a dielectric material; (c) a

conducting grid, cylindrical in shape and placed

concentrically between the said inner and outer walls. A

voltage consisting of a dc base voltage and superimposed pulse

voltage is applied between the outer conducting wall and the

cylindrical grid lying within said outer wall. The voltage so

applied is of sufficient magnitude to cause electrical corona

in the gas within the outer cylindrical wall. A similar, if

much simpler setup can be utilized in a typical ESP

configuration, where negatively charged discharge wire of

small diameter are located between parallel plates. These

discharge wires are typically very far apart from each other,

thus creating a very non-uniform electrical field around the

wire and within most of the ESP volume. A more uniform field

configuration could be achieved by placing the discharge wires

much closer to each other.

The large particles are charged just prior to their entry

into the agglomeration zone by the above described means. The

fine particles which are already charged by traversing the

preceding electrostatic precipitator sections (s) will also

receive a booster charge just prior to their entry to the

agglomeration region.

The idea of deliberately adding particles to the exhaust

stream, when the objective is to remove particles is contrary

to conventional wisdom. However, (1) the proposed

particle-agglomeration scheme includes the recovery and re-use

of the particles deliberately introduced as scavengers.

Therefore, fresh particles will be added at only a fraction of

the rate at which the scavenger-particles are actually used in

agglomeration with fine particles. (2) The large particles

deliberately introduced in the proposed scheme having in many

instances been obtained from the collected ash and do not

contribute another waste material which must be disposed in

some way.