EPA-650/2-73-035
October 1973
Environmental Protection Technology Series
-------
EPA-650/2-73-035
FIELD MEASUREMENTS
OF PARTICLE SIZE DISTRIBUTION
WITH INERTIAL SIZING DEVICES
by
0. D. McCain,
K. M. Gushing, and A. N. Bird, Jr.
Southern Research Institute
2000 Ninth Avenue, South
Birmingham, Alabama 35205
Contract No. 68-02-0273
Program Element No. 1AB012
ROAP No. 21ADM11
EPA Project Officer: D. B. Harris
Control Systems Laboratory
National Environmental Research Center
Research Triangle Park, North Carolina 27711
Prepared for
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON, D.C. 20460
October 1973
-------
This report has been reviewed by the Environmental Protection Agency and
approved for publication. Approval does not signify that the contents
necessarily reflect the views and policies of the Agency, nor does
mention of trade names or commercial products constitute endorsement
or recommendation for use.
ii
-------
TABLE OF CONTENTS
Section Page Number
I INTRODUCTION 1
II BACKGROUND 1
III THE TEST CONDITIONS, PROCEDURES, AND
EQUIPMENT 2
General Description 2
Test Procedure 5
IV TEST RESULTS 7
The McCrone Parallel Cyclone Inertial
Sizing Device 7
Battelle CIS-6 Cascade Impactor 11
TAG Multiple Split Cascade Impactor 13
University of Washington Mark III Source
Test Cascade Impactor 13
Andersen Stak Sampler - Model II 19
Andersen Model IV 21
Andersen Stak Sampler - Model III 21
Brink Impactor 25
Summary of all Impactor Tests 37
Fractional Efficiency of the Electro-
static Precipitator 38
V CONCLUSIONS AND COMMENTS 38
REFERENCES 46
iii
-------
FIGURES
No.
1 Comparison of Fractionation Sizes (D$Q) for Seven 4
Inertial Separators
2 Representative Cumulative Size Distributions Obtained 9
with the McCrone Parallel Cyclone Sampler
3 Differential Mass Distributions Obtained from the 10
Cumulative Distributions Shown in Figure 2
4 Differential Inlet Mass Distribution Obtained with the 12
Battelle CIS-6 Impactor
5 Differential Inlet Mass Distribution Obtained with the ERC 14
TAG Sampler
6 Differential Outlet Mass Distribution at 94% Load as 15
Obtained with the ERC TAG Sampler
7 Differential Outlet Mass Distributions Obtained with the 16
ERC TAG Sampler at Loads of 85% and 100%
8 Differential Inlet Mass Distributions Obtained with the 17
U. W. Mark HI Impactor with Ungreased Substrates
9 Differential Inlet Mass Distributions Obtained with the 18
U. W. Mark HI Impactor with Greased Substrates
10 Outlet Differential Mass Distributions Obtained with the 20
U. W. Mark HI Impactor at 94% Load
11 Differential Inlet Mass Distributions Obtained with the 22
Andersen Model II Impactor at 100% Load
12 Outlet Differential Mass Distribution Obtained with the 23
Andersen Model II Impactor at 80% Load
13 Differential Inlet Mass Distributions as Obtained with 24
the Andersen Model IV Impactor
14 Differential Inlet Mass Distributions Obtained with the 26
Andersen Models n and HI Impactors
15 Differential Inlet Mass Distributions Obtained with the 27
Andersen Model HI Impactor Preceded by a Prototype
Scalping Cyclone
16 Differential Outlet Distribution at 60% Load Obtained with 28
the Andersen Model IE Impactor
17 Differential Inlet Mass Distribution Obtained with Brink 30
Impactor Modified for In-Stack Sampling Preceded by
Cyclone C2, and Corresponding Data Obtained Using
Brink BMS 11 Kit
18 Differential Inlet Mass Distributions Obtained Using Brink 31
Impactors Modified for Ins tack Sampling Preceded by
Cyclones Cl and C2
19 Differential Inlet Mass Distributions Obtained with Brink 32
Impactors Modified for Instack Sampling Preceded by
an Additional Stage, SO, and Cyclone C3
iv
-------
Figures (Continued)
No.
20 Comparison of Averaged Differential Inlet Mass 33
Distributions Obtained with Brink Impactors Using
Greased and Ungreased Impactor Substrates
21 Differential Outlet Mass Distribution at Full Load 34
Obtained Using Brink Impactors Modified for In-Stack
Sampling Using Greased and Ungreased Substrates
22 Differential Outlet Mass Distributions at 94% Load Obtained 35
Using Brink BMS 11 Kits and Brink Impactors Modified
for In-Stack Sampling
23 Differential Outlet Mass Distribution at 55% Load Obtained 36
Using Brink Impactors Modified for Instack Sampling
24 Precipitator Efficiency at 94% and 100% Load as Measured 39
with Modified Brink, Brink BMS 11 Kit. Andersen Model
m, and U.W. Mark TO. Impactors
25 Precipitator Efficiency at 55% of Full Load as Measured 40
with Modified Brink and Andersen HI Impactors
26A Fly Ash Deposits on Ungreased Impaction Substrates 42
Without NH3 Injection Into Flue Gas
26B Fly Ash Deposits on Ungreased Impaction Substrates 42
WithNH3 Injection
27 Inlet Size Distribution as Measured with Three Out- 44
of-Stack Samplers
28 Inlet Size Distribution as Measured with Four In-Stack 45
Samplers
-------
I. INTRODUCTION
The purpose of this research contract is to devise and
evaluate various techniques for field measurements of the
fractional efficiency of particulate control devices. The
primary particle size range of interest is from 0.2 to 2.0 ym
diameter but techniques for measuring larger particles are also
being evaluated. Inertial classification is the basic sizing
technique that is being considered under this contract. Most
of this report is devoted to a description of a comprehensive
particle size measurement program that was conducted at a coal-
fired electric generating station earlier this year. The report
also includes some information that was obtained from other field
tests and from laboratory work so that it provides a summary of
the most significant work during the first funding period of
the contract.
II. BACKGROUND
Because of the current interest in the control of small
particles, the emphasis in this work is on the development of
measuring techniques suitable for the 0.2 to 2.0 vim* particle
size range. Under the proper operating conditions, inertial
sizing devices such as cascade impactors and multiple cyclones
can be used to measure the size distribution of particles in
the general size range of interest for air pollution work. Research
*A11 references to particle size in this report are given as
particle diameter in ym (microns),
-------
-2-
is in progress by others1'2 to develop cascade impactors that
will measure the particle size distribution down to about 0.02 pm.
A minimum particle size of 0.3 ym is the approximate lower limit
for all existing impactors that are suitable for field work. Most
of our work has been devoted to improving field measurement
techniques with cascade impactors.
Inertial sizing devices do not give real-time particle
size information and particle size errors can result from reentrain-
ment and other factors; however, inertial sizing devices have
many features that make them worthy of consideration; these devices
are useful over a wide range of particle sizes and concentrations;
the aerodynamic size distribution obtained with cascade impactors
is useful for the evaluation of most control devices; if needed,
physical size can be computed from average particle density and
shape information; the measured sizes are not affected by the
optical or electrical properties of the particles; and, finally
cascade impactors frequently can be used to provide mass
concentration data, "grain loading", simultaneously with particle
size data.
The work has been a combination of laboratory and field
work with emphasis on field measurements made on a variety of
stationary emission sources.3 As a result of work on this project
and several other recent and current Institute projects, which
had a requirement for particle size determination, measurements
have been made on the following emission sources: a kaoline kiln
with a pilot-scale, gravel bed filter; two sulfate pulp mills
with direct contract evaporators both equipped with a combination
venturi scrubber and cyclone; a sulfate pump mill without direct
contact evaporation equipped with an electrostatic precipitator;
and eight coal-fired electric generating stations all equipped
with electrostatic precipitators.
Most of this report is devoted to a summary of a comprehensive
field test conducted over a period of about three weeks at a
coal-fired electric generating station.
III. THE TEST CONDITIONS, PROCEDURES, AND EQUIPMENT
General Description
Between January 10 and January 24, 1973, particle size
measurements were made at the inlet and outlet of an electro-
static precipitator that was being used on a coal-fired electric
generator rated at 68 megawatts full-load output. A mechanical
collector was in use upstream of the precipitator. The efficiency
of the mechanical collector was approximately 75% so that the
-------
-3-
grain loading in the large particle size range was reduced
and cyclone precollectors were not needed ahead of the impactors.
Five sampling ports were installed in both the inlet and
outlet ducts of the precipitator. These ports were 0.15 m (6
inches) in diameter.
Particle size measurements were made with inertial classifiers
built by six different manufacturers. Including modifications
and prototype designs that resulted from work done on this project,
a total of eleven (11) different sizing devices were evaluated.
These devices are listed below and the stage size cut points,
D50*, for some of the devices are shown in Figure 1.
1. A Battelle CIS-6 impactor which had been modified
by Battelle to include an additional stage with a
D50 of about 0.25 ym. (Battelle)t
2. A parallel cyclone sizing device designed and built
by McCrone Associates. (McCrone)t
3. A University of Washington Mark III Source Test
Cascade Impactor manufactured by Pollution Control
Systems, Inc. (U. W. Mark III)r
4. A conventional Brink BMS-11 sampling kit manufactured
by Monsanto Environchem. (Brink BMS-11)t
5. & 6. Two Brink impactors that have been modified to provide
size information for particles larger than those
caught in the stage of the commercially available
device. These modifications include an additional
stage and cyclone precollectors. (Brink)t
7. A TAG sampler (multiple slit cascade impactor)
manufactured by Environmental Research Corporation.
(TAG)t
8. An Andersen Stak Sampler (Andersen Model II)t
9. A modified Andersen Stak Sampler with glass fiber filter
substrates (Andersen Model III)t
10. A second modified Andersen Stak Sampler with glass
fiber filter substrates and cyclone precollector.
(Andersen Ill-cyclone)t
*D,0 is the particle size (aerodynamic diameter in ym) collected
with 50% efficiency by the inpactor stage.
-------
-4-
ANDERSEN STACK SAMPLER
(14 LPM)
BATTELLE CI-S-6
(12,5 LPM)
BRINK MODEL B (DASHED
LINES ARE CYCLONES AND
AN ADDITIONAL STAGE)
(2,8 LPM)
CASELLA (17 LPM)
McCRONE PARALLEL
CYCLONE (77 LPM)
UNIV, OF MASH, MARK III
(11 LPM)
E.R.C, TAG SAMPLER
LPM)
I I I I I I I
I I I I I II
I I I I I I II
I I I I I I II
.2 ,4 ,6 ,8 1 1,5 2 1 68 10 15 20
AERODYNAMIC FRACTIONATION SIZE (MICROMETERS)
Figure 1. Comparison of Fractionation Sizes (D50)
for Seven Inertial Separators. (The Sizes
Indicated are the 50 Percent Stage Penetra-
tion Points [at Typical Flow Rates] for each
Stage of Fractionation.)
-------
-5-
11. An instack version of the Andersen Non-Viable Sampler
using glass fiber filter substrates. (Andersen
Model IV)t
Test Procedure
Upon arrival at the steam plant, a laboratory was set up
in the plant results laboratory. Two balances were used in this
test. A Cahn Electrobalance was used for weighing the filter and
foil substrates, as well as the other catches of less than about
30 mg. The Cahn weighing accuracy, as used for these tests, was
about ±0.02 mg. For the heavier cyclone catches, a Ventron pan
balance with a sensitivity of about 0.1 mg was used. Due to
vibrations from the turbines, generators, motors, etc., the pan
balance was moved out of the steam plant to one of our motel rooms.
The Cahn balance is insensitive to vibration and was used in the
plant laboratory.
To reduce the tare weight of the impactor stages, glass fiber
filter or aluminum foil collection substrates were used for many
of the impactor tests. The substrates were cut, disiccated and
weighed prior to their placement in the appropriate device. The
Andersen Model II and McCrone were the only devices which were not
tested with the substrates on at least some of the tests.
After preparation of the impactor, an appropriate nozzle and
sampling probe were chosen. The inlet and outlet ducts to the
precipitator were approximately 2.5 m (8 ft) in depth with the air
flow in a horizontal direction. When sampling with the instack
samplers—Andersen, Brink,U. W. Mark III, and TAG probes were
used that placed the inlet to the impactor about 1.5 m (5 ft) from
the top of the duct. The McCrone, Battelle, and Brink BMS 11
sampled through conventional probes and nozzles. These probes
were approximately 1 m (3 ft) long. The Battelle was operated in
the heated enclosure that was used for the McCrone; the same probe
used for the McCrone was used for the Battelle.
A wide aray of nozzle sizes were available to permit near
isokinetic sampling at fixed flow rates with each impactor. During
each test, the Brink impactors were run in the vertical position
while the Andersen, U. W. Mark III and TAG were usually run in a
horizontal position. On a few occasions, buttonhook nozzles were
used and the latter impactors were operated in a vertical position.
It was found that operation in the vertical tended to result in
less material being dislodged from the plates during removal from
the duct.
tKey words shown in parentheses have been used throughout this
report to simplify the identification of the various sizing
devices.
-------
-6-
With the Battelle, there was evidence of material lodging
in the long probe without getting to the impactor. This
problem was evident in low grain loadings at all sizes based
on calculations to compare the data obtained with the Battelle
to the other impactors. The average mass concentration obtained
from the Battelle data was also lower than that measured with
the full traverse ASME stack sampling procedure. The extent
of the probe losses for out-of-stack samplers appear to be
related to flow rate, loading, and particle size distribution.
It is interesting to note that data obtained with the McCrone
parallel cyclone used out of the stack and under about the same
conditions as the Battelle showed general agreement with the
data obtained with the instack samplers. The important difference
was that the sampler probe flow rate for the McCrone was much
higher than that used for the Battelle.
For the Brink tests, the only auxiliary equipment used
were nozzles, a vacuum pump, manometer, and a stopwatch. The
flow rate needed to achieve isokinetic sampling, based on
average gas velocities, was maintained by monitoring the
pressure drop across the impactor. For the Andersen, U. W.
Mark III, and TAG impactor tests, a vacuum pump, a gas meter,
an orifice for monitoring instantaneous flow, and stopwatch
were used. The McCrone system was equipped with two gas meters,
pumps, and flow metering controls as necessary, in a self-
contained unit. The flow rate for the Battelle impactor was
determined with a conventional dry gas meter.
During some of the early tests with the U. W. Mark III,
TAG and Andersen impactors, it was noticed that material was
accumulating on the underside of the plates at locations where
material would not be deposited by such things as particle bounce,
etc. These deposits appeared to be related to the large negative
pressure in the duct (relative to ambient pressure) which caused
a reversed air flow through the impactors when the vacuum pumps
were turned off. This caused loose material in the impaction
deposits to be reentrained and impacted on the underside of the
plates. Competely sealing the flow system at the end of a sampling
run seemed to solve this problem.
As soon as possible after the impactor was removed from
the duct, the impaction substrates were removed from the impactor
and desiccated overnight before weighing. Handling of backup
filters proved to be a problem. On approximately half of the
tests these filter weights were negative, even when it was visibly
evident that material had accumulated on the filter. This
problem was observed on all of our tests and appears to be related
to the design of the filter holder; that is, some of the filter
material is lost when the filter is removed from the holder. On
-------
-7-
about 50% of the tests, the last stage showed weight losses when
greased substrates or glass fiber filter substrates were used.
It should be noted that the correct weight gain of the final
s'tages and filter is typically quite small; however, the weight
losses that were observed appear to be due to handling problems
more than weighing accuracy.
IV. TEST RESULTS
The McCrone Parallel Cyclone Inertial Sizing Device
Six tests were performed with the McCrone parallel cyclone,
all at the precipitator inlet. Because no outlet measurements
were made, fractional efficiencies could not be obtained solely
from McCrone data. The McCrone device consists of 6 major pieces;
two condensing units, two gas pump and metering boxes, the
enclosure for containing the Parallel Cyclone, and the sampling
probe. The cyclone enclosure and probe are maintained at stack
temperatures by means of appropriate heaters. Due to the bulk
and weight of the device, it would be extremely difficult to
run inlet and outlet tests on the same day with only one system.
The parallel cyclone assembly consists of a large scalping
cyclone, three smaller cyclones, and a filter all sampling from
a manifold at the outlet of the scalping cyclone. Each small
cyclone has its own back up filter. The D50's for the four
cyclones are reported to be 6.0, 4.0, 1.0, and 0.5 urn when operated
under the conditions used for these tests. Thus in the 0.2 to 2.0
Mm range of particle sizes, the McCrone unit provided only two
fractionation points. It would be possible, however, to modify
the McCrone to provide additional size cuts in the particle
size range of interest.
For these six tests, sampling times were from 9 to 15
minutes which resulted in sample volumes between 0.6 and 1.3 m3
(22 to 45 ft3). The total volume sampled with the impactors in
comparison was typically about 0.03 m (1 ft3). During operation,
it is the increasing pressure drop across the cyclone back up
filters that limits the sampling time. It is worth noting here
that when the McCrone system was operating, the pitot tube built
into the probe consistently showed a 10% velocity increase over
the value obtained with the sample flow off.
The McCrone sampling probe extended about 1 m (3 ft) into
the duct. The connection from the stack to the sampling box was
not heated. We did not directly measure probe losses; however,
the grain loadings computed from the cyclone and filter catches
agreed very well with those determined by conventional methods
and the size distributions were found to agree reasonably well
-------
-8-
with those obtained with the instack classifiers. Cumulative
size distributions for representative McCrone runs are shown in
Figure 2.
Prior to each run, four filters were desiccated and weighed,
then placed in their holders. The cyclones were then heated to
stack temperature after which sampling was started with flow
rates set by cyclone and metering .orifice pressure drops. (The
flow rates must be held constant.) As the test continued, the
back up filters began to clog and after some time, the filter
pressure drops exceeded the pump capacities at which time sampling
was terminated. The 47 mm filters used in the present unit
limit the sampling time rather than the capacity of the cyclone
reservoirs. After cooling, the cyclone catches and back up
filters are removed, desiccated, and weighed. There was some
problem in removing the filters from their holders. On many
occasions, a considerable amount of the filter material stuck to
the filter holder. As mentioned previously, this was a common
problem with back up filters for all the devices tested which
used rubber rings to seal the back up filter.
Differential size distributions for each run were obtained
by numerical differentiation of cumulative size distributions.
For example, Figure 3 shows the differential distributions obtained
from the cumulative data shown in Figure 2. Each curve tends to
agree fairly well with inlet distributions obtained with the other
devices. Below 0.5 ym, the differential distributions shown are
extrapolations based on an analytical curve fitted to the cumulative
distribution data points.
For all differential distributions given in this report,
the ordinates have the units of mass concentration; that is,
the "M" in the expression dM has the units of g/m3 or gr/ft3.
d log D
Since the main interest here is in the shape of the various
distributions rather than the actual value of the points,
all of the differential distributions in this report are shown with
the more familiar units of gr/ft3. For conversion to metric
units, the number in gr/ft3 should be multiplied by 2.288 to
obtain g/m3.
The lack of information in the particle size range of
interest, 0.2 to 2 ym, makes it difficult to compare data obtained
with the McCrone with the other devices which provide a greater
number of size cuts. The large size and weight of this prototype
-------
g
•V
M
•H
Q
u
-H
-P
^
ID
CM
Test Designation
D M4
O M5
M6
10 20 30 40 50 60 70 80 90 95 98 99
Percent Smaller than Indicated Size
99.8
i
CD
I
99.99
Figure 2. Representative Cumulative Size Distributions Obtained
with the McCrone Parallel Cyclone Sampler.
-------
-10-
10.0
i.o
Test Designation
O M4
O M5
M6
Figure 3.
1 10
Particle Diameter, ym
Differential Mass Distributions Obtained from
the Cumulative Distributions shown in Figure 2
100
-------
-11-
model of the McCrone made this particular device a poor
choice for the measurement of fractional efficiencies. The
concept, however, appears to have merit and the prototype
devices tested here may in fact be useful for determining
efficiencies from 0.5 to 10 ym, especially where inlet grain
loadings are very high and sampling times for impactors are
impractically short. During these tests, the parallel cyclone
was able to sample for about 2 to 6 times as long as the
impactors and sampled 20 to 50 times the gas volume sampled by
the impactors. The sampling time could be substantially
increased by using back up filters with a larger surface area.
Battelle CIS-6 Cascade Impactor
Five tests were conducted using a Battelle CIS-6 cascade
impactor. The model used in these tests had been modified
to include an additional multiple orifice stage which was
reported to have a D50 of about 0.25 ym. The Battelle was
mounted external to the duct and the McCrone probe and heater
were used. The impactor was heated to stack temperature before
the sampling was started. Since the entire seven stage impactor
would not fit in this enclosure, stages 1 and 2 were removed and
not used for four of the five runs. On the fifth run, stage
seven was omitted and stage two was used. This omission caused
a loss of information at some sizes but in this case the
information was of small consequence because of the mechanical
separator upstream of the sampling location. Furthermore, the
fractionation sizes for the stages omitted were much larger than
the upper limit of our primary range of interest.
To improve weighing accuracy by reducing the tare weight
of the collection surfaces, foil substrates were used rather
than the glass plates supplied with the impactor. On four of
the five runs, grease was used on the foils.
Particulate mass concentrations calculated from the Battelle
data do not agree with measurements obtained with the other
devices. The fact that these results are very low indicates
a possible loss of material in the sampling nozzle and probe.
This may have been aggravated by a relatively low gas velocity
that must be used in the probe. Differential mass distributions
determined from the Battelle data show pronounced deficiencies
at larger particle sizes. This occurred in spite of the use of
an oversized inlet sampling nozzle which should have resulted
in some over sampling of large particles. Differential size
distributions as measured with the Battelle impactor are shown
in Figure 4.
-------
-12-
1.0
0.1
M-l
O
Q
O
H
X.
0.01
0.001
0.1
1 10
Particle nianeter, ym
100
Figure 4. Differential Inlet Mass Distribution Obtained
with the Battelle CIS-6 Irapactor.
-------
-13-
TAG Multiple Split Cascade Impactor
The ERG TAG impactor consists of 10 impaction stages and
a back up filter. Each stage weighs about 20 g. This is too
heavy for use with the high precision balance; therefore, for
heavily loaded stages, the material was removed from the plates
and transferred to a small foil boat for weighing. ERG
recommends the use of grease on stages 1 and 2; however, these
stages are for particle sizes outside the range of interest
for our work so these stages were not weighed. Based on
visual appearance and weighing, stages 7, 8, 9, and 10 showed
poor particle retention. For this reason, the bare metal
substrates were covered with foil substrates that had been
coated with a thin layer of grease for most of the runs. It
appeared that the use of foil on stage 10 was probably the
only way to accurately determine the catch on this stage by
use of gravimetric methods.
As shown in Figure 5, the data from the six inlet runs
with the TAG tend to agree fairly well with that obtained with
the remaining devices in the 0.2 ym to 2.0 ym range. In
general, they do lie a little lower in this range.* In the
region above 2.0 ym, the differential distributions obtained
from TAG data were low when compared with the distributions
obtained with the other devices.
Figure 6 shows the results of several tests at the precip-
itator outlet that were made with the TAG during times when the
generator was operating at 94% load. General agreement was
found with the other impactors in the range from 0.2 ym to 2.0 ym;
however, above 2.0 ym there was very poor agreement; the TAG data
generally gave loadings above 2 ym that was lower than the other
sizing devices.
Figure 7 shows the results of measurements made with the
TAG when the generator load was 85% and another performed with
100% load. There does not appear to be enough data from these
individual tests to make a meaningful comparison with the other
devices tested during these load conditions.
University of Washington Mark III Source Test Cascade Impactor
Figures 8 and 9 show the data from the U. W. Mark III
inlet runs with and without greased foil substrates. Four
runs were performed with grease on stage 1 only. Thirteen runs
were performed with grease and foils on all stages. It can
be seen that the apparent .loading at sizes from 0.3 ym to 2.0 ym
*Subsequent to the preparation of this report a revised
calibration was obtained which reduced but did not entirely
eliminate the difference between the TAG and the other impactors.
-------
-14-
1.0
0.1
1 10
Particle Diameter, ym
100
Figure 5. Differential Inlet Mass Distribution Obtained
with the ERG TAG Sampler.
-------
-15-
10.0
1.0
8
w
-H
I
0.1
0.01
0.1
1 . 10
Particle Diameter, ym
100
Figure 6. Differential Outlet Mass Distribution at 94%
Load as Obtained with the ERG TAG Sampler.
-------
-16-
1.0
0.1
0.01
0.001
0.1
1 T 10
Particle Diameter, ym
Figure 7. Differential Outlet Mass Distributions Obtained
with the ERC TAG Sampler at Loads of 85% and
100%.
100
-------
-17-
10.0
Particle Diameter,
100
Figure 8.
Differential Inlet Mass Distributions Obtained
with the U. W. Mark III Impactor with Ungreased
Substrates .
-------
-18-
10.0
u
rd
X
M
(5
•H
(fl
H
Q
O
i-H
T3
£
1.0
100
Particle Diameter,
Figure 9- Differential Inlet Mass Distributions Obtained
with the U. W. Mark III Impactor with Greased
Substrates.
-------
-19-
increased substantially when greased substrates were used. At
the same time, there was much better agreement with the other
devices when grease was used. The difference in fine particle
load, the change in weight gain and appearance of the filter,
and the change in appearance of the impactor deposits, all
indicated that, without grease, particles were bouncing off
the last few stages and ending up on the filter.
The grease used was a Dow-Corning high-vaccum silicone
grease suitable for temperature up to 200°C (400°F). The
pregreasing and preweighing required careful handling of the
impaction substrates, especially when foil substrates were
used, so that grease is not lost due to poor handling.* The
donut shaped foils that were required on stages 2 through 7
are somewhat difficult to put in place when greased. The
foils must be cut a little larger than the final diameter of
the substrate to allow for forming a small lip. This lip
holds the foil in place when the impactor is assembled. Care
must be taken to contour the foil to the shape of the plate
to insure proper stage to plate spacing. Although use of the
foils is time consuming, the increased weighing accuracy
appears to justify the additional time.
During all runs, the impactor was held in a horizontal
position parallel to the gas stream, pointing downstream during
the 20 to 30 minute warm up period and upstream during the
test.
Figure 10 gives the results of additional tests that were
conducted with the generator at 94% of full load. Within the
scatter of the data, these size distribution data obtained
at the precipitator outlet with the U. W. Mark III impactor
agree with that obtained with the other devices, in the 0.2 ym
to 2.0 ym region.
Andersen Stak Sampler - Model II
Four runs were performed with the Andersen Model II.
(Three inlet tests and one outlet test.) The only way to
accurately weigh the stage catches was to try to brush the
material into a foil cup. This proved unsatisfactory because
of poor transfer efficiency. Using the Ventron pan balance
to weigh the catch by determing the weight gain of the plates
*The results of laboratory work after the field tests have shown
that evaporation of the silicone grease at high temperatures (200°C)
can produce a weight loss of 1-2% during the first hour of operation.
This loss can be significant in the fine particle stages but it
can be reduced by baking the greased foils for one hour at 200°C
before the test.
-------
-20-
1.0
0
id
c
•H
(0
N
D>
p
o
rH
•o
0.1
0.01
0.001
0.1
1 10
Particle Diameter, urn
100
Figure 10. Outlet Differential Mass Distributions Obtained
with the U. W. Mark III Impactor at 94% Load.
-------
-21-
was unsatisfactory because of insufficient sensitivity and
accuracy in weighing the small catches on the last few stages.
Because of these problems and the difficulty in using grease
to minimize bounce and reentrainment in the impactor very few
tests were performed with this device.
A sampling method similar to the U. W. Mark III was used.
The impactor was held in a horizontal position, 1.5 m (5 ft) into
the duct. Care had to be taken to not jar the device after a
run and to clean the nozzle properly after each run.
Figure 11 shows the three inlet tests with the Model II
Andersen. Effects of reentrainment and bounce are evident in
the relatively low loadings for sizes caught on the first
stages, the broad peak at about 3 ym and the relatively
low apparent loadings near 0.4 ym. Figure 12 shows the size
distribution at the precipitator outlet at 84% load as obtained
with the Model II Andersen. The Andersen Models III and IV,
which are described next, are similar to the model II but have
been designed to use glass fiber filter collection surfaces.
Andersen Model IV
Figure 13 shows the inlet differential mass distribution
data obtained with the Andersen Model IV. The Model IV is an
instack version of the Andersen non-viable sampler using
glass fiber disks as impaction substrates.
As can be seen by comparing the data shown in Figure 13 with
that in Figure 11, the inlet data obtained with the Andersen
Model IV does not agree well at all with data from the other
devices. It was obvious that severe reentrainment occurred
in the upper stages. Visual inspection of the appearance of
the deposits on the upper stage impaction substrates indicated
that the reentrainment resulted from blow off due to high gas
velocities parallel to the surface of the substrates. This
reentrainment resulted in a pronounced apparent shift in the
measured size distribution toward the lower sizes. Data from
samples obtained at the precipitator outlet showed the same
effects.
It appears that this particular design, even with the
glass fiber inserts, is not suited for sampling aerosols such
as fly ash because of this severe reentrainment problem.
Andersen Stak Sampler - Model III
The Andersen Model III is a modified Model II with provision
for the use of bulls-eye shaped, glass fiber filter impaction
substrates. This particular design was the most useful of the
-------
-22-
10.0
1.0
0.1
0.01
0.1
10
100
Particle Diameter, ym
Figure 11. Differential Inlet Mass Distributions
Obtained with the Andersen Model II Impactor
at 100% Load.
-------
-23-
1.0
O
m
s
•H
I
en
O
I
0.01
0.001
0.1
10
100
Particle Diameter, ym
Figure 12.
Outlet Differential Mass Distribution Obtained
with, the Andersen Model II Impactor at 80%
Load.
-------
-24-
10.0
1.0
0.1
0.01
0.1
10
100
Particle Diameter, ym
Figure 13. Differential Inlet Mass Distributions as
Obtained with the Andersen Model IV Impactor.
-------
-25-
three Andersen models tested. The reentrainment problems
associated with Models II and IV were not evident with this
modification.
Care must be taken in handling the filter substrates. Over
tightening of the impactor case during assembly can cause the
steel o-rings to cut into the filters. It was found that one
of the two possible orientations of the sealing ring was less
likely to cut the glass fiber substrates.
As can be seen in Figure 14, the Andersen Model III inlet
data agrees very well with that obtained with the other classifiers
Five inlet tests were also run with the Andersen 2000 Inc.
prototype scalping cyclone. Data from these runs are shown in
Figure 15. There is no noticeable difference below 6.0 ym in
the differential size distributions measured with and without
the cyclone. Above this point, the data tends to fall a little
lower indicating that some of the material that would have
ended up on stage 1 had been caught in the cyclone. In the 0.2 ym
to 2.0 ym region, the cyclone has no apparent effect.
The outlet tests using this impactor gave generally good
agreement with the other classifiers, although on several
occasions not enough data was collected to give a clear
comparison. Representative data from outlet runs are shown
in Figure 16.
Brink Impactor
This impactor was run in a number of configurations. Four
cyclones were used singly or in combination to remove the
larger particles when the Brink was used at the inlet to the
precipitator. Approximate D5o's for the four cyclones under
these test conditions were 11 ym (SRI Model I, Cl), 4.32 ym (SRI
Model II, C2), 6.44 ym (SRI Model III, in line, C3), and 7 ym
(cyclone supplied with the Brink BMS 11 kit). Also an auxiliary
stage, "0", was designed and fabricated to obtain a fractionation
point at a size about 1.5 times the D50 of the standard Brink
first stage. The zero stage was generally used in conjunction
with the inline cyclone, C3. In addition to three versions of
the Brink impactor modified for instack sampling, the Brink
BMS-11 sampling kit was mounted external to the duct so that it
sampled through conventional probes. Cyclones were not used for
outlet runs because of the absence of large particles at the
outlet. In addition to tests of the several configurations of
the Brink impactor, a limited number of tests were run at both
the inlet and outlet using greased impaction substrates.
-------
-26-
10.0
• Andersen Model III
X Andersen Model II
0.1
0.01
Figure 14.
1 10
Particle Diameter, \im
Differential Inlet Mass Distributions Obtained
with the Andersen Models II and III Impactors.
100
-------
-27-
10.0
•ul.O
u
-------
-28-
1.0
0.1
0
"x
c
•H
in
Q
O
0.01
0.001
0.1
1 10
Particle Diameter, ym
Figure 16. Differential Outlet Distribution at 60% Load
Obtained with the Andersen Model III Impactor.
100
-------
-29-
All Brink impactor measurements were made with the
impactor in the stack except for the tests using the Brink BMS-11
kit. All tests were run with the impactors in a vertical
position with nozzles chosen for isokinetic sampling. Back up
filters were used on all runs.
The three cyclones were run in the following configurations
at the inlet. (The cyclone designations were explained
previously; the numbers 1, 2, 3,... refer to Brink stages; and
"F" stands for the glass fiber back up filter)
Cl, C2, 1, 2, 3, 4, 5, F
C2, 1, 2, 3, 4, 5, F
C3, 1, 2, 3, 4, 5, F
C3, 0, 1, 2, 3, 4, 5, F
Brink Kit Cyclone, 1, 2, 3, 4, 5, F
Figures 16 through 19 show results for inlet tests with these
configurations. As can be seen there were no significant
differences in size distribution results for any of these set
ups, with the exception of those obtained with the BMS-11 kit.
The data obtained with the kit indicated significantly lower
loadings for sizes larger than 2 ym as can be seen in Figure 17.
In fact, the cyclone collected negligible amounts of material.
Presumably most of the material it would have collected was
lost in the probe.
One series of tests were conducted to see if the use of
foil substrates and grease would cause any significant changes
in the Brink data. Figure 20 shows the results of these tests.
As expected, the greased foils tend to retain the material
better on the small particle stages. The use of grease does
not seem to significantly affect retention of particles on
the upper stages. Subsequent to the tests, microscopic examination
of the particles caught on several sets of ungreased substrates
revealed the presence of large quantities of over-sized
particles on the substrates of stages 4 and 5, and on the back
up filters. It was also noted that particles were not collected
immediately under the jet on runs with ungreased substrates
for stages having jet velocities greater than 35 to 40' m/sec.
No such evidence of bounce and reentrainment were observed when
greased substrates were used even with jet velocities as high
as 75 m/sec.
The Brink impactors were also used for measurements at
the precipitator outlet for a variety of load conditions; these
results are shown in Figures 21 through 23. A comparison of
-------
-30-
10
0.1
1 10
Particle Diameter, ym
Figure 17. Differential Inlet Mass Distribution Obtained with
Brink Impactor Modified for In-Stack Sampling
Preceded by Cyclone C2, and Corresponding Data
Obtained Using the Brink BMS 11 KIT.
100
-------
-31-
10.0
1.0
O
nJ
n
•a
s,
tn
0
0.1
0.01
0.1
1 10
Particle Diameter, ym
100
Figure 18. Differential Inlet Mass Distributions
Obtained Using Brink Impactors Modified for
Instack Sampling Preceded by Cyclones
Cl and C2.
-------
-32-
0.1
1 10
Particle Diameter, ym
100
Figure 19.
Differential Inlet Mass Distributions
Obtained with Brink Impactors Modified for
Instack Sampling Preceded by an Additional
Stage, SO, and Cyclone C3.
-------
-Sa-
1.0
0.1
o
(8
\
w
q
•H
Q
O
-a
0.01
0.001
X Greased
• Ungreased
.1
10
100
Particle Diameter,
Figure 20. Comparison of Averaged Differential Inlet
Mass Distributions Obtained with Brink Impactors
Using Greased and Ungreased Impaction Substrates.
-------
-34-
0.001
0.1
1 10
Particle Diameter,ym
Figure 21. Differential Outlet Mass Distribution at Full Load
Obtained Using Brink Impactors Modified for In-Stack
Sampling Using Greased and Ungreased Substrates.
100
-------
-35-
0,1 1 10
Particle Diameter, ym
Figure 22. Differential Outlet Mass Distributions at 94%
Load Obtained Using Brink EMS 11 Kits and
Brink Impactors Modified for In-Stack Sampling,
100
-------
HC-CV:
-36-
1.0 -m^
10
100
Particle Diameter,
Figure 23. Differential Outlet Mass Distribution at
55% Load Obtained Using Brink Impactors
Modified for Instack Sampling.
-------
-37-
these results for greased and ungreased foils, under identical
load conditions, emphasized the importance of using grease at
the outlet as well as the inlet.
Summary of all Impactor Tests
A total of 192 individually evaluated tests were performed
at the inlet and outlet of the precipitator in both high and low
efficiency modes. The precipitator efficiency was varied by
load changes or by addition of fly-ash conditioning agents. This
allowed for both course and fine aerosols to be sampled. A list
of the number of individual tests using each device and each
different combination of stages and cyclones is given below.
This list includes only the tests that appeared to be valid; that
is, some tests have been deleted because of errors that may have
resulted from operational problems, severe reentrainment, etc.
Where appropriate, these test results were used to compute the
fractional efficiency of the electrostatic precipitator as
described in the next section.
Particle Sizing Device
Andersen Model II
Andersen Model III
Andersen Model III with Andersen Cyclone
Andersen Model IV
Battelle - Greased Foils
Battelle - Ungreased Foils
Brink - C3,0,1,2,3,4,5,F - Ungreased
Brink - C2,1,2,3,4,5,F - Ungreased
Brink - C2,1,2,3,4,5,F - Greased
Brink - C3,1,2,3,4,5,F - Ungreased
Brink - C3,1,2,3,4,5,F - Greased
Brink - BMS-11-C,1,2,3,4,5,F - Ungreased
Brink - Cl,C2,1,2,3,4,5,F - Ungreased
Brink - C1,C2,1,2,3,4,5,F - Greased
Brink - 0,1,2,3,4,5,F - Ungreased
Brink - 1,2,3,4,5,F - Ungreased
Brink - 1,2,3,4,5,F - Greased
McCrone
U. W. Mark III - Grease on 1 only
U. W. Mark III - Grease and foils on all
TAG - Grease on 1 and 2 only
TAG - Grease on 1 and 2, grease and foils on
7,8,9,10
Number
of Tests
4
25
5
18
4
1
15
8
2
5
2
8
13
1
12
23
5
6
4
13
4
Total number of valid tests = 187
-------
-38-
Fractional Efficiency of the Electrostatic Precipitator
Examples of the efficiency of the precipitator as a
function of particle size are shown in Figures 24 and 25 for
two operating conditions, 94 to 100% of full load and 55 % of full
load. Each set of points shown on these figures represents an
average value calculated from inlet and outlet mass-size
distributions that were measured with the same type of impactor;
that is, one efficiency is as measured with the Andersen, another
with the Brink, etc. The data used for Figures 24 and 25 were
collected at various times over a period of three weeks and for
some of the tests inlet and outlet size distributions were
not obtained simultaneously. Based on these facts and the
apparent differences in the size calibrations of the various
impactors, the fractional efficiencies shown here should be
regarded as approximate. There is less scatter of the data for
the fractional efficiencies shown in Figure 25 than in Figure 24.
The information shown in Figure 25 was all based on measurements
made with the Brink and Andersen impactors. More measurements
were made with these impactors because more of the impactors
were available for these tests. Because of the additional
measurements, the efficiencies shown in Figure 25 are probably a
better representation of average performance than those shown
in Figure 24.
V. CONCLUSIONS AND COMMENTS
Several general conclusions can be drawn from the results
of our field evaluation of inertial sizing devices:
A. Of the inertial classifiers tested, no single device
was found to be useful over the entire range of particle size
and concentration needed for evaluation of a wide variety of
particulate control devices. In general, the inertial classifiers
designed to operate at low sample flow rates are useful at the
inlet of a control device where the particulate loading is high;
the low-flow-rates permit reasonably long sampling times to be
used. If these same low-flow-rate classifiers are used at the
outlet of a control device, the sampling time may be impractically
long, so that a high-flow-rate device is usually a better
choice at the outlet.
B. One possible solution to the problem of measuring fine
particles in the presence of high concentrations of large particles
is to use one or more small cyclones in front of the impactor.
Modified Brink impactors and the Andersen Model III were successfully
used in this manner during our field tests.
C. Most of the impactors have collection stages that are too
heavy for obtaining accurate measurements of the weight of
particles collected in a size fraction. It is helpful to cover
-------
99.99
99.9
99.0
98.0
tl
r
50
u
c
)
Ijnpactors.
CO
p
-------
99.99
99.9
99.0
98.0
o\°
.. 50.0
o
5
I
o
-------
-41-
the stage with a lightweight collection substrate made of
aluminum foil, teflon, glass fiber filter material, or other
suitable lighweight materials depending on the particular
application. With such an arrangement, it is possible to make
accurate weight determinations of collected samples that are
small enough to prevent or at least minimize reentrainment.
Weighing to a precision of at least 30 yg appears to be required
for impactor size determination for submicron particles and 10 yg
precision is desirable.
D. When impactors are operated at flow rates higher than some
critical value, particle bounce can lead to incorrect sizing
with non-cohesive particulates. This is especially true in the
lower (fine particulate) stages. In these lower stages, the high
jet velocities cause scouring of the plates which are quite
noticeable and is an almost certain indication of reentrainment;
Figures 26 A and B illustrate this effect. Experience with the
University of Washington Mark III and the Brink impactor indicate
that the maximum jet velocities that can be used for impaction
on ungreased substrates without severe reentrainment resulting
from particle bounce is about 40 m/sec. Also, the physical
properties of the particles can effect the maximum loading that
can be safely obtained on an impaction stage. For example, with
some emission sources, a small quantity of cohesive particulate
materials will form a pile that will plug the stage orifice. For
other types of emission sources with dry particulates, reentrain-
ment and particle bounce problems can be severe. The use of a
silicone grease that is stable at high stack temperatures appears
to be helpful for measuring dry particulates. It has been our
experience that even with this grease, errosion and scouring can
occur on the lower stages with high jet velocities, velocities >
75 m/sec. The scouring and reentrainment can be reduced by
reducing the flow rate in the impactor. For example, if the
Brink sample flow is reduced from the recommended value of
2.8 1/min (0.1 cfm), to approximately 1.4 1/min (0.5 cfm) the
scouring effect can be eliminated. This change in flow rate
increases the cut point of each size fraction and decreases the
amount of information that can be obtained in the small particle
size range. These comments are appropriate for the Andersen
and other impactors as well as the Brink. However, the addition
of an additional stage to the impactor with the proper combination
of velocity and jet diameter would make it possible to regain
the information lost at the lower flow rate.
E. Glass fiber "back up" filters have generally been used
after the final stage in our impactor tests. For some emission
sources, the back up filter frequently showed a weight loss
instead of a weight gain. This result appears to be related to
filter handling problems and poor design of filter retainers and
seats. More work is needed to solve this problem.
-------
Figure 26A.
Fly Ash Deposits on Ungreased Impaction Substrates
without NH3 Injection into Flue Gas. (Effects of
Particle Bounce and R-entrainment are Especially
Pronounced on the Last Three Stages.)
NJ
j
Figure 26B. Fly Ash Deposits on Ungreased Impaction Substrates
with NH3 Injection. (Comparison with Figure 26A
Reveals a Pronounced Change in the Adhesive and
Cohesive Properties of the Particulate.)
-------
-43-
F. Particle size measurements have been made with the
impactors mounted outside the stack. Sampling probes and lines
were used to transport the sample to the impactor. Based on
simultaneous size determination with the instack and out-of-stack
sampling arrangements, it seems almost certain that a significant
quantity of particles are lost in the probe and sampling lines,
particularly in sizes larger than 2 ym. Figure 27 shows average
differential inlet distributions as obtained with out-of-stack
classifiers and illustrates this effect.
G. A consideration of all of the fractional efficiency curves
shows that measurements with the modified Brink impactors, Brink
with grease, Brink kit, Andersen Model II, and U.W. Mark III
all seem to result in about the same calculated efficiency.
Although most of the classifiers used during these tests
show general agreement in the differential size distributions
obtained with each, significant differences were noted in
detailed comparisons. Figure 28 shows average inlet size
distributions as obtained with four instack classifiers. A
revised calibration for the E.R.C. TAG sampler was received
after the field work and was used in the preparing of Figure 28.
Differences of as much as a factor of two in the loading at any
one particular size exist between the data obtained with some
of the classifiers. These differences appear to be related to
the lack of good calibration of the stage penetrations under
conditions that are appropriate for field sampling. Simultaneous
calibrations under identical conditions, that simulate field
measurements, appear to be mandatory if reliable field test
results are to be obtained with a wide variety of impactors. This
comparison or mutual calibration will be particularly important
if two different makes of impactors are to be used in determining
the fractional efficiency of a control device. As described
previously in this report, the use of two impactors, a low flow
rate impactor at the inlet and a high flow rate impactor
at the outlet, is often required for measurement of field
emissions sources.
-------
-44-
1
,8
,6
CO
cc O
LU 'fc
Q_
(=3
,08
,06
,02
,01
,2 ,4 ,6 ,8 1 2 i* 6 8 10
PARTICLE DIAMETER (MICROMETERS)
Figure 27.
Inlet Size Distribution as Measured with
Three Out-of-Stack Samplers. Battelle —.--
Brink BMS u——— . McCrone Parallel
Cyclone——•—.-; Solid Line is the
Distribution Obtained with the V.W. Mark III
In-Stack Sampler for Comparison.
-------
-45-
,2 ,1 ,6 ,8 1 2 4 6 8 10
PARTICLE DIAMETER (MICROMETERS)
Figure 28. Inlet Size Distribution as Measured with
Four In-Stack Samplers. E.R.C. Tag,Q;
Brink,9 ; Andersen,^. ; Mark III,Q. (Each
Point Represents the Average of Several
Measurements.)
-------
-46-
REFERENCES
1. M. J. Pilat, "Submicron Particle Sampling with Cascade
Impactors", Presented at the 66th Annual Meeting of the
Air Pollution Control Association, Paper Number 73-282,
(June, 1973) .
2. Private communication, Carl Erickson, Andersen-2000, Inc.,
Salt Lake City, Utah, (July, 1973).
3. A. N. Bird, J. D. McCain, D. B. Harris, "Particulate
Sizing Techniques for Control Device Evaluation", presented
at the 66th Annual Meeting of the Air Pollution Control
Association, Paper Number 73-282, (June, 1973).
2923-X Ic
SORI-EAS-73-299
(10:1:2:15)
-------
-47-
BIBLIOGRAPHIC DATA
SHEET
1. Report No.
EPA-650/2-73-035
3. Recipient's Accession No.
4. Tule and Subtitle
Field Measurements of Particle Size Distribution with
Inertial Sizing Devices
5. Report Date
October 1973"
6.
7. Author(s)
Joseph D Mpf!ain Kenneth M. Cashing Alvin N. Bird, Jr.
8. Performing Organization Rept.
No- SORI-EAS-73-299
9. Performing Organization Name and Address
Southern Research Institute
2000 Ninth Avenue South
Birmingham, Alabama 35205
10. l->ro|ect/Task/Work Unit No.
ROAP 21ADM11
11. Contract/Grant No.
68-02-0273
12. Sponsoring Organisation Name and Address
EPA, Office of Research and Development
NERC-RTP, Control Systems Laboratory
Research Triangle Park, North Carolina 27711
13. Type of Report
Covered
Period
Report & PC
Special
3/1/73 - 5/23/73
14.
IS. Supplementary Notes
16. Absiracts The report describes a comprehensive particle size measurement program
conducted at a coal-fired electric generating plant early in 1973. It also includes
information obtained from other field tests and laboratory work. The primary part-
icle size range of interest is from 0..2 to 2.0 jum diameter, but techniques are also
evaluated for measuring larger particles. Inertial classification is the basic sizing
technique considered. Among the 11 different commercial and modified sizing devices
evaluated are: Andersen Models E, m, and IV; Battelle CB-6; Brink BMS-11: Mc-
Crone; TAG; and University of Washington Mark HI. The program is part of a project
to devise and evaluate various techniques for field measurements of the fractional
efficiency of particulate control devices.
17. Key Words and Document Analysis. 17o.
Air Pollution
Measurement
Measuring Instruments
Fines
Particle Size
Particle Size Distribution
Size Determination
Impactors
Cyclone Separators
17b. IdcntifKTi/Opcn-Ended Terms
Air Pollution Control
Stationary Sources
Fractional Efficiency
Particulate Control Devices
Inertial Classification
IVscriptorb
Electrostatic Precipitators
17c. COSAT1 Field/Group
Cascade Impactors
Mass • Concentration
Grain Loading
Battelle CIS-6 Impactor
McCrone Parallel-Cyclone
Sizing Device
University of Washington
Mark in Cascade
Impactor
Brink BMS-11 Sampling
Kit
TAG Sampler
18. Availability Statement
,140,136 Anderson Models II,
Unlimited
ers
19. Security
Report)
UNC1
Class (This
.ASSLFIF.l
20. Security
Pagi.
UNCI
Class (This
.ASSIFIED
21. No. of Pages
52
22. Price
rORM NTIS-35 (REV. 3-721
USCOMM-DC I4B5Z-P72
------- |