-------
-478-
REFERENCES
(1) Hesketh, H.E., A.J. Engel and S. Calvert, "Atomization - A New Type
for Better Gas Scrubbing", Atmospheric Environment, Vol. 4, No. 6,
p 639 (1970).
(2) Nukiyama, S. and Y. Tanasawa, "An Experiment on the Atomization of
Liquid by Means of an Air Stream", Trans. Soc. Mech. Engrg.
(Japan), Vol. 4, p 86 (1938).
(3) Calvert, S., "Venturi and Other ATomizing Scrubbers Efficiency and
Pressure Drop", Air Pollution (ed. by A.C. Stern), 2nd Ed., Vol.
Ill, Ch. 46, Academic Press, NY (1968).
(4) Hesketh, H.E., "Atomization and Cloud Behavior in Venturi Scrubbing",
JAPCA, Vol. 23, No. 7, p 600 (1973).
(5) Lane, W.R., "Shatter of Drops in Streams of Air", I§EC, Vol. 43, No. 6,
p 1312 (1951).
(6) Behie, S.W. and J.M. Beeckmans, "Trajectory and Dispersion of Transverse
Jets of Water in a Turbulent Air Stream", prepared for AIChE
meeting, Tulsa, March 1974.
(7) Ingebo, R.D., NASA Tech. Note 3762 (1956).
(8) Marshall, W.R., Jr., "Atomization and Spray Drying", Chan. Eng. Pro-
gress Monograph Series, No. 2, Vol. 50 (1954).
(9) Bughdadi, S.M., "Effects of Surfactants on Venturi Scrubber Particle
Collection Efficiency", M.S. Thesis, Southern Illinois University
at Carbondale, School of Engineering and Technology, Carbondale,
111. (March 1974).
-------
-479-
Paper No. 20
REMOVAL OF CARBON BLACK FROM INDUSTRIAL GASES
by
Valery P. Kurkin
STATE RESEARCH INSTITUTE
OF INDUSTRIAL AND SANITARY GAS CLEANING
Moscow
-------
-480-
-------
-481-
REMOVAL OF CARBON BLACK FROM INDUSTRIAL GASES
by
V. P. Kurkin
The removal of carbon black or soot from industrial gases
has become very important. The development of such industries
as tire production, printing and publishing, and the production
of paint and varnish, demands an increased output of various
industrial carbon blacks.
Industrial carbon black as the end product of the operation
in which it produced must be removed from the gas and black
mixture issuing from the soot burner. Also the effluent gases
from carbon black production must be cleaned in order to
prevent air pollution.
Carbon black is one of the most highly dispersed materials
and its removal from the gas is quite difficult. Modern
plants are equipped with various systems for removal of the
black, which comprise devices for conditioning and cooling
of the mixture of gas and black, removal of the black, and
purification of the gas.
The removal of carbon black from gases is also important
in acetylene production based on electric cracking of methane
and in the thermal oxidative pyrolysis of methane. In
recovering the final product—acetylene—from the gases, prior
thorough filtering of the gases is required. Acetylene is
used as the raw material in the synthesis of many products.
-------
-482-
The output gases from the gasification of oil that are used
as fuel in gas turbines and gas furnaces on steam boilers
require prior removal of soot and ash.
The basic types of carbon black used in Soviet industry
are: channel black, gas-furnace black, oil-furnace black,
lampblack, semi-active and active carbon blacks (from liquid
raw materials), and anthracene black (from mixtures of coal
and coke-oven gases). They are classified according
to the relative surface areas of their particles. Soviet active
carbon blacks TM-70 and MT-100 have foreign counterparts HAF,
ISAF, and CFR; TM-50 semi-active carbon black, oil-furnace
black, and gas-furnace black are similar to FEF and GPF; and
channel black is similar to CR.
I. Recovery of Lampblack and Oil-based Furnace Blacks
Lampblack and oil-based furnace blacks are produced by
the injection of liquid raw materials into furnaces and their
combustion. SG-type electrofilters are used for recovery of
lampblack and oil-based furnace blacks after the mixture of gas
and black is cooled in a scrubbing tower to a temperature of
180-230°C. Water evaporation not only lowers the temperature
but also improves the collection efficiency of the electro-
filter and reduces the risk of explosion.
Higher temperatures are undesirable because they lower the
electrical resistance of the inter-electrode space and increase
the danger of fire, which can deform the internal metal construction
-------
-483-
At lower temperatures, the carbon black is deposited on the
electrodes along with moisture which leads to corrosion. SG-type
electrofilters have a collection efficiency of up to 98-99%
when the gas velocity does not exceed 0.5 - 0.6 m/sec.
The "dry" method of carbon black recovery consists of using
cyclones and a filter of glass fiber fabric as bags or sleeves;
all such installations operate under positive pressure up to
70 mm in the cyclones and up to 250 mm in the bags. Gases
from a water spray cooler flow through 4 consecutively mounted
cyclones of large diameter. The filter bags are sectioned and
their regeneration takes place sequentially by blowing the
purified gas back through them. Regeneration of the filter
fabric requires 12,000-14,000 m3/h*. °f gas. The unit load
of the filter fabric corresponds to a direct gas flow of about
0.35 m3/m2 min. The exit residual dust load is about 100 mg/m3.
II. Removal of Carbon Black from Industrial Gases
1. Removal of carbon black from gas emitted from the
decomposition of hydrocarbons in arc furnaces. This involves
the following operations. First, the cracked gas goes to 3
cyclones (each 1200 mm in diameter) set in sequence and then
to a common cyclone which represents the final stage of the
dry part of the collection system. Each bunker has a steam
jacket to heat the walls in order to prevent water vapor and
organic compounds from condensing on them and thus to prevent
the carbon black from adhering to the walls.
-------
-484-
The residual content of carbon black after 4 cyclone stages
is 4 g/m3; thus the collection efficiency (filtration index)
of dry filtration is about 78%.
The gas flows from the cyclones to a drain-type device
with 3 trays, in which it undergoes additional filtering (to
50-100 mg/m3). A very high index of filtration is obtained
by condensation of steam, by which the black is washed out
of the gas. Water is injected into the gas flue at the front
end of the apparatus in order to prevent clogging of grids
or spray nozzles by substances inclined to polymerization.
The temperature of the gas is reduced to about 100°C, and it
is preliminarily washed to remove polymerizing substances.
The final filtering of the gas to 1-5 mg/m3 takes place
in a turbulent washer with a gas velocity in the throat of the
pipe of up to 110-120 m/sec and 100 mm water pressure differential,
2. Removal of carbon black in the thermal-oxidative
pyrolysis of methane. In the production of acetylene by the
thermal oxidative pyrolysis of methane, carbon black is a by-
product. It is formed by the decomposition of acetylene.
Pyrolysis gases from the reactor are piped to a scrubber,
where they are cooled from 90°C to 60°C, and then to an
electrofilter where they undergo final cleaning. The input
temperature of an SPM-8 electrofilter is kept constant
automatically by varying the volume of water sent to the scrubber.
Partial removal of the black and resins takes place in the
scrubber simultaneously. The cooled gases go to electrofilters
-------
-485-
for final purification. The SPM-8 electrofilter is a vertical
steel single-section device of rectangular form. The gases
are moistened and the carbon black, together with the resins,
is removed by spray from fine injection nozzles located in the
top of the electrofilter. Continuous water spray is under a
pressure of 6 atm. A rotatable system fitted with sprayer
nozzles can be placed in the lower part of the electrofilter
in order to provide periodic washing (once in 8-10 days).
The content of carbon black in the gases coining from the
electrofilter does not exceed 6 ing/m3 and therefore the pyrolysis
gases can be used industrially.
3. Removal of carbon black from synthesis gas from steam-
oxygen gasification of residual oil (mazut) without pressure.
The product of this process is the most effective raw material
for production of ammonia, alcohol, and other organic chemicals.
Since carbon black contaminates expensive catalysts, it must
be removed from the gas before use. The carbon black content
must not exceed 3-20 mg/m3. The simplest way of filtering
synthesis gas is the following: the output gas goes from the
gas generator through a flue with a water-cooled jacket to a
heat exchanger and to an air cooler, in which its temperature
is lowered to 350-500°C and the black content is decreased to
1-5 mg/m3.
From the air cooler the gas flows to the final filtering
system, which consists of a spray apparatus and a venturi
-------
-486-
scrubber; in this system, the gas is cooled to 45-60°C and the
final level of black content is achieved, i.e., less than
5 mg/m3.
4. Thorough removal of carbon black from synthesis gas
from oxygen gasification of residual oil under 15 atm pressure.
Synthesis gas as produced by partial combustion and gasification
of residual oil contains 6 g/m3 of carbon black. The gas at
300°C flows from the exhaust-heat boiler to a cooler through
which is dripped oil at 90°C, where the temperature of the
gas is reduced to 140°C, and then the gas flows through two
venturi tubes in sequence via an intermediate connecting
separator and a scrubber with a spray of oil heated to 90°C.
The residual black (1-1.5 mg/m3, maximum 3.5 mg/m3) is
collected in a venturi tube with a cold water spray. Light oil
vapor is also condensed, together with water vapor. The oil
is cooled in pipe water coolers during its circulation. The
residual black content can be reduced to 1 mg/m3. The
scrubbing oil, which absorbs practically all the black, is
burned in the boiler.
III. The Outlook for Scientific Research
on the Collection of Carbon Black
Problems of collecting carbon black are pressing because
of the continued growth in production of highly dispersed
active carbon blacks.
-------
-487-
Gas-cleaning equipment of the bag-filter type is being
designed. The construction of such bag filters must satisfy
the following requirements: it must have a large throughput
capacity (100,000 m3/hr) and high filtering efficiency; the
dimensions of the equipment must fit the conditions of modern
carbon black production; and the operating temperature of the
equipment must be 350-400°C.
The problems of energy-technological uses of high-sulfur
residual oil can be solved by gasification. From this point
of view, high-temperature filtering of synthesis gas and its
subsequent industrial use seem to be the most expedient.
The collection of carbon black in residual oil gasification
is complicated by the pressure of the gas; filtration appears
to be the most promising method of solving the problem.
The large amount of carbon black collected and its activity
makes it necessary to control the material collected by gas
cleaning. Such control is based on complete combustion of
the carbon black and registration by automatic control devices
to maintain continuous monitoring of the effluent.
This report presents the main problems which we face in
our country in the collection of carbon black.
-------
-488-
-------
-489-
Paper No. 21
THE APPLICATION OF WET ELECTROSTATIC PRECIPITATORS
FOR CONTROL OF FINE PARTICULATE MATTER
by
Even Bakke
UNITED STATES FILTER CORPORATION
Summit, New Jersey
-------
-490-
-------
-491-
Abstract
With the enforcement of much more stringent emission
codes requiring the removal of condensable materials that
form very small droplets in the sub-micron range, and with
heavy emphasis on removal of solid particles smaller than
1 micron, the wet electrostatic precipitator has proven
itself to be a highly efficient and economic alternative
to high energy scrubbers. The recent development of a
continuous sprayed, parallel plate, and horizontal flow wet
electrostatic precipitator is described in detail.
The performance is not dependent upon the dust resistivity.
The particle parameters that must be considered are the relative
dielectric constant of the material and its size. The
reentrainment loss is negligible and the cleaning (rapping)
losses are non-existent. Particles with low dielectric
constant, i.e., less than 10, have been shown both theoretically
and experimentally to need a longer distance for collection.
In a three field wet electrostatic precipitator, the removal
efficiency of solid particles has been measured to exceed
99.5% even if 80% of the particles were less than 1 micron.
Removal efficiencies higher than 95% have been measured on
condensable hydrocarbons (tar fumes). The removal efficiency
does not seem to change significantly with changes in the
dust particle size distributions.
The Deutsch-Anderson equation for collection efficiency
predictions and equipment sizing applies only for a limited
range of operating parameters.
-------
-492-
-------
-493-
Introduction
The use of wet electrostatic precipitators for control
of emission from industrial sources is almost as old as the
use of the dry wire-tube, Cottrell type. However, until
the mid sixties, its application was generally restricted to
rather specialized applications such as on acid mist, coke
oven off-gas, blast furnaces and detarring applications. The
method of cleaning was in most cases intermittent and of the
wetted wall type.
As a result of much more stringent local, state and
federal emission codes, condensable materials forming small
droplets or fumes are now being added to the total partic-
ulate loading. Hence, numerous applications have opened up
for the wet electrostatic precipitator (WEP). In order to
meet the codes, the energy consumption for scrubbers has
increased exponentially. Also the removal of organic
condensables which are very difficult to wet and form small
droplets in the 0.1 to 2 micron range, requires scrubber
pressure drops in the range from 40 to 60 inches of water
gauge. Since the WEP is always operated at saturation
temperature (100% relative humidity) it will remove organic
materials with a condensation temperature higher or equal
to the gas saturation temperature. It will also remove
solid dust particles in the submicron range, and gaseous
contaminants soluble in the spraying liquor. This removal
is done with very low energy consumption; the pressure drop
is usually less than 0.5 inches water gauge and the electric
power input through the high voltage power supplies is quite
modest, such as from 0.5 to 0.8 KW/1,000 ACFM.
The recent development of a continuously sprayed, parallel
plate, frame electrode and horizontal flow design has provided
industry with a realistic alternative to high energy scrubbers.
The theory of operation, description of the design, range
of applications with detailed discussion of the performance
of the WEP on Horizontal Stud Soderberg pot line, the
limitations of the method for performance prediction, the
power consumption and its economics will be discussed below.
Theory of Operation
The corona generation, the charging and discharging
processes in the wet electrostatic precipitator are, in general
terms, similar to what takes place in a conventional dry
electrostatic precipitator except for some important
differences as described below.
-------
-494-
Since the gas in the wet precipitator is always
saturated with water vapor, the corona current and voltage
relationship is somewhat different from the same relationship
in the dry precipitator. With increasing amounts of water
vapor, the sparkover voltage increases; i.e., the voltage at
which the field breaks down, but the corona current at a given
voltage is lower (1). When solid particles and droplets
enter the electrostatic field, they will cause a local
distortion of the electrostatic field between the electrode
and the collecting plate. Some of the electric field lines
intersect the particles and ions generated by the corona
discharge will tend to travel along lines of maximum voltage
gradient or along the field lines, and therefore, some of
the ions will collide with the particles and the charge
gradually builds up on the particles.
This process will continue until the charge on the
particles is so high that it diverts the electric field
lines away from the charged particles preventing new ions
from colliding with the dust particle. When this state has
been reached, the particles are said to be saturated with
charge. Theory shows (1) that the saturation charge value
and charging time is dependent upon electric field strength,
size of the particle, the dielectric constant of the particle
and the relative position of the particle in the field. This
charging process is said to be field dependent and is the
dominant process down to a particle size of 0.2 ym (2). For
smaller particles, the so-called diffusion charging process
is the dominant mechanism and is governed by the random
thermal motion of the ions and is not limited to a saturation
charge.
As soon as the charging process of the particle starts,
the resulting electrostatic force will pull the particle
towards the collecting plate. This force, together with the
gravitational and the drag forces, and the gas flow
distribution in the field, determine the particle trajectory
and its point of collection.
In a dry electrostatic precipitator, the dust buildup
on the collecting plate limits the maximum voltage at which
the precipitator can operate. For dust layers with high
resistivity (greater than 2 x 1010 ohm-cm) the voltage drop
can be from 10 to 20 KV (2). This condition lowers the field
strength in the space between the electrode and the dust
deposit surface, and therefore results in a lower saturation
charge which again gives a lower electrostatic force. If, on
the other hand, the resistivity of the dust layer is lower than
107 ohm-cm (2), the electrostatic force holding the dust
-------
-495-
particle on the plates is low and reentrainment can become
a serious problem during the electrode and plate cleaning
(rapping) cycle and also during the steady operation, having
the overall effect of lowering the precipitator collection
efficiency.
For a continuously sprayed wet electrostatic precipi-
tator, the abovementioned problems do not exist. The
spray liquid drops form a film on the collecting plates which
continuously washes off the dust that is being collected, and
the resistivity of the water film is the governing factor in
the dust discharging process and not the resistivity of the
dust layer itself. Reentrainment problems are also non-existent,
since the collected particles are instantaneously and contin-
uously removed from the point of collection and are washed
down as a light slurry. The exit loading is, therefore,
much more stable and does not have the characteristic sharp
increase as the dry electrostatic precipitator has during
the collection plate and electrode rapping cycles.
Therefore, for a wet electrostatic precipitator, the
operation is not influenced by the resistivity of the dust
layer, and the major particle parameters to consider are
their dielectric constant and size.
In order to get a better understanding of the effect of
low dielectric constants on horizontal migration distance of
the particle, a mathematical model of the particle collection
mechanism was developed. The analysis was based upon a field
charging process and a particle or droplet which had to
traverse the whole net field spacing (one half of the plate
to plate spacing) . Particles of different sizes with
dielectric constants of 2, 10 and 78 were investigated.
The unit consists of parallel collecting plates with a
separation of 2r. The velocity profile between the plates
is assumed to be flat (plug flow) and turbulent drag forces
are neglected. Centered between two plates is an electrode
frame with electrode spacing assumed sufficiently close to
provide an approximately uniform electrostatic field near
the plate surface. The field strength is approximately 70%
of the field which would be produced by a solid discharge
plate electrode (1) , or
= -0.70 dv/dr
-------
-496-
The current density under no-load condition will be (2)
j = i/Ac (2)
The ionic space charge can be determined from the current
density - electric field equation (2) .
J = N0euiE (3)
The saturation charge for a nonconductive particle is (2)
qs = 12 -iy Tre0a2E (4)
The relative dielectric constant, e, for a conducting particle
approaches infinity and is equal to one for a perfect
insulator.
The expression for the charge as a function of time is (2)
where T is a charging time constant or
T = 4e0/N0ey (6)
The particle size range examined is larger than 0.2 ym, so
the diffusion charge can be omitted (3) .
If we start with a particle entering the field halfway between
two plates and without any charge, the force balance is
divided into three different components:
x - axis, the direction of the electrostatic field
(transverse to gas flow)
y - axis, the direction of the gravitation force
(vertically down)
2 - axis, the direction of the gas flow
(horizontal and axial)
-------
-497-
The force balance is then as follows:
ZFx = Fqe - Fx - Fix = 0 (7)
IFY = Fg - Fny - Fiy = 0 (8)
ZFz = Fnz - Fiz = 0 (9)
The electrostatic force can be expressed as
Fqe = qE (10)
Substituting eqs. 1, 4 and 5 in eq. 10 gives
Fqe " 12 eT7ffeoa—4T^-(5rV °'49
t+N0ey
which shows the influence of the dielectric constant, the
particle size and the field strength on the electrostatic
force.
The gravitational force is
Fg = mg (12)
The viscous force is, assuming Stoke's Law applies (laminar
flow)
Fn = 6iranw (13)
and the inertia force can be expressed as
Fi = m dw/dt (14)
If we assume a spherical particle with a radius "a" is
moving in this field, it will be charged to carry an amount
of q (coul) charges and the force balance in the transverse
direction becomes after substituting eqs. 10, 13, and 14 in
eq. 7:
qE - 6iranwx - m dwx/dt =0 (15)
-------
-498-
Substituting eq. 5 into eq. 15 gives
Sis t+T ^ dt
let
A = 6iran/m and B = qsE/m (17)
Substituting this in eq. 16 and integrating gives
r -At
The term \- - dt cannot be integrated but using a series
Jt+T J 4':
solution Jt+T (by Jolly (4))':
= e-ab [in (b+x) + ? Ca(b+x);jn/n-n:]
n=l
Then by using this expression in eq. 18 and integrating it
once more with the following initial conditions:
t = 0, wx = wxo = 0 and sx = sxo = 0
the travel distance sx becomes
T
ex = wdt - (t -
(19)
»fln ^±1 + ? ((A(t+T))n -
-------
-49y-
where (wgas - wz) is the relative velocity between the particle
and the gas. Integrating eq. 20, using the constants given
by eq. 17 gives
~ - (21)
where wzo is the initial particle velocity along the z-axis.
The horizontal travel distance becomes then
sz = Jwzdt = wgag (t + I (e-At - 1)) - I w2Q (e-At - 1) (22)
o
Then by using the travel time calculated from eq. 19 , the
horizontal traveling distance can be calculated as a function
of particle (droplet) size and dielectric constant. This
is shown in Fig. 2. With two 5 ym particles or condensed
droplets, one with a dielectric constant of 2 (e.g. a
condensed hydrocarbon droplet) and one with a dielectric
constant of 78 (e.g. pure water droplet), for these two
particles to migrate across a field spacing of 6 inches with
an applied voltage of 50 kv and a gas velocity of 3 ft/sec,
will take a horizontal distance of 7.2 ft. (2.2 m) and 3.9
ft. (1.2 m) respectively. Therefore, the low dielectric
particle takes almost twice the horizontal distance before
being collected and this analysis points to the fact that
condensable hydrocarbons (tars) and other materials with a
low dielectric constant will be much more difficult to
collect than conductive particles, and this has been confirmed
by measurements.
When considering the removal of condensable hydrocarbons
(tar mist) , it should be remembered that the dielectric
constant for petroleum distillates are quite low, i.e. around
2. For example, hexane (C^E^^) has a dielectric constant of
2 and a boiling point of 69° C., toluene (CyHs) has a
dielectric constant of 2.15 and a boiling point of 110° C.,
and naphthalene (CiQHo) has a dielectric constant of 2.54
and a boiling point of 218° c. Other organic liquids like
phenol formaldehyde resin has a dielectric constant of 6.6.
Pure water has a dielectric constant of 78.
The removal efficiency of the WEP on a given gas and
dust stream is a function of six basic parameters:
Collection Area
Operating Voltage
Discharge Current
Liquid to Gas Ratio
Treatment Time
Local Average Velocity
-------
-500-
The performance is often stated by the so-called
migration velocity; the higher the migration velocity, the
better the particulate removal efficiency or the smaller
the WEP in terms of collection area needed to treat the
gas flow. The relationship between migration velocity and
WEP performance is given by the following equation, the
so-called Deutsch-Anderson equation (1):
u> = -Q/A 0.508 In(c0/Ci) (23)
The efficiency of the unit is given by
ne - (1-Co/Ci) 100 (24)
and when substituting eg. 23
ne = (1-e(-Aw/°-508Q)> 100 (25)
The migration velocity, w is a performance parameter
that does not in reality relate directly to the speed at
which the particles migrate to the collecting plates. It
is a "catch-all" which also includes all operating parameters
not included in equation 23.
Description of the Wet Electrostatic Precipitator (WEP)
The wet electrostatic precipitator of the MikroPul
design can be characterized as a continuously sprayed,
horizontal flow, parallel plate, and solid discharge electrode
^-ype, and in terms of gaseous absorption it can be characterized
as a combination of a co-current and cross flow scrubber. Figure
3 shows a cut-away view of the internal configuration.
In the application of a wet electrostatic precipitator, it
is very important that the gas to be treated is saturated with
water vapor to prevent that water inside the WEP evaporates
which causes loss of washing water and dry zones on the internal
members. The saturation of the gas can be done in a spray
tower or scrubber upstream of the WEP, or it can be done
in the inlet section of the WEP, or both.
In addition, it is also necessary to obtain a good and
uniform velocity profile across the WEP, and the diffusion
of the flow from the inlet duct velocity down to the WEP face
-------
-501-
velocity has to be performed in the inlet section. Further-
more, by spraying co-current into the inlet section, some of
the coarser particles will be removed and the gas absorption
process will be started. To accomplish this, sections of
baffles and sprays are located in the inlet cone of the WEP.
After passing through the sections of transverse baffles,
the dirty gas stream then enters into the first electrostatic
field. Water sprays located above the electrostatic field
sections introduce the proper amount of water droplets to
the gas stream for washing of internal surfaces. The
particulates and the water droplets in the electrostatic
field pick up a charge and migrate to the collecting plates.
The collected water droplets form a continuous downward
flowing film over all the collecting plates and keep them
clean. The water film and the collected particulates flow
down the collecting plates into the troughs below which are
sloped to a drain.
The transverse baffle gas distribution system combined
with the extended electrode, located upstream and downstream
of each field, insures complete gas flow uniformity from
passage to passage, and collects particulates and droplets
by impingement, and by electrostatic forces. Also the
extended discharge electrode system improves the collection
efficiency by increasing effective collection area. At the
entry of a field, particles not captured by the transverse
baffles are given an advance charge by the forward extended
electrode before they come into proximity of the collecting
plates. Thus charged, the particles start immediately to
migrate toward the leading edge of the plates. It has been
found that the downstream side of the baffles at the exit
of a field collects a considerable amount of material.
The very small charged particles escaping the parallel plate
field are pulled into the wake of the baffles by the
slight vacuum resulting from the turbulent dissipation
of energy. Since the particles have an electrostatic charge,
some of them will be collected on the back side of the baffles.
All baffles systems are arranged so that a walkway runs
across the front and the back of each of the electrostatic
fields. The discharge electrode frames are mounted on
collar-type high voltage support insulators. Insulator
compartments are heated and pressurized to prevent moisture
and particulate leakage into the insulator compartment.
In any particulate and/or gaseous removal process where
a liquid is used, it is important to remove the carry over
liquid drops and mists before the outlet of the equipment.
We have found that doing this electrostatically is highly
-------
-502-
efficient. Hence, the last section is operated dry, thereby
establishing an electrostatic barrier which the liquid
droplets cannot penetrate, and the mist collects on the
front side of the baffles, and the downstream side is dry.
However, some small dust particles can penetrate through
and will collect on the downstream baffles. Therefore,
this surface is washed intermittently to prevent buildup of
particulates.
Range of Applications
During the past two years, many new applications have
been piloted and units have been sold and installed following
successful pilot plant work. The type of applications where
the WEP should be used can be categorized as applications on
gas streams containing relatively light dust loading of
submicron particles and/or condensed organic materials forming
a submicron fume. Ordinarily these applications would
require very high pressure drop scrubbers in order to meet
the current air pollution codes. Although the initial
investment is higher for the WEP compared to a scrubber,
the energy consumption and operating costs are only a small
fraction of what would be needed to operate the scrubbers.
The water treatment requirements would be the same as for
scrubbers.
On some applications where the dust resistivity is either
very high or very low, the WEP can also be applied successfully
in competition with dry electrostatic precipitators.
MikroPul has installed WEP's on the following
applications:
1) On Soderberg aluminum reduction cells (pot lines) both
of the vertical and horizontal stud type cells, for
simultaneous removal of aluminum oxides, solid and
gaseous fluorides, tar mist (condensable hydrocarbons)
and S02•
2) On carbon anode baking furnaces (ring furnaces) for
removal of carbon particles, tar mists and S02.
3) On fiberglass resin application section and forming
lines for removal of short broken glass fibers,
phenolic resins and tars.
4) On molybdenum sulfate roasting, downstream of a scrubber
for removing ammonium sulfite - sulfate aerosols which
forms in the ammonia scrubbing process and S02.
-------
-503-
WEP's are now being manufactured and installed on the
following additional applications:
1) For upgrading of low pressure drop scrubbers on phosphate
rock driers for removal of the submicron particles
and SO2.
2) On coke oven batteries when connected to a continuous
shed or hood along the push side of the battery where
the coke is pushed into the railroad car. Here the
WEP will remove the fine carbon particles and the
condensable hydrocarbons during the push cycle. In
addition, the WEP will eliminate any emission caused by
door leakage on the push side.
MikroPul has a very active pilot plant program and is
now investigating several other new applications.
Detailed Description of an Application
The application of the WEP on horizontal stud Soderberg
aluminum reduction cells (pot lines) will be discussed in
order to compare the experimental results with commonly used
theory and to give some detailed information of the level of
performance of the WEP on very fine particulate matter and
condensed hydrocarbon droplets (tar fumes) which are believed
to be mostly of submicron size.
A 50,000 CFM prototype unit was installed at Reynolds
Metals Company, Longview, Washington plant and started up in
October of 1971 and has been in operation since. The
precipitator was installed so that it could be evaluated
when connected downstream of the existing cyclonic scrubbers
and when connected directly to the duct from the forced draft
fans, i.e., without presaturation before the WEP.
After showing excellent performance, four full size
units were installed and have been in operation since
May of 1973. Twenty-six more units are under construction
which will complete the primary emission control system for
the Longview North and South plants.
The Reynolds Metals Company specifications called for a
maximum total outlet loading of 0.003 GR/SCF (6.9 mg/m3)
when the total inlet loading is 0.05 GR/SCF (114.4 mg/m3) or
less, and if the outlet loading is higher than 0.003 GR/SCF,
the efficiency should stay higher than 95%. This efficiency
-------
-504-
would correspond to an overall migration velocity of 5.18
cm/sec. The loadings are defined as solids plus condensable
hydrocarbons. The inlet loading of 0.05 GR/SCF is downstream
of the scrubber which is the arrangement for the North Plant.
For the South Plant, the scrubbers will be removed and the
inlet loading will increase to 0.15 GR/SCF (340 mg/m3) of
solids and condensable hydrocarbons.
Figure 4 shows a schematic of the arrangement of
equipment in the North Plant. The fans creating sufficient
draft at the pots are connected to the manifold and the fan
outlet ducts are connected to the inlet of the cyclonic
scrubbers. In between potlines, two scrubber outlets were
connected to one precipitator inlet giving a total flow of
100,000 ACFM through the WEP. At each end of the plant,
one scrubber is connected with one WEP with a flow of
50,000 ACFM. The treated gas leaves the WEP through an
outlet conversion piece and stack combination.
The high pH sodium based liquor is piped into the WEP
and scrubbers as shown in Figure 4. The fresh liquor first
passes through the WEP and discharges into a small receiving
tank and is then pumped into the scrubber with a booster
pump. From the scrubber, the liquor passes back to the
clarifiers and the cryolite recovery plant. The liquid
rate through each of the 100,000 CFM units is approximately
500 GPM.
Tne 100,000 CFM wet electrostatic precipitators have
28 passages and 3 electrically independent fields with four
points of electrode suspension per field. The plates are
6 feet by 25 feet high. The specifications are summarized
in Table I for the 100,000 CFM units installed in the North
Plant.
Raemhild (5) performed an investigation to evaluate
the cyclonic scrubbers in the North Plant. His results
gave the scrubber inlet and outlet loadings of solids and
condensables and are summarized in Table II. These are the
average loadings based upon 11 tests performed in the fall
of 1971. The scrubber inlet loading is the inlet loading to
the WEP when no scrubber is used, i.e., in the South Plant,
and the scrubber outlet loading is the inlet loading to the
WEP's in the North Plant. The condensable scrubber outlet
loading of 0.0069 GR/SCFD is low; higher values have been
measured and found to be as high as 0.03 GR/SCFD. Raemhild
also made particle size distribution measurements with an in-
stack impactor. Figure 5 shows the particle size distributions
-------
-505-
TABLE I
Summary of Specifications for the
Wet Electrostatic Precipitators Installed at
Reynolds Metals Company Plant at
Longview, Washington
Gas Flow
Inlet Temperature to Scrubbers
Inlet Temperature to WEP
Total Particulate Inlet Loading (solids
and condensables, excluding water)
No. of Electrostatic Fields
Liquor, Flow Rate at 60 PSI
Liquor pH in
Outlet Loading for an Inlet Loading of
0.05 GR/SCF or less
Minimum Collection Efficiency for Outlet
Loadings Greater than 0.003 GR/SCF
Face Velocity
Maximum Pressure Drop
Treatment Time
Housing Material, Hot Rolled MS, Thickness
Collection Plates, Hot Rolled MS,
Thickness
Discharge Electrodes, Flatbars MS
Piping Materials
Spray Nozzles, SS 316, Type
No. of Transformer Rectifiers
Rectifier Type
Wave Form
Minimum Output per T-R Set
Primary Voltage
100,000 SCFM
250° F.
100-1100 F.
0.05 GR/SCF
3
500 GPM
7-10
0.003 GR/SCF
95%
2.38 FT/SEC
1" W.G.
10.1 SEC
3/16"
10 GAUGE
1" x 1/8"
PVC
Full Cone
3
Silicon
Full
60 KV, 1000 MA
480 V, 60 Hz
Manual and Automatic Voltage and Spark Rate Control
-------
-506-
for the scrubber inlet and outlet particulates. For the
scrubber inlet, approximately 50% of the mass is smaller than
1 urn and for the outlet approximately 80% is smaller than
1 urn.
TABLE II
Reynolds Metals Company, North Plant,
Longview, Washington
Scrubber Inlet and Outlet Loadings (5)
Avg. Std. Dev.
Condensables In (GR/SCPD) .0115 .0069
Solids In (GR/SCFD) .0488 .0236
Total In (GR/SCFD) .0603 .0254
Condensables Out (GR/SCFD) .0069 .0035
Solids Out (GR/SCFD) .0289 .0028
Total Out (GR/SCFD) .0358 .0045
The 50,000 CFM prototype WEP removed approximately 95%
of the hydrocarbons and 97% of the solids. The design
improvements incorporated in the subsequent units gave a
very significant improvement in solids removal efficiency.
Even if the specific collection area was reduced from 315
SQ.FT./1000 ACFM for the prototype down to 295 SQ.FT./1000
ACFM for the new units, the solids removal efficiency
increased to a value higher than 99.5%. The improvement was
so significant that the plate area for the units in the
South plant will not be increased even if the inlet loadings
according to the specifications are three times higher (0.15
vs. 0.05 GR/SCF). The condensable removal efficiency did not
change significantly probably because of the low dielectric
constant of the tars and their small size. It has been
observed that more tars collect in the third field than in
the two upstream fields which tends to confirm the analysis
presented above showing that low dielectric materials need
a longer distance for collection than materials with larger
values for e (e > 10). The HF outlet concentrations were
found to be lower than needed by the codes and significantly
improved when compared with the prototype.
Continuous vertical velocity profiles were made in the
first 100,000 CFM unit with a hot-wire anemometer measurement
system in seven different gas passages across the unit, with
two traverses at each passage. The measurements were made at
the exit of the second field. When analyzing the 14
-------
-507-
continuous vertical velocity profiles, the average velocity
through the unit was found to be 2.37 FT/SEC with a standard
deviation of 0.204 FT/SEC or 8.62% of the average value.
True root mean square (RMS) measurements were made of the
linearized hot wire voltage signals, and the level of
turbulence, (i.e. the ratio of the RMS voltage to the mean
D.C. voltage) was found to be higher in the central part of
the WEP and varying from 0.56 to 0.83 and lower along the
housing walls, varying from 0.23 to 0.29. Hence, the
flow is highly turbulent even if the Reynolds number based
upon the plate spacing is only 13,720 and the Reynolds number
based upon the flow past the transverse baffles at the entry
of the field is only 2,290. Point measurements were also
made in the prototype. For one test, 30 point velocity
measurements were made giving an average of 2.87 FT/SEC
and a standard deviation of 1.03 FT/SEC or 35.7% of the
average.
The gas distribution in the new units is judged to be
very close to ideal flow conditions and is the major factor
contributing to the improved performance.
Comparison of Analytical and Experimental Results
The Deutsch-Anderson expression, eq. 23, is commonly
assumed to be valid for sizing calculations and performance
predictions. However,1 several authors (6, 7) have pointed
out that eqs. 23 and 25 are only valid over a limited range
of operating parameters. Further, the development of the
Deutsch-Anderson equation was based upon several simplifying
and limiting assumptions (e.g., all particles have the same
size, do not reentrain, are uniformly distributed over any
cross section by turbulent diffusion forces, move indepen-
dently, and are fully charged at the instant they enter the
field).
In order to compare data from the potline application
with theory, the measured collection efficiency was compared
with predictions given by equation 25. Figure 6 shows the
specific collection area in SQ. FT. per 1000 ACFM vs.
collection efficiency in percent of solid particulates as
measured on one of the 100,000 ACFM units at Reynolds
Metals Company. The three groups of data at A/Q approximately
equal 100, 200 and 300 represent data when one electrostatic
field, two fields and three fields respectively are in
operation. Two operating curves as calculated from eq. 25 are
shown, i.e., for a migration velocity of 9 and 12 cm/sec. As
it can be seen, the experimental points suggest a curve shown
-------
-508-
as the dotted line and this curve crosses over from the
u) = 12 cm/sec line for A/Q approximately equal to 100 to the
o> = 9 cm/sec curve for A/Q equal to 260 and greater. However,
in a narrow range of A/Q the experimental points follow the
respective theoretical lines quite well.
Another point that should be emphasized is the fact that
the removal efficiency was always better when downstream
fields were shut off, e.g., the first field in operation and
the two fields downstream shut off. The data points with the
downstream field shut off are marked with a "D" as shown in
Fig. 6 and data points with upstream fields shut off are
indicated with a "U". The reason for this difference in
efficiency is obviously that charged particles are escaping
a field in operation and they continue to migrate to some
extent and are being collected in the downstream field
even if it is not energized.
A third point that should be made about Fig. 6 is that
the three collection efficiency data points when the cyclonic
scrubber is by-passed and the particle size distribution of
the incoming dust is much coarser (see the two particle size
distribution curves in Fig. 5) are shown to fall nicely in
with the operating points when the scrubber was in operation.
This indicates that when the precipitator is operating with
three field, an inlet dust loading with a smaller mean
particle size distribution (0.22 ym vs. 0.70 ym) does not
reduce the collection efficiency.
Power Consumption and Economics
The power consumed to operate the wet electrostatic
precipitators can be divided into 4 categories, (1)
electrostatic power, (2) fan power, (3) insulator heating
power, and (4) pump power.
The electrostatic power input is approximately 1.5
KW/1000 SQ. FT. of collection area. The pressure drop
across the WEP is less than 0.5" W.G., and the net fan
power is then 0.06 KW/1000 CFM. The insulator heating power
is 6 KW per field, and if it is assumed that the WEP uses
an L/G of 5 GPM/1000 CFM and a spray nozzle pressure of 50
PSIG, the net pump power would be 0.110 KW/1000 CFM. These
values are summarized in Table III.
-------
-509-
TABLE III
WEP Power Consumption (all net)
Electrostatic Power
Fan Power at .5" W.G.
Insulator Heating Power
Pump Power at 5 GPM/1000 CPM and
50 PSIG
1.5 KW/1000 SQ. FT,
0.06 KW/1000 CFM
6 KW/Field
0.11 KW/1000 CFM
If we consider a unit handling 100,000 CFM with a
collection area of 300 SQ. FT./1000 CFM and having 3 fields,
the total net power consumption is 80 KW.
Now, if we compare this with a Venturi scrubber, assuming
it would have to operate with a pressure drop of approximately
50" W.G. and an L/G of 7 GPM/1000 CFM to give the same
performance in terms of removal efficiencies, the total net
power input is 615 KW. This would be more than 7 times
higher power consumption when compared with the WEP power
consumption.
The installed cost of a mild steel unit, with approximately
300 SQ. FT./1000 ACFM collecting area/ flange to flange, would
be between $3.00 and $4.00 per CFM. This cost also includes
the power supplies. The annual operating cost is shown in
Table IV assuming a cost of $0.01/KWH for electric power. The
annual operating cost of Venturi scrubbers is also shown.
The installed cost of the Venturi scrubber, cyclonic absorber-
separator and high pressure fan and motor was assumed to be
$1.20 per CFM. The fixed charges were assumed to be 15%.
TABLE IV
Annual operating Cost, Flange to Flange,
100,000 CFM
Investment at 15%
Power Costs at $0.01/KWH
Total Annual Cost
WEP
$52,500
7,500
$60,000
Venturi
$18,000
54,000
$72,000
As it can be seen, the annual operating cost for a WEP
is lower than for the Venturi scrubber even if the installed
cost is much higher. The above analysis assumed that mild
-------
-510-
steel could be used, i.e./ that a sufficient water treatment
and neutralization plant is installed and that the maintenance
costs were equal for the two alternatives.
Conclusions
The following conclusions can be drawn:
1. The performance and the collection efficiency of
the wet electrostatic precipitator is not dependent upon
the resistivity of the dust layer. The resistivity of the
water film is the governing discharge parameter and the unit
can handle very efficiently both high and low resistivity
dusts. The reentrainment loss is negligible, the rapping
loss is non-existent and the outlet loading is stable in
magnitude.
2. The dielectric constant of the particle material
and its size are the two most important particle parameters.
Organic condensable materials which form a very fine aerosol
usually have a low dielectric constant, i.e., less than 10
and as low as 2, and it was shown that from a theoretical
standpoint, these particles will take almost twice the
horizontal distance for collection when compared with
particles with dielectric constants larger than 10. This
finding has been confirmed by field observations which show
that the last field has the heaviest buildup of condensable
materials.
3. The wet electrostatic precipitator has been used
with a high degree of success on applications where the solid
particles are of sub-micron size and where condensable
organic droplets also of sub-micron size have to be removed.
On this type of application, the WEP competes favorably with
the high energy scrubbers because of their very high energy
requirements needed to give similar removal efficiencies.
The wet electrostatic precipitator can also be applied in
competition with dry electrostatic precipitator on dusts with
either very high (> 2 x 1010 ohm-cm) or very low (< 107
ohm-cm) dust resistivities.
4. In a three field wet electrostatic precipitator,
removal efficiencies higher than 99.5% on solid particulates
with 80% less than 1 ym size has been measured consistently.
Removal efficiencies of 95% and higher have been measured
on tar mists (condensable hydrocarbons). The wet electro-
static precipitator is therefore a highly efficient device
for removal of very fine particles both in the form of solid
particles and condensable mist.
-------
-511-
5. It has been shown that the Deutsch-Anderson
equation for sizing the WEP can only be used over a relatively
small range of operating parameters.
6. When operating with three electrically independent
fields/ the removal efficiency seems not to be influenced by
a significant reduction in mean particle size of the incoming
solid particulates.
Acknowledgement
The author expresses thanks to Mr. James Shen of the
Research and Development Department of MikroPul for working
out the details of the mathematical model, to Dr. David
Rimberg for many valuable suggestions concerning the paper,
and to Lorraine Simons for preparing the manuscript.
-------
-512-
NOMENCLATURE
A = 67ran/m * Constant
AC = Collection Area
a = Particle Diameter
B - qsE/m = Constant
ci = Particle Inlet Loading
c0 = Particle Outlet Loading
E = Electrostatic Field Strength
e = Electric Charge
F « Force
g = Gravitational Constant
i = Current
j = Current Density
In * Natural Logarithm
m = Particle Mass
No e Number Density of Free Ions
Q = Gas Flow Rate
q = Charge
qs = Saturation Charge
r = Net Field Spacing
SK = Transverse Distance
sz = Horizontal Distance
T = Migration Time for Collection
t = Time
v = Voltage
w = Velocity
Wgas = Gas Average Velocity
wx = Transverse Particle Velocity
wz = Horizontal Particle Velocity
x = Transverse Horizontal Distance
y = Vertical Distance
z m Horizontal Axial Distance
e « Dielectric Constant
e0 = Permittivity of Free Space
n = Viscosity of Gas
ne = Collection Efficiency
yj = Carrier Mobility of the Gas
TT = 3.14
T = Charging Time Constant
w = Migration Velocity Parameter
-------
-513-
References
1. White, H.J. Industrial Electrostatic Precipitation,
ecipj
rrri
Addison-Wesley Publ. Co., Reading, Mass., 1963.
2. Oglesby, S. and Nichols, G.B. "A Manual of
Electrostatic Precipitator Technology, Part I -
Fundamentals", Contract CRA 22-69-73 for NAPCA,
Cincinnati, Ohio.
3. Oglesby, S. and Nichols, G.B. "Electrostatic
Precipitator Technology for Source Emission Control",
AIChE Paper, No. 8E, 72nd National Meeting, St. Louis,
Missouri, May 1972.
4. Jolly, L.B.W. Summation of Series,
5. Raemhild, G.A. "Collection of Aerosols from a
Horizontal Spike Soderberg Aluminum Reduction Plant
by a Wet Cyclonic Spray Scrubber as Related to
Scrubber Operating Parameters", M.S. Thesis,
University of Washington, 1972,
6. Penny, G.W. "Some Problems in the Application of the
Deutsch Equation to Industrial Electrostatic
Precipitators", Journal of the Air Pollution Control
Association, Vol. 19, No. 8, August 1969.
7. Cooperman, P. "A New Theory of Precipitator Efficiency",
Paper No. 69-4, Air Pollution Control Association Meeting,
New York City, 1969.
-------
-514-
i 1—rr
10.0
u
01
H
1.0
§
M
S
o
0.1
e «• 10
I I
~T 1 1—I
PARTICLE DENSITY
FIELD SPACING
APPLIED VOLTAGE
T
E
0.8
6 in.
50 kv
3.6 m-sec.
2.3 kv/cm
I I
0.1
1.0
PARTICLE (DROPLET) SIZE (pm)
10.0
IOC
FIG. 1 PARTICLE SIZE VS. MIGRATION TIME FOR COLLECTION
-------
-515-
10.0
w
u
et,
10
M
a
O 1.0
2
sj
1
s
o
0.1
T—T
T
T
T
e «= 2
e - 10
PARTICLE DENSITY
FIELD SPACING
APPLIED VOLTAGE
T
E
= 0.8
= 6 in.
= 50 kv
= 3.6 m-sec.
=2.3 kv/cm
e = 78
_L
_L
I I L
_L
_L
J I L
_L
J L
o.i
1.0
10.0
100.
PARTICLE (DROPLET) SIZE (pm)
FIG. 2 PARTICLE SIZE VS. HORIZONTAL MIGRATION DISTANCE FOR COLLECTION
-------
FIG. 3 MIKROPUL ELEKTROFIL WET ELECTROSTATIC PRECIP/TATOR
-------
STACK
LIQUOR
MAIN
WEP SPRAYS
VALVE
INLET DUCT
RECEIVING
^ TANK
POT GAS MANIFOLD
SCRUBBER
SPRAYS
BOOSTER
PUMP
/
MAIN FAN
CYCLONIC
SCRUBBERS
(TWO)
LIQUOR
RETURN
1
ui
f—
FIG. 4 SCHEMATIC OF PRIMARY EMISSION CONTROL SYSTEM, REYNOLDS METALS, LONGVIEW, WASHINGTON
-------
-518-
SCRUBBER
INLET
LOG NORMAL APPROXIMATION
OF PARTICLE SIZE
DISTRIBUTION
MAXIMUM
AND NEGATIVE
ERROR BANDS
15 20 30 40 50 60 70 80 85
PERCENT OF MASS LESS THAN STATED (%)
90
95
FIG. 5 SCRUBBER INLET AND OUTLET PARTICLE SIZE DISTRIBUTIONS
REYNOLDS METALS COMPANY, LONGVIEW WASHINGTON PLANT
BY RAEMHILD
-------
-519-
100
90
Avg. Voltage - 50 kv
Current Density - 40 ya/sq. ft.
• With Scrubber Upstream
O Without Scrubber Upstream
D Downstream Fields Off
U Upstream Fields Off
80 -
70
100 200 300
A/Q (SQ. FT./1000 ACFM)
FIG. 6 SPECIFIC COLLECTION AREA VS. SOLIDS REMOVAL EFFICIENCY
-------
-520-
-------
-521-
Paper No. 22
THE INFLUENCE OF ASH CHEMISTRY
ON THE VOLUME CONDUCTION IN FLY ASH
by
Roy E. Bickelhaupt
SOUTHERN RESEARCH INSTITUTE
Birmingham, Alabama
-------
-522-
-------
-523-
ABSTRACT
Research has been conducted to determine the
influence of fly ash chemistry on resistivity at temperatures
above 200°C. A large number of fly ashes having generally
similar physical and structural character while possessing
typical variations in ash chemistry were used. Resistivity
as a function of temperature and transference experiments
were performed.
It was determined that the quantity of
electricity passed was proportional to a mass transfer
and that lithium and sodium ions migrate. For a given
iron concentration, the resistivity was inversely propor-
tional to the combined lithium and sodium concentration.
Also for a constant concentration of lithium and sodium,
the resistivity was inversely proportional to the iron
concentration.
It was concluded that the volume conduction
process was controlled by an ionic mechanism in which lithium
and sodium are the principal charge carriers. The effect of
the iron concentration is presently being investigated.
-------
-------
-525-
THE INFLUENCE OF ASH CHEMISTRY ON THE
VOLUME CONDUCTION IN FLY ASH
BY
ROY E. BICKELHAUPT
Southern Research Institute
INTRODUCTION
It is well known that one of the important considerations
in the design of an electrostatic precipitator is the resistivity
of the material to be collected.l Several of the preceding
speakers at this symposium have emphasized this point. The
literature2 points out that a large number of factors control the
magnitude of resistivity for fly ash produced from the combustion
of coal.
In the absence of water vapor, ash resistivity is controlled
largely by the chemistry, physical characteristics, and the
temperature of the ash. The resistivity of fly ash without the
influence of moisture is called volume resistivity. This property
is of particular interest for precipitators functioning in the
upper end of the operable temperature spectrum.
Moisture and other gases in the effluent can interact with
the ash surface to provide an alternate conduction path. When
measured resistivity is influenced by this process, the ash
resistivity is termed surface resistivity. This effect is usually
observed at lower operating temperatures due to the increased
relative concentration of water vapor.
Precipitators for collecting fly ash normally operate after
an air heater at a temperature where the ash resistivity is
influenced by both surface and volume conduction. Research
-------
-526-
sponsored by the Environmental Protection Agency has been directed
toward the identification of the factors affecting fly ash
resistivity and the quantification of the relationship between
resistivity and fly ash and flue gas chemistry. Knowledge
pertaining to the conduction mechanisms and the chemical species
involved will provide the approach by which resistivity may be
predicted and altered.
This paper describes the first part of this investigation
relating to volume conduction and illustrates the relationship
between volume resistivity and ash chemistry. Additional research
concerned with surface conduction phenomena is underway, but the
results at this time are not conclusive.
BACKGROUND INFORMATION
Approximately twenty-five fly ash samples representing a
reasonable cross-section of the ashes produced by coal-burning,
steam-generating plants in the United States have been examined.
The ashes have been physically, structurally, and chemically
characterized. Two features of the characterization were
especially noteworthy. First, a relatively large range in
concentration for each element reported in chemical analysis occurs
among the ashes. Table I shows the typical ranges found. Specific
ashes possessed concentrations of certain elements outside the
ranges shown. For example, isolated cases show greater amounts
of Na20, CaO, and S03 with lesser amounts of Si02. With respect
to this paper, the range of concentration of Na2O and Fe203 should
be noted. Second, the structural characterization revealed that
-------
-527-
TABLE I
RANGES OF MAJOR CHEMICAL CONSTITUENTS
REPORTED IN WEIGHT PERCENT AS OXIDES
U20 0,01 - 0,07
NA20 0,13 - 2,66
K20 0,28 - 3,90
McO 0,9 - 5,5
CAO 0,3 - 23,5
FE203 3,9 - 23,7
AL203 17,9 - 31,0
Si02 40.2 - 61,0
Ti02 0,8 - 2,3
P205 0,16 - 1,00
S03 0,07 - 1,83
-------
-528-
all the ashes were principally amorphous. Three or four crystalline
compounds could be identified/ but the combined crystalline fraction
was less than 10 to 15%.
•
Using a small group of these ashes selected with discrimination
for chemical constitution, preliminary research was conducted
regarding the volume resistivity-ash chemistry relationship. In
particular, these ashes contained low, uniform concentrations of
iron. The results3 of this investigation will be published in
the near future. The mode of research involved the measurement
of resistivity and chemical transference on ash specimens which
had been pressed and sintered into self-supporting discs. The
primary conclusion from this research was that volume conduction
in fly ash was controlled by an ionic mechanism having sodium
ions as the principal charge carrier.
The present paper reiterates this point and treats another
aspect of the volume resistivity-ash chemistry subject. Before
examining the data reflecting ash resistivity as a function of
ash chemistry, it would be of benefit to observe examples of the
resistivity data. Figure 1 shows a group of curves relating
resistivity to reciprocal absolute temperature. The curves form
moreorless linear, parallel lines in compliance with an Arrehenius
expression for resistivity. The similarity in the slope of the
curves suggests that one conduction mechanism prevails in all
the ashes. The three order of magnitude difference in resistivity
between the upper and low curves indicates the pronouced effect
-------
-529-
TEMPERATURE (°O
10
13
o
X
o
CO
co
LJ
or
10
12
10'
II
10
10
10
8
10
2.0
250
300
350
400
1.8
.6
.4
1000/T (°K)
Figure I.
Resistivity (As Measured) vs Reciprocal Absolute Temperature,
for Several Ashes Differing Principally in Na»0 Content
-------
-530-
of ash chemistry. In the subsequent discussion, the values of
resistivity used are those selected from the intersection of the
resistivity-reciprocal temperature curve and the ordinate
1000/T = 1.6. These data were then empirically corrected to a
constant porosity. The correction was established from the data
for nine ash specimens for which resistivity was determined at two
levels of porosity. Porosity was calculated from bulk and helium
pycnometer density values. It is noteworthy that the correction
used is similar to one which may be established from the data of
Dalmon and Tidy.1*
CORRELATION OF RESISTIVITY WITH CHARGE CARRIER CONCENTRATION
From the research detailed in reference 3 and synopzied
above, it was concluded that an ionic volume conduction mechanism
prevailed and that the principal charge carriers were lithium and
sodium ions. Under these circumstances, the resistivity should
be inversely proportional to the number of mobile charge carriers
available.
Resistivities at a given temperature were plotted against the
combined lithium and sodium concentrations of the respective ashes.
Resistivity values corrected to 35% porosity were taken from plots
of log resistivity versus 1/T for the temperature parameter value of
1000/K0 = 1.6. The combined lithium and sodium concentrations,
calculated from chemical analyses made on specific specimens used
subsequently in the determination of resistivity, were expressed
as molecular percentages. The molecular percentage was selected as
a reasonable relative measure of the number of mobile carriers. The
result of this approach is graphically illustrated in Figure 2.
-------
-531-
10'
£ io'°
CO
O
(T
= o5
(0
CO
CO
10
8
10
JO'
KO
O.I
OL
\
MO
00
PO
1.0 10.0
MOLECULAR PERCENT LITHIUM + SODIUM
100.0
Figure 2. Resistivity vs Molecular Percent Lithium*Sodium for Western
Ashes Differing Principally in Sodium Content
-------
-532-
The constructed line is a least squares interpretation for
eight data points obtained during the earliest part of the research.
These data were obtained for a series of six ashes arbitrarily
selected for minimal variation in overall chemistry and substantial
differences in alkali metal content. These data points statistically
had an excellent linear correlation coefficiency suggesting that
the resistivity of the fly ash is indeed inversely proportional to
the amount of lithium and sodium present. Furthermore, the slope
of approximately two indicates the severity of the effect.
However, as more data were accumulated for ashes selected with
no regard for overall chemistry, the previously indicated correlation
seemed to deteriorate. After about twenty-five ashes had been
examined, the situation shown in Figure 3 prevailed. All of the
data points for Eastern ashes and a few for additional Western ashes
were positioned below the established line. It was concluded that
some other chemical species in addition to lithium and sodium had
either a direct or indirect effect on the magnitude of resistivity.
This effect was strong enough that by inspection of the resistivity
data with reference to the total chemistry of the ashes it could
noted that the higher iron contents were associated with lower
resistivities.
To evaluate this observation, the resistivity data shown in
Figure 3 were normalized to a constant percentage of lithium plus
sodium using the coefficient of correlation indicated by the
constructed line. These normalized data then represented the
resistivity values expected if each ash contained the same molecular
concentration of lithium plus sodium. The normalized data were
then plotted against the molecular percentages of iron present in
-------
-533-
o
I
I
o
_c
>•
H
>
O)
O Western Ash
D Eastern Ash
10
.0 10.0 100.0
MOLECULAR PERCENT LITHIUM + SODIUM
Figure 3. Resistivity vs Molecular Percent Lithium + Sodium
for all Ashes Examined
-------
-534-
the specific ashes. This is shown in Figure 4.
The correlation between resistivity and iron concentration
shown in Figure 4 explains the data scatter in the previous figure
where resistivity was plotted as a function of lithium plus sodium
concentration. In Figure 3, the resistivity varied by two orders
of magnitude at about 0.3 to 0.5 molecular percent sodium plus
lithium. The fifteen data points plotted in this region represent
the entire spectrum of iron concentrations for all the ashes
examined in this laboratory. That no data points occurred above
the constructed line in Figure 3 is not surprising. The data used
to calculated this line came from ashes having a uniform iron
concentration of the lowest level.
From the correlation shown in Figure 4, it becomes apparent
that it is necessary to consider both the iron concentration and
the sodium plus lithium concentration to define the volume
resistivity of the fly ash. It can be seen that data point w does
not conform well to the correlation. It should be noted that
the ash used to obtain point w came from a pilot boiler and pilot
precipitator, while all other ashes were taken from commercial,
full scale equipment.
The Role of Iron
Since the effect of iron concentration on the magnitude of
resistivity was apparently equivalent in severity to that of the
combined sodium and lithium concentrations, research was conducted
to define the role of iron. Two general hypothesis were advanced.
-------
-535-
CO
1 0.
10
II
I0
10
io
g\8 L
°0°K
RO
O WESTERN
D EASTERN
OW
O.I
nC
LJD
1.0 10.0
MOLECULAR PERCENT IRON
100.0
Figure 4. Resistivity Normalized to Constant Lithium+Sodium
Concentration vs Molecular Percent Iron
-------
-536-
nn the one hand, the iron concentration may participate in the
conduction process in a direct manner either by introducing an
electronic component or an additional ionic carrier. From another
viewpoint, the iron may perform in an indirect manner. It may
affect the amorphous structure of the fly ash so that the effective
concentration, the mobility, or the type of alkali metal serving
as a charge carrier will be altered. Also, it could affect the
heterogeneity of the ash so that the continuous conducting phase
of the ash will possess an alkali metal concentration greater
than that revealed by the average composition.
Four ashes containing a relatively uniform concentration of
alkali metals but representing the total spectrum of iron concentra-
tions encountered in this work were selected for additional
experimentation.
Transference experiments were conducted on these four ashes
to evaluate certain facets of the aforementioned hypotheses.
For a description and discussion of the chemical transference
experiments, one may consult reference 3. The type of experiment
used and the character of the materials investigated precludes
the extraction of unequivocal data. The data are meant to be used
only for the qualitative understanding of the conduction process.
The gravimetric data are representated in Figure 5. In this figure,
the mass transferred out of the ash adjacent to the positive
piectrode toward the negative electrode is plotted against the
quantity of electricity passed during the test. The lines
labeled with the names of the alkali metals represent Faraday's Law.
For example, if conduction were entirely ionic and potassium were
the only charge carrier, one would expect a weight loss of about
-------
-537-
50
30 60 90 120
ELECTRICITY PASSED In COULOMBS
Figure 5. Gravimetric Data from Transference Experiments
-------
-538-
50 mg for the passage of 120 coulombs of electricity. The open
circles represent the experimental data points. These data
strongly suggest that the electricity passed is accountable for by
mass transfer and that sodium is the principal charge carrier.
The only other way the data points could occur near the sodium
line would be due to an averaging effect resulting from the migration
of an element heavier than sodium coupled with the migration of
a lighter element or an electronic contribution.
In Table II, the results of the chejnical analyses for the
transference experiments are given. The data show a trend similar
to that experienced previously for transference tests on low iron
specimens and that these data compliment or support the gravimetric
data expressed in the Figure 5. For ashes containing from 5 to
22% weight percent iron, only the migration of sodium and lithium
from the positive to the negative electrode can be detected. The
small variations in potassium and iron concentrations are thought
to be within the data error due to the technique of analysis
and the selection of random samples. From this information, it
was apparent that the iron does not act as an ionic carrier
directly, and its presence in increased amounts does not induce
the participation of potassium.
In carefully examining the data in Table II, it was observed
that the percentage of lithium and sodium that had migrated,
relative to the amounts initially present, increased with increasing
iron concentration. An empirical parameter was devised to
demonstrate this point. For a constant amount of electricity
passed in each test, the percent increase in sodium and lithium
-------
-539-
TABLE II
TRANSFERENCE EXPERIMENTS
CHEMICAL ANALYSES OF SPECIMENS
IN WEIGHT PERCENT
DISC CONTIGUOUS BASELINE
ASH OXIDE TO POSITIVE ELECTRODE COMPOSITION
DISC CONTIGUOUS
TO NEGATIVE
ELECTRODE
B
C
Li20
NA20
K20
FE203
Li20
NA20
K20
FE203
Li20
NA20
K20
FE203
Li20
NA20
K20
FE203
0,013
2,9
21,1
0,03
0,29
3,8
10,0
0,030
0,40
3,9
4,8
0,019
0,29
3,1
21,6
0,024
0,45
2,9
16,8
0,04
0,39
4,1
10,2
0,04
0,48
3,9
4,9
0,027
0,53
3,2
21,0
0,041
0,80
2,9
16,6
0,05
0,48
4,0
10,2
0,049
0,56
4,0
4,9
-------
-540-
content at the negative electrode over that contained initially by
the ash was computed as "relative effectiveness". When this
parameter was plotted against the iron concentrations of the four
ashes studied, the result shown in Figure 6 was obtained. The
increase in the relative effectiveness parameter with increasing
iron concentration suggests that the role of iron is indirect in
that it seemingly enhances the participation of lithium and
sodium in the conduction process.
It is doubtful that the iron concentration affects the
mobility of the subject alkali metals, for no appreciable
deviation in experimental activation energy had been noted among
all the tests run. However, it is conceivable that the iron
concentration could influence the number of mobile charge carriers.
Although one may use the total concentration of a particular ion
species to graphically display data, it is highly probable that
not every ion of the given type is free to migrate. It can be
suggested that the role of iron is to in some way, possibly
structurally or magnetically, alter the amorphous ash to allow
the participation of a greater percentage of the total available
sodium and lithium ions. Also in a manner unrelated to the
amorphous structure, the iron concentration could induce additional
heterogeneity so that the continuous phase that is responsible
for conduction may contain a concentration of sodium and lithium
that is greater than the average amount reported by chemical
analyses.
The potential electronic contribution to the total conduction
process due to iron was also considered. Although the uniformity
of experimental activation energies among the ashes examined, and
-------
-541-
1.0
CO
CO
UJ
z
UJ
o
UJ
u.
u.
UJ
UJ
>
UJ
or
0.8
LITHIUM
SODIUM
0.6
0.4
0.2
0.0
1
4 6
MOLECULAR PERCENT IRON
8
10
Figure 6. Relative Effectiveness of Lithium and Sodium
as Charge Carriers as a Function of Iron Concentration
-------
-542-
the pronouned effect of polarization demonstrated by current-time
curves do not support an electronic contribution, two additional
experiments were conducted. One ash was repeatedly put through a
magnetic separator until half of the original iron concentration
was eliminated and then forirted into a test specimen. Since the
resistivity of this specimen was almost identical to that of the
one without the magnetic fraction removed, it would seem doubtful
that the pronounced effect on resistivity due to iron was related
to an electronic contribution. Also, a single experiment was
conducted in which an ash specimen high in iron was used as a solid
electrolyte in a galvanic cell.5 The results were compared to that
of the same cell with stabilized zirconia as an electrolyte. Only
an ionic contribution to the conduction process was detected.
SUMMARY
From the characterization of a large number of fly ashes
collected from commerical power stations in the U.S. and Canada,
a collected layer of ash is visualized as an assemblage of more
or less spherical, mainly amorphous particles accumulated to some
degree of compaction. At temperatures where volume conduction
predominates, it is concluded that the charge is carried through
the layer by an ionic mechanism in which lithium and sodium are
the principal carriers. An overall conduction process satisfying
the required electrostatic balance was described in Reference 3.
It is also concluded that the iron concentration of the ash in
some unclear manner influences the number of lithium and sodium
ions capable of mobility.
-------
-543-
The conclusion that sodium ions were the principal charge
carriers through the layer of collected ash was tested in a
pragmatic manner in the laboratory and in the field. In the
laboratory, a calculated quantity of sodium was introduced in the
form of sodium carbonate to an ash sample of known chemical
analysis prior to preparing pressed and sintered resistivity speci-
mens. The addition of sodium produced a reduction in resistivity
in excellent agreement with the reduction predicted by the data
presented in this paper.
Sodium carbonate was also used to make predetermined additions
of sodium to an ash of known chemical composition by adding the
material to the coal feed of a commercial boiler. The objective
in this case was to raise the Na2O content of the ash from 0.2%
to 2.0%. Both the insitu measurements made at the precipitator
and resistivity measured made on ash returned to the laboratory
showed a two order of magnitude decrease in resistivity which
was the amount predicted for the addition selected.
From the ancillary tests described above, one can conclude
that the effect of sodium can be quantified and that the effect is
observed in commercial tests as well as in the laboratory.
-------
-544-
REFERENCES
1. White, H. J., Industrial Electrostatic Precipitation,
Addison-Wesley Publishing Company, Reading, Massachusetts
(1963) .
2a. White, H. J., "Chemical and Physical Particle Conductivity
Factors in Electrical Precipitation", Chem. Engr. Progress 52,
244-248 (June 1956).
2b. Shale, C. C., Holden, J. H., and Fasching, G. E., "Electrical
Resistivity of Fly Ash at Temperatures to 1500°F, RI7041,
Bureau of Mines, U.S. Department of the Interior (1968).
2c. Maartmann, Sten, "The Effect of Gas Temperature and Dew Point
on Dust Resistivity and Thus the Collecting Efficiency of
Electrostatic Precipitators", Paper EN-34F, Second International
Clean Air Congress of the International Union of Air Pollution
Prevention Association, December 6-11, 1970, Washington, D. C.,
U.S.A.
2d. Dalmon, J., and Raask, E., "Resistivity of Particulate Coal
Minerals", J. Inst. Fuel 46 (4) 201-205 (April 1972).
2e. Selle, S. J., Tufte, P. H., and Gronhovd, G. H., "A Study
of the Electrical Resistivity of Fly Ashes from Low-Sulfur
Western Coals Using Various Methods", Paper 72-107, presented
at the 65th Annual Meeting of the Air Prevention Control
Association, Miami Beach, Florida (June 18-22, 1972).
3. Bickelhaupt, R. E. "Electrical Volume Conduction in Fly Ash",
J. Air Pol. Con. Assoc.. (March 1974).
4. Dalmon, J. and Tidy, D., "A Comparison of Chemical Additives
as Aid to the Electrostatic Precipitation of Fly Ash",
Atmos. Environ. 6_ (10) 721-734 (1972) .
5. Kiukkola, K. and Wagner, C., "Measurements on Galvanic Cells
Involving Solid Electrolytes", J. of the Electrochemical Soc. 104
(6) 379-387 (1957) .
-------
SESSION 6
NEW CONCEPTS
Chairman: J. K. Burchard
U. S. Environmental Protection Agency
Research Triangle Park, N. C.
Paper No.
23 Basic Processes in Fine Particle Control
James R. Brock
University of Texas
Austin, Texas
24 Systems of Charged Particles and Electric
Fields for Removing Sub-micron Particles
James R. Melcher and
K. S. Sachar
Massachusetts Institute of Technology
Cambridge, Massachusetts
25 Advances in the Sonic Agglomeration of
Industrial Aerosol Emissions
David S. Scott
University of Toronto
Toronto, Ontario, Canada
-------
-545-
Paper No. 23
BASIC PROCESSES IN FINE PARTICLE CONTROL
by
James R. Brock
UNIVERSITY OF TEXAS
Austin, Texas
-------
-546-
-------
-547-
ABSTRACT
Some of the fundamental physico-chemical processes available
for fine particle control are analyzed briefly. These include
primary processes such as impaction, diffusion, electro-
phoresis, thermophoresis, diffusiophoresis, etc., and secondary
processes which alter the particle size distribution such as
coagulation, condensation, and charging. New techniques for
fine particle collection are possible utilizing coagulation and
condensation in conjunction with the primary processes. Only
a relatively small number of primary and secondary processes
have been utilized up to the present time, so that it is likely
that improvement can be obtained through additional investiga-
tion of fine particle control techniques.
-------
-548-
-------
-549-
BASIC PROCESSES IN FINE PARTICLE CONTROL
Introduction
The purpose of this paper is to indicate the very large
number of possibilities which are open in the development of
new fine particle collection methods. Present techniques for
removing suspended particles from industrial gas streams are well
known to be inefficient in removing fine particles—those
particles with equivalent diameters less than the order of 1 vim.
Before discussing these basic collection processes, let
us note some of the rationale for the control of fine particles
and the inconsistency of total mass emission standards in(
achieving a certain mass of suspended particulate matter in the
atmosphere.
It is well known that the fine particles in the size range
of the order 0.1 — 1.0 ym diameter have the maximum probability
of reaching the lower lung where lung clearance mechanisms for
such contaminants are believed to have the least efficiency.
Such particles in the 'respirable range1 are therefore most
heavily implicated in laboratory and epidemiological studies
which indicate adverse health effects of toxic and irritant
particles (1). Particles in this same size range are also known
to be responsible for much of the loss in visibility and solar
insolation in urban areas (2).
Many governmental regulations limit particulate emissions
according to the total mass rate of such emissions. Such regula-
tions are designed to limit the total mass of suspended
particulate matter in the atmosphere but are somewhat inconsistent
inasmuch as they do not recognize another aspect of the nature
of fine particles — their relatively large residence time in the
atmosphere. Table 1 illustrates the inadequacy of total mass
emission standards by way of example. In this table are presented
certain primary sources of particulate matter, their total mass
emission rates for the U.S., the estimated mass median diameters
of particles emitted from such sources, and the total mass of
such aerosol from each source in the atmosphere at a given time.
These last numbers are obtained from estimated mean residence
times.
One can see from Table 1 (3) that even though the automobile
has the lowest total mass emission rate, the mass of particulate
matter in the atmosphere from the automobile is nearly an order
of magnitude greater than a source whose total mass emissions
are themselves an order of magnitude greater than the automobile.
-------
-550-
TABLE 1
ATMOSPHERIC PARTICULATE MASS
1.
2.
Primary
Source
Automobile
Petroleum
FCC Units
Mass Emission
Rate, Tons/yr. ,
U.S.
4x10 5
4. 5x10 5
Mass Median
Diameter, ym
0.4
0.5
Coal Fired
Electric
Utility,
Pulverized
3x106
Crushed Stone,
Sand, Gravel 5x1O6
10
20
Atmospheric
Mass, tons;
2x10"
1.5x10'
5x103
4x103
-------
-551-
V
s
i^M^H
V
Figure 1.
r S S
\
L
1
s s s
1
i
1
,
I
V
Figure 2,
-------
-552-
Thus, control of fine particles is even more critical than that
of larger particles if one seeks to limit mass of suspended
particulate matter in the atmosphere over large areas.
In the discussion which follows the fundamental aspects
of any particle control process are noted. Basic parameters for
particle collection — residence time and capture time are
discussed. Examples of basic processes of particle collection
are listed and some illustrations are given of their relation to
particle size and their possible efficacy for fine particle
control. Specific control devices will not be discussed nor
will reentrainment phenomena be mentioned.
Processes for Fine Particle Control
Consider the basic processes of particle collection.
Such collection involves the removal of the particles from the
suspending gas, usually, but not always, by deposition on a
surface, which could be, for example, the collection plate of an
electrostatic precipitator, a water droplet as in a wet scrubber,
and so forth. Of practical necessity, particle control processes
are dynamic processes and one is therefore concerned with such
variables as residence time of particles in a given device and
the time necessary for collection of the particle. If V is
the velocity of the particle-gas system through a given particle
collection device and S is the distance in the direction of
flow, then we can say that the residence time is of the order of
fa/v for the particles in the device. Clearly if one is to have
an efficient collection method it is necessary that the time for
capture of particles, tc, be substantially less than this mean
residence time for particles in the device. Figure 1 illustrates
the concept of residence time and Figure 2 that of capture time,
tc •
The time for capture of particles in the device can be
represented schematically as the ratio of some distance for
collection, L, to an average speed of migration, , owing to
some external force acting on the particle. The distance L could
be, for example, the distance an average particle must travel
to some collection surface. In an electrostatic precipitator,
this could be the order of the distance between collection plates
or in the case of a wet scrubber, it could be the average
distance between water drops. Therefore, L is a geometric
parameter connected with the design of a particular collection
device. The mean migration velocity , however, is a function
of the properties of the particle and the basic processes or
forces of collection. For these reasons we will focus attention
on and consider how this migration velocity is related to
particle size and the nature of the collection process. One can
then easily imagine a large number of geometrical arrangements
for carrying out particle collection.
-------
-553-
One can classify the basic control processes as either
primary or secondary. The primary control processes are those
which directly produce a particle migration velocity. Examples
are listed in Table 2 and include particle diffusion, sedimentation,
electrophoresis, thermophoresis, diffusiophoresis, etc. This
list is certainly not exhaustive. Secondary processes have
the principal function of altering the particle size distribution
so that the primary processes will be more efficient. Examples
of secondary processes are coagulation or agglomeration,
condensation of some molecular constituent on existing particles,
evaporation wherein one. seeks to evaporate the particle
completely, and finally electrical charging.
We can examine a few of these primary processes in terms
of their relative magnitudes and their dependence on particle
size.
Figure 3 presents the migration velocity associated with
centrifugation in fields of one and fifty times the gravitational
acceleration. It is noteworthy that fine particles have small
migration velocities even in relatively large centrifugal fields.
Figure 4 gives migration velocities associated with particle
diffusion in a particle laden gas flowing through a one-inch
diameter pipe. The migration velocity associated with Brownian
diffusion, which arises from random molecular impacts on a
particle, decreases rapidly with increasing particle size.
Turbulent diffusion of particles produced by transport of
particles by turbulent eddies is presented for comparison at a
Reynolds number of 101*. if Figures 3 and 4 are overlaid, one
observes, regardless of the conditions, a definite minimum at
^0.1 ym in the resultant value of . Therefore, processes
which depend on both diffusion and centrifugation (or impaction)
will have a pronounced minimum in their effectiveness at about
0.1 ym diameter in particle size.
Figure 5 compares the process of thermophoresis with that
of electrophoresis. The curve for electrophoresis, the move-
ment of electrically charged particles in an electric field,
represents the migration velocity for particles having approximately
50% of their saturation charge moving in an electric field of
4 kV/cm. One notes for electrophoresis again a minimum in
at the order of 0.1 ym diameter particle size, and a marked decrease
with decreasing particle size of the magnitude of . This is
to be compared with the thermophoresis of sodium chloride
particles to an 800 ym particle at a temperature difference
between particle and gas of 100 degrees C. The term thermophoresis
refers, of course, to the motion of a particle in a gas owing to
a temperature gradient imposed on the suspending gas. It can be
noted that the magnitude of owing to thermophoresis decreases
-------
-554-
Primary
Sedimentation
Impaction
Interception
Diffusion
Image Force
Phoresis:
Electro-
Magneto -
Thermo-
Diffusio-
Photo-
TABLE 2
CONTROL PROCESSES
Secondary
Coagulation
Condensation
Evaporation
Charging
-------
-555-
10'
T I0
b
-------
-556-
10'
DIFFUSION
u (Re = l04)
Figure 4.
-------
-557-
15
T 10
O
0)
(0
E
o
10"
ELECTRO-/THERMO-PHORESIS
10
ELECTRO (-50% Saturated,
4kV cm'1)
10
-I
10
D
THERMO (NaCI to
SOOMm Droplet,
AT = IOO°C)
I
10
10
Figure 5.
-------
-558-
with increasing particle size, in contrast to the example of
electrophoresis, and that indeed with this large imposed
temperature gradient the magnitude of owing to thermophoresis
is comparable to that of electrophoresis.
Figure 6 compares the magnitudes of for diffusiophoresis
and thermophoresis to an 800 ym water drop suspended in air.
Diffusiophoresis refers to the particle migration process induced
by the presence of a concentration gradient in the suspending
gas. The upper curve indicates owing to diffusiophoresis,
the positive sign indicating that the particle travels toward
the drop. The lower curve indicates the thermophoretic velocity
owing to the temperature gradient associated with the condensa-
tion process occurring on the water drop. This thermophoretic
velocity is in the opposite direction from and much smaller than
that owing to diffusiophoresis.
For these primary processes as well as those not listed
here, the magnitude of is critically dependent on particle
size. Also in most cases, particularly in those presented in
the preceding figures, those processes having large values of
for large particles have small values for small particles.
One may refer in particular to the processes of centrifugation
(impaction is implicit here) and electrophoresis, which form in part
the basis for the important control process for large particles
of electrostatic precipitation, wet scrubbing, and filtration.
Hence, in general, by increasing particle size one can more
efficiently utilize conventional particle collection processes
for fine particle collection. This observation forms the
motivation for the utilization of two of the secondary processes
in Table 2 — coagulation and condensation.
The desired effect from application of condensation or
coagulation in control processes would be to increase the mean
size of the particles in an aerosol. Both processes have the
desired effect, but differ in the rate at which the increase
occurs and in the nature of the alteration of the particle size
distribution of the aerosol.
For the coagulation process one can write the equation
describing the dynamics of change of the particle size distribu-
tion n(r,t) where r is particle radius and t is the time (4):
f r f °°
3n(r,t) = % I b (ri /r»)n(r')n(rll)drl- n / b (r,r' )n (r')dr'
3t J0 JQ
This relation for a spatially homogeneous system indicates that
particles are contributed to the size class r by collisions
between two particles of size less than r, as indicated by the
-------
0.7
-559-
DIFFUSIO-/THERMO-PHORESIS
0.6
Diffusio (+)
0.5
o
S> 0.4
£
o
$0.3
0.2
O.I
Thermo (—)
io-3 io-2
10" 10" 10' 10'
D (jim)
Figure 6.
-------
-560-
first term on the right hand side of the equation, and that
particles are lost from size class r by collision of a particle
of that class with any other particle in the distribution.
The mean size of the particle size distribution will increase
with time according to the approximate relation:
^ (3
for a total. mass concentration of particles y, a mean collision
frequency b, and particle mass density p.
One has possible control over the magnitude of the mean
collision frequency b appearing in this equations. One might
also try to alter the initial mass concentration, y, of particles
by the addition of a large particle fraction. A possible defect
in the utilization of coagulation for increasing the mean size
of particles lies in the fact that usually coagulation acts to
broaden the particle size distribution. That is,
the standard deviation 6 to the mean radius is an increasing
function of time in most situations. This increase in poly-
dispersity owing to coagulation obviously presents problems in
most collection processes inasmuch as we have seen that the
primary collection processes will differ widely with particle
size in their efficiency.
One can also write down a similar rate expression for the
change with time of the particle size distribution owing to
condensation of some molecular constituent on the particles :
3n(r,t) 8
- — = ~ "
, _. . , . .
taf (r)n(r,t)
Here a is a parameter which depends on the supersaturation
of the molecular constituent undergoing condensation, f (r)
is a function specifying the diffusional rate of condensing
molecules to the particles and is dependent on such factors as
the Reynolds and Knudsen numbers of gas dynamics. In contrast
to coagulation, all condensation processes act to diminish the
ratio of standard deviation to mean size for the particle size
distribution. This statement holds for all particle sizes (4).
-------
-561-
Conclusions
Up to the present, particle collection technology has
employed only a relatively small number of the primary and
secondary processes which we have listed in Table 2. However,
even in this somewhat restricted list one can pose hundreds of
possible combinations of primary and secondary processes to
effect fine particle collection. Additional investigation of
such possibilities should be encouraged as it seems probable
that improvement can be obtained in fine particle collection
efficiency.
Acknowledgment
Author wishes to acknowledge the support of the
Chemistry and Physics Laboratory, National
Environmental Research Center, Environmental
Protection Agency.
-------
-562-
REFERENCES
1. Friedlander, S. K., Environ. Sci. Tech.y 7, 1115 (1973).
2. Kerker, M., "The Scattering of Light and Other Electro-
magnetic Radiation", Academic Press, N.Y. (1969).
3. Hidy, G. M. and Brock, J. R., "The Dynamics of Aerocolloidal
Systems", Pergamon Press, Oxford, (1972).
4. Brock, J. R., Proceedings of the Faraday Society, (1973).
-------
-563-
Paper No. 24
SYSTEMS OF CHARGED PARTICLES AND ELECTRIC FIELDS
FOR REMOVING SUB-MICRON PARTICLES
by
J. R. Melcher
and
K. S. Sachar
MASSACHUSETTS INSTITUTE OF TECHNOLOGY
Cambridge, Massachusetts
-------
-564-
-------
-565-
ABSTRACT
Charged droplets, used as collection sites for charged particulate,
result in a class of devices which combine the characteristics of a conven-
tional scrubber and a conventional electrostatic precipitator. An overview
will be given on the fundamental time constants governing the performance of
these particulate control devices making use of charged drops or particles as
collection sites. Experiments will be described that support the time con-
stant picture of electrostatic agglomeration and self precipitation processes
among particles in the sub-micron range, in the super-micron range, and finally
among charged collection sites and the sub-micron particulate. A summary will
be given of what are felt to be the over-riding issues in making electrically
induced agglomeration processes a competitive approach to controlling sub-micron
particulate.
-------
-566-
-------
-567-
I. Introduction
Electrostatic precipitators and wet-scrubbers have been the
subjects of entire sessions in this symposium. Charged drop scrubbers,
which are the subject of this paner, are a hybrid of these devices.
Charged particles, typically exceeding 20 micron in diameter, are used
as collection sites for oppositely charged submicron particles. Al-
though we are also working on schemes that make use of solids as col-
lection sites, in this discussion the large particles will be water
drops. If compared to an electrostatic precipitator, the charged-drop
scrubber replaces the electrodes with drops. If compared to a wet-
scrubber, the effects of fine particle inertia, conventionally responsible
for causing impaction, are replaced by an electric force of attraction.
No attempt is made here to trace the history of charged-droplet
(1 2)
scrubbers since the pioneering efforts of Penney in the 1930's. '
Rather, remarks are aimed at i) giving a fundamental view of the "bottle-
neck" issues that dominate in a wide range of seemingly different devices
making use of electric fields induced by charge on systems of drops and
sub-micron particles, and ii) to describe experiments that demonstrate
the validity of this view. These experiments are designed to control
and vary essential parameters so as to isolate and test the mechanisms
governing scaling. In devices designed to achieve the highest possible
collection efficiency it is by contrast desirable to incorporate other
collection mechanisms such as those at work in the electrostatic pre-
cipitator and in the inertial impact scrubber.
We are fortunate to have as a participant in this symposium Dr.
M. J. Pilat, who is in the forefront of those attempting to show the
-------
-568-
practical feasability of charmed-drop scrubbers.
II. Critical Time-constants
If we key on pas residence-time in a control device as a major
factor in determining its competitive position, then it is possible to
characterize devices in terms of critical time-constants. For example,
in a conventional nrecipitator, the critical time-constant is the time
T required for the charged submicron particles havdnp, mobility b to
travel the distance s between electrodes in an imposed electric field
E . This is true whether the device is in laminar flow or, as is typical
S
of practical devices, in turbuleiat flow. In detail, the collection law
in these devices are significantly different, but basically they are
characterized by this same characteristic time T . For a precipitator
to be practical, T must be short compared to the pas residence time.
The precipitation tine T , as well as three time constants essential
to the performance of charped-drop scrubbers, are summarized in Fip. 1.
For reference, nomenclature is presented in Table 1.
Table 1. Summary of nomenclature
e E 8.85 x 10~12
o
number density charge mobility radius
drop N 0 B R
particulate n q b a
-------
-569-
If a volume is filled with particles charged to the same polarity,
then T* typifies the time required for these particles to self-precipitate
on the walls. This time is based on the system parameters of particle
number density n, particle charge q, and mobility b. Note that it does
not involve the dimensions of the device. As a variation of this con-
figuration, if a volume is filled with regions of positively charged
particles and other regions of negatively charged particles, then T*
typifies the time required for the oppositely charged particles to inter-
mingle.
If, to prevent space-charge fields, positively and negatively charged
particles are mixed, then T* (based on the density of one or the other of
the species) is the time required for self-discharge and perhaps self-
agglomeration of the particles.
The time required for collection of fine particles on oppositely
charged drops is typically T,. This collection time is also based on the
mobility of the fine particles, but the charge-density NQ of the drops
rather than that of the particles (nq). The residence time of the gas
must be at least of this order for the device to be practical.
For a system of charged drops, T plays the role that T* does with
R
respect to the particles. Thus, in a system of like-charged drops, TR
is the self-precipitation time of the drops, while in a system of oppositely
charged drops where there is no space-charge due to the drops, this same
time-constant governs how long the drops will retain their charge before
at least discharging each other and probably agglomerating.
A simple model that motivates the physical significance of T*, T,
d
and TR in the role of inducing self-discharge and narticle collection is
shown in Fig. 2. The equation of motion for this two particle model,
-------
-570-
included in Fig. 2, is easily solved. Then, by recognizing the relation
between initial interparticle spacing and particle densities, the three
time constants are obtained as limiting cases. These three limits are
summarized in Fig. 3.
(2 3)
The theory of Whlpple and Chalmers ' 'gives a detailed description
of the collection of a continuum of charged particles on a charged drop
in an ambient electric field and with a relative gas velocity. The dia-
gram of Fip. 4 shows that there are 12 possible collection regimes, deter-
mined by the net drop charge 0, gas slip velocity w and ambient electric
field E . The electrical current to the collection site associated with
o
the collection of charged particles is either i, or i2> depending on the
regime. These currents depend on (Q,E ). For example, in regimes (1),
(k) and (£) , where the drops have charge Q <-|Q | (Q the saturation
c c
2 +
charge 12TC RE) the electrical current is ij • - bnqO/eQ.
Thus, the rate at which particles are collected is
(1)
dt a (e /bNQ)
o
This equation makes it clear that the characteristic time constant for
the collection of fine particles is T ... The details of the collection
transient depend on the regine, but, so long as the drops are charged
significantly the characteristic time is essentially T ,. (It is useful
to recognize that even if the drops have no net charge, but are polarized
bv the ambient electric field so that they collect particles over a
hemisphere of their surface, Td is still the basic collection time nro-
vided that Q is interpreted as the charge on the collecting hemisphere.)
-------
-571-
The alternative roles of T* and T as self-precipitation tines in
K
systems of like charged particles or drops follows from the laws summarized
in Fig. 5.
Consider now some possible configurations, all making use of fields
induced by charges on fine particles and drops.
a) The volume of the control device is filled with oppositely charged
particles only. Self-agglomeration proceeds at a rate characterized
by T*, but leads to little increase in size with each agglomeration.
Since the agglomerated particles must be recharged to achieve a sig-
nificant increase in size, the process is generally too slow to be
practical.
b) The volume is filled with fine particles charged to a single
(A)
polarity. This is the space-charge precipitator , which can be
regarded as a variant of the conventional precipitator. Hence, It is
competition for the electrically augmented scrubber.
c) Drops charged to one polarity and fine particles charged to opposite
polarity with drops dominating the volume charge density. Then the
drops are lost from the collection volume in time TR. Because the
mobility of the drops is generally much greater than that of the
particles, T « T and hence the drons have a residence time that is
R d
short compared to the time required to remove a significant fraction of
particles. This means that charged drops must be resupplied to the
volume many times during the gas residence time to achieve effective
cleaning.
d) Drops are injected of opposite polarity and particles with oppo-
site polarity. Then, there is no self-precipitation either of the drops
-------
-572-
or of the particles. However, the self-discharge among the drops
occurs with time constant TR, and hence the drops are as effectively
lost as collection sites as in case (c). Again, drops must be
resupplied many times during one pas residence time to achieve
effective cleaning.
e) Oppositely charged drops and particulate are injected with
sufficient particle charge density to achieve space-charge neutrality.
In this case, the drops are used efficiently, since they do not self-
precipitate or self-discharge. Rather, the drops lose their charge
by collecting particles. Such a configuration is possible only if
the fine particles are very dense. Note that space charge neutrality
means that NQ - nq so that T* - T This means that the collection
time is of the same order as would be obtained in a space-charge
precipitator (particles injected without the drops). The difference
is essentially in the space charge precipitator, the particles end
up on the walls whereas in the charged drop scrubber, they are col-
lected by the drops.
III. Theoretical Performance of a Controlled Experiment
The collection volume for a controlled experiment is shown schematically
in Fig. 6. Mbnodisperse drops with acoustically controlled size (50 micron
diameter) and electrically controlled charge are injected vertically at
10 in/sec at the top. Gas is injected horizontally at the top with intrained
fine particle density n. and removed at the bottom where the particle
density is reduced to n . Momentum is transferred from the injected drops
to the gas so that in the absence of drop charge, the drops form a fully
-------
-573-
developed jet by the time they reach the bottom of the central interaction
region. The drops are found to slow to the pas velocity in about 1/3 of
the device length. Mean gas and drop velocities in the center interaction
region are 1.4m/sec. The collection volume is baffled so that the gas
recirculates upward along the sides in a feedback loop that is driven by
the drop jet (c/d - 1). This feedback loop insures that all of the gas
is subject to drop cleaning the same number of times. Because Tn « T,,
R d
the gas residence time is made much longer than the drop residence time.
The gas typically circulates 50 times during one gas residence time.
The experiment is designed so that all of the configurations described
in Section II can be tested. Here, remarks are confined to experiments
aimed at configuration (c), with configuration (b) playing, an inadvertant
role. That is, electrically induced scrubbing is dominant in determining
the one pass particle removal n , = ^K/11 whereas self-precipitation is
dominant in determining particle removal in the feedback sections represented
by ncd = nd/nc.
The system equations representing conservation of particles in the
mixing regions at the top and bottom are summarized in Fig. 6. It is
assumed that the volume rate of gas flow through the device, FV> is small
compared to the recirculation volume rate of flow F . Thus, these system
o
equations can be solved for the overall efficiency expression given in
Fig. 7.
The laws used to determine the "one-pass" efficiency in the scrubber
region, n ,, and one pass efficiency r\ , in the space-charge precipitation
feedback regions are summarized in Fig. 8. These expressions are based on
the model of a fully developed turbulent flow in each of the regions. By
-------
-574-
defining the step- function U.(E ) as unity in the drop-scrubbing region
where the electric field at the wall is positive (here the drops are posi-
tive and the particles are negative) and as zero in the feedback repions
where L is nepative, the same expressions pertain to either region. Of
w
course in the feedback regions there are no drops, and hence N = 0 for
these repions. In writing these expressions, it is assumed that the
collection of particles on drops is described by i_, defined with Eq. (1).
Moreover, it is assumed that the respective regions are in a state of
fully developed turbulent flow so that the simplified quasi-one-diraensional
expressions apply.
With the further assumption that in the drop scrubbing region the
charge density of the drops dominates, the one pass efficiencies summarized
in Fig. 9 follow. Also included in the figure is the expression for the
decay of drops in the interaction region. These expressions illustrate
how the characteristic times discussed in Section II turn up in specific
configurations. The one-pass self-precipitation of particles with position
z measured from the location (c) in Fig. 6 upward is characterized by the
length I* which will be recognized as UT*. Similarly, in the interaction
region, the drop density decays in number density as z increases from
point (a) in Fig. 6 downward, with the characteristic length £_ « UT.,.
K R
The one pass scrubbing efficiency in this region, which has a total length
£, is determined by a combination of £R and a length fc. which represents
the rate of particle collection on the drops....Ui,.
As the drop charge is raised, both the tine constant for particle
collection, T. and that for drop self-precipitation, are decreased. Thus,
it is expected that the one pass efficiency shows an optimum. In fact this
-------
-575-
optimum is characterized by the two charges Q, and Q_, and is obtained
by making the drop charge 20_. The optimum efficiency and drop charge
K
are summarized in Fig. 10.
IV. Experimental Observation
The experimental system is shown schematically in Fig. 11. Included
are diagnostic components for accurately determining particle number
density, mobility, volume rate of flow, drop size, charge and rate of
injection. Not shown is the apparatus used to determine the feedback volume
rate of flow found by making anemometer measurements in the feedback
repion.
The drop charge is induced by weans of inducer bars next to the
acoustically driven orifices. Because the rate of drop injection is con-
trolled acoustically, measurement of the current carried by the drops as
a function of the voltage on the inducer bars, vdro » makes it possible to
determine the drop charge, given the inducer bar voltage. This relation
between drop charge and inducer bar voltage is shown in Fig. 12.
According to the self-precipitation model for the drops, the electrical
current I intercepted by an electrode placed at the position z (measured
from a in Fip. 6) in the interaction region is given by the relation sum-
marized in Fig. 13. The drop charge is normalized to the charge Q[ which
characterized the system. At low values of drop charge, the current should
increase linearly. But then, as 0 is increased beyond Q^, the self-
precipitation should lead to a decrease in the drop current. Measurements,
shown in Fig. 13, give this expected relation. Also, the peak in current
as a function of V, (proportional to Q) shifts to the right as the
measurement position z is increased.
-------
-576-
The theoretically predicted performance of the device in removing par-
ticles is shown in Fig. 15. With the charge density used in the experiments,
the self-precipitation of the particles is an important factor, in fact ac-
counting for about 80% removal of the particles with the particles charged
but the drops uncharged. Thus, with the inducer bar voltage V - 0,
n ../n. - 0.2. In the limit where the fine particles are very tenuous,
out in
n /n. minimizes at about the same value of V, , but goes to unity as
out in drop
V, •* 0.
drop
The experimentally observed removal is also given in Fig. 15. The
observed removal peaks at 93%, as compared to the predicted peak of about 94%.
Refinements to the theory include using as n the number density corrected
for losses due to inertial impact (about 20%) in the entrance region. It is
important to realize that there are no empirical inputs to the theory. Con-
trol over the hydrodynamics is the rnnin difficulty in further refinements of
the experiment. For example, the distance required to have a fully developed
turbulent jet is a function of drop charge, and hence the flow structure
varies somewhat with V. . This mav account for the somewhat slower observed
drop
decrease in n /n. with V, than would be theoretically expected. By
out in drop
far the most important confirmation is of the optimum in the efficiency curve
as a function of drop charge. Both theory and experiment optimize in the
range of an inducer bar voltage of 30 volts. Although not shown, efficiency
measurements were made out to a value of V. exceeding AOO volts, with the
drop
observed efficiency found to continue to slowly degrade. Clearly, far more
charge can be placed on the drops than is desirable.
With particles in the size range of 2 micron, the device functions
efficiently as an inertial scrubber. In the size range of 0.6micron used
for these experiments it is clear that the electrically induced scrubbing
can dominate inertial scrubbing.
-------
-577-
This apparatus can also be used to test our understanding of the
other configurations outlined in Section II. For example, oppositely
charged drops can be injected either by making the inducer bars of opposite
polarity so that they induce charges of opposite sign on the two rows of
drops injected or by making V. a symmetric square-wave. The dual sig-
nificance of the drop self-precipitation time or self-discharge time makes
it clear that the limitations on particle removal efficiency should be
very similar to those found here with drops of a single polarity.
-------
-578-
V. Concludinp Remarks
The experiments support the assumption made at the outset in this
discussion that in the submicron range, the electrically induced impaction
can easily dominate that due to Inertia. So, when compared to the inertlal
impact scrubber, the electrically augmented device is indeed attractive.
The fact that self-precipitation can make such an appreciable contribution
to the removal efficiency supports the view that the charged drop scrubber
is poor competition for the conventional precipitator. What we have
emphasized here is that although the charped-drop scrubber appears closelv
related to the conventional precipitator, in fact it is subject to different
limitations. The drops, unlike the electrodes of the conventional precipi-
tator, are not fixed, either in charge or in mechanical position!
It is felt that in applications where the conventional precipitator
is the competition, the charged-drop device is a poor contender. But, in
applications where scrubber technology is appropriate, electrical augmenta-
tion of the scrubbing is attractive.
-------
-579-
Finally, it should be remembered that our remarks have been con-
fined to charged drops as collection sites. Fundamentallv, it is diffi-
cult to make charp.ed drop devices compete with the electrostatic pre-
cipitator because of limitations on the drop density resulting from the
tendency of the drops to self-discharge or self-precipitate. Breakthroughs
in residence time for efficient cleaning are possible by usinp dense
systems of particle collection sites in an imposed electric field. Either
the sites are continuously recharped by electrically induced collisions, or
they function as sites through polarization in the imposed electric field.
Such devices could outperform the electrostatic nrecipitator in terms of
residence time, while providing new alternatives in solving problems such
as the removal of high resistivity particles.
-------
-580-
-------
-581-
References
1. Penney, G. VJ., U.S. Patent 2,357,354, "Electrified Liquid Spray Dust
Precipitators," (1944).
2. Melcher, J. P.. and K. S. Sachar, "Charged Droplet Technology for
Removal of Particulates from Industrial Cases," Final
Report under Task No. 8, E.P.A. Contract #68,002-0018,
(1971).
3. Whipple, F. J. W. and J. A. Chalmers, "On Wilson's Theory of the
Collection of Charge by Falling Drops," Quart. J. of the
Roy. Met. Soc.. 70, (1944), pp. 103-119.
4. Hanson, D. K. and C. R. Wilke, "Electrostatic Precipitator Analysis,"
I & EC Process Design and Development, ji, #3 (1969), pp.
357-364.
-------
-582-
Desiftnation
Description
Schematic
TP = bE,
T* =
— -
bqn
Precipitation time
for particle in
conventional preci-
pitator having electrode
spacing s
Particulate self-
agglomeration or
self-precipitation
time.
q
q q
t t
GO Gh)
d = bQN
T = °
R ~ BQN
Particle-particulate
collection' time
Drop discharge or
self-precipitation
time
00
Q
Fig. 1 Summary of time-constants governing performance of conventional
electrostatic precipitator and a wide range of agglomeration devices.
-------
-583-
Two-Particle Interactions
Charge Q
Charge q
(qB + bQ) 1
dt
6-rrna ' " 6irnR
Fig. 2 Two-particle model for electrical interactions
-------
Initial Spacing £ » (R + a)
o
•ff
6 Td
T -
2ir
l
Cn
oo
>£>.
I
Fig. 3 Limits of collection time for two particle model that pive the
three time constants when these are interpreted as self-discharpe and
drop-particle agglomeration times.
-------
-585-
particles enter z -»• + °°
particles enter z -*•
4 Whipple and Chalmers model for collection of positively charged particles
having mobility b on drop carrying charge 0 in ambient electric field E0 with
relative gas velocity wo. E0 and wo are respectively defined as positive and
negative in the z direction. With increasing gas velocity, the vertical line
of demarcation indicated by wo moves to the right. Initial charges, indicated
by . , follow the trajectories shown until they reach a final value given by x.
If there is no charging, the final and initial charges are identical, and are
indicated by ^ . The inserted diagrams show particle trajectories.
-------
-586-
Self-Precipitation
T T * t * t » t . t * t
+ + +
J, if ^ * ^ I
Gauss' Law: E
w 2e
o
Conservation of Charge:
It (nq8) " * 2nqbEwall
Together, for System of Charged Particles
dn n_ * o
dt " " * ; T " nqb
For System of Charged Drops:
M. S_ . . £o
dt " ~ T_ ' TR " NQB
R
Ewall
Fig. 5 Summary of simple model used to illustrate role of T* and T_ as
R
self-precipitation times for like-chareed particles and like-charped
drops respectively.
-------
-587-
n. part/m"
F m /sec
depth w
n n.
a d
U
"in Fv - 2wcUna -
n F - w2cUn. - w2d(f U)n
out v b d c
n
n
out
nout part/m
3
F m /sec.
Fig.
6 Schematic cross-sectional view of cleaning
volume. Drops are injected at top to form
turbulent jet that drives gas downward in the
central interaction region. Drops are removed
by impaction at the bottom, while the baffles
provide a controlled recirculation of gas and
particles. The equations describe the ba-
lance and mixing of gas streams at the inlet
and outlet regions.
-------
-588-
nout nab
n. F
in i+_£(i- n n
ab cd
F = 2cwU = 2dw(y U)
g d
n =
nab - n
a
ncd E
nd
i
Flp. 7 Summary of equations which follow from those given in Fip. 6 in
describing the system performance in terms of one-pass efficiencies
nab md ncd"
-------
-589-
Drop Conservation
.„ NQE
dN _ w
dZ
Drop Charge Conservation
dO nobO
dZ * ~ e U
o
Particle Conservation
j L™ U , (-E )
dn nbQN w -1 w
dZ e U cU
o
Gauss' Law
• (NQ - nq) |
o
Fig. 8 Summary of lav/s used to determine the one-pass collection of
particles on drops in the interaction region and self-precipitation
of particles in the feed-back regions. Model assumes a fully
developed turbulent flow in each region.
-------
-590-
Space Charge Precipitation
•M * O
Particulate n . " J 5 & • r
cd i , z ban
1 + ~* «
4
Drops
„ e u
"o
Fig. 9 Summary of one-pass efficiencies and drop density distribution
found using the model summarized in Fij». 8.
-------
-591-
In Region NO » nq
Optimum
-(1/2) (^)
QR
0, = 6irnRb
d
£N
o
Qopt - 20R
Fig. 10 Optimum one-pass efficiency of drop-particle scrubbing. For optimal
collection, the drop charge should be 2Q.,, and In that case the
R
efficiency is the value of n given.
flD
-------
Condensation
Aerosol
Generator
o
o
"Owl1
v :- //
R'
I 1
.Transducer and
Orifice Plate
Feedback Regions^
Scrubbing Region —.
I I
Laser
Extinction Cells
FiR. 11 Schematic of experiment, showing system for controlled generation
of charged particles and measurement of essential parameters.
J
n
I
Cn
-------
4-.
Drop Charge
Q(C x 1014) 3 - .
H H-
H r
0
H -*-
20
AO
60 80
100
I
ui
VD
00
Vdrop
-------
-594-
Drop Precipitation
I - C
1 +
; o,
U6irnR
N
Fi£. 13 Schematic of electrode used to intercept some of the drop current
by impaction. Theory predicts the dependence on drop charge Q and
electrode position z shown.
-------
-595-
I (Ma)
0.1 .
100
150
200
drop
Fi«». 14 Measured current as function of V, , which is proportional to
drop
Q at four different positions z.
-------
0.20 --
0.15 --
nout/nin
experimental
0.10 __
0.05
drops
40
(volts)
50
60
70
80
Fig. 15 Theoretical and observed overall efficiency of svstem uti-v, *-h
-
corrected for diluation and inertial inaction in the entice region?"
1
Ln
-------
-597-
Paper No. 25
ADVANCES IN THE SONIC AGGLOMERATION OF
INDUSTRIAL AEROSOL EMISSIONS
by
David S. Scott
UNIVERSITY OF TORONTO
Toronto, Ontario, Canada
-------
-598-
-------
99-
ABSTRACT
The agglomeration o£ industrial aerosols by finite-
amplitude sound is well known, although the process has enjoyed
limited commercial application resulting primarily from high
specific power requirements (and to some degree high capital
costs). These limitations can be attacked by either more effec-
tive acoustic field configurations, and/or more efficient sound
generation. The present paper outlines some of the results of
recent studies carried out at the University of Toronto and the
Ontario Research Foundation on both approaches to the solution.
A brief review of the first-order mechanisms of
acoustic agglomeration is followed by an indication of the
theoretical and practical advantages of progressive saw-tooth
waves as compared with the conventional standing-wave config-
uration. The suitability of using a "resonant pulse-jet" to
generate sound is discussed from the point of view of sound
generation efficiency, waveforms, sound intensities and reli-
ability. Estimates of annual operating costs (agglomeration
plus subsequent collection) using a "pulse-jet" agglomerator
are given. A very brief outline of the work of the Braxton
Corporation on acoustic agglomeration is also presented.
-------
-600-
-------
-601-
INTRODUCTION
The phenomenon of acoustical enhanced aerosol agglom-
eration has been known for about four decades. For example, as
i
early as 1935, Brandt and his colleagues had studied acoust-
ically enhanced agglomeration of tobacco smokes, ammonium
chloride vapours, and paraffin oil fogs. Following the second
2
world war, several American investigators, such as St. Glair
3
and Boucher , carried out laboratory and field studies. More
recently, a number of Soviet scientists have conducted contin-
uing and extensive research on the phenomenon. Of the many
excellent Soviet scientists working in the area, probably the
i»
best known in the English-speaking world is Mednikov , whose
well known monograph on the subject was published in English
in 1965. Since the preceding is the barest outline of the
substantial literature in this field, it is clear that I could
not presume to present acoustic agglomeration as a "new concept",
in spite of this paper finding itself placed in such a session.
Nor could I claim to be capable of telling you about all the
recent advances, or even the most significant recent advances, -
as these could well have occurred in the Soviet Union. I will,
however, discuss what we believe are some new approaches to
sonic agglomeration which we have been following at Toronto. And
while so doing, I will use this discussion as a vehicle to review
the "motivation for", "a bit of how-it-works", and "certain other
aspects" of the process.
-------
-602-
The presence of finite-amplitude sound in an aerosol
acts to increase the coagulation rate of that aerosol. More-
over, under fairly readily achieved conditions, this rate
increase can be substantial. A finite-amplitude acoustic field
can be said to exist when the acoustic mach number, e ,
becomes non-negligible. This might be said to occur when
-2
e » 10 , which corresponds approximately to a sound intensity
level of 154 db under standard atmospheric conditions. Acoust-
ically enhanced agglomeration is achieved by bringing about a
manyfold increase in particle-particle contacts or, what is the
same thing, particle-particle collision frequency. As such, it
is a process by which a fine particle, high number density
aerosol is changed into a coarse particle lower number density
aerosol. As such, I believe we should think of acoustic agglom-
eration as a dust conditioning process. If the title of this
paper had referred to "acoustic dust conditioning" , it would
have better indicated where the process fits into a "systems
approach" to air cleaning technology, but would have less well
indicated "what it does".
I believe the advantages of acoustic dust conditioning
fall into three primary categories:
(a) The process is especially suitable for high dust load
ultrafine particulate matter.
(b) In principle, there are no restrictions on resistivity,
explosiveness, temperature, stickiness, or other aerosol charac-
teristics which sometimes cause difficulties with conventional
-------
-603-
equipment.
(c) The process lends itself to installations which supple-
ment existing equipment, thereby improving the overall process
efficiency at a fraction of the cost of an entirely new system.
In view of the preceding technical advantages of the
process, in my opinion, the limited commercial application of
acoustic agglomeration has been due to high operating costs
resulting from high specific power requirements and high capital
costs.
The solution to the high power requirements:would
appear to fall into two general categories. First, we might
seek more effective acoustic field configurations in order to
bring about a greater increase in the coagulation rate for the
same acoustic intensity. Secondly, we can direct attention to
sound generation devices, in the hopes of bringing about an
improvement in the efficiency of sound generation.
The solution to capital costs appears more difficult
to subdivide. Rather, is probably best approached by noting
that when we are seeking advances in either, the acoustic field
configuration, or energy conversion efficiency of finite-amplitude
sound generation, that we do so with an eye to capital cost
implications. At least that has been our approach at Ontario
Research Foundation, with the result that the remainder of this
paper primarily subdivides into two headings; firstly, "Acoustic
Field Configuration" and secondly, "Sound Generation". Capital
cost implications will be discussed within these sections as we
proceed and where appropriate.
-------
-604-
ACOUSTIC FIELD CONFIGURATION
To date, the accepted acoustic configuration has been
standing
an essentially/* sinusoidal wave "tuned" to the aerosol being
treated in order to maximize the particle-particle differential
or "orthokinetic" motion. In my view, orthokinetic interactions
may be regarded as the predominant "first order" mechanism
responsible for agglomeration, and occur when two or more
particles of different sizes, which are close together, are
located with their line of centers substantially parallel to
the gas vibration (i.e., orthogonal to a wave plane). Clearly,
particles of different sizes vibrate with different amplitudes
and phases. Hence, there is a differential motion established
between such particles which increases their collision prob-
ability.
A simple way of looking at this process is to consider
a particle sufficiently small that it moves with the gas and a
particle so large that it is essentially unperturbed by the
gas motion. In this case, the "largeness" or "smallness" of
a particle should be defined in terms of the non-dimensional
group GOT , where u> is the acoustic frequency in radians per
second, and T is the particle dynamic relaxation time.*
Corresponding to a very small particle WT •*• 0 , and correspond-
ing to a large particle WT + «» .
However, the sinusoidal wave imposes limitations on
the effectiveness of such interactions due, primarily, to the
rather sharp transitioij from E •»• 0 to E -*• 1 which occurs
over a particle radius change of order 10 as seen from Fig. 1.
-------
-605-
These limitations are as follows:
(a) It is very important that the acoustic frequency be
chosen to match the aerosol being treated, in order to "split"
the particle size distribution at a point which maximizes the
orthokinetic differential motion. This requires aerosol diag-
nostic and control equipment which can substantially add to
capital costs.
(b) Even when the frequency is "tuned" such that E - 0
(essentially equivalent* to WT - 1) at the "mean" particle size,
the frequency necessarily becomes "mistuned" as the size distri-
bution evolves as a consequence of the acoustic treatment. In
7
response to this difficulty, Mednikov has proposed multiple
stage agglomeration.
(c) Finally, and even if the optimum "tuning" could be
achieved, each of the two sets of particles outside the set of
particles for which 0.05 < E < 0.95 experience little differ-
ential motion among themselves. Although this is probably the
least significant limitation of standing sinusoidal acoustic
fields, it could conceivably become important in aerosols exhib-
iting very broad size distributions.
In view of the preceding, we proposed the use of a
series of progressive low-amplitude shocks (saw-tooth waves).
We believe that, from the point of view of optimizing orthokinetic
interactions, this alternate waveform exhibits two advantages
over the standing sinusoidal waveform. These advantages can be
illustrated by reference to Fig. 2, which gives the dynamic
response of different size particles to the passage of a step
-------
-606-
velocity change in the bath gas., for the case of negligible
*
particulate mass loading, that is M -»• 0 . The figure also
assumes Stokes drag, which is, of course, inappropriate for
the finer particulate sizes. Although this physical circum-
stance differs from that of a series of saw-tooth waves and
finite particulate loading, and further deviates for the fine
particulate sizes when Stokes drag law is inappropriate, the
essential features with regard to optimizing the orthokinetic
differential motion between different particle sizes are the
same, and are better illustrated in this more simple case.
The advantages of the progressive saw-tooth wave train are as
follows:
(a) All particles initially experience a differential
motion with respect to the gas as a consequence of the dis-
continuity in the gas velocity with the passage of each wave-
front. But since the different size particles have different
dynamic relaxation times, T , all particles have a period
during which they move differentially with respect to all
others of different sizes. Considering Fig. 2, it is seen that
at t - ti , the set of particles of r < 0.04 vim are essen-
tially moving with the gas, while the set of particles of
r > 1.0 pm are essentially stationary. Later, at t = t2 ,
the set moving with the gas has increased to include all par-
ticles of r < 0.2 urn and the stationary set has decreased
to include only those particles r > 5 ym . And so on. Thus,
the passage of each saw-tooth wave "sweeps" the size distribu-
tion, in contrast to "splitting" the size distribution as occurs
-------
-607-
in the standing sinusoidal field.
(b) The second advantage of progressive saw-tooth waves
follows directly from the first. That is, no "tuning" of the
fundamental frequency is necessary. It is sufficient that the
relaxation time of the largest particle be less than the inverse
of the fundamental frequency. In practice, this requires only
that f < 500 Hz.
Cc) Thirdly, in the standing sinusoidal field and neglect-
ing drifts, all particles remain in the same mean position with
respect to the gas, and hence remain in the same mean position
with respect to each other. Conversely, in a progressive saw-
>
tooth wave train, different size particles change position with
respect to the gas by different amounts and in accordance with
their different relaxation times. As such, there is a systematic
shifting of the relative position of different size particles
with respect to each other, with each wave. Such a redistri-
bution has no analog in the conventional standing field.
The preceding illustrate that by changing the wave-
form to a train of finite-amplitude progressive saw-tooth waves,
the orthokinetic sub-mechanism of acoustic agglomeration can be
expected to be optimized and, in addition, a. new sub-mechanism
has been introduced. The new sub-mech is the systematic re-
distribution of different size particles with respect to each
other.
Before leaving the matter of the acoustic field con=
figuration, we should note that there are several sub-mechanisms,
notably acoustic drifts, self-centring processes, etc., and they
-------
-608-
cannot all be independently maximized in a single acoustic field,
However, by introducing two mutually perpendicular fields, the
first a progressive saw-tooth wave train as we have just dis-
cussed, and the second a standing sinusoidal field, it can be
shown by qualitative arguments similar to the preceding that
it is possible to optimize all the known sub-mechanisms which
exist in conventional standing fields, while at the same time
introducing several new sub-mechanisms, which do not exist in
standing single acoustic fields. Very simply, in this system
regions of concentrated aerosol are generated in the nodal/anti-
nodal planes of the standing field (aerosol striations), and
the shocks are run along these striations. Details of this
8
more refined acoustic configuration are given by Scott.
Although this configuration holds certain process advantages,
the primary current use appears to be in experimental isolation
of the various sub-mechanisms of acoustic agglomeration. Early
experimental results from such a facility at the University of
Toronto (shown schematically in Fig. 3) indicate that (a) the
aerosol striations are predominantly acoustic circulation
induced, (b) the progressive saw-tooth wave does not break up
the striations, and (c) for the same sound intensity the
single progressive saw-tooth waveform brings about a greater
coagulation rate increase than a single standing sinusoidal
field. In these experiments, the sound pressure levels were
up to 154 db in the standing field and 149 db in the progressive
field. The basic frequency of the progressive field ranged
between 500 - 2500 Hz. and for the standing field between
980 - 3155 Hz.
-------
-609-
Returning to the matter of capital costs, and con-
centrating specifically on the single progressive saw-tooth
wave train we note the following.
(a) Acoustic frequency tuning control requirements are
eliminated.
(b) There is no need for special agglomeration chambers
which involve adjustable lengths, in order to maintain an
integral number of half wavelengths as the speed of sound
through the aerosol changes with the loading and size distri-
bution of the inlet dust.
(c) Taking the above to its logical conclusion, there
is really no need for an agglomeration chamber at all. One
simply looks for a convenient elbow in the flue gas line, and
uses this point for the insertion of the sound horn.
SOUND GENERATION
In view of the discussion on waveforms, it is apparent
that we would like a device which generated a waveform with the
essential features of a progressive saw-tooth wave train, and
accomplished this at a high energy conversion efficiency. We
have investigated the use of a resonant pulse-jet (not dis-
similar from a World War II VI rocket engine) in both a (a)
valved combustion chamber configuration and (b) a no-moving-
part combustion chamber using aerodynamic valving. The advan-
tages of this acoustic generator for. sonic agglomeration appear
to be as follows:
-------
-610-
(a) Although the wave-form is rough, it contains the
basic saw-tooth features for enhancing the particle-particle
collision frequency.
(b) There is a direct conversion of fuel energy to
acoustic energy which implies that this might be carried out
with a relatively high overall efficiency.
(c) The capital costs of the equipment is very low.
Figures 4 and 5 show an overall view of the facility
used to study the feasibility of a pulse-jet generated acoustic
field at Ontario Research Foundation. The size of the facility
points up the difficulty of scaling the pulse-jet to normal
laboratory dimensions. To-date, a ZnO fume aerosol has been
used exclusively. Figure 6 illustrates typical particle size
distribution results prior to, and subsequent to, sonic treat-
ment. The residence time in these cases was about 2 seconds.
The no-moving-part pulse-jet appears to run indefinitely.
Table 1 presents the range of values of the more important
parameters involved in this study, as well as the indicated
total annual loss for a typical installation including capital,
maintenance and operating costs. These estimates were based
upon using a cyclone as the ultimate collecting device.
In summary, the status of this program appears to be
that both technical and commercial viability are indicated, in
the sense that dusts can be agglomerated to a degree which would
be useful in certain emission control installations and this
can be accomplished with a cost range normally acceptable in
-------
-611-
such circumstances. Work from this point on will concentrate
on three basic areas.
(a) Significant debugging and parameter optimization
must be carried out.
(b) The coagulator must be viewed as a conditioner and
optimization studies with cyclones, wet-scrubbers, and fabric
filters must be conducted. It is speculated that sonic-co-
agulator/wet-scrubber combinations might hold the most promise.
(c) Work must be carried out on determining the most
effective means of accomplishing ambient sound isolation, with
particular emphasis on capital costs.
OTHER COMMERCIAL SCALE INSTALLATIONS
The only other sonic agglomerator development work
in North America which I am aware of,, and which is on the
order of a commercial scale unit, is that which has been carried
out by Braxton Corporation, Medfield, Massachusetts. This
system, which is referred to by Braxton as their AVP (alter-
nating velocity precipitator) uses the traditional standing-
wave acoustic field and agglomeration chamber. From what little
I know of the system, it appears to be an exceptionally well
engineered unit. The system treats both particulate matter
and noxious gases such as S02 and NO and, in most circumstances,
utilizes the addition of a sodium carbonate "fine spray"
solution to enhance the collection efficiency. It is my under-
standing that the Braxton system is capable of treating
-------
-612-
15,000 ac£m at 300° F. Residence times are in the order of
1/2 seconds and the sound intensity varies between 166 - 170 db
at frequencies of 250 - 500 Hz. The installed capital costs
for the unit are estimated at $4.00/acfm and I have no infor-
mation on the power requirements. Undoubtedly, further infor-
mation could be obtained from the Braxton Corporation.
FOOTNOTES
*
The acoustic mach number, OJT product, mass loading ratio
and other non-dimensional groups important in finite-amplitude
acousto-aerosol interactions are discussed in Ref. 5 and 6.
It should be noted, however, that there must necessarily
be a lower limit on the dust loading for which the process is
suitable. Typically, this lower limit can be expected about
1 grain/ft.3, however, is somewhat dependent upon the mean
particle size which is required after coagulation.
REFERENCES
1. Brandt, 0. and Freund, H., "Uber die Aggregation von
Aerosolenmittels Schallwellen", Z. Phys. 94 (5-6)
pp. 348-355; (1935).
2. St. Clair, H.W., "Agglomeration of Smoke, Fog or Dust
Particles by Sonic Waves", Indust. Eng. Chem. 41(11),
pp. 2434-2438 (1949).
3. Boucher, R.M.G., "Acoustic Energy in Fog Dispersal
Techniques", Ultrasonic News 4(1), pp. 11-19 (1960).
-------
-613-
References (continued)
4. Mednikov, E.P., "Acoustic Coagulation and Precipitation
of Aerosols", translation from Russian by Consultants
Bureau, New York, 1965.
5. Davidson, George A., and Scott, David S., "Finite Amplitude
Acoustic Phenomena in Aerosols from a Single Governing
Equation", J. Acous. Soc. Amer., 54(5), pp. 1331-1342
(1973).
6. Davidson, George A., and Scott, David S., "Finite-Amplitude
Waveforms in Aerosols", J. Aerosol Sc., 5(1), (1974).
7. Mednikov, E.P., "Method of Acoustic Coagulation of Aerosols
USSR Patent 149399 (1962).
8. Scott, D.S., "Method of Coagulating Aerosols", U.S.A.
Patent 3,771,286 (1973).
-------
-614-
FIGURE CAPTIONS
Figure 1; Entrainment of particles by the gas motion, E ,
versus particle radius in ym , r , for several frequencies
of sinusoidal waves. The particles are spherical and of
1 gm/cm3 density. The gas is air at approximately STP.
Figure 2: Entrainment of particles by the gas motion, E ,
versus time, t , since the passage of a low amplitude step
velocity change in the gas for several particle sizes. The
particles are spherical and of 1 gm/cm3 density. The gas is
air at approximately STP.
Figure 3: Schematic of U. of T. striated-shock acousto-aerosol
channel. The aerosol enters at upper right and exists at
lower left. A progressive saw-tooth field is produced by the
shock field generator and propagates down the channel to be
absorbed in a relatively inefficient anechoic base of baffels
and steel wool. The agglomeration chamber is 7.6 x 45.7 x 137.2
cm.
Figure 4: View of ORF pulse-Jet agglomerator facility.
(A) ZnO fuming chamber (B) Primary data readout station
(C) Agglomeration chamber (D) Cyclone.
Figure 5: View of ORF pulse-Jet agglomerator facility.
(A) ZnO fuming chamber (D) Cyclone (E) Pulse-Jet inserted in
elbow (F) Fan (G) Baghouse.
Figure 6: Size distribution of treated and untreated ZnO
aerosol, given in terms of cumulative percentage by particle
mass versus particle aerodynamic diameter.
-------
h
ro
0
0.1
0.2 Q3 0.5 0.8 1
5 6 8 10
20
-------
E 1.0
K
ro
i
a\
-------
-617-
SHOCK FIELD
GENERATOR
OPPOSING
STRIATION FIELD
GENERATORS
Figure 3.
-------
-618-
Figure 4
-------
-619-
Figure 5.
-------
-620-
80
70
60
50
40
30
20
10
COARSE PARTICLE AEROSOL
PRIOR TO SONIC TREATMENT
\ \ AFTER 0.8 sec.0.06 BAR SONIC TREATMENT
\\ \ AFTER IBS (or 2.65) sec,0.06 BAR TREATMENT
x^ \ AFTER 2.65 sec,O.I BAR SONIC TREATMENT
J L
I
L5 2
8 10
15
AERODYNAMIC DIAMETER
Figure 6.
-------
-621-
ENERGY CONVERSION (fuel to sound) EFFICIENCY
Flap -valve unit
No-moving-part unit
Comparitive high- % siren
14%
7%
4%
ACOUSTIC POWER AND INTENSITY
Power
Intensity
0(10 k watts)
0(160 db.ref I0~l6watts/cm)
AEROSOL
Flow rate
Dust toad
75 m3/min
H4 gms/m3
TOTAL ANNUAL COSTS
Agglomerator and collector
1.00-1
.25 $/acfm.year
Comparitive equipment:
Baghouse
Venturi scrubber
Electrostatic precipitator
0.71
L3I
2.04
§/acfm.year
$/acfm. year
$/acfm. year
TABLE i
Summary ORF pulse-jet. agglomerator results
-------
-622-
-------
-623-
SESSION 7
ADVANCES IN MEASUREMENT TECHNIQUES
Chairman: Elbert Tabor
U. S. Environmental Protection Agency
Research Triangle Park, N. C.
Paper No.
26 The Present Status of Particulate Mass
Measurements
J. A. Dorsey and
D. B. Harris
U. S. Environmental Protection Agency
Research Triangle Park, N. C.
27 Plume Opacity Measurement
David S. Ensor
Meteorology Research, Inc.
Altadena, California
28 Instrumentation for Dispersion Analysis of
Particulates in Industry
S. S. Yankovskiy and
Valery P. Kurkin
State Research Institute of
Industrial and Sanitary Gas Cleaning
Moscow
U.S.S.R.
29 Technology of Particulate Sampling From
Reactive, Damp, and High-Temperature Gases
V. A. Anikeyev,
V. P. Bugayev,
V. A. Limanskiy,
Ye. N. Andrusenko, and
V. Yu. Padva
(presented by Valery P. Kurkin
State Research Institute of
Industrial and Sanitary Gas Cleaning)
Moscow
U.S.S.R.
-------
-624-
SESSIQN 7 - Continued
Paper No.
30 Measurement of Particle Size Distributions at
Emission Sources with Cascade Impactors
Michael J. Pilat
University of Washington
Seattle, Washington
31 The Chemical Composition of Fly Ash
David F. S. Natusch
University of Illinois
Urbana, Illinois
-------
-625-
Paper No. 26
THE PRESENT STATUS OF
PARTICULATE MASS MEASUREMENTS
by
J. A. Dorsey
and
D. B. Harris
U. S. ENVIRONMENTAL PROTECTION AGENCY
Research Triangle Park, N.C.
-------
-626-
-------
-627-
ABSTRACT
The manual and instrumental determination of particulate
mass loading in process and control equipment in gas streams
are discussed. These measurements remain among the most
difficult required for air pollution stuides. Available
manual methods, while costly and time consuming, are accurate
at higher concentrations that have been reasonably well
standardized. A significant lack of reproducibility exists
at the low concentrations found after many control devices
in work conducted thus far has failed to define the exact
causes. A brief review of the principles available for
instrumental measurements show that several approaches should
be useful in research and development work. Of these, the
adsorption of beta energy is the most general and has been shown
to be applicable to control work. Several other techniques
are promising at least for special situations. However, much
work remains to be done before low costs, reliable particulate
monitors are widely available.
-------
-628-
-------
-629-
Introduction
The majority of techniques presently available for the measurement
of particulate mass were developed for the determination of concentra-
tions that were in the range of 1.0 gr/NM3 or greater and generally
assumed the particulate size to be mostly 1 ym and larger. Not a
great deal has been accomplished in evaluating the applicability of
these approaches to the measurement of lower concentrations of parti-
culate which have a large percentage by mass of submicron particles.
The discussion of techniques that follows must, therefore, be con-
sidered with this in mind. Where data exists that is specifically
applicable to the measurement of fine particulate, it will be pointed
out.
In any discussion of particulate measurements it is necessary to
consider the environment in which the measurements must be accomplished
and what effect it will have on the validity of the data acquired. For
particulate control system evaluations there are obviously two distinct
environments; that of the stream entering the control device and; that
of the stream leaving the device. For baghouses and electrostatic
precipitators these regimes differ primarily in the amount and size
distribution of the suspended particulate. For scrubbers, the exit
regime may also differ drastically from the process stream being con-
trolled in temperature, moisture content and composition of the suspended
particulate. It is also quite possible that the two regimes have time
dependent variations which are not influenced by the same parameters,
-------
-630-
that is, inlet particulate characteristics are process time dependent
\
while the outlet participate is most strongly dependent upon the con-
trol system operating characteristics.
As an attempt to mitigate against these factors, it has become
standard practice to collect a time and spatially averaged sample
through the techniques of traversing and isokinetic sampling. With
this approach the duct or stack is divided into a number of equal
areas, the sampling probe is moved in sequence to the center of each
area and the probe velocity matched to the gas velocity at that point
during acquisition of the sample. The procedures were developed many
years ago and from the data available appear to be reasonably useful
for steady state process streams with moderately efficient control
devices. However, even under these circumstances, such sampling
programs are very expensive and difficult to perform on large industrial
equipment. It has been proposed that traversing and isokinetic sampling
are unnecessary for fine particulate measurements and it would appear
that the verification of these postulations could lead to significant
reduction in the cost of fine particulate measurements. Unfortunately,
the extent to which detailed sampling may be simplified is, for the most
part, unknown and the areas where less complex approaches can be applied
have not been investigated. It is, therefore, mandatory that the
detailed techniques continue to be applied in spite of their cost.
-------
-631-
Manual Sampling
A number of methods have been utilized for manual sampling in
the United States with the most common being those published by the
American Society of Mechanical Engineers (ASME, PTC 27), American
Society for Testing Materials (ASTM, D-2928), Industrial Gas Cleaning
Institute (IGCI Pub. No. 101) and Western Precipitating Company
(Bulletin WP-50). All of these methods use dry filtration for collec-
tion of the particulate, and in general, were developed for "dusts"
greater than 1 ym, There are many differences in the equipment speci-
fied in the methods and quite often several possible configurations
are suggested without any comment as to the best or recommended sys-
tem. Nevertheless, it Is probably safe to assume that consistent
results can be obtained by trained operators when the particulate
concentrations are above 2 gm/NM3 and are not primarily fine particles.
In 1964, investigators in what is now the Control Systems Laboratory,
Office of Research and Development, EPA, devised a sampling train that
incorporated features from a number of other methods. The basic con-
cepts of this equipment were promulgated in 1971 as the official method
for determining particulate emissions from fossil fuel fired steam
generators, incinerators and cement plants. The equipment has a nominal
flow rate of 25 1pm and uses dry filtration as do the previously men-
tioned methods but differs in that it specifies the filtration media
and the minimum temperature of sampling and filtering. It is also unique
in requiring velocity measurements to be made simultaneously with the
-------
-632-
sampling traverse and uses both instantaneous flow rate and integrated
gas volume measurements. This train has been utilized successfully
for sampling many processes and gives reproducible results down to
approximately 0.5 gm/NM3 with either dusts or fine particulate.
Recent data indicates that manual particulate measurements
require improvement if they are to be useful for the very low concen-
trations normally found after high efficiency control devices. For
example, while the coeffecient of variation at a 95 percent confidence
level is about 24% at 2.0 gm/NM3, it increases to 76% at 0.02 gm/rIM3.
a concentration quite typical of the exit from control devices. The
reasons for this rather drastic loss in reproducibility are not evident.
Material handling losses, filter media penetration, adsorption of
moisture and chemical reactions with the low mass, high surface area
fine particulate sample have all been postulated as causes. However,
the limited work conducted thusfar has not demonstrated a significant
effect from any single parameter studied.
Instrumental Methods
The selection of partfculate mass monitoring systems presents
many problems because the characteristics of the particulate and gas
stream have a profound effect on the system response. This situation
makes the discussion of particulate monitoring difficult for it requires
that each type of installation be specifically discussed 1n terms of
the requirements necessary to perform the desired measurement accurately,
reliably and at reasonable cost. Such a discussion of the specific
-------
-633-
effects of each measurement parameter for each possible monitoring
application would require a comprehensive review and the following
discussions will consider the generalized attributes of various analyzers
and sampling equipment presently available.
There are a very limited number of techniques available for
monitoring the mass concentration of particles in a process stream.
A recent review of mass monitoring devices indicates that the only
technique that has been reduced to practice is based on the attenua-
tion of beta radiation by particulate collected on a filter media.
The attenuation of energy is somewhat dependent upon the atomic
number and atomic weight of the elements and the measurement is not
strictly a function of mass. However, the data available for coal
flyash, coal soot, cement dust and gypsum correlates to within 10%
for a given sensor and the variation in the composition of particu-
late from a specific source is probably small enough in most cases
to reduce the actual analysis error to 5% or less. There are also
variations in sensor response introduced into the measurement by
changes in filter media thickness, particulate deposition pattern on
the filter and radiation source-detector geometry. These are all
controllable variations which are fixed for a given sensor-source
installation and can be kept small by proper calibration of the device.
There are commercially available beta attenuation sensors on the
market. Several of these instruments operate at ambient temperatures
and require gas stream cooling prior to filtration of the particulate.
-------
-634-
The sensors have more than adequate sensitivity for most source
concentrations and the cooling is normally accomplished by dilution
in order to prevent liquid condensation. One recently available
instrument filters at an elevated temperature and does not require
cooling below 150°C (300°F). The overall measurement with all beta
monitors is discontinuous in that the system goes through a repetitive
sequence of filtration followed by analysis of the collected material.
The time cycle is a function of particulate concentration, sample flow
rate, dilution ratio and source-sensor properties. Information on the
best combination of these variables for various process streams is
lacking at present but it is reasonable to assume a 1 or 2 minute
cycle can be achieved prior to any control equipment. After low
temperature fabric filters and high efficiency dry electrostatic preci-
pitators, adjustment of the sampling rate and dilution ratio should
yield a cycle time of 2 to 4 minutes. Little is known about the opera-
tion of these devices after wet electrostatic precipitators, wet
scrubbers or other high-moisture, low-particulate streams. Extremely
high dilutions to prevent condensation could result in extended cycle
times and require redesign of the sensing element geometry. Alterna-
tively, filtration at even higher temperatures than presently available
could produce a sensor which does not require any dilution and would
be more suitable for these applications.
Both optical and electrical principles have also been applied
to the analysis of particulate in process gases but only limited
attempts have been made to correlate their response with particulate
-------
-635-
mass. Since these techniques are strongly influenced by many particle
properties, it is virtually impossible to reach generalized conclu-
sions concerning their applicability without acquiring extensive
field data on the characteristics of various process streams.
Several particulate monitoring devices based on sensing the
electrostatic charge on particles have been tested in process streams.
In one system, the particles were passed through a corona discharge
and the charge induced on them detected at a collector electrode
located downstream. Several different design prototypes were constructed
and measurements have been made on coal-fired boilers. No significant
attempts to correlate the device response to particulate mass were
made and the reports indicate that the responses were expected to corre-
late better with surface area. A second type of electrostatic device
is based on the charge transfer developed between moving particles and
a surface with which they are brought into contact. The response of
this type of device is a function of the surface material, particulate
composition and size, and gas flow rate through the sensor. In spite
of these variables, several studies have shown a reasonably good corre-
lation between instrument response and particulate mass in several
different process streams. In each instance, calibrations were established
for the specific source. These empirical calibrations varied considerably
between the different sources. However, on a coal-fired boiler the
correlation remained constant under conditions of full load, partial
load and soot blowing which would indicate a low sensitivity to minor
changes in particulate characteristics. All of the test data available
-------
-636-
has been obtained on effluent gases after particulate control devices
and this would provide a partial answer to the unexpectedly good
correlations since it is reasonable to assume that the mass size
distribution of particles after a control device is fairly narrow.
Devices based on this principle are available commercially and can
reportedly be operated either with dilution or at elevated sensor
temperatures. While the studies cited above indicate potential for
use after all types of control equipment, there is no data available
to define applicability to inlet gas streams where particulate pro-
perties may vary considerably.
No extensive correlations between mass and light transmission
have been reported and the manufacturers of such instruments usually
define the response of the sensor in terms of smoke density, percent
transmittance or equivalent Ringelmann number. In one study, a
reasonable correlation was found between mass and light transmission
after an electrostatic precipitator on a coal-fired utility boiler
under normal operating conditions. However, the same correlation did
not hold under reduced load conditions and still a third response was
noted when soot blowing was in progress. It appears that the complex
nature of the interactions between particles and light will make this
monitoring approach useful only for very homogeneous particulates:
The application of the technique is improbable for general mass measure-
ments.
-------
-637-
The measurement of opacity has not been used in the evaluation
of control equipment efficiency and does not appear to be applicable
because of the complex interactions between particles and light
energy mentioned earlier. The wide differences in particulate con-
centration and size distribution that exist between inlet and
outlet of a control device would make the evaluation of efficiency
based on light transmission data virtually impossible. However,
the measurement of opacity after a control device is of importance
in control equipment evaluations since there are often visible
standards which must be met. Instruments are available for monitoring
the opacity of effluent gases and they can be utilized to indicate
the visual appearance of the plume provided the proper criteria
have been considered in their design. In particular, the spectral
response must be in the visible region and the sensor must be protected
from stray light interference.
Sampling Systems
The techniques presently available for monitoring paniculate
mass require that a sample be extracted from the main gas stream and
transported to the sensor. Normally, stainless steel nozzles and
probes similar to those used in manual sampling systems are used to
convey the sampled gases to the instrument. The use of this type of
probe results in particulate losses by deposition and continuous
monitoring, unlike manual sampling, cannot compensate for these losses.
This may represent a major source of error in the measurement since
-------
-638-
probe depositions of 50% or more of the sampled particulate have not
been uncommon in manual sampling. The losses are a function of many vari-
ables including particulate properties, nozzle design, probe diameter
and length, and gas velocity in the transport system.
Conclusions
The measurement of particulate mass concentration, which was just
beginning to achieve a reasonable level of technology for conventional
sources, is now faced with a new set of problems related to fine parti-
culate control systems. The existing manual methods are not capable of
acquiring reproducible data at very low concentrations and have also
become very expensive due to the long sampling times required to collect
a weighable sample. The possibility of chemical reactions causing
erroneous results has become very significant due to the much higher
concentration of gaseous materials relative to the particulate mass.
On the positive side, the requirements for traversing and isokinetic
sampling can potentially be simplified—although studies must be con-
ducted to demonstrate the validity of this—resulting in lower measure-
ment costs.
The instrumental measurement of particulate may actually be easier
to achieve for fine particulate than it has been for total particulate
because the range of particle characteristics will undoubtedly be
much smaller and this should increase the sensing principles which can
be applied. As with manual methods, if the requirements for traversing
-------
-639-
and Isokinetic sampling can be reduced, the Installation of participate
monitors will be greatly simplified. The major effort required In the
area of Instrumental methods 1s the acquisition of field data that will
define the applicability of available sensors to various process and
control system streams.
-------
-640-
-------
-641-
Paper No. 27
PLUME OPACITY MEASUREMENT
by
David S. Ensor
METEOROLOGY RESEARCH, INC.
Altadena, California
-------
-642-
-------
-643-
ABSTRACT
The visual appearance of plumes has long been used as a
means to regulate the emission of particulate matter from
air pollution sources. The technical interperation of the
visual effects of the plume in terms of other aerosal pro-
penties such as size distribution, particle composition and
concentration will be disucssed. The current approaches
used to measure plume opacity will be covered including
in-stack and remote sensing techniques. The physical
limitations of present approaches as well as possible future
developments will be covered.
-------
-644-
-------
-645-
I. INTRODUCTION
A. Objective
The visual appearance of smoke plumes has been used for
regulation of air pollution sources for over 60 years. Opacity is defined
in the Federal Register (1971) as
"... the degree to which emissions reduce the trans-
mission of light and obscure the view of an object in
the background. "
The generally accepted method for plume opacity measurement is a field
observation by an individual who has been trained to "read" the plume
under specified conditions of sun angle, wind direction, and distance
from the stack. In the last few years, there has been increased interest
in substituting or supplementing instrumentation for this "subjective"
observation. It is the purpose of this paper to review the various
instrumental concepts used to measure opacity.
B. Brief Review of Legal Aspects
The earliest air pollution law regulating smoke emissions
in the United States was the 1881 Chicago smoke ordinance which deemed
the emission of dense smoke a public nuisance (Nicholson, 1905). The
nebulous legal definition of public nuisance or dense smoke in the regula-
tion forced an inspector to make an arbitrary decision at the site as to
the degree of public nuisance, a decision that could be easily challenged
in court.
In 1898, Ringelmann reported a smoke chart devised to quantify
the appearance of smoke. This chart consisted of standards constructed
by drawing black ink lines of various widths on white paper to form a grid.
When the charts are viewed from a distance of at least 50 ft, apparent
shades of gray from white to black may be seen. The chart was simple
and inexpensive and could be reproducibly constructed. The usefulness
of this chart to quantify the public nuisance of dense smoke was quickly
recognized, and the Ringelmann smoke chart was specified in regulations
throughout the world. The first smoke ordinance in the United States
utilizing the Ringelmann smoke chart was passed in Boston in 1910
(Kudlich, 1955). The constitutionality of smoke ordinances was upheld
in the United States Supreme Court in the case of Northwestern Laundry
versus City of Des Moines in 1916 (Edelman, 1970). The paper Ringelmann
chart was used as a smoke standard in most large cities in the United
States for the next 40 years.
-------
-646-
The 1947 State of California enabling act, Health and Safety
Code 24198 to 24341, allowed any county to form air pollution control
districts. Section 24242 contained a provision extending the Ringelmann
number evaluation to non-black smokes of "such opacity as to obscure an
observer's view to degree equal to or greater than does smoke. " This
so-called "equivalent opacity" concept was first used in the Los Angeles
County Air Pollution Control District Rule 50 in 1948. The use of
equivalent opacity for source regulation is still a very controversial
subject and was difficult initially to enforce.
The Los Angeles County Air Pollution Control District started
a "plume reading school" to train smoke inspectors for qualification as
legally recognized expert witnesses. The black smoke school was begun
in 1950, and the qualification of trained smoke inspectors as expert
witnesses for evaluation of smoke emissions without the paper Ringelmann
chart was upheld in People versus International Steel in 1951. The white
plume school was begun in 1954, and the "equivalent opacity" concept was
legally recognized by the courts in People versus Plywood Manufacturers
of California in 1955 primarily on the qualifications of the inspectors as
expert witnesses (Edelman, 1970).
Smoke school field work consists of training inspectors with
plumes of known opacity generated by a standard smoke source, as
described by Weisburd (1962). The smoke source has a short stack about
15 ft high and 12 inches in diameter. A reference transmissometer is
mounted near the top of the stack to measure the in-stack light trans-
mittance. The Ringelmann number and opacity are related to the light
transmjttance as indicated in Table I.
Black smoke is generated by the incomplete combustion of
benzene, and white smoke is usually generated by the injection of fuel oil
into the manifold of a small air-cooled gasoline engine.
Table I
COMPARISON OF PLUME LIGHT ATTENUATION TERMS
Plume Trans-
mittance
(Percent)
100
80
60
40
20
0
Ringelmann
Number
(Black Plumes)
0
1
2
3
4
5
Plume Opacity
(Percent)
(Non- Black Plumes)
0
20
40
60
80
100
-------
-647-
The training for witnessing white and black plumes is conducted
separately. The inspectors are first familiarized with the reading tech-
niques using plumes of known opacities. After the familiarization phase,
the inspectors then evaluate plumes of unknown opacities to the nearest
5-percent opacity generated in random order in runs of 25 readings. For
certification, an inspector must demonstrate a proficiency for plume
evaluation within an absolute deviation of 7. 5-percent opacity measured
in the stack for at least two runs. Any reading in error by more than
15-percent opacity will void the run.
The training and subsequent field reading for enforcement
purposes are usually performed according to the following rules:
1. The observer is at right angles to the plume.
2. The sun is at the back of the observer.
3. The plume is read at the point of greatest opacity.
4. The observer stands at approximately two stack heights
but not more than a quarter of a mile from the stack.
5. The plume is viewed against a contrasting background.
The observation is shown diagrammatic ally in Fig. 1.
The Bay Area Air Pollution Control District adopted a similar
"equivalent opacity" rule in its Regulation 2 in I960. The Bay Area Air
Pollution Control District, instead of using the defined relationship
between the in-stack transmittance and plume opacity reported in Table I,
considered the problem from a different aspect (Coons et al. , 1965). A
group of inspectors, trained with the paper Ringelmann chart, evaluated
white and black plumes of unknown opacities generated with a standard
smoke source. Their readings of plume opacity were then used to develop
"calibration curves" relating the in-stack transmittance to the inspector
plume opacity reading.
These calibration curves were used by the Bay Area Air
Pollution Control District until recently when a new rule was approved
reducing the Ringelmann limit from Number 2 to Number 1. Brennan
(1971) reported that a new "calibration curve" had been developed by expert
observers who had prior training with a smoke generator. This new cali-
bration curve is said to be quite similar to the defined relationship in
Table I.
-------
-648-
The Texas Air Control Board Regulation I adopted in 1969
has a provision for the use of instrumental measurements of in-stack
transmittance to supplement human observers (McKee, 1971). A simple
smoke meter has been used in most of the industrial sources in Texas.
The regulation specifies a minimum transmittance of 70 percent with
standard graphs to correct the instrumental measurements for length of
path and volumetric flow rate of the stack gases.
On December 23, 1971, the standards of performance for new
stationary sources pursuant to Section 111 of the Clean A ir Act as
amended were promulgated. New (after August 17, 1971) steam genera-
tors, cement plants, incinerators, nitric acid plants, and sulfuric acid
plants are regulated nationally to stringent limits of opacity. Also, the
continuous monitoring of particulate matter in steam generators is
required using "a. photoelectric or other type smoke detector and recorder
except where gaseous fuel is the only fuel burned. "
-------
-649-
II. INTERPRETATION OF OPACITY
A. Theory of the Smoke PI tune Observation
The visual luminance contrast of an object against an exten-
sive and uniformly appearing background is given by (Middleton, 1952)
B0 -
Co -
where B0 is the luminance of the object and B^ is the luminance of the
background. Our sense of vision (in the absence of color effects) depends
primarily on the perception by the eye of differences in luminance (light
flux per unit area normal to the direction of observation divided by the
solid angle subtended by the light source at the viewing surface, candles/
m8) between points in the field of view. For an ideally black object,
the object luminance Bo is zero and the contrast equals -1.0. Contrasts
greater than 10 are seldom measured for bright objects during periods
of daylight.
If the visual contrast between the object and its background
is less than the contrast threshold of the eye, the object will not be
visually detectable. The contrast threshold is a function of the subtended
angle of the object (size of the object and distance from the observer),
the background luminance (eye adaptation luminance), the sharpness of
the boundary of the object, and the presence of other interfering objects
in the field of view. The contrast threshold for a 50-percent probability
of visually detecting lighted targets was reported by Blackwell (1946) to
be about 0. 003 for daylight illumination and a subtended angle of the object
greater than 30 minutes. The contrast threshold for black objects is
usually 20 percent lower than for white objects.
The plume contrast is simply related to the light transmittance
of the plume. The luminance of the plume is given by
Bp = Ba + Bb T (2)
where Ba is the plume air light (the light scattered by the plume to the
observer), and BDT is the amount of light transmitted through the plume.
fib is the background luminance, and T is the fraction of background
light transmitted through the plume. Often, the background luminance is
the sky behind the plume, The plume contrast Cp can be obtained by
substituting Eq. (2) for Bo in Eq. (1) as reported by Conner and Hodkinson,
(1967)
-------
-650-
cp - f; - n - T, (3)
_. Plume air light luminance
Plume contrast = —-—: ~ : - Plume opacity
Background luminance r *
This equation illustrates the three main optical effects of plumes:
1. The plume contrast Cp indicates the visual appearance
of the plume against its background (what a person sees).
2. The fractional plume opacity (1-T) is an intrinsic property
of the plume, which is independent of the illumination and
the viewing angles, and indicates the fraction of background
light transmitted through the plume.
3. The plume air light to background luminance ratio, Ba/Bb,
indicates the magnitude of light scattered to the observer.
This ratio is dependent upon the plume light scattering
properties, the angle between the sun and the observer,
and the background illumination.
Smoke inspectors are taught to associate a plume to background
contrast with a given in-stack transmittance. The proper selection of a
background is one which allows a consistent change in plume to background
contrast with changes in the plume transmittance.
B' Relationship of Plume Transmittance to Mass Concentration
The plume transmittance is related to the diameter of the
plume by
T = exp [ - bext L ] (4)
where bex£ is the extinction coefficient and L is the diameter of the
plume.
In a theoretical study reported by Ensor and Pilat (1971), the
mass concentration is related to the extinction coefficient by
M = K p Bext (5)
where K is a parameter which is a function of particle size distribution
and refractive index, and p is the specific gravity of the particles. An
example of these calculated results is shown in Fig. 2 for a refractive
index of 1. 50 and a log normal size distribution. One of the significant
-------
-651-
results of the calculations was the discovery that increasing polydispersity
reduces the effects of size distribution variation on the parameter K.
C. Considerations in Developing Instruments for Opacity
Measurement
The substitution of an instrument for the smoke inspector is
a very desirable advance. The observer-determined opacity is an
interpretation of the plume to background contrast to determine an opacity
reading. Questions are always raised about the suitability of an instru-
ment to measure the degree of public nuisance caused by the plume (Sim
and Borgos, 1973). Fick (1973) has described in detail the many
uncertainties caused by differences in-stack and exterior plume properties.
Whatever instrumental approach is taken, the output will be subtly
different than the legally recognized observation.
Texas Regulation I boldly specified a maximum in-stack light
transmittance over a specified path length as the basis for the regulation.
Thus, many of the objections of the use of instruments for compliance
measurement were solved from a legal standpoint.
-------
-652-
III. INSTRUMENTAL MEASUREMENTS
A. ' Stack-Mounted Sensors
(
I. Considerations in Application
Location of the sensor at the stack has the advantages
of permanent installation and low cost and does not require constant
attention. Stack sensors also are independent of ambient lighting condi-
tions and are capable of continuous use. However, the engineering
problems of measuring a hot, dirty, corrosive gas are often severe.
The sensor electronics are also subject to vibrations and extremes in
atmospheric temperatures.
2. Transmissometers
Stack transmissometers in the form of the "smoke
meter" have been installed on smoke stacks for quite a number of years
to warn operators of possible opacity violations. These instruments are
very simple, consisting of a light source on one side of the stack and a
detector on the opposite side of the stack as shown in Fig. 3. Various
versions of these instruments have been used with hundreds of installa-
tions. However, these instruments vary widely in quality and price andmany
are considered to be qualitative rather than quantitative sensors.
The EPA is currently developing guidelines in the
construction of stack transmissometers. The guidelines call for stan-
dardization of the wavelength response to a photopic or green, a
restriction of the detector acceptance angle and collimation to less than
5 degrees, and minimum levels of reliability. The wavelength standardi-
zation is very important because the extinction is usually wavelength
dependent. The function relationship between light extinction and wave-
length depends on the particle size. The restriction of the acceptance
angle of the transmissometer is important because the detection of
scattered light may cause large increases in the apparent light trans-
mittance. This is a subtle error because it is a function of the angle of
the detector, projector, particle size, and, to a less appreciated extent,
the light transmittance. The theoretical error in the extinction coefficient
was computed by Ensor and Pilat (1971) for various size distributions, as
shown in Fig. 4. The error is reported as the ratio R, the measured
extinction coefficient divided by the real extinction coefficient.
The error in light transmittance from scattered light is
also a function of the magnitude of the light transmittance as a consequence
of Beer's law. This error is shown in Fig. 5 as a function of light trans-
mittance and the parameter R.
-------
-653-
Tomaides and Peterson (1973) reported experimental
determinations of the errors in the measurement of fly ash transmittance.
They compared the light transmittance measured with a transmissometer
having a variable acceptance angle to a reference transmissometer.
The most advanced transmissometer is that developed
by Irwin Sick. This instrument is sold in the United States by Lear Siegler,
Inc. v(Sims and Borgos, 1973). Other manufacturers have recently
announced transmissometers which meet the EPA guidelines.
3. Integrating Nephelometer
A direct way to measure the extinction coefficient as
defined by Beer's law, Eq. (4), is with the angular integrated nephelo-
meter. An advantage of this approach is the inherent sensitivity of the
instrument because the signal is directly proportional to the parameter
of interest instead of a ratio as in the transmissometer. The extinction
coefficient is used for correlations to mass concentration. The integrating
nephelometer as reported by Charlson et al (1969) is proving to be an
extremely useful instrument for measuring atmospheric visibility,
(Samuels et al. , 1973)
The instrument is almost as simple as a transmisso-
meter without the problems of alignment and dirt buildup on optics. The
optical arrangement is shown in Fig. 6. The instrument is a physical
analog to the equation defining the scattering coefficient
T/
bscat = 2TT I(9) sin 6de
•where 1(9) is the angular scattered light.
The light source is behind a diffuse filter which weighs
•flie light as a cosine function. The detector at right angles to the light
source looks past the light into a light trap. The scattered light as
measured by the detector has a sine weighting by virtue of the perpendicular
orientation to the cosine lamp. Thus, the progressive distance away from
the detector in the scattering volume corresponds to a scattering angle of
near forward to far backs cattering. In addition, the scattering angles
are weighed identically because the sample volume increases on the
distance from the detector squared. This exactly balances the distance
squared reduction in brightness of the scattered light.
Meteorology Research, Inc., is developing this concept
into a useful stack monitoring device as described by Ensor and Bevan
-------
-654-
(1973). Potential sources of error with this approach are light losses
at the extremes of the integration "angular truncation" and light attenua-
tion from the sample volume to the detector. The angular truncation
errors from preliminary analysis are about ±15 percent and the error
from light attenuation is less than 6 percent under most conditions.
The instrument has been released in a production run of ten units after
an extensive evaluation of the prototypes and is expected to be used on
additional selected stacks.
The range of operation of the instrument is shown in
Fig. 7. The stack instrument is two orders of magnitude less sensitive
than the ambient instrument but has more than sufficient sensitivity for
monitoring most sources.
It was decided to concentrate on an extractive method to
solve the difficult sampling problems such as high humidity stacks and
condensing materials. The flexibility gained with a sample probe and an
out-of-stack sensor in many cases outweighs the uncertainties caused by
line losses and representativeness of the sample.
B. Remote Methods
1. Implementation Considerations
Measurement of the exterior plume has the advantage
of being more similar to the opacity observations than in-stack techniques
because the plume near atmospheric temperature and dilution can be
measured. All remote methods are affected by the ambient meteorologi-
cal conditions and most are not usable at night. In addition, trained
personnel must be used to operate many of the instruments.
2. Passive Methods
Passive methods utilize ambient lighting as a means of
measurement and thus are not usable at night or under overcast conditions.
Many of these methods are described in detail by Conner and Hodkinson
(1967).
• Smoke Inspection Guide
For at least 60 years, smoke inspection guides utilizing
tinted glass have been used to improve the accuracy of smoke reading.
Modern guides for the evaluation of black smoke have been reported by
Rose et aL (1958) and for white smoke by Conner et al. (1968). The guides
allow the observer to compare the plume transmittance to the light
-------
-655-
transmittance of a reference filter that has light scattering and absorbing
properties similar to many smoke plumes under similar atmospheric
lighting conditions, rather than comparing plume transmittance with light
reflected from a paper chart. However, the guides do not improve the
accuracy of the observation over that of a trained observer sufficiently
to be of general use.
• Sun Photometer
The sun photometer is a simple detector designed to
measure the attenuation of the sun through the plume. Circumstances
must be selected where the plume, sun, and observer are in the correct
position. A sun photometer is currently being sold by Forney Engineering
Company (1973) which is similar to a concept suggested by Conner and
Hodkinson (1967).
• Contrasting Target Method
The brightness of a contrasting target viewed through
the plume is measured either with a telephotometer or photographically.
The points of measurement are shown in Fig. 8. The transmittance of
the plume is given by
Bsp - Bhp
B -
where Bsp is the brightness of the sky and plume, B^p is the bright-
ness of the hill and plume, Bs is the brightness of the sky, and Bh
is the brightness of the hill. The ratio of the differences of brightness is
used to cancel the plume air light. The target may be artificial, a handy
hill, or nearby dark structure. When a telephometer is used to measure
the brightness, the plume should be constant to allow time for the four
measurements. When the plume and background target brightness is
measured photographically, the optical density of the negatives at the
point of interest is measured with a microdensitometer in a laboratory.
The relative brightness is determined from the optical density with the
use of a calibration curve determined for each roll by photographing a
grey scale or a series of neutral density filters. This method is
suitable for measurement of rapidly changing plume opacities.
3. Active Methods
These techniques are limited by the meteorological
conditions but are not dependent on ambient lighting. A number of
methods have been used for research, including external transmisso-
meters, photometry of lights, and lidar (light detection and ranging).
The lidar may have promise because it is a single-ended remote mea-
surement technique. For this reason, extensive engineering efforts have
-------
-656-
been sponsored by the Edison Electric Institute and the EPA to develop
plume lidar. These efforts are reported by Cook et al. (1972) and Bethke
(1973).
The device consists of a pulsed or chopped laser light
source, a telescope detector, and signal processing electronics as
diagramed in Fig. 9. The return is quite similar to that diagramed in
Fig. 10. There is a large backscatter signal from the plume followed
by a light scattered from the atmosphere behind the plume. The light
scattered beyond the plume is attenuated by the plume. The transmit-
tance of the plume is related to the reduction in signal. The error, as
reported by Cook et al. (1972), is less than 7. 5 percent for transmittances
greater than 0. 8 and less than 12 percent for transmittances greater than
0. 50. One of the goals of current plume lidar research is the develop-
ment of an instrument with a price low enough for general use (Conner,
1973).
C. Future Development
1. Instrumentation
The instrumentation for smoke plume measurement is
undergoing marked improvement in design in response to EPA guidelines
and the demands of measurement of cleaner stacks. As regulations
become more stringent, there will be a greater requirement for more
sensitive detectors. If an invisible plume is required, an instrument
must be made sensitive enough to be responsive to process upsets that
may lead to visible emissions.
2. The Physics of Stack Aerosol
The new generation of opacity instruments as well as
other source instrumentation will permit better measurements of source
aerosol. These data will allow better understanding of the physics of the
aerosol.
• Relationship of Plume Opacity to Mass Concentration,
Size Distribution, and other Properties
In a number of research programs, the opacity to mass
concentrations are being determined experimentally. This work is being
done at the University of Washington under an EPA grant and under EPA
contract.
• In-Stack Opacity vs. Exterior Opacity
One of the fundamental assumptions with the use of an
in-stack monitor for compliance testing is that the monitor will correctly
-------
-657-
indicate the opacity seen by people outside. There have been some
studies which indicate that for cool-fired power plants there is good
correspondence between in-stack and out-of-stack opacity (Tomaides and
Peterson, 1973). However, there is evidence that for other sources this
may not be the case. For example, Kester (1972) reported that the
exterior opacity of a hogged fuel boiler was related to the fuel moisture
content rather than in-stack opacity.
• The Effect of Water Vapor on Opacity
The effect of water on plume opacity has been side-
stepped by either exempting the plume from regulation or specifying an
inspector to read the plume downwind of the source. With increased
use of wet scrubbers, it will be imperative to understand this
phenomenon. Yocom (1963) reported sample withdrawal from the stack
and heating prior to measurement with a transmissometer. It appears
that this approach has limited application. From experiments done
with humidity conditioning of the inlet of an Integrating Nephelometer
(Covent, Charlson, Ahlquist, 1972), ambient aerosols exhibit a large
increase in scattering coefficient with increasing relative humidity.
The relationship of scattering coefficient to relative humidity is a strong
function of the composition of the material. The change in visual effects
of a condensing and then evaporating plume should be similar to that
determined for materials of similar composition.
• Relating Source Opacity to Downwind Visibility
The impact of point sources on the visibility of nearby
regions is of great concern. The availability of improved opacity instru-
ments and ambient aerosol and gaseous sensors in instrumented aircraft
(Blumenthal, 1973) makes it possible to gather data that may allow linking
of source opacity to ambient visibility. Predictions of this type have been
reported by Ensor et al. (1973), and MRI is currently doing programs to
gather this type of data for industrial clients.
-------
-658-
IV. SUMMARY
The measurement of plume opacity still is dominated by the use
of trained observers. The recent adoption of Regulation I in Texas
as an alternative to the trained observers and the EPA specification of
a "smoke meter" as a continuous monitor are stimulating the use of
stack instrumentation. With the new proposed guidelines by EPA for
opacity meters, instruments are being sold with known wavelength and
acceptance angles. It is expected that the use of various kinds of
instruments to measure opacity will become more accepted as scientific
measurers.
The development of improved stack and exterior plume instru-
ments of various kinds is expected to stimulate new research in plume
aerosol physics.
-------
-659-
V. ACKNOWLEDGMENTS
Much of the background information for this presentation was
developed during graduate studies at the University of Washington
Department of Civil Engineering. Support at that time was in the form
of EPA Air Pollution Traineeships and Special Air Pollution
Fellowships. Meteorology Research, Inc., internal research funds
were used to write this paper.
-------
-660-
VI. REFERENCES
Air Pollution Control Field Operations Manual. 1962, (M. F. Weisburd,
ed.), P. H. S. Publ. 937.
Bethke, G. W., 1973: Development of range squared and off-gating
modifications for a lidar system. December, EPA-650/2-73-040.
Blackwell, H. R., 1946: Contrast thresholds of the human eye. J. Opt.
Soc. Amer., 36. 624-643.
Blumenthal, D. L., 1973: Measurement of physical and chemical plume
parameters using an airborne monitoring system. Paper 73API 6,
Pacific Northwest International Section Air Pollution Control Assoc.,
Seattle, Wash., November 28-30.
Brennan, T. , 1971: Evaluation of visible plumes. Paper presented at
12th Conf. on Methods in Air Pollution and Industrial Hygiene
Studies, Los Angeles, Calif. , April 6-8.
Charlson, R. J., N. C. Ahlquist, H. Selvidge, and P. B. MacCready, Jr.,
1969: Monitoring of atmospheric aerosol parameters with the
integrating nephelometer. J. Air Poll. Control Assoc. , 19, 937-942.
Conner, W. D. and J. R. Hodkinson, 1967: Optical Properties and Visual
Effects of Smoke Stack Plumes. P. H.. S. Publ. No. 999-AP-30.
Conner, W. D., C. F. Smith, and J. S. Nader, 1968: Development of a
smoke guide for the evaluation of white plumes. J. Air Poll. Control
Assoc.. ^8, 748-750.
Conner, W. D., 1973: Environmental Protection Agency personal
communication.
Cook, C. S., G. W. Bethke, and W. D. Conner, 1972: Remote measure-
ment of smoke plume transmittance using lidar. Appl. Optics. 11.
1742-1748.
Coons, J. D., H. A. James, H. C. Johnson, and M. S. Walker, 1965:
Development, calibration and use of a plume evaluation training unit.
J. Air Poll Control Assoc., 1J5, 199-203.
Covert, D. S., R. J. Charlson, and N. C. Ahlquist, 1972: A study of the
relationship of chemical composition and humidity to light scattering
by aerosols. J. Appl. Meteor., II, 968-976.
-------
-661-
Edelman, S., 1970: The Law of Air Pollution Control. Environmental
Science Services Div., Stamford, Conn., 296pp.
Ensor, D. S. , and M. J. Pilat, 1971: Calculation of smoke plume opacity
from particulate air pollutant properties. J. Air Poll. Control
Assoc. , 2J_, 496-501.
Ensor, D. S. , and M. J. Pilat, 1971: The effects of particle size
distribution of light extinction measurement. Amer. Ind. Hyg. J.,
3£, 287-292.
Ensor, D. S., and L. D. Bevan, 1973: Application of nephelometry to
the monitoring of air pollution sources. Paper 73-AP-14 presented
Annual Meeting Pacific Northwest International Section Air Pollution
Control Assoc., November 28-30.
Ensor, D. S., L. E. Sparks, and M. J. Pilat, 1973: Light transmittance
across smoke plumes downwind from point sources of aerosol
emissions. Atmos. Environ., in press.
Environmental Protection Agency, 1971: Standards of performance for
new stationary sources. Federal Register. 36, 247, December 23.
Fick, O. A., 1973: Compliance vs. control monitoring using optical,
in-stack opacity monitors. Paper presented at Annual Meeting of
Pacific Northwest International Section Air Pollution Control Assoc.,
Seattle, Wash., November 28-30.
Forney Engineering Co., 1973: Forney Opacity Meter. Product Informa-
tion Bulletin, 3405 Wiley Post Road, P.O. Box 189, Addison, Texas
75001.
Kester, R. A., 1972: Hog-fuel boiler plume opacity related to operating
parameters. Paper 72-AP-33 Air Pollution Control Assoc. Pacific
Northwest Section, Eugene, Oregon, November 15.
Kudlick, R., 1955: Ringelmann smoke chart. U. S. Dept. of Interior,
Bureau of Mines, Information Circular #7718.
Mckee, H. C., 1971: Instrumental method substitutes for visual estima-
tion of equivalent opacity. J. Air Poll. Control Assoc. . 21, 488-490.
Middleton, W. E. K., 1952: Vision Through the Atmosphere. University
of Toronto Press, Toronto, Canada, 250 pp.
-------
-662-
Nicholson, W., 1905: Smoke Abatement, Griffinan Co., London.
Rose, A. H., J. S. Nader, and P. A. Drinker, 1958: Development of an
improved smoke inspection guide. J. Air Poll. Control Assoc., 8,
112-116.
Samuels, H. J., S. Twiss, and E. W. Wong, 1973: Visibility, light
scattering and mass concentration of particulate matter. State of
California Air Resources Board Report of the California Tri-City
Aerosol Sampling Project, July.
Sem, G. J., and J. A. Borgos, 1973: State of the art: 1971, Instrumenta-
tion for measurement of particulate emissions from combustion
sources, Vol. IV: Experiments and Final Report. TSI report to
EPA Control Systems Laboratory, EPA-650/2-73-022.
Tomaides, M., and C. M. Peterson, 1973: Practical accuracy of the
particulate emission opacity measurement. Paper presented at
First International Conf. in Particle Technology, Chicago, HI.,
August 21-24.
Yocom, J. E., 1963: Problems in judging plume opacity --a simple
device for measuring opacity of wet plumes. J. Air Poll. Control
Assoc., 13, 36-39.
-------
-663-
CONTRASTING
BACKGROUND
SUN
OBSERVER
Fig. 1. DIAGRAM OF PLUME OPACITY OBSERVATION
-------
-664-
u
DC
UJ
tu
oc
GEOMETRIC
STANDARD
DEVIATION, crg
I
REFRACTIVE INDEX = 1.50
WAVE LENGTH OF LIGHT = 550 nm
I I I I I I 1(1 I I I I II III I III mi
I I I
10
10
10'
10*
GEOMETRIC MASS MEAN RADIUS, rgw (MICRONS)
Fig. 2. THE PARAMETER AS A FUNCTION OF THE GEOMETRIC
MASS MEAN RADIUS FOR A WHITE AEROSOL
-------
CONDENSER
LENS
PINHOLE
LIGHT
SOURCE
COLLIMATOR
LENS
DUST PARTICLES
TELESCOPE
LENS
PHOTOCELL
PINHOLE
RECORDEI
AMPLIFIER
i
(Ti
CTi
(_n
I
Fig. 3. DIAGRAM OF SIMPLE TRANSMISSOMETER
-------
-666-
CL
U_
UJ
az
1-00
-90
.00
.70
.60
GEOMETRIC STflNDflRD
DEVIATION,
.50
.40
flCCEPTRNCE flNGLE = i*
REFRflCTIVE INDEX = 1.50
WflVE LENGTH OF LIQHT = 550
• • *
10P 2 468
6
GEOMETRIC MflSS MEflN RflDIUS, r
Fig. 4. THE EXTINCTION COEFFICIENT CORRECTION FACTOR, R,
AS A FUNCTION GEOMETRIC MASS MEAN RADIUS
-------
-667-
Id
O
o:
UJ
a.
UJ
u
o:
S
UJ
M
UJ
O
10s
8
6
10*
8
6
10'
8
6
4
10
CORRECTION FACTOR, R
6 7 8 9 10"
MEASURED TRANSMITTANCE , I/L
Fig. 5. ERROR IN MEASURED LIGHT TRANSMITTANCE
-------
Apertures
Aperture
Opal Glass
Detector
Light Trap
Opal Glass
Calibrator
i
a\
CO
Fig. 6. DIAGRAM OF OPTICAL ASSEMBLY OF NEPHELOMETER
FOR STACK MEASUREMENTS
-------
-669-
±FJT
i i i •
10
-1
u
V)
10
10
-3
J....»,.;.
J 4
frr
,.,-J.,
H >
1
III
11L
J^-A
iii '"
ft
\
"K
:
X
1
•HI
fill
tj
in
\
-ii,
rtri
K
-*-
V
X,
^
V
i
i40% Opacity
i20% Opacity
10 % Opacity
~"^"~| 5% Opacity
Hi
: ' 2% Opacity
loT
t / S 9 1O.
Stack Diameter (feet)
Fig. 7. RANGE OF OPERATION OF NEPHELOMETER
FOR STACK MEASUREMENTS
-------
-670-
c =
- B
hP
STACK
HILLSIDE
Fig. 8. BRIGHTNESS MEASUREMENT IN PLUME
TO DETERMINE TRANSMITTANCE
-------
-671-
Volume Illuminated
by Laser Pulse
Transmittin
Telescope
Field of View
of Receiver
Path of Laser Pulse
Receiving
Telescope
Fig. 9. DIAGRAM OF TYPICAL LIDAR
-------
_
PLUME
.RETURN
AMBIENT
AIR RETURN
LASER
PULSE
T = /A/B
-------
-673-
Paper Ng^ 28
INSTRUMENTATION FOR DISPERSION ANALYSIS
OF PARTICULATES IN INDUSTRY
by
S . S. Yankovskiy
and
Valery P. Kurkin
STATE RESEARCH INSTITUTE OF
INDUSTRIAL AND SANITARY GAS CLEANING
Moscow
-------
-674-
-------
-675-
Instrumentation for Dispersion Analysis of Particulates
in Industry
S. S. Yankovskiy, V. P. Kurkin
Stokes diameter of particles must be calculated to evaluate
fractional and general efficiencies of particle collection instruments
like cyclons, filters, and scrubbers. The size of diameter may
differ significantly from that determined by normal methods of dispersion
analysis if agglomeration occurs in the gas stream. Measurement for
purpose of determining instrument efficiency should, for that reason,
be conducted directly inside of the gas stream.
Two types of instruments for dispersion analysis were developed
in NIIOGAZ, small size eleven-stage impactor and cyclon separator.
Both instruments analyze particles from 1 to 20-30 microns in size.
Uniqueness of the developed impactor is its small dimensions which
are of great importance in collection of samples from industrial scale
gas streams. Entire body of the impactor is 125 mm long with a
diameter equal to 40 mm. Impactor (Fig. 1) is made up of discs
with perforations as nozzles for one level, and with lining for preceding
level. In odd number levels, nozzles are located in central part of
the disc and lining is located on the outer part. The opposite is
true for even numbered discs. The first three levels have varied
geometry but equal collection efficiency. Such design is more reliable
-------
-676-
for removal of large particles from the gas stream. Eleventh level
of the impactor contains a filter for collection of very fine particles.
Two-phase lubricant or thin layer of nappy material is used
as lining. It is placed in special depression of disc.
Both types of lining, patented, allow for good collection of
dust particles, up to 30 mg. for one level. For use of two-phase
lubricant gas temperature should not exceed 100°C and for use of nappy
material 300°C.
Deposition of particles on each level has its own fractional
efficiency curve. Usually it is a log-normal distribution curve.
Fig. 2 shows that when the curves are drawn on log-probability paper,
they become straight lines. The curves are described by two parameters;
diameter, d50, which has collection efficiency of 50% as its limit of
separation and standard deviation 9 fraction.
In plotting curve of dispersion composition for increases in
individual levels; curve of fractional efficiency for each level usually
is substituted by vertical line (for which & fraction- 0) passing through
Last number is calculated with Stokes equation:
Vd250 . ""'s
M D M D3
«*50 ' V50 . so - const (1,
-------
-677-
where:
Stk5Q - Stokes diameter relating to d50
v - gas velocity through nozzle
if - particle density
M - gas viscosity
D - nozzle diameter
Q - gas volume through the instrument
Nomogram (fig. 3) which is based on equation 1 allows to determine
d50 as a fraction of the number of perforations, n and the size of
perforation's diameter, D. Standard conditions are used in determination
of ^50. Those are: Qst = 10 1/min, Jfa = 1 g/cm3, tst = 20°C.
(temperature stability is related obviously to M=M(+) = const)
Value dso for standard conditions is most indicative in characteri-
zation of impactor levels. It can also be used when measurements are
taken under conditions other than standard. Following correction is
then applied. Standard value for d50 at that level is multiplied by
O£=d50 /d50 (standard), J^ - ^ fi*)*,*** W
Coefficient of is calculated from STKso = const. According to
equation 1 we have;
Where: C - constant
Nomogram shown in fig. 4 is used to determine d$Q for a known
gas volume Q, density of particles r and gas temperature t. The
nomogram allows to first find the size of coefficient cC and then size of
-------
678-
Values of c^g, obtained with normal calibration of impactor under
conditions which assure 100% retention of large particle on
surface, (covered with thin oil layer) cannot directly relate to
linings. The reason is that part of the particles which come into
contact with lining may be carried away. Calibration method was
developed in which two analogous levels are placed sequentially. An
assumption was made that curves for fractional efficiency of levels
and dispersion composition of particles in the instrument are log-normally
distributed.
Two parameters (distribution limit d50, standard deviation d> fraction)
for the fractional efficiency curve are found with this method.
The parameters are found by knowing two corresponding parameters (average
geometric diameter d particle and standard deviation ^>part) of dispersion
composition at inlet to the instrument and from knowing efficiency values
for first JJi and second tin analogous levels of impactor. Nomograms
constructed in generalized coordinates IgOpart./ Ig ^> fraction
part.) / Ig^part. represents relationship between parameters
of fractional efficiency and parameters of dispersion composition of
particulates at the inlet and efficiency of two analogous levels of
impactor .
Two curves , one with experimental value Jfi and other JA o » are
drawn through tracing paper, placed on top of the nomogram, to calculate
parameters d5o and/i fraction. X and Y coordinates at the point where two
curves cross each other allow to calculate d^Q and fraction providing
d particle and ^S particle for distribution by particle size at inlet to
the instrument are known.
-------
679-
Results of measurements conducted at different levels of impactor
are shown in Table 1. Values obtained in calibration, (d50)true and
«2> fraction)true, characterize true curves of fractional efficiency of
studied levels. Calculated values (C^Q)calculated and ( fraction)calculated
are also presented in the Table. Obtained data agrees with other presented
in literature for normal calibration. Relationship between true and
calculated d5Q is also presented in the Table. For impactor under considera-
tion, curves of fractional efficiency are slightly skewed and have a
slightly different slope. To calculate particulate dispersion composition
(d^g)true should be used.
Cyclon separator consists of three sequentially connected cyclons
with diameters 30,32,16 mm and filter at the outlet. Dimensions of the
instrument are: 150 x 100 x 50mm. Rate of gas sampling is 10^/min.
Increase in the amount of dust in the cyclons and on the filter is used
for dispersion analysis.
Cyclon efficiency values are presented as .two parameters of log-normal
distribution. The parameters are average geometric diameter, dp,
and distribution dispersion of particles Ig^p. Both parameters are
estimated by nomograms with equal value of efficiency curves on which
amount of dispersion is shown on the ordinate and diameter of particles
dp on the abscissa. Separate nomogram was constructed for each of the
cyclons with gas flow rate equal to 10^/min. Fractional characteristics of
cyclons were determined by calibration in laboratory with particles of
known size, for purposes of nomogram development. Parameters of dispersion
analysis are easily determined with nombgrams.
Cyclons were investigated in great detail under experimental
conditions.
-------
Table 1
Comparison of Calculated and Experimental
Characteristics of Impactors
No. of
Levels
1
2
3
4
5
6
7
8
9
10
Diameter of No. of
Perforations Perforations
10
slit 5 x 30
8.5
3.5
5
2
1.4
1
1
0.8
1
-
1
8
1
6
8
8
5
4
dso Calculated d50 Experimental
Microns Microns
15
15
12
8.
5.
3.
2.
1.
1.
0.
5
5
2
3
4
0
65
15
15
12
8
5
4
3
2
1
1
.5
.5
.0
.2
.1
.6
.1
d50 exp/
d50 cal
1
1
1
1
1
1.25
1.4
1.5
1.6
1.6
a
ex
1.
1.
1.
1.
1.
1.
1.
1.
1.
1.
E
3
3
3
3
4
5
6
7
8
8
i
00
0
1
-------
-681-
Fractional characteristics of cyclons for particle concentration
from 10 mg/m to 40 g/m^ remain the same. This was established by
measuring particle dispersion by Stokes diameter at inlet and outlet of
laboratory cyclons. Differences in cyclon efficiency observed by other
authors are explained by phenomena of particle aggregation at inlet and in
some cases their disaggregation in the cyclon. In determination of
fractional cyclon efficiencies gas flow rates from 5 to 30 1/min were
used. Gas velocity at inlet varied from 0.4 m/sec for cyclon with
50 mm diameter to 40 m/sec for cyclon with 16 mm diameter.
In all cases fractional efficiency curves were log-normally distributed.
(fig. 6)
Analytically these lines are discribed by two parameters: distri-
bution limit, d/particle) f or which efficiency eguals 50% and distribution
dispersion, IgO particle.
Changes in above parameters indicate changes in fractional efficiency
for cyclons with varied geometry and varied operational systems. Stokes
criteria function was used in evaluation of data on fractional efficiency
of laboratory cyclons. Gas turbulence in cyclons, relationship ^ cyclon/D cycloi.
(^-length,D-diameter)
relationship inlet/Fp^ane (inlet cross-section/cross-section of cyclon body)
were accounted for in evaluation.
Following empirical relation is used to calculate first of two
parameters which characterize cyclon efficiency.
-------
682-
Where:
M - gas viscosity
D - cyclon diameter
.r - density of particles
Vin - velocity at inlet nozzle
Re - Reynolds number accounting for flow turbulence
A - constant, non-unit, for example equal to 6
Following empirical is used to calculate the second parameter Ig cyclon:
ft'}
l /
This parameter is determined by cyclon geometry and does not depend
on aerodynamic parameters.
Equations 3 and 4 can be used in design of laboratory cyclons in
making choices for geometric and system parameters which assure desired
fractional characteristics.
Nomogram of equal efficiency values for cyclons with known fractional
characteristics can be constructed with use of a single generalized
nomogram, shown in fig. 7.
Coordinates Ig (d particles/ d cyclon) and Ig^^particle + Ig^Qcyclo
Following problems can be solved with this nomogram and with construction
of individual nomograms for cyclons of known fractional characteristics :
a) When fractional characteristics for two different cyclons are
known dispersion composition of particulates can be calculated with values
of general efficiency.
b) When parameters of dispersion composition are fixed fractional
characteristics of cyclon can be calculated with values of general
-------
-683-
efficiency for two different aerosols.
c) When fractional characteristics of cyclon are known with
fixed dispersion composition general efficiency of cyclon can be
determined.
Parametric representation of dispersion composition in industrial
aerosols is advantageous in calculation and preliminary evaluation of
efficiency of dust collectors. Log-normal distribution of dispersion
composition is used most widely. It was determined in the past that
aerosols of different origins characterized by same chemical composition
are distributed log-normally.
It should be added that when fractional efficiency of apparatus
an d dispersion composition of aerosols at inlet is log-normal then
dispersion composition of aerosols in the hopper and at the outlet of the
collector is also log-normal. Kolmogorov criteria can be used to determine
amount of deviation from log-normal distribution. According to this criteria
size of maximum deviation, D, of two probability functions from each
other is a measure of their deviation. D ranged from 3 to 5% for aerosols
of log-normal dispersion composition in wide interval of dispersion
values. (0.2 to 0.4) This was true for aerosols collected in cyclons
and for those which passed through.
Errors resulting from approximation of dispersion composition are
small when general efficiency of the collector is evaluated with curves
of fractional efficiency and dispersion composition, (on probability scale).
As an example: when D=3%, difference between calculated and true values
-------
-684-
of general efficiency in area close to 50% is about 1% and in area close
to 95% is about 0.3%.
Measurements of dispersion composition were conducted for various
aerosols in a number of industrial set-ups with a single cyclon separator
and impactor. Table 2 shows some of the results. It can be observed that
regardless whether the sample is taken at the inlet or outlet to the
apparatus, aerosols of different origins, coagulated are log-normally
distributed. This can be explained by the fact that during technological
process (burning, drying, grinding, etc.) as well as during gas flow
through pipes over-pulverization of aerosols occurs due to influence of
gravity, inertia, turbulent diffusion of stream etc. The enumerated
influences, none of which assures sharp separation of aerosols, should aid
in normalization of distributions. Aerosols are also log-normally
distributed by Stoke's diameter. The average diameter on distribution
curve is skewed to an increase when dispersion distribution is retained
or somewhat decreased.
For some aerosols measurements with impactor and cylindrical
separator compared to results of dispersion analysis conducted in the
laboratory with gonell instrument air separating deposition extracted
from cyclon separator.
Results of measurements are presented in Table 3. Measurements of
aerosols of magnesite, coal ash, and cement agree well for all three
methods. For zinc oxide and dolomite significant changes were observed.
For first three cases aerosols were larger than 10 microns and did not
coagulate during flow through pipes. Fine oxide and dolomite coagulated
which resulted in marked particle size increase.
-------
TABLE 2
Results of Measurements of Dispersion Composition of Aerosols
at Inlet and Outlet of Dust Collectors
t
Industrial
Set-Up
Furnace for
Aluminum
Electrolysis
M
Jet Mill
Fire Kiln
CURES (?)
Sampling
Location
Inlet to
Foam
Apparatus
Outlet to
Foam
Apparatus
Cyclon
Outlet
Inlet to
ESP
Outlet
Foam
Aerosol
Classification
Electrolytic
ii
Hydrophobic
Aerosols
Dolomite
Fly Ash
Cyclon Separator Impactor
Particle Microns | cr Particle Microns a
35 2.3 30 2.1
4 1.9 5 2.1
i
00
32 2.2 28 2.3 V
26 2.6 24 2.8
10 3.8 12 3.5
-------
TABLE 3
Comparison of Aerosol Dispersion Composition Measurement with Different Methods
Classification
of Aerosol
Magnesite
Coal Ash
Cement
Zinc Oxide
Dolomite
Cyclon Separator
Particle Microns 1
14
20
9
3
4
Imp act or
a Particle Microns
2.3 12
1.8 17
3.5 9
2.3 2.6
1.6 3.4
Gonell Apparatus
| 0 Particle Microns
2.1 15
1.9 22
3.6 12
2.1 15
1.8 13
1 "
2.1
1.7
3.3
2.7
1.7
l
00
1
-------
-687-
fil^m Tir7j/iiifM^rJ^tg*TlTrft
Figure 1. Eleven-stage impactor.
1. Body
2. Nozzle perforations
3. Lining (two-phase lubricant or
nappy material)
4. Glass fiber filter
5. Outlet pipe
6 . Inlet pipe with partition nozzle
-------
-688-
80
60
40
20
r
i/D
0.38
'^ I I
'I St
I I I
' I I
U
O.HU
Oft 1.1 1.SVAZA
C
•>/Stk
5 \O \S
e
4
f
ft
95
90
70
50
30
10
12 ».5I
a
U
Vstk
Figure 2. Experimental curves for fractional efficiency
in different levels of impactors
a - d from: T. T. Mercer, R. G. Stafford, Ann.
Occup. Hyg., 12, 1969
d - f from: W. E. Ranz, W. B. Wong, Ind. Eng.
Chem., 44, 1952
Upper set - linear scales; lower set - log probability
coordinates
-------
-689-
0.5
20
D,mm
Figure 3. Nomogram for determination of dso
for various levels of impactor as
it depends on nozzle diameter Dn
and on number of perforations n
with gas volume 10 1/min.
-------
-690-
dstd/
Figure 4. Nomogram for determination of d50
for various levels of impactor with
predetermined values of gas volume
Q, particle density p, and gas
temperature t.
-------
-691-
c
o
-p
u
M
t>
0)
•H
u
•H
D
O
If) CVJ N IO CM
'
Iog(d50/d particle)
log a particle
Figure 5. Curves for equal efficiency values for two
analogous sequentially set up impactor
levels.
equal efficiency values, curves for
first level
equal efficiency values, curves for
second level
-------
-692-
2 -
Figure 6. Curves of fractional efficiency
for three cyclons which compose
cyclon separator.
1. for first cyclon
2. for second cyclon
3. for third cyclon
-------
-693-
0)
0
u
o
tJI
o
cu
rH
o
•H
-p
M
(0
p.
-1.0-0.9-0.8-0.7-0.6-0.5-0.4-0.3-0.2-0.1 0 O.I 0.20.30.4 0.5 0.6 0.70.8 0.9 1.0 I.I
log (d particle/d cyclone)
Figure 7. Generalized nomograjm of equal values for
cyclon efficiency.
-------
694-
-------
-695-
Paper No. 29
TECHNOLOGY OF PARTICULATE SAMPLING FROM
REACTIVE, DAMP, AND HIGH-TEMPERATURE GASES
by
V. A. Anikeyev,
V. P. Bugayev,
V. A. Limanskiy,
Ye. N. Andrusenko,
and
V. Yu. Padva
(Presented by Valery P. Kurkin,
STATE RESEARCH INSTITUTE OF
INDUSTRIAL AND SANITARY GAS CLEANING)
Moscow
-------
-696-
-------
-697-
Technology of Particulate Sampling from Reactive,
Damp, and High-Temperature Gases
(Presented by V. P. Kurkin)
V. A. Anikeyev, V. P. Bugayev, V. A. Limanskiy,
Ye. N. Andrusenko, V. Yu. Padva
Aerosol sampling to obtain data like weight concentration, dispersivity
and chemical compsotion of solid, liquid and gaseous phase is important
in periodical efficiency inspection of the collector instruments in chemical,
metallurgical and other branches of industry. The theoretical bases,
classical sampling methods and instrumentation are described in literature.
(1,2,3).
Sampling from damp, reactive and hiqh temperature qases with the
use of classical methods does not always give reliable results as was
shown in the studies of Zaporozhe division of NIIOGAZ. This lack of
reliability was explained by the presence of several reactive components in
the gaseous phase, (H, HC, C2, 02, P203 and others) which reacted with
elements of filtering materials, and also by the presence of liquid spray
in the wet scrubber gases. One of the most difficult and complex problems
during efficiency inspection of the collectors is calculation of particulates
concentration in gas by weight. Automatic instruments for calculation of
particulates concentration, which were developed recently, have had rather
limited application for several reasons. The technology for particulates
control in majority of cases uses instruments of intermittent functioning
which require large number of servicing personnel.
In sampling for damp, high temperature and reactive gases, thermally
and chemically stable designs and filtering materials were used in
-------
-698-
Zaporozhe NIIOGAZ.
Improvement of the design of sampling instrumentation for intermittent
functioning was also considered.
The use of the glass adapters filled with glass wool in sampling from
wet scrubbers where the gas not only contains solid and liquid phases
but also hydrogen fluoride in concentrations greater than 30 mg/nm3 is
not recommended.(2) The hydrogen fluoride reacts with the glass to form
silico-fluorides. The original weight of the glass adapter decreases
significantly. The resulting mistake, depending on the hydrogen fluoride
concentration and time of sampling may reach tens of percent.
A number of synthetic fibers and a number of construction materials
were experimented with in the laboratories and in the industry (3) for
particulate sampling from electrolytic production of aluminum and from
melting of fluoride flux in the furnaces. The model samples were subjected
to hydrogen fluoride with concentration up to 700 mg/nm^ and to the
solutions of hydrofluoric acid (0.2 g/Ji) .
The recommendation was made to use stainless acid resistant steel
and teflon sampling instruments with carbon or polyphene fibers in
combination with porcelain clay wool.
To decrease the amount of time spent for sampling NIIOGAZ developed
several cassette models of particulate sampling units. With these
cassettes spray concentration up to 5 g/nm3 and particulate concentration up to
10 g/nm3 can be determined simultaneously or individually. The mist and
particulate concentration is determined by weight.
-------
-699-
Figure 1 shows a scheme of the aerosol sampling unit with manual
control of filter exchange (Patent No. 380980, I. OIP 1/22). It is made
up of the sampling nozzle 1, transport pipe 2, screw 3, turning socket 4,
cassette 5 (with three filters), body 6, nut 7, which joins body 6 with
hood 10, fingers 8, placed on the hinged disc 9, conical reductor 11,
pipe 12, connecting pipe for gas outlet 13, handle for control of filter
exchange 14, filters 15, and the sealing liner 16. The appropriately
prepared filters are earlier placed in the cassettes. On the sampling
location cassette is placed inside the sampler which is secured at the
sampling point. The aerosol sample flows through one of the three filters.
(gas does not flow into the other two at this time). When the earlier
decided upon hydraulic resistence of the filter is reached, the cassette
is moved with the fixed turn of the control handle. The dirty filter
is exchanged for a clean one.
Other modifications of the aerosol sampling unit include different
numbers of filters in the cassette and a different system of their exchange.
Three types of filters are used in the aerosol sampling unit. They
are: for determination of particulates concentration, for determination
of only spray concentration following wet scrubbers and for determination
of total concentration of particulates and spray. The filters are porous
teflon, metalloceramic, ceramic and other materials. Depending on the
specific properties of the aerosols possibility exists that the filters can
be used repeatedly. The filters are regenerated with acid or alkaline
treatment. The regeneration procedure is determined experimentally for
each specific case.
-------
-700-
When the samples of damp gases are taken spray liquid also containing
collected particulates separates on the surface of the sampling unit.
The separated liquid flowing onto the surface of the sample unit is
directed into the collector nozzle and on to the filter. In this way,
the true concentration of the spray and the particulates is increased. This
phenonenon is especially true for the vertical and sloped gas pipes with the
uptake gas stream. The use of the gas collecting nozzle (Patent No. 393640
I. DIP 1/24) allows to exclude that increase from the measurements and allows
highly reliable aerosol sampling for the determination of spray and
particulate concentration. The separated spray liquid is removed from the
surface surrounding the collector nozzle and in this way excludes possible
(false) increase of true aerosol concentration. The aerosol sampling units
are also fitted with such nozzles.
In sampling with the use of the cassettes, non-productive time loss
is decreased by 25-30% in comparison to the sampling done with the singular
filters in the conventional sampling units. The sampling instrumentation is
constructed with non-corrosive materials which allows its use in sampling
from damp, reactive gas with temperature range 0-250°C. The weight of the
sampling unit is, depending on the modifications, from 3 to 4 kg, diameter
60-75 mm, length (with control handle) up to 1.5m. The dimensions allow
for the sampling from the gas pipes with a diameter up to 2.5m.
The method developed for the internal sampling (Patent No. 326131,
I. SOI v 17/02) allows for the determination of sulfur concentration in
the gas stream with temperatures ranging from 550 to 500°C. The collected
gas sample is filtered preliminarily in the adapters of molybdenum glass
-------
-701-
filled with porcelain clay wool. It is measured then under the gas
stream conditions with a diaphragm. The elementary sulfur vapors
condense in the storage and the second measurement of the gas sample is
taken. The difference is indicative of the sulfur vapor concentration in
the technological gases.
Concrete applications of the sampling units for the measurement of
particulates concentration in gases, under raised pressures are looked at
in the following examples.
1. Sampling from surfaces of reactors and regenerators of catalytic
cracking installation and also following cyclon collectors in such installations,
Methods of external gas filtration are used. The sampling tube is
introduced through the valve unit,. (Fig. 2) The valve unit is constructed
using standard fasteners with conventional inlet, 50-80 mm.
In sampling for reactor gases the valve unit has additional inlets
for water vapor and drainage outlet for water. The use of the vapor is
necessary for the control of salt compounds density before the sample is
taken.
Depending on the concentration of particulates and the dispersion
composition of the particulates.
a.) only laboratory cyclon with a hopper;
b.) laboratory cyclon and filtering adapter;
c.) only filtering adapter
are used.
The particulates have to be separated from the products of refining
when the particulates from the reactors are sampled for. To accomplish
that the samples and the residues of the products of refining and coke are
washed and burned in the muffle furnace.
-------
-702-
To take the gas samples following the cyclons, where stream velocities
and particulates concentration are unevenly distributed, inside of the gas
pipe devices are placed which equalize stream velocity and particulates
concentration in the cross-section (gas line axis) before the gas sample
is taken.
The measurements were conducted under the experimental as well as
under the industrial conditions of catalytic cracking installations and
butane dehydrogenation.
2. Taking of the samples following soot removal from gasification of
sulfurous oils under pressures 2.5-2.0 atm.
With pressures higher than 2.5 atm. sampling with valve units becomes
more complicated because special arrangements have to be made for inserting
and removing the sampling tube.
Special units with intercepting rods were found to be more appropriate
under those circumstances. (Fig. 3) The metal adapter for filtration is
placed inside of the intercepting rod (internal filtration) or inside of
the gas line (external filtration). All phases of the sampling (gas
interception, purging of the sampling nozzle with inert gas, adapter
installation, gas flow through the adapter.) are secured by different
locations of the locking rod as it relates to the nozzle opening.
Each sampling unit contains a pneumatic drive of the intercepting
drain, which allows for automatic change of the rod to different positions.
The adapter is changed with the handle. The sampling time is determined
by the particulates concentration. The units sample soot concentration in
-------
-703-
the gas stream from 2 mg/nm to 10 mg/nm . The design studies and
the preparation of the sampling unit models showed that the type with the
intercepting rod can collect a sample automatically with pressures
up to 100 atm.
3. With small concentration of particulates and temperatures reaching dew
point (blast furnace gases P=2.5 atm., following passage through ESP) only
internal filtration method of sampling is possible.
When particulates concentration reaches tens of mg/m^ of gas glass
adapters with glass wool are used. When the concentration of particulates
is below 10 mg/m gas is filtered through very thin cellulose fiber filters
which were developed by Karpov Institute. The hydrophobic filter fibers
eliminate the need for a long process of bringing the filter to constant
weight. To determine the quantity of coarsely dispersed spray in the gas
from the ESP, adapters which accumulate moisture are used. The spray
settles in the adapters because of the gas colliding with the surface.
All sampling from the blast furnace is conducted with the use of the
above described sampling units with valve interceptors.
-------
-704-
REFERENCES
1. G. M. Gordon, T. L. Peysakhov, Inspection of Particulate Collectors,
Moscow, 1961
2. V. P. Bugayev, V. A. Limanskiy, V. N. Sajonov, V. P. Klyushkin,
Industrial and Sanitary Gas Cleaning, 1972, No. 4, 25-26.
3. I. Ya. Boyev, Ye. G. Levkov, V. A. Limanskiy, V. P. Bugayev,
A. S. Levkova, Industrial Laboratory, 1972, No. 3 278-281.
-------
-705-
10
Figure 1. Sampling Unit.
-------
-706-
Figure 2.
1. regenerator
2. fastener
3. gasket packing
4. sampling tube
5. measurement diaphragm
6. metal adapter
7. cyclon with hopper
-------
Figure 3. Schematic presentation of a set up for particulates sampling
with increased pressure.
-------
-708-
-------
-709-
Paper No. 30
MEASUREMENT OF PARTICLE SIZE DISTRIBUTIONS AT
EMISSION SOURCES WITH CASCADE IMPACTORS
by
Michael J. Pilat
UNIVERSITY OF WASHINGTON
Seattle, Washington
-------
-710-
-------
-711-
ABSTRACT
Cascade impactors can be used to measure the size distribution of parti-
cles in ducts and stacks at emission sources. The cascade impactor is
usually inserted inside the duct or stack to enable isokinetic sampling, to
minimize losses of particles to the sampling probe wall, and to reduce pro-
blems with water condensation. Traditionally, cascade impactors have pro-
vided particle size data in the 0.2 to 30 micron diameter general size range.
Recently high pressure drop cascade impactors have been used to measure the
particle size distribution down to about 0.02 microns diameter. Solutions
to the problems of particle re-entrainment, sampling gases with entrained
water droplets, and simultaneous sampling at the inlet and outlet of parti-
culate control devices are discussed.
-------
-------
-713-
!. Introduction
Cascade impactors have been used for some 30 years to measure the size
distribution of aerosol particles. May (1945) reported particle size
data in the 1 to 20 micron diameter range measured with a 4-stage cas-
cade Impactor with rectangular jets. Since then many different types of
cascade Impactors have been developed^ mainly for sizing aerosol particles
in atmospheric and Industrial hygiene studies.
First et al (1952) and Gussman and Gordon (1966) reported on the modifi-
cation of a Casella Impactor so that it could be used to sample particles
in ducts and stacks. The Casella was modified such thet the first im-
pactor stage was located in the sampling probe elbow, thus preventing
the loss of large particles to the elbow wall,, Brink (19580 1963) deve-
loped a five stage cascade impactor with single round jets to size mist
droplets in the 0.3 to 3 micron diameter size range. The mist aerosol is
sampled from the stack or duct into the Brink impactor located in a heated
sampling box. In 1968 the need for size distribution date concerning par-
ticles in exhaust gas streams stimulated the development of the Mark I
University of Washington Source Test Cascade Impactor. As reported by
Pilat, Ensor and Bosch (1970), the Mark I UW impactor (has six stages with
multiple round jets on each stage) had problems with the loss of particles
onto the top of the first jet stage (particles too large to follow the
gas stream lines through the jets) and with loss of particles,to the walls
between the stages. Based on the experience with the Mark I UW impactor9
a Mark II model was developed. The Mark II UW impactor has a single jet
first stage followed by six multi-jet stages and a 47 mm diameter filter
holder. The sampling nozzle includes the single jet for the first stage
and therefore, the problem of particle loss upon the top of the first
multi-jet stage was eliminated.
The Mark El UW Cascade Impactor was developed in order to further reduce
the loss of particles to the walls. The Mark III UW impactor has a single
jet first stage followed by six multi-jet stages and a 47 mm filter holder
as shown in Fig. 1. The particle collection plates have a hole in their
centers for the gas to flow through and this center hole eliminates the
need for the gas to flow around the outside edges of the plate. The
various parts of the. Mark III UW impactor are shown in the photo in Fig.
2. The annular particle collection plates have successfully solved the
problem of particle loss to the cylindrical walls of the outer casing as
the gas and aerosol particles have no contact with.these walls. The only
surfaces upon which the particles can be deposited (other than the tops
of the particle collection pVates)- and become "lost" from the sample are
the top and bottom of the jet stages and the bottom of the collection
plates. By the proper selection of the gas sampling flow rates through
the Mark III UW impactor the deposition of particles upon the jet stage
tops and collection plate bottoms only seldom occurs (possibly due to rough
handling of the Impactor between the sampling location and the laboratory„
to overloading the collection plate with too large a sample, or to very
high electrostatic charges on the particles). However, at times the par-
ticles do bounce off the particle collection plates and deposit upon the
bottom side of the jet stages. The problems with particle re-entrainment
are covered in the discussion section of this paper.
-------
-714-
Fig. 1 u.w. MARK HI
SOURCE TEST CASCADE IMPACTOR
(CROSS SECTION)
NOZZLE
INLET SECTION
COLLECTION PLATE NO. I
JET STAGE NO. 2
COLLECTION PLATE NO. 2
JET STAGE NO. 3
COLLECTION PLATE NO. 3
JET STAGE NO. 4
COLLECTION PLATE NO. 4
JET STAGE NO. 5
COLLECTION PLATE NO. 5
JET STAGE NO. 6
COLLECTION PLATE NO. 6
JET STAGE NO. 7
COLLECTION PLATE NO. 7
FILTER COLLAR
FILTER
FILTER SUPPORT PLATE
OUTLET SECTION
-O-RINGS
-------
U OF W S
-------
-716-
Since the development of the UW Source Test Cascade Impactors, other cas-
cade impactors have been also developed or adapted to stack sampling.
Downs and Strom (1972) reported on the modification of a Brink Mist
Sampler by adding a cyclone at the inlet, two additional jet stages and a
filter at the outlet for use in sampling with the impactor inside the
stack. Holland and Conway (1973) have reported on three in-stack sampling
cascade impactors; the Andersen Stack Sampler, the TAG impactor, and the
Mark III UW Source Test Cascade Impactor. Bird, McCain, and Harris (1973)
reported on a comprehensive particle size measurement program involving
the use of eleven different commercial and modified particle sizing de-
vices at a coal -fired electric generating plant.
II. Calibration of UW Cascade Impactors
A. Theoretical
Based on a solution to the particle equation of motion the Stokes iner-
tlal parameter i|> is given as
where C is the Cunningham correction factor pp the density of the particles,
d the particle diameter, Vj the velocity of the gas in the jet, y the gas
viscosity, and Dj the diameter of the gas jets in a given stage. The magni-
tude of the Stokes inertial parameter at the particle diameter that is col-
lected with 50% efficiency by a given jet stage has been reported to range
between 0.12 to 0.17 for circular jets (the square root of ^50 ranges from
0.35 to 0.41). Note that these studies assumed impaction upon a flat plate
and the particles upon impaction are collected (does not take into account
particle bounce, particle re-entrainment, or particla impaction onto glass
fiber filters).
Using equation (1) an equation can be developed which relates d5Q to the
cascade impactor design and operating parameters. Solving for the particle
diameter in equation (1) gives
|) ,/? .
Substituting a value of 0.145 for \IUJQ (corresponds to the square root of
of 0.38) provides the equation
2.61nD.
The velocity of the gas in each jet is calculated by
V, - -- (4)
0
-------
-717-
Where Q in the gas volumetric flow rate at stage conditions, N the num
ber of jets on a given jet stages and Dj is the jet diameter for that
same stage. Substituting equation (4) for Vj ira equation (3) gives
2.05yD.3N
(5)
Equation (5) can be used to calculate the dgQ of any cascade impactor
stage. The d5Q magnitudes calculated for a certain set of stages for the
UW Mark III impactor are presented in Fig, 3, Note that the gas flow rate
Q is at the stack conditions (actual flow rate passing through the impac-
tor).
B. Experimental Measurements
The experimental calibration of cascade impactor stages is sometimes de-
sired in order to verify the impactors sizing capabilities (even though
its design may be based upon experimentally verified criteria). The UW
Mark I Cascade Impactor was calibrated ming mcirtodi sparse Dew latex spheres
of 1.9 and 3.5 microns diameter as reported by Bosch, Pi'lat and Hrutfiord
(1971). The quantity of particles collected on each collection plate was
determined using an optical microscope and the stage collection efficien-
cies calculated from these quantities. The stage collection efficiencies
and the square root of the Stokes number are plotted in Fig. 4 and show
good agreement with the square root of ^(-n of 0038 reported by Ranz and
Wong (1952) for circular jet impactors. au
t!
It is recommended that photomicrographs of the particles sampled on the
collection plates be periodically taken in order to illustrate the size,
color, and shape of the particles being sampled. Please note that the
size of the particles on a given collection plate should range from about
the dso of that stage to the d§o of the next upstream stage. For more
detailed information on the interpretation of impactor data9 please refer
to Mercer (1965).
III. Sampling Procedure for UW Source Test Cascade Impactors
The entire particle size distribution measurement procedure includes
three phases; pre-test preparation, source test sampling of the particu-
lates, and analyses of the collected samples and recorded data. The pre-
test preparation includes cleaning the impactor s placing a thin layer of
grease on the collection plates if solid particles are to be sampled,
weighing the plates and filter (or weighing the insert foils placed on
top of the plates). The source test involves first determining the gas
velocity profile in the stack (measure the gas temperature and pressure
drop profile with type S pitot tube) and then calculating the nozzle size
for isokinetic sampling. The sampling train is set up as shown in Fig. 5
with the cascade impactor on a sampling probe (1/2 inch diameter stainless
steel probe) followed by a 1/2 inch diameter Teflon lined flexible hose,
four Greenburg-Smith impingers (first two with 100 ml of water, the third
is dry, and the fourth has silica gel), a leak! ess vacuum pump, and a dry
gas meter. Sometimes a 47 mm diameter glass fiber filter is placed down-
stream of the dry Greenburg-Smith impinger to collect particles condensed
in the impingers. The UW Cascade Impactor is preheated to prevent conden-
-------
-718-
Particle density = 1.0 gm/cm
2.60 3
4 B 6 7 8 9 10°
FLOW RflTE.Q.tCUBIC FEET /HIM.)
Jet
Dia. No.
Stage (inches) Jets
1
^ 2
: 5
2 .GO
0.7180
0.2280
0.0960
0.0310
0.0135
0.0100
1
12
90
0.0200 110
110
90
F1g. 3 Calculated dgo of Mark III-F Stages
-------
-719-
V,
O
uu-
90-
80-
70-
60-
50-
40-
30-
20-
1O-
0-
o
o
A
u
Ro.ni ?3nd V^o^g found v^ "0.38
.Xot 50% efficiency
o •
Stage 3 A A
Stage 4 o •
Stage 5 a •
A Stage 6 0 •
o
A
* 1 1 1 1 1 I 1 I
0.2 0.4 0.6 0.8 1.
0
Fig. 4 Measured Particle Collection Efficiency versus Stokes
Number for Mark I UW Cascade Impactor
-------
-720-
Impactor
]
I
Exhaust Gas
Flow
Teflon Lined
Flexible Hose
S^=^ jr-i
rt?—i
LhCVB T "'"..•nV-nx.'^i.Ti'ii^ •*
THERMOMETERS FINE ADJUST VALVE
DRY TEST METER
Fig. 5 Sampling Train for U.W. Cascade Impactors
-------
-721-
sation problems by placing it into the stack with the nozzle faced down-
stream for about 10-15 minutes prior to sampling. After preheating the
impactor, a particle sample is obtained by facing the impactor nozzle up-
stream and turning on the vacuum pump. The gas sampling rate is main-
tained at the constant isokinetic rate throughout the test (typical flow
rates are in the 0.3 to 1.5 cfm range). As can be seen in Fig. 3 chang-
ing the gas sampling rate will also change the magnitude of the stage
d5QS. The gas sampling time, temperature, volumetric readings from the
dry gas meter, etc. are recorded on a data sheet. After obtaining the
particle sample, the UW impactor is removed from the stack being careful
not to hit the impactor or probe against the stack or sampling port
(particles can fall off the collection plates if the impactor is bumped
around too much). After removal from the stack, the impactor is dis-
assembled and the particle collection plates are removed (preferably in
a clean wind-free location) and weighed. In some cases (for example when
sampling downstream of a wet scrubber) it may be necessary to dry the
collected particles by heating in an oven and cooling to room temperature
in a desiccator. The weights of the particles collected on the plates
and in the filter are used to calculate the cumulative particle size
distribution,
IV. Discussion
A. Particle Re-entrainment
The problem of the re-entrainment of particles impacting upon the collec-
tion plates of cascade impactors was reported many years ago by May (1945).
May suggested that cascade impactors use low gas velocities in the jets
and an adhesive (such as 3 parts castor oil to 1 part rosin) to reduce
particle bounce and blow-off. By referring to equation (3)
2.61uD. ,.,
H - r Ji '/z
d50 L CpV. J (3)
r J
it can be seen that if the gas velocity in the jet Vj is restricted to
lower magnitudes, then it is necessary to use small jet diameters Dj in
order to obtain the lower magnitudes of the stage dsgs. To achieve suffi-
cient gas volumetric sampling rates we have found it necessary to use
many jets per stage (unfortunately the single large jet per stage impac-
tors have excessive particle bounce and blow-off problems at the smaller
particle dcgs).
The design of the UW Mark III Cascade Impactor includes considerations for
reducing the particle blow-off caused by excessive gas velocities. Also
adhesives such as Dow silicone high vacuum grease are used on the particle
collection plates.
The magnitude of the apparent problems of particle blow-off is illustrated
in Fig. 6 which presents particle size distributions simultaneously mea-
sured by four cascade impactors, as reported by Bird, McCain and Harris
(1973). The measurements were made at a coal fuel power plant at the out-
let of a mechanical collector and the inlet to the electrostatic precipi-
tator. As is shown In Fig. 6 the peak in the particle size distribution
1s at about 4.5 microns diameter for the UW Mark III impactor (commercially
-------
-722-
-I—I Mill
Bird, McCain, anc
1 1 T
Harris (1973)
1 I I IT
O U.W. Mark III
O Brink
Andersen
Q E.R.C. TAG
I I
I I I I
.4 ,6 ,8 1 2 468 10
Particle Diameter (microns)
Fig. 6 Comparison of Simultaneous Cascade Impactor Size Distribution
Measurements
-------
-723-
avaiTable model as purchased from Pollution Control System Corp.), at
about 3.0 microns diameter by the Brink impactor (specially modified for
in-stack use with an additional stages an internal filter holder, and
one or two cyclone precollectors), at about 2.8 microns by an Andersen
impactor (specially constructed prototype using ylass fiber filter im-
paction substrates)9 and at about 2.3 microns diameter by an Environmental
Research Corporation TAG (commercially available from ERC) with greased
foils on the last four stages. It appears that the cascade impactors
with the size distribution peaks at the lower sizes suffer from the pro-
blem of particles bouncing or blowing off and being collected on down-
stream plates or by the filter (and thus indicating a smaller size dis-
tribution than actually exists). It should be tnot^d that only the UW
Mark III and the ERC TAG impactors were commercially available models,
whereas the other two were specially assembled for tills EPA funded evalu-
ation (this research project concerning the svaluation of cascade impac-
tors for measuring the size distribution of particles at emission sources
is being conducted by Southern Research Institute).
B. Simultaneous Sampling at the Inlet and Outlet of Particulate Control
Equipment
For a single cascade impactor particle sample9 the stack gas sampling
time need only be as long as is necessary to obtain a weighable particle
sample. Excessive sampling times can result In the overloading of the
particle collection plates which can contribute to the particle re-entrain-
ment problem. Note that a number of particle samples obtained with short
sampling times (we have used as short as 4 minute sampling times) is
superior to one sample of longer times, as reported by Kahnweld (1966).
Of course, for characterizing the emissions from a process that is vari-
able over small time periods, it is mandatory to use short sampling times.
When simultaneously sampling at the inlet and outlet of a particle collec-
tion device it is common to have the outlet concentration about 1 to 10%
of the inlet particle concentration. Therefore, by sampling at the same
gas volumetric flow rate with identical cascade impactors, the weight of
particles collected on the outlet impactor plates is about 1 to 5% of the
inlet impactor plates. This approximately one hundred-fold difference in
the weight of the collected particles can result in overloading of the in-
let impactor and insufficient particle sample weights with the outlet
impactor.
There are a number of approaches to solving the simultaneous inlet-outlet
sampling problem. One is to operate the inlet impactor at a lower gas
sampling rate than the outlet impactor. However, this causes the impactor
stage dijgs to be different between the inlet and outlet samples and can
result in strange particle collection efficiency curves. A second approach
is to use light weight foil inserts on the particle collection plates thus
lowering the plate tare weight and increasing the weighing precision. Also
using weighing balances with greater sensitivity (say to 0.010 milligrams)
will enable accurate weighing of the outlet impactor plate samples (i.e.,
at about 0.5 milligrams particles per plate) with the inlet impactor plates
not overloaded with particles (i.e., at about 50 milligrams particles per
plate). A third approach is to use cascade impactors designed to operate
at different gas sampling flow rates. The low sampling rate impactor would
be used at the inlet and the high sampling rate impactor at the outlet.
-------
The research and development of such a pair of cascade impactors Is under-
way at the University of Washington (primarily for use with our research
programs concerning the development of high efficiency particle collection
systems). A fourth approach is to use a cyclone with a cascade impactor
at the inlet of control equipment. A BCURA cyclone with a Mark III UW
Impactor is shown in Fig. 7. The cyclone serves to prevent the impactor
from becoming overloaded.
C. Sampling of Gases Containing Water Droplets
When sampling downstream of wet scrubbers it is possible to encounter
water droplets in the gas stream. The presence of these water droplets
is usually the result of inefficient (or the lack of) rnist eliminators.
As these water droplets usually contain particulate matter, it is actually
not good practice to allow these water droplets to exhaust into the atmos-
phere (assuming that the objective is to reduce the particulate emissions).
An in-stack cascade impactor will classify the water droplets into size
fractions in a manner similar to any other aerosol particle. However, it
is possible to flood the impactor when high water droplet concentration
exists.
If the water droplet concentration is low, filters may be placed on the
particle collection plates to absorb the water and thus prevent it from
washing off the particles, seeping from plate to plate, and in general
making a mess out of the impactor samples. The particle collection plates,
the filters, and the collected particle and water sample can be weighed
"wet" to provide information concerning the size distribution of the wet
aerosol (particles plus water droplets).
With high water droplet concentrations and large diameter water droplets,
it is very difficult to isokinetically sample the gases as the droplets
flood the inlet particle collection plates. The use of a long heated in-
let nozzle may serve to evaporate the water droplets sufficiently to pre-
vent this flooding. However, in some cases it may be necessary to sample
at right angles in order to obtain an indication of the size distribution
and mass concentration of the smaller aerosol particles.
D. Use of UW Cascade Impactors to Measure In-Stack Particle Mass
Concentration
The UW Cascade Impactors have been used to measure the particle mass con-
centrations at in-stack conditions. The sum of the particle weight? col-
lected by the plates and the filter divided by the volume of gas sampled
provides the particle mass concentration. Simultaneous particle sampling
with UW Cascade Impactors and in-stack alundum thimbles lined with glass
fiber and followed by a 47 mm diameter glass fiber filter have shown good
agreement in the particle mass concentration measurements. In Fig. 8 a
plot of alundum thimble and UW Mark III Cascade Impactor mass concentra-
tion measurements made on particles emitted from a hog fuel boiler show a
correlation coefficient of 0.94 between these measurements. However, it
should be noted that to obtain the total particle mass concentration (in-
stack particles plus condensible particles) it is necessary to add the
weight of the residue from the probe and hose washings, the residue from
the impinger solutions, and the particles collected by a filter located
downstream of the third impinger in Fig, 5.
-------
I
-J
to
<_n
I
Figure 7. BCURA Cyclone at Inlet to Mark III UW Cascade Impactor
-------
-726-
0.25-
Hog Fuel Boiler
June-August 1973
0.05 0.10 0.15 0.20
UW Cascade Impactor Particle
Mass Concentration (grains/act)
0.25
Fig. 8 Comparison of Particle Mass Concentrations Simultaneously
Measured with Mark III UW Cascade Impactor and with Alundum
Thimble Followed by Glass Fiber Filter
-------
-727-
E. Variation in Measured Particle Size Distributions
The size distribution of particles emitted from different sources varies
considerably. The size of particles emitted from non-continuous processes
can change greatly. Hanna and P1lat (1972) reported that the measure-
ments of particle emissions from the exhaust of a horizontal spike
Soderberg aluminum reduction cell with a UW Mark II Cascade Impactor
showed the particle mass mean diameter ranged from about 0.5 to 100 mic-
rons and the particle geometric standard deviation from about 5 to 1S000.
However, the size distribution of particles emitted by a continuous pro-
cess (such as a kraft recovery furnace) remains fairly constant as long as
the operating conditions remain the same. But the size distribution of
particles emitted by similar continuous processes are not necessarily
similar.
Of course, there is considerably more size distribution data measured by
the UW impactors than can be presented in this paper. For this additional
information it is possible to refer to publications,, The Mark I UW
Cascade Impactors have been used to measure the particle size distributions
at the UW coal-fired power plant, as reported by Pilatj Ensor, and Bosch
(1970); at the exhaust of the number 3 kraft recovery furr.sce at the St.
Regis Paper Co. pulp mill in Tacoma, as reported by Bosch, Pilat, and
Hrutfiord (1971); and at the exhaust of a steam heated veneer drier at
the U.S. Plywood-Champion Paper plant in Seattles as reported by Larssen,
Ensor, Sparks, and Pilat (1970). The UW Mark II Cascade Impactor has been
used for sizing particles emitted from kraft recovery furnaces, as reported
by Larssen, Ensor, and Pilat (1972); and for sizing the particles emitted
from fluidized bed sewage sludge incinerators, as reported by Liao and
Pilat (1972).
F. Measurement of Submicron Particle Sizes
Pilat (1973) reported on the use of cascade impactors to measure the size
distribution of submicron particles down to about 0.02 microns diameter.
Research concerning the development of the Mark IV UW Source Test Cascade
Impactor for sizing these submicron particles has been underway since
1971. The Mark IV impactor utilizes low absolute gas pressures in the out-
let jet stages to increase the Cunningham correction factor which, as shown
in equation (3), can result in lower magnitudes of d§n for a given jet
diameter (about 0.010 inch), gas velocity (less than Mach I), particle
density, and gas viscosity. The magnitude of the Cunningham Correction
factors at low gas pressures using an equation reported by Davies (1945)
is shown in Fig. 9. Laboratory and field tests have demonstrated that this
approach will provide submicron particle size data. The major problems
with this method appear to be the requirements for a good portable vacuum
pump (light enough for stack sampling) and particle bounce at the high gas
velocities.
-------
-728-
1000
I
I
1
100
10
I
I III!
Particle Diameter
(microns)
Temperature = 70° F
10 100 1000
Pressure (millimeters of mercury)
Fig. 9. Cunningham Correction'Factor as a Function
of Absolute Gas Pressure
-------
-729-
References
Bird, A. N., J. D. McCain, and D. B. Harris (1973). "Particulate Sizing
Techniques for Control Device Evaluation," Paper No. 73-282 presented at
APCA Annual Meeting, Chicago, 111.
Bosch, J. C., M. J. Pilat, and B. F. Hrutfiord (1971). "Size Distribution of
Aerosols from a Kraft Mill Recovery Furnace," TAPPI 54 1871-1875.
Brink, J. A. (1958). "Cascade Impactors for Adiabatic Measurements," Ind.
Engr. Chem. 50_ 645-648.
Brink, J. A. and W. F. Patton (1963). ''New Eqjipn;er,'; ard Techniques for
Sampling Chemical Process Gases, •' JAPCA 13 162-"66.
Davies, C. N. (1945). "Definite Equations for tne Fluid Resistance of Spheres,"
Proc. Phys. Soc. 57_ 259-270.
Downs, W. and S. S. Strom (1972) "New Parvicie Size Measuring Probe - Applica-
tion to Aerosol Collector and Emissions £va"iua-:lors,'! J, Engr. Power
(Transactions of ASME) 94 117-126,
First et al (1952). "Air Cleaning Studies Progress Report," NYU-1586, Harvard
University.
Gussman, R. A. and D. Gordon (1966). "Notes on the Modification and Use of a
Cascade Impactor for Sampling in Ducts," Am. Ind. Hyg. Ass. J. 27 252-255.
Hanna, T. R. and M. J. Pilat (1972). "Size Distribution of Particulates Emitted
from a Horizontal Spike Soderberg Aluminum Reduction Cell," J. APCA 22_
533-536.
Holland, W. D. and R. E. Conway (1973). "Three Multi-stage Stack Samplers,"
Chem. Engr. Progress 69 93-95.
Kahnweld, H. (1966). "Dust Measurements in Flowing Gases," Staub 2j[ 20-22.
Larssen, S., D. S. Ensor, L. E. Sparks, and M. J. Pilat (1970). "Size
Distribution of Particulate Emissions from a Veneer Drier," Presented at
PNWIS-APCA, Spokane, WA.
Larssen, S., D. S. Ensor, and M. J. Pilat (1972). "Relationship of Plume
Opacity to the Properties of Particulates Emitted from Kraft Recovery Furnaces,"
TAPPI 55_ 88-92.
Liao, P. B. and M. J. Pilat (1972). "Air Pollutant Emissions from Fluidized Bed
Sewage Sludge Incinerators," Water and Sewage Works, 68-74.
May, K. R. (1945). "The Cascade Impactor: An Instrument for Sampling Coarse
Aerosols," J. Sci. Instr. 22_ 187-195.
Mercer, T. T. (1965). "The Interpretation of Cascade Impactor Data," Amer. Ind.
Hyg. Assoc. J. 26,236-241.
-------
-730-
Rcferences - Cont.
Pllat, M. J., D. S. Ensor, and J. C. Bosch (1970). "Source Test Cascade
Impactor," Atmos. Envir. 4.671-679.
Pllat, J. J. (1973). "Submicron Particle Sampling with Cascade Impactors,"
Paper No. 73-284, Presented at APCA Annual Meeting, Chicago, 111.
Ranz, W. E. and J. B. Wong (1952). "Jet Impactors for Determining the Particle
Size Distribution of Aerosols," A.M.A. Arch. Ind. Hyg. Occup. Med. 5_ 464-477.
-------
-731-
Paper No. 31
THE CHEMICAL COMPOSITION OF FLY ASH
by
David F. S. Matusch
UNIVERSITY OF ILLINOIS
Urbana, Illinois
-------
-732-
-------
-733-
ABSTRACT
Almost all naturally occuring elements are represented in
fly ashes and in airborne participates derived from industrial
processes. Special significance is, however, attached to toxic
species containing elements such as As, Sb, Pbs Tl, Hg, Cd, Se,
V, Ni, Cr, S, C and Be, In fly ash derived from coal combustion
many of these elements are found to increase in concentration
with decreasing particle size. This is possibly due to vola-
tization of the element or one of its compounds followed by
preferential adsorption or condensation onto the small particles.
Evidence in support of this hypothesis is presented«, The
significance of toxic element dependence on particle size is
discussed in terms of human respiratory intake, existing and
potential control technology and emission inventories. Tech-
niques for the sampling,', size differentiation and chemical
analysis of particles are briefly reviewed.
-------
-734-
-------
-735-
INTRODUCTION
From an engineering viewpoint the control of particu-
late emissions from stationary sources implies collection
of as many particles as possible from the effluent gas
stream. This is a good approach because, if successful, it
ensures that all particles large and small, toxic and non-
toxic, corrosive and non-corrosive -will be prevented from
reaching the atmosphere. Unfortunately many particles, es-
pecially those of sub-micrometer (pm) size, do reach the
atmosphere and it is meaningful tc find out what these
particles are and what their environmental effects might be
in order to make decisions about the need for tneir control
and how this control might be achieved.
The intention of this paper is to show how a knowledge
of the chemical and physical character of particulate matter
is of fundamental importance in deciding which particles are
most hazardous and how these hazards might be reduced. In
short, it develops the thesis that control strategy should be
more firmly based on an understanding of the chemical and
physical nature of particulate material.
REASONS FOR PARTICULATE CONTROL
There are four major reasons why it is undesirable for
particulates to reach the atmosphere. Not all have equal
importance.
1. Soiling
Particles larger than about 5 l-ini can be seen by
the naked eye and, as a consequence, give rise to visible
-------
-736-
soiling when they fall upon contrasting surfaces such as
paint, clothing and the like. Their removal is therefore
important for aesthetic reasons. However, current control
technology is, in large part, capable of preventing emis-
sion of the large particles responsible for soiling. For
example the mass median diameter of airborne particles in
the San Francisco Bay area in 1960 was estimated to be 201am1
whereas today urban particulates commonly have a mass median
diameter of about l|jm or less2. Prevention of soiling can
thus no longer be considered as a major goal for control
strategists.
2. Light Scattering
While the mass median diameter of U.S. urban
aerosols has decreased over the last few decades, Table I
shows that there has been no corresponding decrease in total
mass3. Thus, the net effect has been to move the aerosol
mass into smaller and smaller particles and promote the
amount of material present in the size range 0.1 - 1.0 p.m
responsible for light scattering. It should be noted that
light scattering is related primarily to particle size and
number and is not effected by the chemical composition of a
particle.
~5, Atmospheric Interactions
A number of atmospheric reactions involving both
natural and pollutant species are influenced catalytically
by airborne particles. Many particles can also provide ad-
sorption, condensation and nucleation surfaces for gaseous
-------
-737-
species (eg formation of watei droplets). While such
processes are not necessarily detrimental to the environ-
ment they do represent a significant environmental pertur-
bation and, until more is known about them, they should be
prevented where possible. The main point, however, is that
these catalytic or reactivity effects depend upon the chem-
ical nature of the particle.
4. Health Effects
The fourth, and in most instancies the major, rea-
son for particulate control involves both occupational and
environmental health. Thus many sources emit particles which
can produce adverse health effects when inhaled or when de-
posited on skin or sensitive tissues such as the membranes
of the eye. In the great majority of cases (asbestos sap-
phires provide a notable exception) the toxicity of particu-
late matter resides in its content of more or less toxic or-
ganic and inorganic compounds. Consequently, the health ef-
fects of particulate matter are closely related to the chem-
ical composition of the particles.
Collective consideration of these four adverse effects
of particulate matter leads one directly to the conclusion
that the need for control depends primarily on the chemical
and physical character of the particles in question. In de-
termining control strategy, therefore, it is clearly appro-
priate to consider the chemical composition of particulate
emissions, their size distribution and the relationship be-
tween chemical composition and size distribution.
-------
-738-
CHEMICAL COMPOSITION OF FLY ASH
The chemical composition and physical behavior of par-
ticles emitted to the atmosphere obviously depends upon the
source or process of origin. Consideration of a variety of
sources is, however, beyond the scope of this paper. The
following remarks, therefore, apply to only one major par-
ticulate emission viz fly ash emitted from coal fired power
generating plants. Such emissions are of major current con-
cern (Table I) and, with the probable increase in coal usage
in future, can be expected to remain so for some years to come
Ideally, one would like to determine the actual chem-
ical compounds present in fly ash both as major matrix con-
stituents and at trace levels. However, while determination
of elemental composition is relatively straightforward, actual
compound identification is often difficult even for matrix
species (> 1$ by weight) and, in many cases, may be virtually
impossible for species present at trace levels. In general,
identification of trace organics can be achieved by sophis-
ticated chromatographic-mass spectroscopic combinations4 but
speciation of trace inorganics (which include many highly
toxic compounds) is still a matter for research. Consequently
most data on fly ash composition is presented in terms of
elemental abundance.
The matrix elements present in fly ash are normally Si,
Fe, Al, C, Ca, Na and K. Relative concentrations vary widely
depending primarily on the type of coal burned as shown by
-------
-739-
Bickelhaupt who also lists the probable compounds present.
It should be stressed that these compounds are almost cer-
tainly not the only ones containing so called matrix elements.
For example, substantial mass fractions of fly ash are ferro-
magnetic; an attribute which cannot be due to Fe20 . It is
also worth pointing out that fly ash is an extremely heter-
ogeneous material so that the composition of individual par-
ticles may differ dramatically from the average composition
of an integrated sample.
At the trace level (> lug/gin) almost every element in
the periodic table is found in fly ssh as shown by the mass
spectrographic analysis in Table 11°. Again, the actual
concentrations found can vary considerably (by as much as 1000
times) depending on coal type so the values in Table II are
representative only. Indeed the trace levels in this par-
ticular sample are relatively low.
DEPENDENCE OF ELEMENTAL COMPOSITION ON PARTICLE SIZE
Probably the most important single characteristic of
a particle which determines both its ability to elude con-
ventional control equipment and its atmospheric residence
time is aerodynamic particle size. Consequently it is im-
portant not only to know the chemical composition of bulk
fly ash but also to know how this composition depends on
particle size. This dependence was therefore determined
in a number of fly ash samples representing a variety of
-------
-740-
U. S. coal types. The results are presented for a single power
plant equipped with cyclonic precipitators and utilizing a
Southern Indiana coal.
Two types of sample are represented:
(a) Fly ash retained in the precipitating system.
This was collected in bulk and size differentiated in the
laboratory using a Roller particle size analyzer.
(b) Fly ash emitted to the atmosphere. This
was collected and size differentiated in situ, using an
Anderson Stack Sampler. No backup filter was used so par-
ticles less than about 0.5 Mm were not retained. The sam-
pling point was about ten feet from the base of the stack
where the temperature of the gas stream was approximately
350°F (~ 175°c).
Analyses were performed by spark source mass spectro-
metry, DC arc emission spectrometry, X-ray fluorescence spec-
trometry, atomic absorption spectrometry, differential pulse
anodic stripping voltammetry and by colorimetry using the
Weisz Ring Oven7, with the exception of carbon and sulphur,
all elements were determined by at least two distinct tech-
niques. Procedural details are given elsewhere8.
The twenty five elements determined are classified roughly
into three groups. In Table III are listed those elements
which showed convincing dependences of concentration (|ag/gm)
on particle size in all samples analyzed. Table IV contains
those elements which exhibited concentration trends in some, but
not all, samples. Table V lists those elements which showed no
-------
-741-
evidence of particle size dependences. Multiple analyses
indicated that the apparently random variations which are
superimposed on the size dependences are probably due to
poor sampling statistics. These variations have not been
removed by averaging the raw data which is presented for a
single set of size fractions analyzed by a single technique.
The sulphur concentrations are considered to be only qualita-
tive due to difficulties in obtaining a standard having a ma-
trix composition and sulphur distribution similar to that of
fly ash. It should be noted that quantitative comparison of
results obtained for fly ash retained in the plant and that
leaving is not justified since the two sample types represent
material collected over quite different integrated time periods
SURFACE DISTRIBUTION HYPOTHESIS
The results presented in Tables III-V show that many
highly toxic elements are most concentrated (on a ug/gm
basis) in the smallest particles emitted. The reasons for
this size dependence are obviously important both in terms
of environmental impact and potential counteractive control
strategy.
One attractive explanation is that certain elements,
or their compounds, are volatilized in the coal combustion
zone and then either adsorb or condense (possibly via a
nucleation process) onto the surface of entrained particles
composed of non volatilizable materials. The mass deposited
is thus greater, per unit weight, for small particles than
for large. Three pieces of evidence are presented in support
of this tentative hypothesis.
-------
-742-
1. All the elements (with the exception of Ni
and Cr) in Table III have boiling points comparable to or
below the temperature of the coal combustion zone ( 1300-
1600°C) so would be capable of volatilizing. (This is
also true for Ba, Sr, and Rb which show similar size de-
pendences in fly ash.)9 The statement implies that metal
compounds can be reduced to the element before volatiliza-
tion; however, while reduction in the combustion zone is
certainly feasible, such reduction is not necessary to the
basic hypothesis. Indeed, neither Ni nor Cr could exist
as stable elemental vapors (Table VI). It is suggested
that these elements may have access to the gas phase as
sulphides or, conceivably, as highly transient carbonyls
whose formation has been postulated"!? Mercury, of course,
is known to volatilize as the element and is predicted to
show a dependence of concentration on particle size for
that fraction associated with fly ash.
2. Consideration of a simple volatilization-
surface deposition model for a single particle containing
an element, X, present both uniformly within the particle
matrix and deposited additionally on the particle surface
leads to a relationship of the form.
cx -co + eag-p-1 D- (1)
C is the total average concentration of X (|ag/gm) in a
X
size fraction with mean particle diameter D and density p.
Cn and Cc are, respectively, the average concentration of X
u s
-------
-743-
in the particle interior and deposited on the particle sur-
face. Data from Table III for As, Ni, and Cd are plotted
according to this relationship (l) in Figure 1. It can be
seen that within the sampling and analytical errors, Equation
(1) provides sufficient description of the particle size
dependence to offer support for the proposed mechanism. Further-
more, although there are too few data points to establish firm
statistical relationships, correlation between elements is
indicated at the 95^ confidence level suggesting that the same
mechanism applies for all elements listed in Table III.
~5. If, indeed, the proposed surface deposition
hypothesis is correct, one would expect elemental concentra-
tions to be higher on the particle surface than in its in-
terior. This has been found. Fly ash particles were etched
with a stream of argon ions so as to expose the particle in-
terior and then subjected to X-ray spectrometric analysis
9
under a scanning electron microscope . Of the metals listed
in Table III only Zn, Cr and Ni were present in sufficient
concentrations for detection by this analytical technique,
however, these elements were found to be present only on
the particle exterior as predicted.
While the foregoing results do not constitute scientific
proof of the volatilization-surface deposition hypothesis they
do offer considerable supporting evidence10 . In any event,
it seems probable that the observed dependences of element con-
centration on particle size are due to a surface phenomenon.
-------
-744-
If, in fact, the volatilization-surface deposition hy-
pothesis is correct one would expect particle size dependences
to be exhibited by any species capable of being volatilized
by a high temperature combustion process and then retained
in some fashion (eg adsorption, condensation, reaction) on
a solid surface. Indeed, the process could well persist
in the atmosphere. In this connection it is noteworthy
that many species whose boiling points lie in the 100 - 400°C
range are thermodynamically capable of existing in signifi-
cant concentrations in the vapor phase at much lower tem-
peratures once volatilized. For example, thermodynamic datan
show that SeO and As 0 can exist in the vapor at 25°C up
223 ^
to concentrations as.high as 80|jg/m3 of Se and 70(ag/m3 Of As.
The effect is even more pronounced for organic species. Con-
sequently the exposure time of these vapors to particulate
matter need not be limited to the period during which a mole-
cule is at a temperature greater than its boiling point.
One might expect, therefore, to find many highly toxic
i
organic species (eg the polyaromatic hydrocarbons) present
at highest concentrations in the smallest fly ash particles.
Indeed the size dependence should be very pronounced since
^ in Equation (l) should be small due to the low boiling
points of organic compounds. No direct evidence in support
of these suggestions exists although measurements of the de-
pendence of hydrocarbon concentrations on particle size12'14
in the Los Angeles aerosol do conform to Equation (1). Since
-------
-745-
organics present in fly ash may very well be responsible
for health effects15 their possible preponderance in small
respirable particles emitted from stationary sources should
be seriously investigated.
SIGNIFICANCE OF SIZE DEPENDENCES
Preferential concentration of certain species in the
smallest emitted fly ash particles is of significance for
the following reasons:
1. Control
The results in Table III show that existing con-
trol devices, whose efficiency falls off16 below about l|om,
are least effective for collecting those particles contain-
ing the highest concentrations of undesirable species. Further-
more, existing control legislation, which specifies the
total particulate mass which may be emitted, is least stringent
for the most undesirable material. The data further indicate
that estimates of toxic emissions which are based on analyses
of fly ash retained in the plant, can be expected to be 10
to 20 times too low for many elements of interest.
2. Atmospheric Enrichment
The residence time of a particle in the atmos-
phere depends upon its effective aerodynamic size. Con-
sequently, those elements which predominate in small par-
ticles can be expected to remain in the atmosphere longer than
others which are not preferentially distributed with respect
to size. One would therefore expect marked enrichment of
-------
-746-
the elements in Table III (and possibly some of those in
Table IV) in urban aerosols when compared with crustal dusts.
1*7 1 ft
This is indeed found ' in that the elements Zn, Ni, As, Cd,
Sb, Pb, Se, S, Sn, Na, Cl and Br are enriched by at least
1000 times when normalized to Al. It should be noted that
particles derived from any combustion source (eg. cement kilns,
municipal incinerators, metal smelters and blast furnaces)
should preferentially concentrate certain elements with de-
creasing particle size. The particular elements will vary
with source depending on its operating temperature.
3. Health Effects
Probably the most important consequence of the observed
elemental size dependences in fly ash is that many toxic
elements, and probably organic species, are most concentrated
in particles which will deposit in the human respiratory
system ' . The actual region of deposition depends markedly
on particle size as illustrated in Figure 2. Furthermore,
the potential health impact of toxic species present in re-
tained particles depends upon the region of deposition21. Thus,
particles deposited in the naso-pharyngeal and tracheobronchial
regions of the respiratory tract are normally removed quite
rapidly to the pharynx, often by cilial action22 and swallowed
within a matter of hours. Consequently, extraction of toxic
species from these particles takes place predominantly in the
stomach where residence time is short. On the other hand, par-
ticles deposited in the pulmonary region may remain there for
-------
-747-
weeks or even years23 in intimate contact with approximately
J>Q m3 of alveolar membrane which separates the bloodstream
from-inhaled air^ The net result is that many species are
extracted much more efficiently in the pulmonary region than
in the stomach. For example only 5-15$ of the lead present
is extracted from particles in the stomach whereas in the
lung the corresponding efficiency24*25 is 6o-8o$0
The foregoing remarks make it clear that toxic species
present in particles which can be-deposited in the pulmonary
region of the lung will have a much greater potential for
producing adverse health effects than.if the same particles
were deposited earlier in the respiratory tract. Consequently
those species having an equivalent mass median diameter of
a micrometer or less constitute the greatest health hazard.
The influence of preferential surface deposition of toxic species
on small particles can be demonstrated quantitatively if the
total particle mass emitted from a power plant, for example, is
assumed to be log-normally distributed with respect to particle
size. (This assumption is reasonable in many cases). Thus,
the mass, M, in a given size fraction of mean particle diameter,
D, is given by the normalized expression
dM i
d(lnD) /2rif lnar
exp
1/2
(2)
g
where ag is the geometric standard deviation of the size dis-
tribution and D is its equivalent mass median diameter.
If all chemical species were equally distributed in con-
centration (in jag/gm) and showed no particle size dependence then
-------
-748-
all would have the same mass median diameter, D . However,
O
if some species are subject to surface deposition according
to Equation (1) then the mass distribution of a species, X,
is obtained by combining Equations (1) and (2) to give21
dMY _ J- I 5 . exp I -1/2 -
d(lnD) /2^ ina- (. L (lno_)
[-
L
g,
r "h
r, /« , 0 1 (lnD/D_ + In^an-)2 /
[1/2 In2agj- exp -- - - — & - -&: — V
L 2lna2 J
,
+ Cs exp
2(lnag)2
Here Dg and ag refer to the distribution of the substrate
particles as in Equation (2) and C and Cc are as defined
0 b
for Equation (2). Equation (3) is not normalized.
Extraction of an explicit expression for the mass median
diameter of X, Dg(x) from Equation (5) is tedious, however,
in the case where GO « GS ( ie most of X is present on the
surface as expected for an organic species) it can readily
be shown that26
In D(x) = In D - In2a
g g g
This equation (3) demonstrates the profound effect of sur-
face deposition of a species X in reducing its effective
mass median diameter even though that of the substrate par-
ticles is unaltered. For example21 if the total mass dis-
tribution emitted has Dg =2.7 (am and cg = 2.9 jjjn the data for
Zn in Table III, when incorporated into Equation (3), show
that Zn has a mass median diameter of approximately 1 (am.
Similarly, an organic species distributed such that CQ « Cg
would have a mass median diameter < 0.1 urn.
-------
-749-
POSSIBLE PROTOCOL FOR DETERMINING PARTICULATE TOXICITY
Just because a toxic species may be present, in particles
emitted to the atmosphere does not necessarily imply that
this species will have an adverse health effect when inhaled,
Before this can be established the following factors must be
considered.
(a) The mass distribution of the species, X, emitted,
(b) The contribution of this distribution to that
inhaledo (From the standpoint of environmental health the
inhalable distribution is an essentially stable urban aerosol.
Occupational health considerations will involve a more localized
distribution),
(c) The particle size deposition profile in the
respiratory tract„
(d) The efficiency of transport of X from particle
to target organ or molecule as a function of particle size. This
factor can be subdivided in terms of
(i) Particle clearance rate
(ii) Extraction rate of X
(ill) Elimination rate of X
(iv) Rate of transport of X to target
(e) The effective toxicity of X at a target site.
At first sight quantitative evaluation of (a) through
(e) seems a formidable and barely worthwhile task. However,
such a protocol can be greatly simplified by recognizing
that the data required for (d) and (e) can be determined directly
by bioassay in which the appropriate particle size distribution
-------
-750-
of all X is presented for inhalation. (Determination of
the toxicity of fly ash to bacteria or tissue cultures has
little merit in this context). (a) can be determined ex-
perimentally and recent work27 has shown that step (b) can
be mathematically modelled surprisingly well for contribu-
tions to an urban aerosol.
As an example of the utility of this type of approach
consider the case of an urban aerosol for which Dg andog
have been experimentally determined for the total mass, and
O Q • O Q i
for Pe, Zn, Pb, and the carcinogen Benzo-a-pyrene. (Table VII)
Assuming a log-normal distribution of these species (a bi-
modal distribution with two log-normal components is more
realistic) the deposition efficiency, E, in Figure 2 can
be incorporated into Equation 2 such that
dM
d(log D)
= 2-303 E
deposited
dM
d(ln D)
inhaled (5)
Equation (5), with appropriate experimental parameters from
Table VII, thus describes the retention of inhaled species
in each respiratory region and has been used to generate
Figures 3> ^, 56 and 7 which show the retention of total aerosol
mass, Fe, Zn, Pb and Benzo-a-pyrene. The 'fractions of inhaled
mass retained in each region are presented in Table VIII.
It is possible to go one stage further and include (d)
(i), (ii) and (ill) in the case of lead whose efficiency of
extraction into the blood stream is 60-80$ in the pulmonary
region and 5-15$ in the stomach. ' Assuming mean values of
-------
-751-
70$ and 10$ the data in Table VIII show that 22$ of the
total inhaled lead enters the bloodstream by absorption
through pulmonary membranes while only 2.3$ enters from
the stomach. Based on these figures an average adult in-
haling 20 m3 of air per day containing 2|ag/m3 Of ]_ead (a
typical urban aerosol loading) would absorb 10 |_ig of lead
per day. Absorption from food and water amounts to about
30 ug per day?3'3°Recent Pb isotope labelling studies31 in-
dicate that approximately 30$ of the daily lead intake
comes from inhaled aerosols; a figure wnich is considered to
be in excellent agreement with that estimated using the
partial protocol.
Obviously considerable information is required before
such a protocol could be used to assess the effective toxicity
of a given source emission. However, it is clear that this
procedure can easily include the collective influence of
a variety of both known and unknown toxic species and their
synergisms. Insofar as it provides at least a semiquantitative
basis for establishing the environmental health significance
of chemical species present inurban aerosols and emission
sources its importance in terms of source -control strategy
is clear.
CHEMISTRY AND CONTROL PROCESSES
The foregoing sections have presented information re-
lating to the chemical composition of fly ash, the possible
processes which establish trace element distribution, and the
-------
-752-
significance of this distribution in terms of control, en-
vironmental perturbation, and health. They have illustrated
the suggestion that more emphasis should be placed on con-
trolling undesirable particles and less on controlling par-
ticles per se. However a knowledge of the chemistry of par-
ticles and of particle production can assist not only in
delineating the least desirable particle fractions but also
in indicating possible new methodology for control. The follow-
ing examples illustrate how such a principle might be realized
in practice. It should be strongly stressed that the examples
are primarily illustrative and do not, at this time, con-
stitute practical propositions.
1. Preliminary studies have shown that substan-
tial fractions of most coal flyashes- are ferromagnetic and
that even higher proportions of many toxic species are associ-
ated with this magnetic fraction. In view of the recent
advances made in magnetic collection the possibility of em-
ploying magnetic collection for preferential collection of
toxic particulates is viable?4 Before such a process could
be considered, however, the relationships of toxic species
to the ferromagnetic fraction must be established.
2. If many toxic species are, indeed, preferentially
concentrated in small particles by surface deposition it may
be possible to provide an alternative surface for deposition.
For example entrainment of activated carbon particles and fly
ash through cadmium vapor produces highly preferential associa-
tion of Cd with the large, easily collected, carbon particles32,
-------
-753-
Carbon may have an additional advantage in that recent
o o
work has indicated it to be extremely active in catalyzing
the formation of solid sulphate from gaseous SO . Certainly
2
adsorption of volatile organics is entirely feasible. Al-
ternatively, injection of magnetic Fe304 particles may pro-
vide a suitable surface for preferential deposition. In
essence such a procedure would take advantage of the volatili-
zation—deposition of toxic species to increase their effec-
tive mass median diameter to a range which can be more efficiently
collected by existing, or future, control equipment.
~5. A similar preconditioning process can be envisaged
for increasing the resistivity of certain fly ashes to a
level at which more efficient electrostatic collection can
be achieved. In principle, at least, modification of com-
bustion chemistry or subsequent chemical conditioning could
be employed to reduce the surface concentrations of Na and
Li which appear to be responsible for electrical conduction
in fly ash.
PROTOCOL FOR CHEMICAL CHARACTERIZATION
In none of the examples cited above is sufficient basic
information available to assess even their potential feasi-
bility. This amply demonstrates the need for a deeper chemical
understanding of fly ash and its production. Although a number
of independent studies have been, and are being, conducted
the tremendous variability of fly ashes makes it very difficult
to correlate results. It is considered vitally important, there-
fore, that a coherent investigational protocol be developed for
-------
-754-
the chemical and physical characterization of a single fly
ash sample. Having established parameters of major interest
these can then be effectively and efficiently investigated in
other samples .
Such a protocol, which is being followed in our laboratory,
is presented in the form of an experimental matrix, in Table
IX. The fly ash is physically separated into fractions on
the basis of aerodynamic particle size, ferromagnetism, particle
density and solubility all of which differentiate material
having different chemical composition, physical properties and
practical significance. Each fraction is then processed to
obtain, where possible, the parameters listed. Completion
of such a matrix is clearly a major undertaking. However it
is considered that this type of approach provides the most
efficient and coherent method for obtaining basic data re-
lating to existing and potential control methods for tne
most environmentally significant fractions of coal fly ash.
CONCLUSION
Particulate control processes have improved considerably
over the last few decades to the stage where further im-
provement necessitates consideration of more and more subtle
parameters. On the one hand it is appropriate to assess the
justifications for more stringent control very seriously; on
the other it is necessary to seek ways of improving collection,
of reducing costs, and of accomodating to new processes and fuels
resulting from a changing energy production profile. In all cases
practical considerations require a much better understanding of
-------
-755-
of the chemistry of particulates from their production
to their eventual removal to an environmental sink.
-------
-756-
REFERENCES
1. W.R. Grouse, G.D. Aase, H.C. Johnson, J.D. Coons and
J.E. Yocom, J. Air Pollution Control Assoc. _10, 285 (I960).
2. R.E. Lee and S. Goranson, Environ. Sci. Technol., 6>, 1019
(1972).
3. W. J. Moroz and V. Withstandly, paper presented to 66th
Annual A.I.Ch.E. Meeting, Philadelphia, November 1973.
4. R.C. Las, R.S. Thomas, H. Oja and L. Dubois, Anal. Chem.,
15, 909 (1973).
5. R.E. Bickelhaupt, paper presented to US-USSR Working
Group Symposium on Control of Fine Particulate Emissions
from Industrial Sources, San Francisco, January 1974.
(Included in these Proceedings)
6. R. Brown, M.L. Jacobs and H.E. Taylor, American Laboratory,
29, (November 1972).
7. H.H. Weisz, "Microanalysis by the. Ring Oven Technique",
Pergamon Press, N.Y., 1961.
8. R.L. Davison, D.F.S. Natusch, J.R. Wallace and C.A. Evans,
Environ. Sci. Technol. (in press).
9. Sparks, Oak Ridge National Laboratory (personal communication).
10. D.F.S. Natusch, J.R. Wallace and C.A. Evans, SCIENCE, 183,
202 ( 1974).
11. Handbook of Chemistry and Physics, 44th ed., The Chemical
Rubber Publishing Co., Cleveland, Ohio pp. 2430-2445 (1963).
12. P.K. Mueller, R.W. Mosely, L.B. Price, J. Colloid Interface
Sci., 32, 235 (1972).
13. S.L. Heisler. S.K. Friedlander, R.B. Husar, Atmos. Environ.,
I, 633 (1973).
14. R.B. Husar, and K.T. Whitky, Environ. Sci. Technol. J,
241 (1973).
15. H.R. Menck, J.T. Casagrande and B.E. Henderson, SCIENCE,
183, 210 ( 197^).
-------
-757-
16. G.B. Nichols, paper presented to US-USSR Working Group
Symposium on Control of Fine Particulate Emissions from
Industrial Sources, San Francisco, January 1974 (included
in these Proceedings)
17. G.E. Gordon and W.H. Zoller, paper presented to Oak Ridge
Trace Elements Conference, Oak Ridge, Tennessee, August 1973.
18. K.A. Rahn, M. Demuynck, R. Dams and J. Degraeve, Proc.
Third International Clean Air Congress, Dusseldorf,
Germany, 1973, C8l.
19. T.F. Hatch and P. Gross, "Pulmonary Deposition and Retention
of Inhaled Aerosols," Academic Press, N.Y., 1964.
20. U.S. Dept. of H.E.W., National Air Pollution Control Board,
Pub. No. A849, "Air Quality Criteria x'or Particulate Matter,"
Washington D.C., January 1969.
21. D.F.S. Natusch and J.R. Wallace, SCIENCE (in press).
22. H.A. Schroeder, Environment, 13, 18 (1971).
23. P.E. Morrow, Ind. Hyg. Assoc. J., 25, 213 (19^5).
24. L. Dautreband, "Microaerosols," Academic Press, N.Y.,
25. C.C. Patterson, Arch. Environ. Health, _11, 344 (1965).
26. S.S. Butcher and R.J. Charlson, "An Introduction to Air
Chemistry," Academic Press, N.Y. 1972 (Chapter 9).
27. S.K. Friedlander, Proc. Third International Clean Air
Congress, Dusseldorf, Germany, 1973, 078.
28. R.E. Lee, SCIENCE, 178, 567 (1972).
29. M. Kertesz-Saringer, E. Meszaros and T. Varkonyi, Atmos.
Environ., _5, 429 (1971).
30. G. Nagelschmidt, E.S. Nelson, E.J. King, D. Atygalle
and M. Yoganathan, A.M.A. Arch. Ind. Health, lo, 188
(1957).
31. G. Wetherill, M. Rabinowitz and J. Kopple, SCIENCE (in press)
32. D.F.S. Natusch, Unpublished results.
33. T. Novakov, Lawrence Berkeley Laboratory, personal com-
cunication.
34. Chemical and Engineering News, Jan. 28th (1974) page 21.
-------
-758-
ACKNOWLEDGEMENTS
The assistance of Dr. C. A. Evans, Materials Research
Laboratory, University of Illinois, Mr. J. Kuhn, Illinois
State Geological Survey, and Mr. R. Fry, Environmental
Analytical Laboratory, University of Illinois, in performing
some of the analyses reported herein is gratefully acknow-
ledged. Stimulating discussions with Professor G.E. Gordon,
Dr. W. Pulkerson, Dr. T. Novakov, Dr. C.A. Evans and with
the author's research group are also acknowledged.
The original work reported was supported in part by
N.S.F. grants GI 31605 and GH 33634.
-------
TABLE 1
NATIONWIDE ESTIMATES OF PARTICULATE EMISSIONS 1940 - 1970
(106 tons/year)
Sourcs Category
Fuel combustion in stationary
sources
Transportation
Solid waste disposal
Industrial process losses
Agricultural burning
Miscellaneous
Total
Total controllable*
1940
9.6
0.4
0.4
8.8
1.6
6.4
27.1
20.7
1950
9.0.
0.4
0.6
10.8
1.8
3.3
25.9
22.6
1960
7.6
0.5
1.0
11.9
2.1
2.1
25.3
23.2
1968
6.5
0.8
1.4
13.8
2.4
1.7
26.6
24.9
1969
6.4
0.7
1.4
14.3
2.4
2.1
27.3
2S.2
1970
6.8
0.7
1.4
13.3
2.4
1.0
25.6
24.6
I
-~J
en
V£>
I
Miscellaneous sources not included
Reference: Nationwide Inventory of Air Pollutant Emission Trends 1940 - 1970 U.S. E.P.A.
-------
TABLE II
Typical trace element analysis of fly ash, concentration, in ug/gm.
Element
Thorium
Uranium
Bismuth
Lead
Thallium
Mercury
Tungsten
Tantalum
Hafnium
Lutetium
Ytterbium
Thulium
Erbium
Holmium
Dysprosium
Terbium
Gadolinium
Europium
Samarium
Neodymium
Praseodymium
Cerium
Cone.
' 40
30
0.50
60
0.8
<0.01
6.0
0.50
9.0
0.50
5.5
3.0
3.4
1.2
7.0
2.3
13
3.0
11
55
38
100
Element
Lanthanum
Barium
Cesium
Iodine
Tellurium
Antimony
Tin
Cadmium
Silver
Molybdenum
Niobium
Zirconium
Yttrium
Strontium
Rudidium
Bromine
Selenium
Arsenic
Germanium
Gallium
Zinc
Cone.
40
2400
20
0.20
0.20
4.5
60
0.70
0.50
36
60
300
100
2000
140
0.70
0.30
22
2.4
20
260
Element
Copper
Nickel
Cobalt
Iron
Mangane s e
Chromium
Vanadium
Titanium
Scandium
Calcium
Potassium
Chlorine
Sulfur
Phosphorus
Silicon
Aluminum
Magnesium
Sodium
Fluorine
Boron
Beryllium
Lithium
Cone.
200
50
7.0
Major
400
110
50
1600
3.7
Major
7000
420
Major
8000
Major
Major
Major
Major
150
230
8.0
42
o
I
-------
-761-
TABLE III
Elements showing pronounced Concentration trends
Pb
Tl
Sb Cd
Se As
Ni
Cr
Zn
Particle
ditiTflcter — "— ^~™~~— ~
(lira)
UR/R
A. Fly ash
retained in
the
Sieved fractions
>74
44-74
140
160
7
9
1.5 <10
7 <10
<12 180
<20 500
100
140
100
90
plant
500
411
S Mass
frsction
(wt%) (%)
66.30
1.3 22.89
Aerodynamically sized fractions
>40
30-40
20-30
15-20
10-15
5-10
<5
Analytical
90
300
430
520
430
820
980
method
b
5
5
9
12
15
20
45
b
8 <10
9 <10
8 <10
19 <10
12 <10
25 <10
31 <10
b b
<15 120
<15 160
<15 200
<30 300
<30 400
<50 800
<50 370
b b
300
130
160
200
210
230
260
d
B. Airborne fly
11.3
7.3-11.3
4.7-7.3
3.3-4.7
2.1-3.3
1.1-2.1
0.65-1.1
1100
1200
1500
1550
1500
1600
29
40
62
67
65
76
17 13
27 15
34 18
34 22
37 26
53 35
_—
13 680
11 800
16 1000
16 900
19 1200
59 1700
460
400
440
540
900
1600
70
140
150
170
170
160
130
d
ash
740
290
460
470
1500
3300
—
730
570
480
720
770
1100
1400
b
8100
9000
6600
3800
15000
13000
—
<0.01 2.50
0.01 3.54
3.25
0.80
4.4 0.31
7.8 0.33
.08
c
8.3
7.9
25.0
48.8
Analytical method
abb a
(a) Atomic Absorption Spectrometry
(b) Spark Source Mass Spectrometry
(c) X-ray Fluorescence Spectrometry
(d) DC Arc Emission Spectrometry
-------
-762-
TABLE IV
Elements showing limited concentration trends
Particle Fe
diameter (wt %)
Sieved fractions
>74
44-74 18
Mn
(ug/g)
A.
700
600
Si Mg
(wt %) (wt %)
Precipated at base of
__
18 .39
C
(wt %)
stack
—
—
Be
(Pg/g)
12
12
Al
(wt X)
~
9.4
Aerodynamically sized fractions
>40 50
30-40 18
20-30
15-20
10-15 6.6
5-10 8.6
<5
Analytical method c
150
630
270
210
160
210
180
d
3.0 .02
14 .31
__
—
19 .16
26 .39
__
c c
.12
.21
.63
2.5
6.6
5.5
~
e
7.5
18
21
22
22
24
24
d
1.3
6.9
—
~
9.8
13
~
c
B. Airborne material
>11.3 13
7.3-11.3
4.7-7.3 12
3.3-4.7
2.06-3.3 17
1.06-2.06 —
.65-1.06 15
Analytical method c
150
210
230
200
240
470
—
d
34 .89
—
27 .95
_
35 1.4
_
23 .19
c c
.66
.70
.62
.57
.81
.61
—
e
34
40
32
55
43
60
—
d
19.7
—
16.2
—
21.0
--
9.8
c
(a) Atomic Absorption
(b) Spark Source Mass Spectrometry
(c) X-ray Fluorescence Spectrometry
(d) DC Arc Emission Spectrometry
(e) Oxygen Fusion
-------
-763-
TABLE V
Showing no concentration trends
Particle Bi
diameter (yg/g)
Sieved fractions
>74 >2
44-74 >2
Aerodynamically sized
>40 >2
30-40 >2
20-30 >2
15-20 >2
10-15 >2
5-10 >2
>5 >2
Sn
(Mg/g)
A.
>2
>2
>2
>2
>2
>2
>2
>2
>2
Cu
(ug/g)
Fly ash
120
260
220
120
160
220
220
390
490
Co
(pg/g)
retained in
28
27
75
76
55
50
55
46
54
V
(ug/g)
plant
150
260
250
190
340
320
320
330
320
Ti Ca K
(wt 7,) (wt %) (wt 3D
—
.61 5.4 1.2
.01 2.5 2.54
.64 6.3 6.26
—
4.5 4.46
.66 4.0 4.04
1.09
_.
B. Airborne material
>11.3
7.3-11.3 >3.5
4.7-7.3 >4.0
3.3-4.7 >4.8
2.06-3.3 >4.5
1.06-2.06 >4.4
.65-1.06
Analytical method b
7
11
18
19
16
18
—
b
270
390
380
—
330
300
—
b
60
85
90
95
90
130
—
d
150
240
420
230
310
480
—
b
1.12 4.9 4.9
—
.92 4.2 4.2
—
1.59 5.0 5.0
_
1.08 2.6 2.6
c c c
(b) Spark Source Mass Spectrometry
(c) X-ray Fluorescence Spectrometry
(d) DC Arc Emission Spectrometry
-------
-764-
TABLE VI
Boiling points of possible inorganic species
evolved during coal combustion
Species boiling or Species boiling or
subliming _< _1550*C subliming > 1550°C
As, As205, As203, Aa2S3 Al, Al^
Ba Be, BeO
Bi Bi203
Ca C
Cd, CdO, CdS CaO
Cr(CO)6, CrCl3, CrS(1550) Co, CoO, CoS
K Cr, Cr00,
*« O
M8 Cu, CuO
N* Fe^, FeO
Pb MgO, MgS
Rb Mn, MnO,
s Ni, NiO
Se, Se02, Se03 Si,
Sb, Sb2S3, Sb203(1550) Sn,
SnS Ti, Ti02, TiO
Sr U, U02
Zn, ZnS
Tl, T120, T1203
-------
TABLE VII
Typical Size Parameters of Urban Aerosol Components
Mass
Median
Species
Total mass
Pe
Pb
Zn
Ba
Noncarbonate
Carbon
Benzo-a-pyrene
so/
Diameter (|om)
.6
2.7
.56
1.05
1.95
.6
.15
.45
Deviation
7.0
2.9
4.1
2.06
5.54
-
-
7.8
Reference
28
28
28
28
28
12
29
21
Comment
Year Average for 6
Eastern U.S. Cities
11 11
11 ii
Year Average for Chicago
Denver Quarterly Average
Los Angeles Photo-Chemical
Smog, 90$ < .6 nm
Budapest
Week Average for Cincinnati
i
-------
TABLE VIII
Species
Total mass
Fe
Pb
Benzo-a-pyrene
Percent Deposition of Inhaled Aerosols.
Size distribution parameters are listed
in Table I.
Nasopharyngeal
%
23
48
17
5
Tracheobronchial
%
6
7
6
7
Pulmonary
%
30
22
32
39
i
-j
CTi
O>
I
-------
-767-
Solubility
Separation
Density
Separation
Magnetic
Separation
Aerodynamic
Sizes
2
H
FV
w
n>
TJT3
1-i p
O 4
0 (B
CD C+
p, H-
l± O
4 D
0)
Particle Density
Particle Mosphology
Particle Size Distribution
B.E.T. Surface Area
Resistivity
Magnetic Susceptibility
Matrix Elements
Matrix Compounds
Trace Elements
Trace Compounds
Volatilizable Organics
Extractable Organics
Reversible Adsorption
Irreversible Adsorption
Surface Characterization
D.T.A.
Anions
Natural and Biological Extrac
tion
TABLE IX
Investigational Matrix
for particles
-------
-768-
Figure Captions
Figure 1. Dependence of the average concentrations of As,
Ni and Cd on airborne particle size in coal fly
ash.
Figure 2. Respiratory deposition profiles for inhaled
20
particles .
Figure 3. Respiratory deposition efficiency of particle mass
a typical urban aerosol.
Figure 4. Respiratory deposition efficiency of iron in an
urban aerosol.
Figure 5. Respiratory deposition efficiency of zinc in an
urban aerosol.
Figure 6. Respiratory deposition efficiency of lead in an
urban aerosol.
Figure 7. Respiratory deposition efficiency of benzo-a-
pyrene in an urban aerosol. (A typical value of
a = 3.0 was assumed since no experimental values
were available).
-------
-769-
1500 -
o
M—
O
en
"D
O
O
CD
O
O
O
1000 -
500
a
15
^
a>
Jt
-o
O
O
CD
O
O
0.25 0.50
(Particle Diameter)' (microns)'
Figure 1.
-------
-770-
Q
UJ
(7)
O
QL
LJ
Q
O
§
NASOPHARYNGEAL-,
TRACHEO-
BRONCHIAL
0
icr
10
MASS MEDIAN DIAMETER (p.)
Figure 2.
-------
-771-
0.25
0.20
S 015
_o
TD
010
0.05
Pulmonary
Nasopharyngeal
Tracheo-
bronchial
I
0.01
0.10 10 10
Particle Diameter (p.)
100
Figure 3,
-------
-772-
0.60
Q50
0.40
0.30
0.20
QJJO
Nasopharyngeal
Tracheo-
bronchial
Pulmonary
0.01
0.10 10 10
Particle Diameter
100
Figure 4.
-------
-773-
0.30
0.20
Q
§
"O
010
0.0
Nasopharyngeal
Pulmonary
Tracheo-
bronchial
0.01
0.10 1.0 10.0
Particle Diameter (/JL)
100.0
Figure 5,
-------
0.25
-774-
Pulmonary
0.20
T3
a>
S 0.15
_o
•O
0.10
0.05
0.01
Nasopharyngeal
V.
Tracheo-
bronchial
0.10 1.0 10
Particle Diameter (/JL)
JOO
Figure 6.
-------
-775-
Tracheobronchial
\
Nasopharyngeal
0.01
o.io 1.0
Particle Diameter
10.0
100.0
Figure 7.
-------