United States
Environmental Protection
Agency
Environmental Sciences Research
Laboratory
Research Triangle Park NC 277 11
EPA-600 2 80-048
February 1980
Research and Development
&EFK
Use and Limitations of
In-Stack Impactors
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S. Environmental
Protection Agency, have been grouped into nine series. These nine broad cate-
gories were established to facilitate further development and application of en-
vironmental technology. Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The nine series are:
1. Environmental Health Effects Research
2. Environmental Protection Technology
3. Ecological Research
4. Environmental Monitoring
5. Socioeconomic Environmental Studies
6. Scientific and Technical Assessment Reports (STAR)
7. Interagency Energy-Environment Research and Development
8. "Special" Reports
9. Miscellaneous Reports
This report has been assigned to the ENVIRONMENTAL PROTECTION TECH-
NOLOGY series. This series describes research performed to develop and dem-
onstrate instrumentation, equipment, and methodology to repair or prevent en-
vironmental degradation from point and non-point sources of pollution. This work
provides the new or improved technology required for the control and treatment
of pollution sources to meet environmental quality standards.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.
-------
EPA-600/2-80-048
February 1980
USE AND LIMITATIONS OF IN-STACK IMPACTORS
by
Dale A. Lundgren and W. David Balfour
Department of Environmental Engineering Sciences
University of Florida
Gainesville, Florida 32611
Grant No. R803692-02
Project Officer
Kenneth T. Knapp
Emissions Measurement and Characterization Division
Environmental Sciences Research Laboratory
Research Triangle Park, N.C. 27711
ENVIRONMENTAL SCIENCES RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK, N.C. 27711
-------
DISCLAIMER
This report has been reviewed by the Environmental Sciences Research
Laboratory, U.S. Environmental Protection Agency, and approved for
publication. Approval does not signify that the contents necessarily reflect
the views and policies of the U.S. Environmental Protection Agency, nor does
mention of trade names or commercial products constitute endorsement or
recommendation for use.
11
-------
ABSTRACT
A systematic evaluation of the operating parameters for four
commercially available in-stack cascade impactors was carried out with
polydisperse test aerosols. The impactors tested included a MK III
University of Washington source test cascade impactor, an Andersen MK III -
stack sampler, a Sierra Model 226 source cascade impactor and a modified
Brink Model B cascade impactor. Test aerosols used, classified according to
their collection characteristics, were hard and bouncy (polystyrene latex
spheres), hygroscopic and medium bouncy (uranine and sodium chloride), or a
sticky liquid (dioctyl phthalate and dinonyl phthalate). The effect upon the
apparent measured size distribution of each polydisperse test aerosol was
noted for various gas sampling rates (flow rates), types of impactor
collection surfaces (glass fiber, uncoated aluminum, and aluminum coated with
silicone), stage loadings and interstage losses. Collection surfaces were
further characterized as to their weight loss during exposure to elevated
temperatures and their tendency to be blown off by an impinging jet of air.
Based upon these observations the spray silicone was the only "grease"
type collection surface coating found suitable for use at temperatures of up
to 400°F and incident jet velocities up to 120 m/sec. At temperatures of
500°F and greater the only collection surfaces which gave acceptable
results were the uncoated aluminum and the glass fiber. The type of
collection surface coatings used was shown to have a definite effect upon the
apparent measured size distribution. The observed increase in mass mean
diameter (mmd) noted with the glass fiber collection surface was shown to be
due, at least in part, to the increased collection efficiency of submicron
particles for the upper stages of the impactor. Both the silicon spray and
glass fiber collection surface coatings provided stable collection
characteristics over a range of stage loadings up to 10 mg per stage.
Measurements revealed that interstage losses may amount to 30% of the
total collected mass; however, there is little effect upon the apparent
measured size distribution when these losses are ignored. The useful range
of flow rates available for the impactors was defined at the lower end by a
loss of useful sizing data and at the upper end by the presence of particle
bounce off the latter stages. In general, the impactors were found to give
similar apparent measured size distributions when operated at various flow
rates within this useful range.
Recommendations were made for: 1) optimum operation of the impactor
when sampling various types of aerosols, and 2) accounting for observed or
known errors in the data.
111
-------
CONTENTS
Abstract iii
Figures vi
Tables x
Symbols xii
!• Introduction 1
2. Discussion and Conclusions 3
A. Introduction 3
B. Collection Surface Coatings 3
C. Flow Rate 5
D. Stage Loadings 7
E. Interstage Losses 8
F. Summary and Recommendations 8
3- Theoretical Background and Literature Review . . 12
A. Introduction 12
B. Impactor Theory 15
C. Particle Adhesion 18
D. Previous In-Stack Impactor Studies 20
4. Experimental Apparatus, Methods and Procedures 25
A. Introduction 25
B. General Experimental Setup 25
C. Description of the In-Stack Impactors 27
1. MK III University of Washington Source Test
Cascade Impactor (Model D) 27
2. Sierra Model 226 Source Cascade Impactor 27
3. Andersen MK III Stack Sampler 33
4. Modified Brink Model B Cascade Impactor 33
D. Impactor Collection Surface Coatings 43
E. Aerosol Generation 43
1. Description of Test Aerosols 43
2. Vibrating Orifice Aerosol Generator 48
3. Three Jet Collison Atomizer 49
F. Techniques for Mass Determination 49
1. Fluorometric Technique 49
2. Gravimetric Technique 50
G. Determination of Collection Efficiency 50
H. Measurement of Flow Rate, Temperature, and
Relative Humidity 50
5. Experimental Results 52
A. Introduction 52
B. Analysis of Errors 52
C. Collection Surface Coatings 54
1. Suitability of Use at Elevated Temperatures and
High Jet Velocities 54
-------
2. Effect of Collection Surface Coating on the
Measured Size Distribution 59
D. Effect of Flow Rate on the Measured Size Distribution . . 63
E. Stage Loading 80
1. Effect of Stage Loading on Stage Collection
Efficiency 80
2. Effect of Stage Loading on the Measured Size
Distribution 89
F. Interstage Losses 95
1. Interstage Losses as a Function of Particle Size . . 95
2. Effect of Interstage Losses on Size Distribution . . 95
References 107
Appendix 110
A. Guide for the Use of In-Stack Cascade Itnpactors 110
VI
-------
FIGURES
Number
\ Effect of collection surface on measured size
distribution for polydisperse uranine aerosol
using SRI Dp,... cut-points calibrated for grease
and glass fiber collection surfaces (Brink) 6
2 Total interstage loss versus particle
diameter (U. of W.) 9
3 Basic design of an in-stack cascade impactor 13
4 Principle of operation of an impactor 14
5 Typical impactor efficiency curve 14
6 Impactor efficiency curves showing the effect of
jet-to-plate distance, Reynolds number and
throat length 17
7 Particle sizing data presented as differential
and cumulative plots 19
8 General arrangement of test equipment 26
9 MK III University of Washington source
test cascade impactor 28
10 MK III University of Washington source test cascade
impactor nomograph for determining Dprn cut-points 31
11 Sierra Model 226 source cascade impactor 34
12 Andersen MK III stack sampler 38
13 Modified Brink Model B cascade impactor 42
14 Test set-up for dynamic testing of collection
surface coatings 57
15 Effect of collection surface on measured size
distribution for polydisperse uranine aerosol (13. of W.) 60
VI1
-------
Figure
16 Effect of collection surface on measured size
distribution for polydisperse oil aerosol (U. of W.) ____ 61
17 Effect of collection surface on measured size
distribution for polydisperse oil aerosol
of increased mmd (U. of W. ) ............................. 62
18 Deposition of uranine aerosol onto glass
fiber collection surface at 0.5 cfm ..................... 64
19 Effect of collection surface on measured size
distribution for polydisperse uranine aerosol (Brink) ... 66
20 Measured size distributions for a polydisperse
uranine aerosol obtained with the University
of Washington, Andersen, Brink and Sierra impactors ..... 67
21 Effect of flow rate on measured size distribution
for polydisperse uranine aerosol sampled
onto spray silicone (U. of W. ) .......................... 68
22 Deposition of uranine aerosol onto
spray silicone at 1.0 cfm ............................... 69
23 Effect of flow rate on measured size distribution
for polydisperse uranine aerosol sampled
onto glass fiber (U . of W. ) ............................. 71
24 Effect of flow rate on measured size distribution
for polydisperse oil aerosol sampled onto spray
silicone (U. of W.) ..................................... 72
25 Effect of flow rate on measured size distribution
for polydisperse oil aerosol sampled onto
glass fiber (U. of W.) .................................. 73
26 Deposition of oil aerosol ............................... 74
27 Effect of flow rate on measured size distribution
for polydisperse oil aerosol of increased
mmd sampled onto spray silicone (U. of W. ) .............. 76
28 Effect of flow rate on measured size distribution
for polydisperse oil aerosol of increased mmd
sampled onto glass fiber (U. of W. ) ..................... 77
\
29 Deposition of polystyrene latex spheres ................. 78
30 Effect of flow rate on measured size distribution
for polydisperse uranine aerosol collected
on glass fiber (Andersen) ...................... ......... 81
v
-------
Figure Page
31 Effect of flow rate on measured size distribution
for polydisperse uranine aerosol collected
on glass fiber (Sierra) 82
32 Effect of flow rate on measured size distribution
for polydisperse uranine aerosol collected
on glass fiber (Brink) 83
3? Effect of collection surface on measured size
distribution of polydisperse uranine aerosol
sampled at 1.0 cfm (U. of W.) 84
34 Effect of collection surface on measured size
distribution of polydisperse uranine aerosol
sampled at 0.25 cfm (U. of W.) 85
;: i
35 Effect of stage loading on penetration
for polydisperse uranine aerosol 86
36 Effect of stage loading on penetration for
polydisperse oil aerosol 87
37 Effect of stage loading on penetration
for polydisperse oil aerosol 88
38 Effect of stage loading on penetration
for polydisperse salt aerosol 90
39 Effect of stage loading on measured size
distribution for polydisperse uranine aerosol
sampled onto spray silicone (U. of W.) 91
40 Effect of stage loading on measured size
distribution for polydisperse uranine aerosol
sampled onto glass fiber (U. of W.) 92
41 Effect of stage loading on measured size
distribution for polydisperse oil aerosol
sampled onto glass fiber (U. of W.) 93
42 Effect of stage loading on measured size
distribution for polydisperse oil aerosol
sampled onto spray silicone (U. of W.) 94
43 Total interstage loss versus particle diameter(U of W). 96
44 Interstage losses for a polydisperse uranine
aerosol at a total loading of 0.3 mg (U. of W.) 97
IX
-------
Figure
45 Interstage losses for a polydisperse uranine
aerosol at a total loading of 10.0 mg (U. of W.) 98
46 Effect of interstage losses on measured size
distribution of polydisperse uranine aerosol sampled
onto spray silicone at low stage loading (U. of W.)... 99
47 Effect of interstage losses on measured size
distribution for polydisperse uranine aerosol sampled
onto spray silicone at high stage loading (U. of W.)..100
48 Effect of interstage losses on measured size
distribution for polydisperse uranine aerosol sampled
onto glass fiber at low stage loading (U. of W.) 101
49 Effect of interstage losses on measured size
distribution for polydisperse uranine aerosol sampled
onto glass fiber at high stage loadings (U. of W.)....102
50 Interstage losses for the Andersen, Sierra and Brink..103
51 Effect of interstage losses on measured size
distribution for a polydisperse uranine
aerosol (Andersen) 104
52 Effect of interstage losses on measured size
distribution for a polydisperse uranine aerosol
(Sierra) 105
53 Effect of interstage losses on measured size
distribution for a polydisperse uranine
aerosol (Brink) 106
-------
TABLES
Number Page
1 Commercially Available In-Stack Cascade Impactors 2
2 Collection Surface Coatings Tested 4
3 Comparison of Measured Total Interstage Loss
for Impactors Tested 10
4 Experimental Studies of In-Ststck Impactors 21
5 Collection Surface Coatings Used with In-Stack
Cascade Impactors 23
6 MK III University of Washington Source Test
Cascade Impactor Critical Dimensions 29
7 MK III University of Washington Source Test Cascade
Impactor Measured Pressure Drop at Various Flow Rates.... 30
8 Reported Values of Dp _ for the MK III University of
Washington Source Test Cascade Impactor 32
9 Sierra Model 226 Source Cascade Impactor
Critical Dimensions 35
10 Sierra Model 226 Source Cascade Impactor
Measured Pressure Drop at Various Flow Rates 36
11 Reported Values of Dp _ for the Sierra Model 226
Source Cascade Impactor 37
12 Andersen MK III Stack Sampler Critical Dimensions 39
13 Andersen MK III Stack Sampler Measured
Pressure Drop at Various Flow Rates 40
14 Reported Values of Dp__ for the Andersen
MK III Stack Sampler..: 41
15 Modified Brink Model B Cascade Impactor
Critical Dimensions 44
xi
-------
Number
16 Modified Brink Model B Cascade Impactor
Measured Pressure Drop at Various Flow Rates 45
17 Reported Values of Dp,... for the Modified
Brink Model B Cascade Impactor 46
18 Test Aerosols 47
19 Errors Associated with Mass Measurement 53
20 Collection Surface Coating Weight Loss for 1-Hour
Exposures to Temperatures of 200°F and 500°F 55
21 Maximum Jet Velocities for Impactors Operated
at Design Flow Rate 56
22 Observed Tendency of Collection Surface Coating
to be Blown Off by an Impinging Jet of
Varying Velocity and Temperature 58
23 Percent of Total Collected Mass Per Stage
for a Polydisperse Uranine Aerosol 65
XII
-------
SYMBOLS
C Cunningham slip correction factor, dimensionless
C.E. collection efficiency
D, droplet diameter, cm
D particle diameter, cm
D particle diameter associated with 50% collection efficiency, cm
P50
1. stopping distance, cm
M. mass collected on Stage i, gm
AM mass collected on stage n, gm
mind mass mean diameter, cm
P atmospheric pressure, cm mercury
3
AP pressure drop at the dry gas meter, cm mercury
Q gas sampling rate, cfm
Re Reynolds number, dimeiisionless
STK Stokes number, dimensionless
STK,.,, Stokes number associated with 50% collection efficiency,
dimensionless
S/W ratio of the jet-tq-plate distance to the diameter or width of
the jet, dimensionless
T temperature of the aerosol stream, C
cL
T temperature of the aerosol stream at the dry gas meter, °C
m
T/W ratio of jet length to the diameter or width of the jet,
dimensionless
V ,. actual volume of aerosol stream sampled, cm
act
V volume of aerosol stream measured by the dry gas meter, cm
xiii
-------
V single component, of velocity, cm/sec
W width or diameter of the jet, cm
3
0 particle density, pi/cm
y viscosity of the media, poisa
6 geometric standard deviation, diniensionless
o
XIV
-------
SECTION 1
INTRODUCTION
Design of particulate control equipment necessitates a knowledge of the
chemical and physical nature of the aerosol stream, including the particle
size distribution present in the process gas stream at the point of control.
This requirement results because the effectiveness of any given physical
mechanism for particle removal is a function of particle size. Thus,
knowledge of the process stream particle size distribution allows for
optimum design or selection of particulate control equipment.
At present, size distribution measurements of process gas streams are
obtained mainly through the use of in-stack cascade impactors (1), a wide
variety of which are commercially available (Table 1). However, many
questions still exist concerning the most effective operation of these
devices for a given set of stack conditions. Earlier studies have provided
information as to the theoretical design, performance, and limitations of
cascade impactors operating at ambient conditions, and recent reports have
presented generalized guidelines for in-stack cascade impactor calibration,
testing procedures and data analysis (1-6).
The wide variety of process streams (temperature, relative humidity,
mass loading, stack gas velocity, type of particles, etc.) adds variables
whose effects upon the operation of in-stack impactors have not yet been
fully documented. This study was an attempt to evaluate systematically V
several operating parameters for four different models of in-stack impactors,
by using polydisperse test aerosols of various composition. The impactors
tested were a MK III University of Washington source test cascade impactor,
an Andersen MK III stack sampler, a Sierra Model 226 source cascade impactor,
and a modified Brink Model B cascade impactor. The parameters evaluated
included: 1) gas sampling rate (flow rate); 2) impactor collection surface
coating; 3) stage loading limits; 4) interstage losses; and 5) the effects
of these above parameters upon the "indicated or apparent" particle size
distribution of a polydisperse aerosol.
-------
TABLE 1. COMMERCIALLY AVAILABLE IN-STACK CASCADE IMPACTORS
1. University of Washington Source Test Cascade Impactor
Pollution Control Systems Corporation
Renton, Washington 98055
2. Andersen Stack Sampler
Andersen 2000, Inc
Atlanta, Georgia 30320
3. Brink Model B Cascade Impactor
Monsanto Enviro-Chem Systems, Inc
St. Louis, Missouri 63166
4. MR1 Model 1502 Cascade Impactor
Meteorology Research, Inc
Altadena, California 91001
5. Sierra Model 226 Cascade Impactor
Sierra Instruments, Inc
Carmel Valley, California 93924
6. Tag Sampler
Sierra Instruments, Inc
Carmel Valley, California 93924
-------
SECTION 2
DISCUSSION AND CONCLUSIONS
A. INTRODUCTION
This study has attempted to evaluate experimentally the effect of
several parameters upon the operation of in-stack cascade impactors,
including: type of collection surface, gas sampling rate (flow rate), stage
loading and interstage losses. It was accomplished by sampling polydisperse
aerosols of various composition and observing the effect of each of these
parameters upon the resulting measured size distribution. Results were
obtained based upon extensive testing using a MK III University of Washington
source test cascade impactor. Further testing utilized an Andersen MK III
stack sampler, a Sierra Model 226 source cascade impactor and a modified
Brink Model B cascade impactor. In addition, tests were conducted to
establish the suitability of various collection surface coatings to operation
at: (1) elevated temperatures and (2) high jet velocities.
B. COLLECTION SURFACE COATINGS
Of the collection surfaces tested (Table 2) only the uncoated aluminum
and glass fiber were suited to operation at temperatures of 500°F or greater.
At temperatures below 400°F the industrial spray silicone was the only
"grease-type" coating which had low, reproducible weight loss and showed no
tendency to be blown off the surface to which it was applied. This coating
is highly recommended when working temperatures permit its use.
When the uncoated aluminum, glass fiber or spray silicone collection
surfaces were used, results of this study show a true need for proper
preconditioning of the surface coatings. Several techniques for
preconditioning surfaces have been previously cited (4,7,8) and will not be
repeated here. With proper preconditioning, both the absolute weight loss
and the variability of the coating are minimized. This decreases the
collected mass required for given statistical accuracy of the sizing.
When glass fiber collection surfaces were used, the measured size
distribution had mmd's greater than distributions obtained when uncoated
aluminum or spray silicone were used. This effect was observed for various
types of aerosols (uranine, OOP, DNP) at various flow rates (0.25, 0.5, 1.0
cfm) and at various total loadings 5, 10, 15 and 30 mg). In all of the above
cases the difference between distributions was approximately 30%. It is not
known at this time if such a correction factor could be applied to aerosols
of different mmd and 6g. Results of sampling a polydisperse aerosol which
-------
TABLE 2. COLLECTION SURFACE COATINGS TESTED
1. White Petroleum Jelly
Andrew Lewis Distributing Corporation
Rochester, N.Y. 14604
2. High Vacuum Grease
Dow Corning Corporation
Midland, Michigan 48640
3. 200 Fluid
Dow Corning Corporation
Midland, Michigan 48640
4. Apeizon Grease T
James G. Biddle Company
Plymouth Meeting, Pennsylvania 19462
5. Industrial Spray Silicone
Hercules Packing Corporation
Alden, N.Y. 14004
6. Gelman Type A
Gelman Instrument Company
Ann Arbor, Michigan 48106
-------
should have primarily penetrated to the final filter suggest that the
observed difference be attributed at least in part to an increased collection
efficiency for submicron particles by the upper stages when the glass fiber
surfaces are used. It is believed that this increase in collection
efficiency can be attributed to a filtering effect produced by the
penetration of the boundary layer into the fibrous mat (9,10).
Several investigators have previously cited such differences when
sampling monodisperse aerosols and thus reported the use of a different set
of Dp50 cut-points when interpreting particle sizing data. From the data
obtained in this study, it appears that the differences between the
distributions obtained with spray silicone and glass fiber collection
surfaces could not be accounted for through the use of such experimentally
derived monodisperse calibrations. It was neither the purpose of this study
nor did'time allow for such calibrations to be conducted for the impactors
tested. A set of Dpso ciit-points experimentally derived for the Brink
impactor with both "grease" and glass fiber collection surfaces has been
reported by SRI (3), and thus allowed for the plotting of the sizing data
with these Dp50 cut-points. Although the impactor used in this study was
not the one calibrated at SRI, all modifications were the same, and testing
was carried out under similar conditions (flow rate, temperature, pressure).
Figure 1 shows the distributions obtained when plotting the sizing data using
the SRI cut-points calibrated for both glass fiber and "grease" coated
collection surfaces. The figure suggests that such a monodisperse
calibration does not account for the difference noted for size distributions
obtained with glass fiber collection surfaces.
It is recommended that further investigation of this effect be conducted
in an attempt to quantitatively account for the observed differences in
distributions for various mmd's and 6g. If an initial calibration is to be
used to account for the shift in the measured distribution, it is suggested
that a polydisperse calibration technique be developed. At present,
investigators should be aware that distributions obtained using glass fiber
collection surfaces may have mmd's 30% greater than those obtained using
grease coated collection surfaces.
C. FLOW RATE
Theoretically, the size distribution obtained from a given impactor
should not be affected by operation at various flow rates. There are,
however, practical limits within which an impactor will operate most
effectively. The lower limit is established at the flow rate for which
useful sizing data can no longer be obtained. For example, when operated at
0.125 cfm the Sierra has a stage 6 cut-point of approximately 1.5 ym.
Operation at this low flow rate provides minimal information for an aerosol
of mmd = 0.5 urn. Thus, an increased flow rate would be desired in order to
allow for collection of particles in this size range. The upper limit is
set at the flow rate for which particle bound and reentrainment becomes
significant, since at this point the size distribution will become skewed.
Within this range, experimental results showed little effect of the flow rate
on the measured distribution when a uranine aerosol was sampled. There was,
however, an apparent effect when an oil aerosol was sampled. A suitable
-------
99.9
99
o
CO
u
•I-1
T3
•H
K
JS
•Jl
o
(ft
«n
90
80
70
60
50
40
30
20
10
0.1
0.1
glass fiber
spray sllicone
•4-1
J_
1.0 10.0
Aerodynamic Diameter (ym)
100*. 0
Figure l. Effect of collection surface on measured size distribution for a polydisperse uranine aerosol
with SRI DP_- cut-points calibrated for spray and glass fiber collection surfaces (3).
(Modified Brink Model B cascade impactor, 0.03 cfm, 70°F, 30.06 "Hg, 15.0 rag total loading,
gravimetric analysis).
-------
explanation for the observed shift in 6g is not available at this time. The
effect cannot be described as a result of particle reentrainment.
Experimental results show that particle reentrainment is influenced by
not only jet velocity, but also the type of particle and collection surface.
Examination of the data reveals that for hygroscopic solids (uranine) and
liquids,particle bounce begins at jet velocities between 75-100 m/sec for
submicron particles collected onto spray coated or glass fiber collection
surfaces. This supports the findings of previous investigators (4,7). The
specification of submicron particles is made, since larger particles may
show a pronounced tendency to be reentrained at these high jet velocities,
resulting in lowered collection efficiency. If the cascade impactor is
operating in its normal configuration these larger particles will be removed
from the stream onto the stage with the appropriate jet velocity and not be
subjected to reentrainment.
Solid particles such as PSL spheres show a much more pronounced
tendency to reentrainment at lower velocities. This was observed at
velocities as low as 10 m/sec with uncoated aluminum surfaces. Spray
silicone and glass fiber surfaces both showed the same effect between 40-50
m/sec for the PSL spheres.
The relative humidity of the test aerosol streams was approximately 45%.
Previous studies have shown that at relative humidities of 75% or greater,
the available moisture may act as an adhesive, preventing reentrainment at
higher velocities or from uncoated surfaces. At the current test conditions,
results may be interpreted as the lower threshold for reentrainment related
problems. Increased humidity of the airstream may allow for increased
collection efficiency at higher velocities or upon uncoated surfaces.
D. STAGE LOADINGS
Experimental results have shown that increased collection efficiency
(decreased reentrainment) may be expected after initial loading of particles
on smooth collection surfaces; i.e., uncoated aluminum and spray silicone.
Similar results were reported by Rao (9). It is assumed that the initial
loading results in a roughened surface providing increased probability of
interception and thus an overall increased collection efficiency.
The opposite may be assumed for the glass fiber surface, which is
initially rough, and does in fact provide increased collection efficiency as
both these and previous results have shown (9,10). Increased loading may
tend to smooth over the surface, limiting the penetration of the impinging
airstream into the fibrous mat and thus gradually decreasing the collection
efficiency.
A practical upper limit for individual stage loadings appears to be 5-7
mg for hygroscopic solids. Uncoated aluminum, spray silicone and glass
fiber surfaces seemed to provide stable collection characteristics up to this
limit if other restrictions were met (i.e., velocity, etc.). Increased
loading resulted in excessive losses onto the backside of the nozzle plate
-------
and nonuniform deposition. The above limit on stage loading is slightly
lower than the previously recommended value of 10 mg (4). when sampling a
liquid oil aerosol, the uncoated aluminum surface is considered to be
unacceptable due to highly variable collection characteristics. At stage
loadings greater than 15-20 mg, the liquid deposits were observed to soak
through the glass fiber surface. This suggests that loading be maintained
below this amount. The spray silicone coating provides a stable collection
surface for liquid aerosols when loadings were kept below 10-15 mg.
E. INTERSTAGE LOSSES
Results of measuring the effect of particle diameter upon interstage
loss agree closely with those obtained by SRI (3) (Figure 2). Interpretation
of these results in predicting the interstage losses for a polydisperse
aerosol having a mmd of 0.5 urn suggest approximately 2% of the total
collected mass. The measured interstage losses for such an aerosol were
found to be approximately that, when the total mass loading was 0.3 mg.
However, upon increasing the total mass loading to 10.0 mg, a loading likely
to be obtained during field testing, the interstage losses were found to
increase to 10% of the total collected mass. It appears that a simple
correlation cannot be assumed between the measured losses for a given sized
particle, and those measured for a distribution having a similar mmd.
Of primary importance was the observation that exclusion of these losses
from the particle size distribution did not appear to significantly alter the
distribution. It was observed that losses occur primarily on the stages
collecting the greatest mass. It is therefore not known if distributions of
increased mmd will be affected in the same manner when excluding interstage
losses from the calculations. Past operations of the impactors in the field
have included painstaking efforts to recover all interstage losses. The
present results indicate that such procedures may not be necessary to obtain
a valid size distribution measurement. Because of the problems and errors in
recovering losses in a field test, it may be more accurate to neglect the
losses. It should be noted that if, however, the testing is to supply
information as to the total mass output of the source, it does become
necessary to recover these losses and include them as collected mass.
Present testing for interstage losses was only conducted at design flow
rates. However, there was an indication that these losses tended to increase
with increasing flow rates. It is not known if the high interstage losses
recorded for the Brink are characteristic of single jet impactors in general,
or due to its specific design. Similar consideration is true for the
rectangular jet Sierra impactor, which exhibited very low losses at the flow
rate tested. Interstage losses shown in Table 3 compare favorably to the
results obtained by SRI (3), with the Brink showing the greatest losses.
F. SUMMARY AND RECOMMENDATIONS
A guide to the use of current commercially available in-stack impactors
is found in Appendix A. Recommendations are based upon the results presented
-------
70
60
50
&
40
V)
O
00
5 30
0)
+J
20
10
O - Present Study
• - SRI (3)
• o
•°
o
1 I I 1 I III
0.5 1.0 10.0 20
Particle Diameter (ym)
Figure 2. Total interstage loss versus particle diameter for the
University of Washington MK III source test cascade impactor.
-------
TABLE 3. COMPARISON OF MEASURED TOTAL
INTERSTAGE LOSS FOR IMPACTORS TESTED
Total Interstage Loss*
Impactor (% total collected mass)
MK III University of Washington
Source Test Cascade Impactor
(uncoated aluminum) 20%
(spray silicone) 13%
(glass fiber) 12%
Sierra Model 226 Source Cascade
Impactor
(glass fiber) 4%
Andersen MK III Stack Sampler
(glass fiber) 9%
Modified Brink Model 6 Cascade
Impuctor
(glass fiber) 36%
(spray silicone) 31%
*Measurements for a polydisperse uranine aerosol of mind =056 \m at
total loadings of 15 mg.
10
-------
in this study and include information pertaining to: 1) the selection of a
suitable collection surface coating, 2) operational flow rate, 3) limits on
stage loadings, and 4) interstage losses. Investigators should be aware of
the discrepancy in the particle size distribution obtained when using glass
fiber collection surfaces. Further research is needed in order to correct
accurately for this observed difference. Visual examination of individual
collection surfaces can supply information to the operator of an impactor as
to problems of particle bounce (flow rate) and/or overloading of the stage.
Data obtained in this study indicate that interstage losses can be excluded
from particle sizing data without significantly affecting the measured size
distribution. It is recommended that testing be extended to include
distributions of larger mmd in an attempt to further justify exclusion of
interstage losses from sizing calculations. Recovered losses must, however,
be included if information concerning the mass output of the source is
desired. It is hoped that these findings will serve to allow for the more
effective operation of in-stack impactors and thus provide a more valid
instrument for assessing the particle size distributions of process gas
streams.
11
-------
SECTION 3
THEORETICAL BACKGROUND AND LITERATURE REVIEW
A. INTRODUCTION
The cascade irapactor as originated by May (11) classifies particles
according to their aerodynamic diameter. This distinction is made since
particles having the same physical size may behave differently in an
airstream due to differences in shape "and density. An irregular particle
can be represented by an equivalent unit density sphere which, when moving
through still air at low Reynolds numbers, will attain the same terminal
settling velocity as the irregular particle. The diameter of this spherical
particle is said to be the aerodynamic diameter of the actual particle under
consideration.
The basic design of a cascade impactor consists of a series of
alternating nozzles and impaction plates (Figure 3). In addition, an in-stack
impactor is equipped with an array of entrance nozzles to allow for isokinetic
sampling of the moving airstream, in order to minimize particulate sampling
bias.
If one considers the movement of a particle-laden airstream through a
typical impactor stage (Figure 4), it is seen that particles which have
gained the required inertia will cross air streamlines and impact onto the
collection plates. Particles with less inertia will not be impacted and will
be carried by the airstream to the following stage. By increasing the
velocity of the airstream in the succeeding stages, progressively smaller
aerodynamic diameter particles are impacted and an inertial classification
results. Ideally, for each stage all particles with an aerodynamic diameter
greater than a critical diameter would be collected, while those smaller
would not be collected.
Impaction efficiency as used in this report is defined as that fraction
of incident particles of a given aerodynamic size which strike an impaction
surface. This parameter can be calculated from theoretical considerations,
or measured under ideal experimental conditions (12). The term collection
efficiency is used to mean that fraction of the total aerosol mass actually
collected (retained) by the impactor collection surface. The difference
between the theoretical impaction efficiency and the experimental collection
efficiency is often quite large and can be attributed to nonideal flow fields
within the impactor and/or failure of the particles to adhere to the
impaction surface.
12
-------
Entrance Nozzle
Alternating Nozzles and
Collection Plates
Backup Filter
Figure 3. Basic design of an in-stack cascade impactor (MK III University of
Washington source test cascade impactor).
-------
Streamlines
r
T
L
A
/
/
/
/
/
1 1
t
!
1 I
;
//
T
1
\
I
\i
\v .
/>
s
\
s
/
«' . .
Jet
Jet Exit
^^X' ^ Jr* ^ S ^ ^ ^ ^ S S/S^ ^4*
Impact ion /
Plate — — J
\ Trajectory of
*""** Impacted Particle
/, Trajectory of Particle
Too Small to Impact
Figure 4 . Principle of operation of an impactot, showing
commonly referred to dimensions.
1.0
o
c
<0
•H
o
OJ
J_
Actual
Ideal
0 /sTic
Figure 5. Typical impactor efficiency curve.
14
-------
B. IMP ACTOR THEORY
In the theoretical analysis of inertial impaction, it is necessary to
describe the flow field and then calculate the trajectories of particles
within the flow field. Most earlier investigators assumed a simplified flow
field when calculating impactor efficiency (13-17) . By solving the Navier
Stokes equations for the impactor flow field and then calculating particle
motion within it, Marple (12) was able to obtain more exact solutions which
he compared to experimental results. This method of analysis enabled him to
account for the effects of several variables upon the overall performance of
an impactor including: 1) the ratio of the jet-to-plate distance to the
throat width (S/W) ; 2) the throat length to throat width ratio (T/W); 3) the
jet Reynolds number (Re); and 4) the shape of the nozzle (see Figure 4). The
corresponding effects of each of these parameters on both the flow field, and
subsequently the impaction efficiency is discussed later in this section.
A particle trajectory in a flow field is calculated by equating Newton's
second law with the fluid media resistance acting on the particle. While
following a fluid streamline, a particle can be displaced by a distance equal
to its stopping distance (Jlj) , defined by Fuchs (18) as the distance a
particle with initial velocity, V0, will travel in still air while that
component of velocity (Vo) is lost due to the resistance of the media. When
the fluid resistance force can be described by Stokes law, the stopping
distance can be defined as:
p V CD 2
- P ° P
i 18u
where :
&i = stopping distance (cm)
C = Cunningham slip correction factor (dimensionless)
Pp = particle density (gm/cm3)
V0 = particle velocity (cm/sec)
Dp = particle diameter (cm)
y = viscosity of the media (poise)
If one assumes a uniform particle concentration across the jet plane, a
given impactor geometry, and a constant flow rate, the impaction efficiency
becomes a function of the ratio of a particle's stopping distance to the
radius or half width of the nozzle; this is an inertial impaction parameter
commonly referred to as the Stokes number (STK) .
The Stokes number is then described as:
p V CD 2
P°P r-TX
STK = — - r- (2)
18y W/2
15
-------
where:
W = width for rectangular nozzle
= diameter for circular nozzle
The square root of the Stokes number (/STK) is a dimensionless particle size.
The impactor stage can then be characterized bya plot of impaction efficiency
versus the square root of the Stokes number (/STK) (Figure 5) .
As mentioned earlier, deviation from ideal impactor behavior results when
there is a nonideal flow field. Marple (12) stated that when the velocity
profile is uniform across the jet exit, ideal impactor behavior can be
obtained. However, the presence of a boundary layer at the walls of a nozzle
prevents this criteria from being perfectly met and results in an actual
impaction efficiency curve similar to the one shown in Figure 5.
Marple (8) investigated the effects of impactor geometry upon impaction
efficiency curves for both round and rectangular nozzles by varying the three
dimensionless parameters: S/W, T/W and Re, (Figure 6a,b,c) and has
characterized these effects as stated below. Inspection of Figure 6b shows
that except for low Reynolds numbers (Re < 500) and very high Reynolds numbers
(Re > 25000), the shapes of the impactor efficiency curves are very similar.
In the case of the low Reynolds numbers, the increased viscous boundary layer
at the nozzle walls results in the poorer cut-offs. For Re = 25000 the knee
in the curve appears to be due to the presence of a thin boundary layer in the
area of the stagnation point at the impaction surface. The effect is to allow
smaller particles to come in contact with the surface relative to areas where
the boundary layer is thicker.
The effect of varying the S/W ratio is to alter the velocity profile at
the jet exit. This is minimal for round nozzles with S/W ratio values
greater than one-half and rectangular nozzles at values greater than one. For
values of S/W smaller than those mentioned above it is seen that the position
of the efficiency curve shifts towards smaller values of /STK, while the
general shape of the curve remains the same (Figure 6a).
Variations in T/W over a range of values from 1 to 10 are seen to have
little effect on either the shape of the efficiency curve or the position
along the /STK axis (Figure 6c).
The above considerations have been applied to a variety of impactor
designs. Because of the inherent design of a rectangular nozzle, there are
definite end effects. Ends of the rectangular nozzles act (more or less) as
individual round nozzles, giving rise to variations in collection efficiency
along the length of the nozzle. This effect can be minimized by limiting
radial flow at the ends of the nozzle and by having a large aspect ratio
(length-to-width ratio). Whether an impactor utilizes rectangular or round
nozzles, if it has multiple nozzles it must be assumed that the velocity
profile through each nozzle is identical.
16
-------
100
dG
-.60
£
u.
"20
3 0.1 0.2 0.3 04 OS 0.6 07 06 09 10 I.I
JSTK
la) tFFECT OF j£T TO PLATE DISTANCE (R«-3.000I
12
too
01 02 0.3 0.4 OS 0.6 07 Ob 0.9 tO I.I
Ibi EFFECT OF JET REYNOLDS NUMbEH { T/W-II
0 0.1 0.2 03 0.4 09 Ob 0.7 0.8 0.9 1.0 U 12
-ffK
(C) EFFEtl OF THROAT LENGTH (H«»i.OOO)
Figure 6. Impactor efficiency curves for the rectangular (broken
line) and the round (solid line) impactors showing the
effect of jet-to-plate distance, Reynolds number and
throat length. (8)
17
-------
Size distributions as measured by an impactor are, for the most part,
based upon the Dp50 method of data reduction. This method assumes an ideal,
sharp cut point about the 50% efficiency, with all significant quantities of
particles having aerodynamic diameters greater than the critical aerodynamic
diameter assumed collected on that stage. The aerodynamic diameter
corresponding to the 50% efficiency (Dp50) may be calculated based upon theory,
or obtained through an experimental calibration.
Theoretical calibration of Dp50 involves assuming a 50% efficiency value
for the Stokes number (STK50) (see Equation 2). The STK50 value is based upon
considerations of the geometric design parameters of the impactor stage, as
previously discussed. Thus, for a cascade impactor whose geometry differs
from stage to stage, the STK50 value would also vary (2,6). Agreement between
theoretical and experimental values of Dp50 would only be expected when the
impactor flow field approaches ideal conditions.
Graphical presentation of impactor data may be based on either a
differential or a cumulative particle size distribution. The differential
particle size distribution is customarily plotted on either log-log or
semi-log paper with AM/A(log Dp) as the ordinate and log Dp as the abscissa.
The mass collected on stage n is designated by AM and represents the aerosol
mass with diameters between (Dp50)n and (Dp50)n_i. The term A(log Dp) is then
defined as log (Dp50)n-l - log (Dpso)n- The result is a histogram which is
frequently represented by a best fit smooth curve. The cumulative particle
size distribution is obtained by summing the stage mass catches plus the final
filter and plotting the fraction of mass less than a given size versus
particle size (Dp50). The plot is made on log-normal probability paper and
results in a straight line when the distribution is of a single mode and
log-normally distributed. Figure 7 shows the same particle sizing data
presented on both the differential and cumulative basis described above. In
both cases the resulting distributions (since they are approximately log
normal) may be characterized by a mass median diameter (mmd) and a geometric
standard deviation (6g).
The nonideal characteristics of an impactor necessarily result in a
cross-sensitivity, which means that particles of a given size may be collected
on more than one stage. Attempts have been made to characterize this
cross-sensitivity based upon mathematical models (19-21). Natusch and Wallace,
however, concluded that only when a large weight fraction of the collected
mass is not made of particles close in size to the mmd do there occur any
significant errors and that the mmd and 6g are valid parameters for
describing particle size distributions (21).
C. PARTICLE ADHESION
The previous assessment of impaction theory has assumed that once a
particle comes in contact with the impaction surface, it adheres to it.
However, the greater the velocity of the incident particle, the greater the
probability that the particle will not adhere. Theoretical estimates of the
critical velocity at which a given size particle will reentrain rather than
adhere have been made by Jordan (22) and Dahneke (23). The important forces
18
-------
30
w>
o
20
10
0
0.1
1.0
10.0
100.0
Aerodynamic Diameter (ym)
a. Differential particle size distribution plot.
99.9
99
g 90
o
SH
(U
CX
« 50
>
•H
•!->
ea
•3 10
§
CJ
0
> 11»1
0.1 1.0 10.0 100.0
Aerodynamic Diameter (pm)
b. Cumulative particle size distribution plot.
Figure 7. Particle sizing data presented as differential and
cumulative plots.
19
-------
in the adhesion of the particle to the collection surface include: 1)
intermolecular forces between the particle and surface (Van der Waal forces
2) electrostatic forces; and 3) capillary forces in liquid bridges. In an
investigation of the adhesion of particles, Loffler (24) calculated that a
10 urn diameter particle with 1000 elementary charge units would still have a
Van der Waal force 100 times greater than the electrostatic force. Thus,
Van der Waal forces are assumed to be the primary adhering force between
particle and surface.
To prevent bounce one must either decrease the velocity of the incident
particles or increase the adhesive forces between the particle and collection
surface. A decrease in stage velocity decreases the collection
characteristics of the impactor; therefore, the desired alternative is to
increase the adhesive forces between the particles and collection surface.
In most cases this can be effectively accomplished by coating the collection
surface with a grease or some other nonvolatile viscous substrate. While the
coating may serve as nothing more than a cushion (24) for absorbing the
kinetic energy of the particle, Buchhloz (25) and Berner (26) feel that the
increased area of contact caused by the deformation of the coating does, in
fact, increase the adhesion force. As a monolayer of particles accumulate
upon the coating, the initial collection characteristics of the coating may
be lost unless the coating substrate sufficiently wets the particles to
achieve a coating of the newly formed surface (25). Excessive loading of the
collection surface can lead to drastic losses in collection efficiency (27) .
Past studies indicate that when loadings exceed that of a monolayer deposit,
reentrainment of the particles may result due to fluid drag forces (23,28).
In a study of atmospheric aerosols, Winkler (29) determined that for relative
humidities greater than 75%, high collection efficiencies were maintained
without the use of an adhesive coating. The collection efficiency of
uncoated surfaces was found to decrease rapidly as the relative humidity
fell below 75%. For a more complete review of adhesive coatings and nonideal
collection characteristics of impactors, the reader is referred to Rao (9).
D. PREVIOUS IN-STACK IMPACTOR STUDIES
There are a large number of published papers concerning the use of
impactors, and for a survey of basic impactor studies the reader is referred
to the earlier works of Marple (12) and Rao (9). This review will consider
studies of in-stack cascade impactors, a summary of which is presented in
Table 4. This listing indicates the impactor model, the specific studies
undertaken, and the type of test aerosols, if any, used.
As noted, several studies have compared the theoretical stage cut points
of various commercial in-stack impactors with those obtained experimentally
using a monodisperse test aerosol (2,3,6). Air Pollution Technology (A.P.T.),
Incorporated has published a set of cascade impactor calibration guidelines
which describes a calibration method involving atomization of polystyrene
latex (PSL) spheres and particle detection utilizing an optical particle
counter (2). Other methods of calibration include gravimetric or
fluorometric analysis of stage catches after sampling of a monodisperse test
aerosol (3,6). One important consideration pointed out by Willeke and
20
-------
TABLE 4. EXPERIMENTAL STUDIES OF IN-STACK IMPACTORS
Reference
Studies Conducted
Impactor Model
SRI (5)
SRI ( 8 )
SRI ( 7 )
A.P.T., Inc (2)
SRI (3)
SRI (31)
Impactor Models:
1
a)
b)
c)
d)
e)
5,6,7 a-,b,d,e
4
4,7,3 a,b,d
1,5 a,f
,2,3,4 a,b,c,d,g
4
University of Washington
Andersen Stack Sampler
Sierra Model 226 Cascade Impactor
Brink Model B Cascade Impactor
TAG sampler
Studies Conducted:
f) A.P.T. M-l Cascade Impactor
g) MRI Model 1502 Cascade Impactor
1) calibration against test aerosol
2) interstage losses
3) range of flow rates
4) collection surface coatings
5) data reduction
6) stage loadings
7) field tests
Aerosol Used
Flyash
Ammonium Fluorescein,
DOP
PSL Spheres
Ammonium Fluorescein,
DOP
-------
McFeters is that, in general, a single stage calibration will not give the
stage characteristics of the completely assembled impactor due to the
corresponding variations in the internal flow field (30). Thus, it is
suggested that a calibration be conducted with the impactor assembled as it
will be used in the field.
D. B. Harris has presented a set of nomographs which allows for the
selection of isokinetic sampling conditions and sampling time based on the
process stream mass loading (4). Included on the nomograph are the maximum
recommended flow rates for several commercially available impactors ranging
from 0.03 scfm for the Brink to 0.8 scfm for the University of Washington
MK III. The upper limit for flow rate was based upon the maximum jet velocity
at which deposition may be obtained without bounce or reentrainment. Studies
at Southern Research Institute (SRI) suggest that jet velocities should be
maintained below 65 m/sec for greased collection plates and below 35 m/sec for
ungreased plates. Optimum results are reported at jet velocities of less than
10 m/sec (4,10).
Minimum single stage loadings are limited by the weighing accuracy
attainable. Increased weighing accuracy may be gained by the use of light-
weight collection plate inserts which reduce tare weights. Mass measurements
with an accuracy of 0.01 mg or less are possible with several currently
available microbalances. Maximum single stage loadings of 10 mg are
suggested to avoid excessive reentrainment or loss of collection efficiency
(3,4). Pre-cutters (cyclones) are often used upstream of the first impactor
stage to avoid overloading the upper stages, yet allow sufficient sampling
volumes to collect enough mass on the lower stages to maintain the desired
weighing accuracy. *
Suitability of the impactor collection plate surface to the sampling
conditions is extremely important. Table 5 lists various surface coatings
which have been suggested for increasing the collection efficiency of in-stack
impactors. Previous studies indicate that collection efficiency is dependent
upon the type of surface coating used (30,31). Several methods of
preconditioning these surfaces have been suggested to prevent weight loss due
to exposure to the process gas stream (3,4,7).
SRI has reported that various makes of glass fiber filters may gain
weight when sampling a process stream with high SOa concentrations due to
sulfate formation on the filter (3,7,8). Minimization of this effect can be
accomplished through proper conditioning or choice of filter. At present,
glass fiber filters are the only substrates commonly used at temperatures
greater than 400°F (3,4).
SRI has also assessed the interstage losses for five different in-stack
impactors (3). Their results show that for 15 urn particles interstage losses
can amount to as much as 70% of the total stage catch. These losses are due
mainly to deposition in the stage nozzles, and decrease to a minimum for
2 urn diameter particles.
In field work it is common practice to include any recovered stage
losses with the catch for that stage (3,4). Results are then normally
22
-------
TABLE 5. COLLECTION SURFACE COATINGS USED WITH
IN-STACK CASCADE IMPACTORS
Reference
1. SRI (3)
2. APT» Inc (2)
3. SRI ( 7)
4. Harris (4)
5. SRI (31)
Collection Surface Coatings
Teflon, Whatman GF/A, Whatman GF/D,
Reeve Angel 934 AH, Gelman Type A,
Gelman Spectro Grade Type A, MSA 1106 BH,
Ree^e Angel 900 AF, Dow Molykote III
Compound
Dow Corning High Vacuum Silicone Grease
Dow Corning High Vacuum Silicone Grease,
K-Y Jelly, Vaseline
Polyethylene Glycol 600, Apiezon L,
Apiezon H
Vaseline, Molykote III Compound,
Stopcock Grease, Dow Corning II
Compound, Polyethylene Glycol 600,
High Vacuum Grease, UCW:98, Apiezon L,
Kiln Bearing Grease, 200 Fluid,
Fluorolube GR-362, Fluorolube GR-544,
Fluorolube GR-470, Fluorolube GR-290,
Fluorolube GR-660, Carbowax 1000,
Carbowax 4000, Carbowax 20M, STP Oil
Treatment, Reeve Angel 934 AH,
Gelman AE, Gelman Spectro Glass,
Whatman GF/A
23
-------
reported with the Dpso method of data reduction previously described,
preferably based upon calibrated cut-points. The resulting particle size
distribution can then be defined by the mass median diameter and the
geometrical standard deviation, assuming the distribution is log-normal.
A more involved method of data reduction has been proposed by Picknett
(32). This method calculates the mass contributions of a set of particle size
intervals which would result in the actual collection efficiencies measured
and is based upon a previous knowledge of the stage collection efficiencies
for the chosen size intervals. A comparison of the Picknett and Dpso methods
for a test aerosol size distribution show that while the indicated mmd's are
approximately the same, the Picknett method gives a better approximation of
the true distribution (4) .
SRI has also conducted a number of field tests with various models of
in-stack impactors (5,7). Results show that the mass loadings obtained from
cascade impactors commonly differ as much as three times that of EPA Method 5
loadings.
24
-------
SECTION 4
EXPERIMENTAL APPARATUS, METHODS AND PROCEDURES
A. INTRODUCTION
Initial information was gained concerning the effects of various
operating parameters for in-stack cascade impactors based upon extensive use
of the MK III University of Washington source test cascade impactor. After
establishing a general test procedure with the MK III, other commercially
available in-stack cascade impactor units were tested. They were: a Sierra
Model 226 source cascade impactor; an Andersen Mark III stack sampler; and a
modified Brink Model B cascade impactor. Basic differences between models are
the impactor nozzle design and the unit's compatability with alternative
collection surfaces. Each individual device is described in later sections.
B. GENERAL EXPERIMENTAL SETUP
The basic experimental arrangement involved equipment for generating a
reproducible (monodisperse or polydisperse) test aerosol of a variety of
materials; means of determining collection efficiency and interstage losses;
pumps, valves and metering devices for controlling and measuring flow; a
variable temperature environment; and an air supply. The general arrangement
of this test equipment is shown in Figure 8. The building compressed air
system was filtered and used to operate the aerosol generator and as a source
of dilution air. Monodisperse test aerosols were generated with a vibrating
orifice type aerosol generator, similar to the one described by Berglund and
Liu (33), while polydisperse test aerosols were generated with a three jet
Collison atomizer (34). The conditioned, neutralized and diluted aerosol
stream was then sampled using an in-stack impactor assembly operated at a
predetermined set of conditions (flow rate, impactor collection surface
coating and sampling time). Flow rate was monitored by rotameters and the
sampled volume measured with dry gas meters. A parallel control filter was
tun to determine the mass output of the aerosol generator and serve as a
comparison for the total mass collected by the impactor. An oven was used to
run impactor tests at elevated temperatures. Collection efficiencies were
determined by gravimetric or fluorescent techniques. Major components are
further discussed in the following section.
25
-------
Compressed
Air
Dilution Air
NJ
Ionizer
Collison
Atomizer
Conditioning
Chamber
Manometer
Absolute
Filter
In-stack
Impactor
Manometer
By-Pass
Rotameters
Pumps
By-Pass
Figure 8. General arrangement of test equipment.
-------
C. DESCRIPTION OF THE IN-STACK IMPACTORS
1 • MK III University of Washington Source Test Cascade Impactor (Model D)
The MK III University of Washington source test cascade impactor is a
round jet impactor consisting of seven stages plus a final filter. It is
designed to be inserted directly into the ducting or stack to obtain particle
size distributions. Construction is of machined stainless steel throughout,
with both the alternating nozzles and collection plates and the final filter
assembly enclosed by a threaded cylindrical casing (Figure 9). A variety of
isokinetic sampling nozzles are provided which thread into the inlet section.
All of the isokinetic nozzles taper to form the first stage nozzles of
approximately 1.865 cm (Table 6). Viton o-rings are fitted about each of the
nozzle stages and the final filter collar, as well as onto the threaded outlet
section to prevent leakage. These o-rings are recommended for use at
temperatures up to 500°F. For operating temperatures in excess of 500°F the
manufacturer suggests that the o-rings be removed and an extra collection
plate be used as a spacer after stage seven.
The collection plates accommodate aluminum foil, stainless steel foil,
polymer films and glass fiber collection surfaces which are easy to cut due to
the simple design of the collection plates. The final filter assembly accepts
a 47 mm glass fiber filter which is supported by a mesh screen and perforated
plates, and held in place by a collar.
Table 6 compares the reported geometrical dimensions of each of the
seven stages with the measured values of the impactor purchased for this
study. The reported ratios of jet-to-plate/jet diameter (S/W) varies from
0.78 to 12.50, while the jet depth/jet diameter (T/W) varies from 0.55 to 4.03
for the seven stages of the impactor. The measured pressure drop through the
complete impactor is presented for flow rates of 0.25, 0.5 and 1.0 scfm in
Table 7.
The manufacturer supplies a nomograph which relates calculated Dpso
cut-points for each impactor stage to volumetric flow rates over a range from
0.25 to 2.5 scfm for three given temperatures of the process gas stream. It
is stated that the particle density for the calculated values is assumed to be
1.0 g/cm3; however, further basis for the calculations are not presented
(Figure 10). Table 8 lists other published Dp50 values, both calibrated and
calculated, for various flow rates. The basis for the calculated values is
seldom given.
2. Sierra Model 226 Source Cascade Impactor
The Sierra Model 226 cascade impactor is an in-stack, multi-stage cascade
impactor of radial slot design. It consists of six common nozzle-collection
plates and a final filter assembly (Figure 11). A selection of isokinetic
nozzles can be attached to a cylindrical inlet section with a 5/8 inch
Swaglock fitting; this inlet section and the six nozzle-collection plates are
held together by means of two studs and thumb nuts protruding from the final
filter stage. The backside of the final filter stage has 1/2"NPT pipe threads
which accommodate the sampling probe. All construction is of stainless steel
27
-------
Figure 9 . MK III University of
Washington source
test cascade impactor,
Complete unit and
individual nozzle
and impaction plates.
-------
TABLE 6. MK III UNIVERSITY OF WASHINGTON SOURCE TEST CASCADE IMPACTOR CRITICAL DIMENSIONS
Stage No. No. of Jets Jet Width (in.)
Jet-to-Plate
Distance (in.) S/W T/W
Reynolds Number
@ 0.5 cfm
IT n "isft
•!• 1 \J . t 1OU
(0.7340)
2 A n ^ocn
(0.2240)
3 12 0.0960
(0.0910)
4 90 0.0310
(0.0292)
5 110 0.0200
(0.0200)
6 110 0.0135
(0.0133)
7 90 0.0100
(0.0099)
OCf.
* DO
(0.568)
Oocc
. zoo
(0.226)
0.125
(0.125)
0.125
(0.127)
0.125
(0.126)
0.125
(0.123)
0.125
(0.125)
07ft
(0.77)
1 80
(1.01)
1.97
(1.37)
4.03
(4.35)
6.25
(6.30)
9.26
(9.25)
12.50
(12.63)
2f\Q
(1.70)
1 f\(\
JL . OU
(0.58)
1.97
(1.38)
4.03
(4.32)
3.15
(3.03)
2.22
(1.92)
3.00
(2.32)
(1070)
(584)
(719)
(299)
(357)
(536)
(883)
Note: Manufacturer's reported values listed with measured values in parentheses.
-------
TABLE 7. MK III UNIVERSITY OF WASHINGTON SOURCE
TEST CASCADE IMPACTOR MEASURED PRESSURE DROP
AT VARIOUS FLOW RATES
Flow rate (scfm) Pressure Drop ("Hg)
1.0 5.6"
0.5 1.6"
0.25 0.45"
30
-------
Particle density a 1.0 gm/cm
2.50
4 6 6 7 B 9 10°
FLOW RflTE.Q.(CUBIC FEET /f1IN.)
Figure 10. MK III University of Washington source test cascade impactor
nomograph for determining D cut-points (supplied by the
manufacturer). P50
31
-------
TABLE 8. REPORTED VALUES OF Dps^ FOR THE MK III
UNIVERSITY OF WASHINGTON SOURCE TEST CASCADE
IMPACTOR OPERATED AT 0.5 CFM
References
Stage #
1
2
3
4
5
6
7
*
SRI (3>
14.0
11.2
4.1
1.86
1.57
0.67
...
+
SRI (3)
10.5
10.0
4.4
1.86
1.60
0.66
0.30
— **
APT (2)
1.85
1.22
0.60
0.35
H
Present Study
28.3
11.5
4.2
2.1
1.3
0.65
0.35
experimental values, exclude losses
experimental values, include losses
**
experimental values
+4-
calculated values based upon a *^TK__ value of 0.44 obtained based
upon the impactor's critical dimensions and Marple's curves (12).
32
-------
and the manufacturer claims the unit is capable of operation in corrosive
environments at temperatures of up to 1500°F.
Slotted glass fiber collection surfaces are sandwiched between stages and
double as pressure seals. For operation at temperatures in excess of 500°F,
stainless steel collection plates are available or the impactor stage itself
may be used. The final filter assembly accepts a 47 mm glass fiber filter
supported by a screen and held in place by the collection plate for stage 6.
Table 9 compares the measured geometrical dimensions of the Sierra Model
226 obtained in this study with the values reported by the manufacturer.
Pressure drop across the complete impactor is reported by the manufacturer as
about 33 inches of water at 0.75 cfm, 25°C and 760 mm Hg, and measured values
are listed in Table 10.
The reported Dp50 for unit density spheres at 0.75 scfm are given in
Table 11 along with other published cut-points. The impactor is designed to
operate at flow rates below 0.5 scfm.
3. Andersen MK III Stack Sampler
The Andersen MK III stack sampler consists of eight stages plus a final
filter. Each stage has a number of concentric round nozzles, offset on each
succeeding stage such that the one plate serves both as nozzle and collection
surface. The impactor stages accept precut glass fiber collection surface
coatings which are held in place by cross bars. Stainless steel rings serve
both as spacers between stages and as gaskets. The assembled stages and final
filter are held in a plate holder which is enclosed in a stainless steel
housing with a 1/2 inch female pipe fitting to allow attachment to the
sampling probe (Figure 12). Both straight and "gooseneck" isokinetic sampling
nozzles are available to allow for operation with a variety of stacks. With
stainless steel plates the unit is reported to operate at temperatures up to
1500°F under corrosive conditions.
Table 12 lists the measured dimensions of the unit used in this study and
compares it to manufacturer's values. The measured pressure drops across the
unit at flow rates of 0.25 cfm, 0.5 cfm and 1.0 cfm are reported in Table 13.
The manufacturer provides a nomograph and set of correction factor figures for
obtaining the stage cut-points (Dp50), reportedly based upon calibration with
unit density spheres. These and other reported values of Dp50 are listed in
Table 14.
4. Modified Brink Model B Cascade Impactor
The Brink impactor is designed for operation at low flow rates (0.04 to
0.07 cfm) and is therefore well suited for sampling gas streams with high mass
loadings. Each stage has a single round nozzle and separate collection
surface. The impactor obtained for this study had been previously modified to
include both a 0 and 6th stage and a final filter, providing a total of seven
impactor stages. The unit is shown in Figure 13. The collection surfaces are
suited for use with either foil or glass fiber inserts.
33
-------
Figure 11. Sierra Model 226
source cascade
impactur. Complete
unit and combination
nozzle impaction plate.
5 I
-------
TABLE 9 . SIERRA MODEL 226 SOURCE CASCADE IMPACTOR CRITICAL DIMENSIONS
Jet-to-Plate Reynolds Number
Stage No. No. of Jets Jet Width (in.) Distance (in.) S/W T/W @ 0.25 cfm
tn
1
2
3
4
5
6
4
4
4
4
4
4
0.141
(0.137)
0.0782
(0.074)
0.045
(0.042)
0.025
(0.024)
0.014
(0.012)
0.010
(0.011)
0.250
(0.252)
0.125
(0.124)
0.094
(0.103)
0.094
(0.105)
0.094
(0.104)
0.094
(0.104)
1.78
(1.8)
1.60
(1.7)
2.08
(2.5)
3.75
(4.4)
6.67
(8.7)
9.40
(9.5)
2.59
(2.7)
3.04
(3.4)
5.30
(5.7)
2.04
(2.3)
3.64
(4.25)
5.10
(5.0)
(308)
(308)
(409)
(412)
(404)
(685)
Note: Manufacturer's reported values listed with measured values in parentheses.
-------
TABLE 10. SIERRA MODEL 226 SOURCE CASCADE IMPACTOR
MEASURED PRESSURE DROP AT VARIOUS FLOW RATES
Plow Rate (scfm) Pressure Drop ("Hg)
1.0 5.3"
0.5 2.25"
0.25 0.95"
0.125 0.3"
36
-------
TABLE 11. REPORTED VALUES OF DPso FOR THE SIERRA
MODEL 226 SOURCE CASCADE IMPACTOR
Reference
++
Stage #
1
2
3
4
5
6
SRI (3)
18.0
11.0
4.4
2.65
1.70
0.95
SRI (3)*
14.5
12.0
4.2
2.55
1.65
0.95
**
Sierra Instruments
16.0
8.6
3.9
2.4
1.2
0.61
Present S
20.3
11.0
5.25
3.0
1.5
1.0
**
experimental values, exclude wall losses at 0.25 cfm
experimental values, include wall losses at 0.25 cfm
reported values at 0.75 scfm
calculated values at 0.25 cfm based upon a /STK5Q value of 0.7
obtained based upon impactorti critical dimensions and Marple's
curves (12).
37
-------
Figure 12. Andersen MK III stack
sampler. Complete
unit and combination
nozzle impaction
plate.
-------
CM
TABLE 12 . ANDERSEN MK III STACK SAMPLER CRITICAL DIMENSIONS
Jet-to-Plate Reynolds Number
Stage No. No. of Jets Jet Width (in.) Distance (in.) S/W T/W @ 0.5 cfm
2
3
4
5
6
7
8
264
264
264
264
264
264
156
OAAC
(0.057)
Of\A Q
(0.044)
Ofi^7
(0.036)
(0.0292)
On^i
(0.021)
0 0146
(0.0135)
0 0107
(0.0095)
0 0096
(0.0095)
01 nj
. 1UZ
(0.101)
01 no
(0.100)
01 fi9
(0.101)
01O?
• XUfe
(0.102)
0 102
(0.101)
0 102
(0.099)
0 102
(0.100)
0 102
(0.100)
1 57
(1.75)
2 no
• uo
(2.27)
2 76
(2.81)
^1 46
(3.49)
4 86
(4.81)
(7.3)
Q "^
(10.5)
10 6
(10.5)
(1.09)
(0.82)
(0.97)
(1.16)
(1.71)
(1.41)
(1.89)
(1.89)
(52)
(68)
(83)
(102)
(142)
(221)
(313)
(530)
Note: Manufacturer's reported values listed with measured values in parentheses.
-------
TABLE 13. ANDERSEN MK III STACK
SAMPLER MEASURED PRESSURE DROP
AT VARIOUS FLOW RATES
Flow Rate (scfm) Pressure Drop ("Hg)
1.0 3.55"
0.5 1.15"
0.25 0.4"
40
-------
TABLE 14. REPORTED VALUES OF Dp50 FOR THE
ANDERSEN MK III STACK SAMPLER
References
Stage ff
1
2
3
4
5
6
7
8
*
SRI (3)
14.0
10.4
6.1
4.0
2.35
1.20
0.76
0.46
i
SRI (3)
10.5
9.4
5.8
4.4
2.20
1.20
0.70
0.43
** J--
Andersen 2000, Inc
13.0
8.6
5.7
4.0
2.55
1.25
0.80
0.54
Present Study
13.4
8.8
5.5
3.8
2.15
1.24
0.75
0.47
experimental values, exclude losses at 0.5 cfm
+experimental Values, include losses at 0.5 cfm
**
manufacturers reported values from nomograph at 0.5 cfm
calculated values at 0.5 cfm based upon a /STK5Q value of 0.52
obtained based upon the impactorfe critical dimensions and Marple's
curves (12).
41
-------
Figure 13 . Modi fied Brink
Mode] B cascade
impactor. Complete
unit and individual
nozzle and irapaction
plate.
42
-------
The impactor's measured dimensions are listed in Table 15 and are
compared with the manufacturer's values. Measured pressure drops through the
unit for flow rates of 0.025., 0.05 and 0.1 cfm are reported in Table 16.
Table 17 lists the various Dp50 values found in the literature and those used
for this study.
D. IMPACTOR COLLECTION SURFACE COATINGS
A variety of collection surface coatings (Table 2) were assessed as to:
1) their suitability for use at various operating temperatures and flow rates,
and 2) their collection efficiency for several types of aerosols. Although
the commercial impactors did not necessarily come with either aluminum and/or
glass fiber collection plate inserts, it was entirely feasible to make them
for use with most of the impactors tested.
Before applying any coating substrate to the aluminum inserts, they were
first rinsed in chloroform to remove any residual coatings, then washed in
warm soapy water, rinsed in distilled water, and allowed to dry. Several
application methods were tried for the Dow Corning 200 fluid, all involving a
volatile solvent to make possible a thin coating of the highly viscous
substrate. Applications of both the 1% and 10% (by weight) solutions in
benzene resulted in a nonuniform coating upon drying. One ml of a solution of
1% Dow Corning 200 fluid in hexane resulted in a uniform coating of
approximately 2-4 um thickness (calculated from coating weight). The
resulting lowered viscosity made it possible to spread the substrate easily.
Solutions of 10% in hexane were much more viscous, making it impossible to
achieve uniform 2-4 um thickness coatings. However, for applications of
coatings of 10-20 um thickness it was found that the 10% solution was easier
to work with due to the excessive amount of 1% solution required to achieve
the desired thickness. It is best to keep the volatile fraction to a minimum
since incomplete drying could lead to weight loss during testing.
Application of the Dow Corning High Vacuum grease, the Apiezon grease and
the White Petroleum jelly was accomplished by initially spreading the grease
manually onto the aluminum surface, followed by wiping it smooth with a
tissue. After several trials the proper amount for the desired thickness
could be easily duplicated, making the application a simple matter.
The high viscosity industrial silicone spray was sprayed directly onto
the aluminum inserts. Three passes with the spray at arm's length resulted in
a 3-5 um thickness coating (calculated from coating weight). This method of
application was very easy and reproducible. Initially the surface appeared
uneven; however, within five minutes the surface coating obtained a uniform
thickness.
E. AEROSOL GENERATION
1. Description of Test Aerosols
Laboratory test aerosols generated for use in this study (Table 18) were
broadly classified as to three basic types, including: 1) liquid aerosols,
43
-------
TABLE 15 . MODIFIED BRINK MODEL B CASCADE IMPACTOR CRITICAL DIMENSIONS
Jst-to-Plate Reynolds Number
Stage No. No. of Jets Jet Diameter (in.) Distance (in.) S/W @ .03 cfm
0
1
2
3
4
c
1
(0.134)
1 n OQK
(0.091)
i n n?o
(0.065)
1 o o1*1;
(0.050)
1 n 0^7
X V . vO I
(0.033)
i n O?Q
A V . U£ ?
(0.028)
i
(0.021)
(0.395)
(0.299)
(0.200)
(0.168)
(0.103)
(0.079)
(0.078)
(2.95)
(3.29)
(3.08)
(3.36)
(3.12)
(2.82)
(3.71)
(353)
(483)
(728)
(860)
(1279)
(1632)
(2253)
Note: Manufacturer's reported values listed with measured values in parentheses.
-------
TABLE 16. MODIFIED BRINK MODEL B CASCADE IMP ACTOR
MEASURED PRESSURE DROP AT VARIOUS FLOW RATES
na. Rate fscfm) Pressure Drop ("Hg)
0.1 17'4"
0.05 2'2"
0.025 °-9"
45
-------
TABLE 17. REPORTED VALUES OF Dp50 FOR THE
MODIFIED BRINK MODEL B CASCADE IMPACTOR
**
Stage #
0
1
2
3
4
5
6
SRI (3)
9.3
5.8
3.7
2.30
1.05
0.78
0.46
SRI (3)
9. ,3
5.4
3.7
2.35
1.10
0.76
0.46
Present Study
9.1
5.6
3.35
2.32
1.24
.84
.49
**
Experimental values, excludes losses at 0.03 cfm
Experimental values, includes losses at 0.03 cfm
Calculated values at .03 cfm based upon a /STK__ value of 0.46
obtained based upon the impactor's critical dimensions and Marple's
curves (12) .
46
-------
TABLE 18. TEST AEROSOLS
1. Oil (liquid)
A. 1% and 2% solutions of OOP in ethanol
B. 100% DNP
2. Soft "hygroscopic" Aerosol
A. 1% solution of uranine-methylene blue (90:10) in
water - ethanol (50:50)
B. 30% solution of sodium chloride in water
3. Hard (solid) Aerosol
A. Polystyrene Latex Spheres dispersed in water
47
-------
2) relatively "soft" hygroscopic aerosols, and 3) "hard" solid aerosols. The
liquid aerosols can be characterized as "sticky", being generated from both
dioctyl phthalate (OOP) and dinonyl phthalate (DNP). Such an aerosol would
be characteristic of the condensation aerosol stream produced by processes
which involve petroleum products and resins. The "soft" hygroscopic aerosols
are characterized as having a low to medium potential for bounce, and were
generated from solutions of both a uranine-methylene blue mixture and sodium
chloride (NaCl). These "soft" hygroscopic aerosols reflect the properties of
many aerosol streams. The "hard" solid aerosols can be considered as having
a high potential for bounce; these aerosols were generated from a suspension
of polystyrene latex spheres (PSL) and served to characterize mineral dust
aerosol streams.
All of the above aerosols were generated with a three-jet Collison
atomizer (34). They were chosen based on the fact that they are accepted as
standard test aerosols which result in a consistent, reproducible aerosol
stream. Through the use of test aerosols with such a variety of surface
characteristics it was possible to assess the degree to which the type of
aerosol defines the optimum operating parameters for an in-stack impactor.
2 . Vibrating Orifice Aerosol Generator
A vibrating orifice type aerosol generator similar to that described by
Berglund and Liu (33) was used to generate monodisperse particles from 2 to
13 urn diameter. The generator consisted of a constant rate liquid feed
system, an orifice-piezoelectric crystal assembly, a signal generator and an
aerosol flow system incorporating dispersion air and dilution air for drying
and neutralization of the aerosol stream.
A solution of a nonvolatile solute in a volatile solvent was delivered
at a constant rate by a syringe pump to the orifice assembly. By applying a
periodic disturbance of an appropriate frequency, the liquid jet formed at
the orifice-piezoelectric crystal assembly was broken into uniform droplets.
The volume of each droplet was calculated by liquid feed rate and the
frequency of the disturbance. The droplets were dispersed to prevent
coagulation, and upon evaporation of the volatile solvent, an aerosol of the
nonvolatile fraction was formed. The final diameter of the aerosol particle
formed was defined by the nonvolatile volume concentration of the solution
as:
1/3
Dp = Dd(Conc) (3)
where:
Dp = particle diameter
D^ = droplet diameter
Cone = volumetric concentration of the solution
Thus, a range of particle sizes was obtained simply by varying the solute
concentration. The electrical charge of the aerosol stream was brought to a
Boltzman's Equilibrium through the use of a 10 me Kr-85 radioactive source.
48
-------
Monodisperse particles of uranine, a fluorescent dye, were generated
having an average geometrical standard deviation of 1.02 as determined by
optical sizing from a light microscope equipped with a filar eyepiece. A
50:50 solvent mixture of distilled water and ethanol was found upon drying to
produce the most spherical particles. Concentrations of uranine ranged from
10 ppm to 1%. A 20 urn diameter orifice gave trouble-free operation over the
range of particle sizes desired with the mass output of the generator being
stable throughout periods of up to 3 hours.
3. Three Jet Collison Atomizer
A three jet Collison atomizer as described by Green and Lane (34) was
used to generate polydisperse test aerosols over a range from 0.35 urn mmd to
1.8 urn mmd. The atomizer was operated at 30 psig, resulting in an 11 £/min
aerosol flow rate. The aerosol stream was then increased to 76 Vmin upon
dilution by the charge neutralizing airstream.
The ionizer used with the Collison atomizer was similar to that
previously described by Whitby (35). The ion generator was operated at a
3,000 volt A.C. potential with a constant air flow of 61 £/min. A 69 £
capacity conditioning chamber provided adequate time for both the evaporation
of the solvent and contact with the stream of bipolar ions to insure
achievement of an equilibrium change distribution.
Test aerosols were generated from a variety of materials, including
uranine, OOP, DNP, sodium chloride and polystyrene latex spheres. In all
cases samples of the aerosol were viewed with an optical microscope to further
characterize the particles. The mass output of the atomizer was found to be
constant over the periods of time required to achieve the desired loading.
F. TECHNIQUES FOR MASS DETERMINATION
1. Fluorometric Technique
Fluorometric analysis was employed for collection efficiency measurements
(when applicable) because of the method's excellent sensitivity and ease of
use. The uranine dye was detectable within ±0.0001 ug for most applications
and thus was suited for detection with low mass loadings and measurement of
wall losses.
Initial background fluorescence was determined for all the collection
surface coatings used and this correction factor was applied to obtain the
final mass values. After a given impactor run, collection surfaces and both
nozzles and collection plates were individually washed with 20 mil of distilled
water and allowed to set for 12 hours to insure that all the uranine had gone
into solution. The samples were then diluted (if necessary) with distilled
water and the fluorescence measured with a Turner Model 110 fluorometer. The
mass concentration (yg/fc) of uranine was determined from a prepared standard
calibration of measured fluorescence vs. uranine mass concentration and
multiplied by the appropriate dilution factor (fc) to obtain the collected
mass (yg).
49
-------
2. Gravimetric Technique
Gravimetric measurement of mass was achieved with a Cahn Model 4700
automatic electrobalance. Weighings were made to the nearest .001 mg with a
typical 95% confidence interval of ±.005 mg. Prior tests showed that if
initial and final weighings were made within several hours of each other,
preliminary desiccation was not required to obtain stable weighings.
G. DETERMINATION OF COLLECTION EFFICIENCY
As previously stated, the term collection efficiency will be used in this
study to characterize the mass fraction collected by each of the impactor
stages. The mass collected upon individual collection plates was determined
by either fluorometric or gravimetric techniques (see Section F) and reported
as the mass collected for that stage. The total mass collected by all seven
stages plus the final filter was reported as the total impactor catch. Thus,
the collection efficiency for any given impactor stage can be defined as:
M.
i
C.E. . = ^ i=F .. (4)
s-i V . M.
where:
C.E. . = collection efficiency for stage desired
M. = mass collected upon desired stage
53-1, M. = total impactor stage catch
^•^x—x i
The cumulative stage collection efficiencies were then plotted against the
corresponding Dpso for the given stage on log-normal probability paper to
obtain a graphical representation of the measured particle size distribution.
Deposition onto surfaces other than the impactor collection surface
coating was considered interstage loss and not included as mass collected by
a given stage unless otherwise noted. Both interstage loss per stage and the
total interstage losses are reported as a percent of the total impactor
catch.
H. " MEASUREMENT OF FLOW RATE, TEMPERATURE, AND RELATIVE HUMIDITY
For laboratory testing, the impactor and parallel filter were followed
directly by calibrated dry gas meters used to record the sampled volume of the
aerosol stream. The flow rate was monitored by calibrated rotameters and
maintained throughout the sampling period by means of a bypass valve in line
with the pump. The pressure drops at the dry gas meters were measured
directly by mercury manometers. The temperature of the gas stream was
\
50
-------
measured by bimetal dial thermometers. The volume of gas sampled at
laboratory conditions was then calculated as:
T P - AP
IF ,, a a m ,._,
V - = V x — x (5)
act m T P
m a
where:
Vacf = actual cubic feet of aerosol stream sampled
Vm = cubic feet of sampled aerosol stream measured by
the dry gas meter
Ta = actual temperature of the sampled aerosol stream (°K)
Tm = temperature of the sampled aerosol stream at the
dry gas meter (°K)
Pa = actual pressure of the sampled aerosol stream ("Hg)
APm = pressure drop at the dry gas meter ("Hg)
This volume over the elapsed sampling time resulted in the volumetric sampling
rate (flow rate) reported. The relative humidity of the gas stream was
determined by the wet-bulb-dry-bulb technique.
51
-------
SECTION 5
EXPERIMENTAL RESULTS
A. INTRODUCTION
Results of the impactor testing are presented in sections corresponding
to the parameter evaluated; i.e., collection surface coatings, flow rate,
stage loading and interstage losses. Results of the in-depth testing
performed with the MK III University of Washington source test cascade
impactor were followed by results of tests from the Andersen MK III stack
sampler, the Sierra Model 226 source cascade impactor and a modified Brink
Model B cascade impactor for each section. A discussion of these results and
comparison with prior findings is detailed in Chapter II.
B. ANALYSIS OF ERRORS
The size distributions were calculated (Chapter IV, Section G) based upon
Dpso cut-points which were derived from the impactorfs critical dimensions and
the impactor efficiency curves of Marple (12). The intent of this study was
not to calibrate the impactors, but rather to observe relative changes in the
measured distributions due to the varied parameters. The mass output of the
Collison atomizer has been previously characterized as log-normal, therefore,
the apparent distributions should approximate straight line plots. The fact
that the calculated cut-points do provide such apparent distributions suggests
that they are accurate relative to themselves, although their absolute values
may vary.
The errors associated with the individual mass measurements are outlined
in Table 19. It was noted that handling losses and variability were different
for the various types of collection surfaces used. The resulting confidence
intervals corresponding to the individual size distributions are represented
on the plots as percent mass by the use of error bands. Thus, the magnitude
of the error band is inversely proportional to the amount of mass collected
on the given stage. Nominal error bands ranged from ± 1% to ± 10%. The
relative magnitudes of the error bands may appear to be distorted due to the
probability function on which they were plotted. Significant differences
between size distributions were defined by the above confidence intervals,
and reproducibility of experimental results.
52
-------
TABLE 19. ERRORS ASSOCIATED WITH MASS MEASUREMENT
1. Fluorometric Analysis
reproducibility + 0.68%
impactor surfaces +_ 0.0001 yg
spray silicone +_ 0.0017 yg
glass fiber +_ 0.0202 yg
uncoated aluminum +_ 0.0001 yg
2. Gravimetric Analysis
reproducibility +_ 0.005 rag
spray silicone +_ 0.02 mg
glass fiber +_ 0.04 mg
uncoated aluminum +^0.05 mg
Values shown are for 95% confidence interval.
53
-------
C. COLLECTION SURFACE COATINGS
1. Suitability of Use at Elevated Temperatures and High Jet Velocities
Collection surface coatings (Table 2) were subjected to various elevated
temperatures for periods of one hour. The collection surfaces were weighed
before and after exposure, allowing a measure of weight loss of the coating
due to volatilization at the elevated temperature. After an initial one-hour
exposure period, the identical coatings were re-exposed at the same
temperature for an additional one hour in order to evaluate the extent to
which preconditioning of the coatings would decrease weight loss during a
testing period.
Results of this testing are summarized in Table 20. The values shown for
weight losses and their corresponding variations were based upon two separate
trials of three or more samples. In most cases, the measured weight losses
during the second hour of exposure to the elevated temperature were decreased
by a factor of ten (10) or more, and were accompanied by decreased variance
between samples. Whereas the Dow Corning 200 Fluid, Apiezon grease and White
petroleum jelly all were suitable at temperatures of 200°F, at 300°F and
above, these coatings were observed to decompose (color change, splotchy
appearance and running) and have highly variable weight losses, even after the
initial one-hour exposure period. Exposure of both the Dow Corning High
Vacuum grease and the industrial spray silicone after the initial one-hour
period resulted in low weight losses and variability at test temperatures of
200°F, 300°F and 400°F, but were found to be unsatisfactory at 500°F, as
evident by the greatly increased weight losses (Table 20). At temperatures
of 500°F only the uncoated aluminum and glass fiber filter were found to have
stable mass. No further tests were made at higher temperatures.
The above testing was conducted in an oven in still air and could be
classified as a static test. Dynamic testing was performed by directing a jet
of hot air at the coatings. The jet velocities of the final stages for the
impactors fell within the range from 18 m/sec (Sierra at 0.25 cfm) to 100
m/sec (modified Brink at 0.05 cfm) when the units were operated at their
design sampling rate (Table 21). Thus, the coatings were subjected to a
similar range of incident velocities in order to evaluate their suitability.
The test setup is shown in Figure 14. The nozzle diameter was 0.0368 cm and
the jet-to-plate distance was 0.318 cm. These dimensions are approximately
those of stage 6 of the MK III University of Washington source test cascade
impactor. Both jet velocity and temperature were varied and the coatings
checked as to their tendency to be blown off the collection surface to which
they were applied. Each coating was exposed to an equal volume of air for
varying velocities. Results are summarized in Table 22.
The Dow Corning 200 fluid was blown off under all the test conditions,
resulting in a crater-like appearance. The thicker coating (-15 urn) of Dow
Corning High Vacuum silicone grease also proved unacceptable at all conditions
tested; however, the end result was less drastic than for the Dow Corning 200
fluid. At room temperature and a velocity of 120 m/sec, the thinner coating
of Dow Corning High Vacuum grease began showing signs of blow-off, with a
slight build-up of coating around the perimeter of the area covered by the
54
-------
TABLE 20. COLLECTION SURFACE'COATING WEIGHT LOSS FOR 1-HOUR EXPOSURES TO TEMPERATURES OF 200°F 8 500°F
Ol
01
Collection
Surface Initial Depth
Coating of Coating
Uncoated
Aluminum
Dow Corning High 4 pm
Vacuum Silicone
Grease 15 um
Industrial Spray 4 usi
Silicone 15 ym
Gelman Type A
(mg) Weight
1st Hour
.12
.27
.11
.37
•>a
Loss @
2nd
.01 +
.04 +
^M>
.01 +
.03 jf
n-> *.
200°F
Hour
m
. Ui
.01
.02
.01 "
.01
(mg) Weight
1st Hour
1 1
. io
.46
1.18
.44
.76
: TC
Loss @
2nd
m j.
. Ul +•
.27 -f
.55 -f
.30 +
1.00 +_
nn
500°F
Hour
ni
. Ul
.05
.10
.10
.30
nr
Glass Fiber Filter
Note:
1) Initial Depth of coating estimated according to coating weight
2) All weights are "dry" (after desiccation) weights
3) Dow Corning 200 fluid, Apeizon Grease and White Petroleum jelly all showed signs of
decomposition and highly variable weight loss at temperatures above 300°F
-------
TABLE 21. MAXIMUM JET VELOCITIES FOR IMPACTORS
OPERATED AT DESIGN FLOW RATE
Impactor
MK III University of Washington
Source Test Cascade Impactor (0.5 cfui)
Maximum Jet Velocity
53 m/sec
Andersen MK III Stack Sampler (0.5 cfui)
Sierra Model 226 Cascade Impactor
(0.25 cfm)
33 m/sec
18 m/sec
Modified Brink Model B Cascade
Impactor (0.05 cfm)
107 m/sec
56
-------
By-Pass
r
Rotameter
Pump
o
Absolute Filter
Test
Nozzle
Coil
Collection
Surface
Coating
L
Figure 14. Test set-up for dynamic testing of collection surface coatings.
-------
C/l
oo
TABLE 22, OBSERVED TENDENCY OF COLLECTION SURFACE COATING TO BE BLOWN OFF BY AN IMPINGING JET OF
VARYING VELOCITY AND TEMPERATURE
Observed Blow Off of Coating Substrate
Coating
Substrate
Dow Corning
200 Fluid
Dow Corning
High Vacuum
Silicone Grease
Industrial Spray
Silicone
Initial Depth
Of Coating
15 ym
4 um
15 ym
4 ym
15 ym
Room Temperature
26 m/sec
Yes
No
Yes
No
No
50 m/sec
Yes
No
Yes
No
No
120 m/sec
Yes
Yes
Yes
No
No
200"?
50 m/sec
Yes
No
Yes
No
No
500~F
50 m/se<
Yes
Yes
Yes
Yes
Yes
Note:
1) initial depth of coating estimated according to coating weight
2} a 0.0368 cm nozzle was used with a corresponding jet-to-plate distance of 0.318 cm
(S/W =8.64)
-------
impinging jet. Both thicknesses of the industrial spray silicone showed only
slight signs of the impinging jet at room temperature and 120 m/sec incident
velocity, with the majority of the coating undisturbed. At 200°F and an
incident velocity of 50 m/sec, both the Dow Corning High Vacuum grease and the
industrial spray silicone proved suitable, with little or no visible sign o'f
the impinging jet. When the temperature was raised to 500°F, all coatings'
exhibited areas from 0.5 to 2.0 cm in diameter over which the coatings were
blown off, and some showed signs of decomposition, as previously noted. From
these series of tests the industrial spray silicone was chosen as the most
suitable representative of a "grease" type coating. Further testing of
collection surface coatings involved only the spray silicone, glass fiber and
uncoated aluminum.
2. Effect of Collection Surface Coating on the Measured Size Distribution
The MK III University of Washington source test cascade impactor was
operated at 0.5 cfm during the testing of the following collection surfaces:
1) spray silicone, 2) glass fiber, and 3) uncoated aluminum. Various types
of polydisperse test aerosols were sampled with each of these collection
surfaces. Total impactor loadings were held constant for all tests compared.
A comparison of the collection characteristics of each of the surface coatings
was made based upon the observed differences in the resulting cumulative
particle size distributions.
The effect of the collection surface coating on the measured size
distribution for a polydisperse uranine aerosol is shown in Figure 15. The
apparent measured size distribution obtained using the uncoated aluminum and
the spray silicone collection surface coatings is not significantly different.
However, when the glass fiber collection surfaces were employed, the apparent
measured distribution shifted to an increased mmd and a slightly increased
V
When a similar sized aerosol of dioctyl phthalate (OOP) was tested, once
again the apparent measured size distribution obtained with glass fiber
collection surfaces was shifted toward an increased mmd relative to the
distribution obtained for the spray silicone coatings (Figure 16). The
geometric standard deviations (6g) for the surface coatings were observed to
be approximately the same, with the distribution for the uncoated aluminum
surface appearing to "tail off" with increased particle size. Such an effect
may be attributed to several factors; however, since in this case it involved
only 2% of the total mass collected, the data points were not weighted heavily
when interpreting the apparent distribution.
Dinonyl phthalate (DNP), another oil aerosol having a larger size
distribution, was also sampled. Again, the distribution obtained with the
glass fiber collection surfaces was observed to have an increased mmd
relative to the distributions for both the uncoated aluminum and the spray
silicone (Figure 17). In the case of all the aerosols sampled (uranine, DOP
and DNP), increased deposition of particles on the upper stages of the
impactor was visually observed when glass fiber collection surfaces were used.
This deposition was not confined just to the areas of the primary deposit
beneath the impinging jet, but encompassed the entire collection surface as
59
-------
99.9
99
in
OS
90
80
70
60
50
40
30
20
10
0.1.
0.1
fit
1.0
O - Glass Fiber Filter
Spray Silicone
D - Uncoated Aluminum
I-
Typical 95% confidence inter-
vals for all collection
surfaces
10.0
-*—4™!l_UlJ
100.0
Aerodynamic Diameter
Figure 15. Effect of collection surface on measured size distribution for polydisperse uranine aerosol.
(MK III University of Washington source test cascade impactor, 0.5 cfm, 70°F, 30.19 "Hg, 10.0 mg
total loading, fluorometric analysis).
-------
M
•H
in
*o
o>
ri
o
•r-t
T3
•r-i
g
X
4J
IT,
rt
99
90
80
70
60
50
40
30
20
10
0.1
0.1
1.0
&
D
O - Glass Fiber Filter
D
- Spray Silicone
- Uncoated Aluminum
- Typical 95% confidence
interval for all collection
surfaces
_L
Aerodyanmic Diameter
10.0
(urn)
.
100.0
Figure 16. Effect of collection surface on measured size distribution for polydisperse oil aerosol (MK III
University of Washington source test cascade impactor, 0.5 cfm, 70 F, 30.20 "Hg, 5.0 mg total
loading, gravimetric analysis).
-------
Q>
N
•8
u
OS
O
•o
c
CO
m
v,
a
99.9
99
90
80
70
60
50
40
30
20
10
0.1
0.1
O - Glass Fiber
- Spray Silicone
— Uncoated Aluminum
T - Typical 95% confidence
interval for all collection
surfaces
..L
1.0 10.0
Aerodynamic Diameter (urn)
mo.o
Figure 17. Effect of collection surface on measured size distribution for polydisperse oil aerosol
(increased mmd) (MK. Ill University of Washington source test cascade impactor, 0.5 cfm, 70 F,
30.06 "Hg, 30 mg total loading, gravimetric analysis).
-------
seen in Figure 18. Similar secondary deposition was not apparent for
either the uncoated aluminum or the spray silicone collection surface
coatings.
As a result of these observations and in an attempt to determine if the
glass fiber collection surfaces were selectively filtering out submicron
particles, it was decided to sample an aerosol of such a size distribution
that greater than 95% of the total mass should penetrate through to the final
filter. A polydisperse uranine aerosol was sampled resulting in the data
presented in Table 23. As shown, approximately 98% of the total collected
mass penetrated through to the final filter when employing either uncoated
aluminum or spray silicone collection surface coatings. When the impactor
was operated with the glass fiber collection surfaces, only 79% of the total
mass penetrated through to the final filter, thus significant increases in
deposited mass were detected for all but the initial stage of the impactor.
The modified Brink was operated at 0.05 cfm with both spray silicone and
glass fiber collection surface coatings. A polydisperse uranine aerosol was
sampled and resulted in the apparent measured size distributions shown in
Figure 19. As .seen with the MK III University of Washington source test
cascade impactor, when glass fiber collection surfaces were used, an increased
size distribution was observed. Similar tests were not run with the Andersen
and Sierra samplers. However, when these impactors were each operated at
their nominal flow rates with the pre-cut glass fiber collection surfaces and
sampling the same polydisperse uranine aerosol, the apparent measured size
distributions obtained were all very similar, as shown in Figure 20. Thus,
it is suspect that the size distributions obtained with any impactor would
exhibit an increased mmd due to the use of a glass fiber (or a filter media)
collection surface as opposed to use of an uncoated or grease coated surface.
D. EFFECT OF FLOW RATE ON THE MEASURED SIZE DISTRIBUTION
The MK III University of Washington source test cascade impactor was
operated at flow rates of 0.25 cfm, 0.5 cfm, and 1.0 cfm, with 0.5 cfm
reported as the design flow rate. Both spray silicone and glass fiber
collection surface coatings were used over the complete range of test flow
rates. Various types of polydisperse test aerosols were sampled in an attempt
to determine: 1) if the operational flow rate had any effect upon the
resulting particle size distribution, and 2) an upper and lower limit of the
operational flow rate with respect to both the aerosol being sampled and the
collection surface coating being used.
Figure 21 presents the measured size distributions obtained when sampling
a polydisperse uranine aerosol at various flow rates onto spray silicone
collection surfaces. Sizing data obtained at the three different flow rates
(0.25 cfm, 0.5 cfm and 1.0 cfm) resulted in virtually the same cumulative
distribution, although specific observations were made associated with each
flow rate. At the lower end of the distribution there was evidence of mass
carry-over onto the final filter when the impactor was operated at 1.0 cfm.
This was reflected on the plot as a displaced data point. Visual observation
of the 7th stage collection surface deposition (Figure 22) revealed halo-like
deposits which have been previously associated with particle bounce. Such
63
-------
Figure 18. MK III University of Washington source test cascade
impactor stage 4 deposition of uranine aerosol onto
glass fiber collection surface at 0.5 cfm.
:
-------
TABLE 23 PERCENT OF TOTAL COLLECTED MASS PER STAGE FOR A POLYDISPERSE URANINE AEROSOL (MK III
UNIVERSITY OF WASHINGTON SOURCE TEST CASCADE
1.0 rag TOTAL LOADING, FLUOROMETRIC ANALYSIS)
UNIVERSITY OF WASHINGTON SOURCE TEST CASCADE IMPACTOR, 0.5 CFM, 70°F, 30.18"Hg,
Collection Surface S-l S-2 S-3 S-4 S-5 S-6 S-7 S-F
Uncoated Aluminum 0.03 0.07 0.15 0.21 0.23 0.53 1.22 97.56
Glass Fiber 0.05 0.21 0.53 1.17 1.75 5.22 11.87 79.28
01 Spray Silicone 0.03 0.04 0.09 0.14 0.20 0.43 1.35 97.71
-------
ON
99.9
99
M
•H
in
CO
y
—i
"O
t
•H
2
in
ca
£
90
80
70
60
50
40
30
20
10
0.1
0.1
O - glass fiber
£ - spray silicone
J L.
_L
i.o 10.0
Aerodynamic Diameter (pro)
100.0
.Figure 19- Effect of collection surface coating on measured size distribution for polydisperse uranine
aerosol (Modified Brink Model B cascade impactor, 0.05 cfm, 70°F, 30.06 "Hg, 15.0 mg total
loading, gravimetric analysis).
-------
99.9 r
99
N
•H
tfi
•H
•a
c
VI
90
80
70
60
50
40
30
20
10
0.1
DO
D
O
O - University of Washington
0 - Andersen
Q - Brink
• - Sierra
J L. t I.
J_
JL
t t 1
* '*'*».
0.1
1.0 10.0
Aerodynamic Diameter (urn)
100.0
Figure 20. Measured size distributions for a polydisperse uranine aerosol, using glass fiber collection
surfaces obtained with the MK III University of Washington source test cascade impactor, Andersen
MK IV stack sampler, modified Brink Model B cascade impactor and Sierra Model 226 source cascade
impactor (design flow rates, 70 F, 30.10 "Hg, 15.0 mg total loading, gravimetric analysis).
-------
00
Figure 21.
99.9
99
"O
rt
o
w
(/>
t/:
90
80
70
60
50
40
30
20
10
-
particle
bounce
1
MB
-
0.1
D
O - 1.0 cfm
$ - 0.5 cfm
- 0.25 cfm
- Typical 95% confidence
interval at all flow rates
0.1
-*-«.
L
_L
\ .. I I .JL
100.0
1.0 10.0
Aerodynamic Diameter (urn)
Effect of flow rate on measured size distribution for polydisperse uranine aerosol sampled onto
spray siiicone collection surfaces (MK III University of Washington source test cascade impactor,
70°F, 29.93 "Hg, 0.3 mg total loading, fluorometric analysis).
-------
•
Figure 22. MK III University of Washington source test cascade
impactor stage 7 deposition of uranine aerosol onto
spray silicone collection surface at 1.0 cfm.
69
-------
deposits were not observed when sampling at flow rates of 0.5 cfm or 0.25
cfm.
When this same aerosol was sampled onto glass fiber collection surfaces
at these flow rates, the same effect was observed (Figure 23). Equivalent
distributions were obtained when sampling at either of the three flow rates.
Tailing was observed at the upper end of the distributions, however, only 1%
of the total collected mass was involved. Once again, there was evidence of
particle bounce off, resulting in an increase of mass associated with the
final filter when the impactor was operated at 1.0 cfm.
A similarly sized liquid aerosol (OOP) sampled under the same conditions
(flow rate, loading and collection surfaces) resulted in distributions having
approximately the same mmd's but with apparently different Sg's (Figures 24
and 25). Distributions derived from collection onto both spray silicone and
glass fiber surfaces all showed the similar trend of increasing 6g with
increased sampling rate. This testing was repeated several times, always with
the same results. The sizing data for the oil aerosol reflects particle
bounce problems at the 1.0 cfm flow rate as previously seen with the uranine
aerosol. Once again inspection of the collection surfaces themselves revealed
evidence of this problem (Figure 26). As the photographs show, the well-
defined deposits collected upon stage 6 at 0.5 cfm are quite different from
the highly splattered stage 7 deposits collected at 1.0 cfm. At 0.5 cfm the
stage 7 deposits exhibited only slight splattering, with no significant problem
evident. Close inspection of the stage 7 DOP deposition (1.0 cfm) revealed
primary deposits surrounded by rings of secondary deposits, similar to the
halo-like effect previously noted with the uranine aerosol.
Sampling of liquid aerosol (DNP) having a distribution of somewhat larger
mmd onto both glass fiber and spray silicone collection surface coatings did
not show the above mentioned trend of increasing 6g with increasing sampling
rate (Figures 27 and 28). Distributions obtained from both the glass fiber
and spray silicone collection surfaces at 1.0 cfm showed signs of increased
mass penetration onto the final filter and visual observation revealed
similar splattered deposits on the stage 7 collection surfaces (as previously
cited).
Where there was no apparent problem with particle bounce off on stage 7
when operated at 0.5 cfm for either collection surfaces and sampling the
uranine and oil aerosols, the polystyrene latex (PSL) spheres were observed
to bounce excessively. The PSL spheres sampled were monodisperse and thus
should have been predominantly collected on a single stage (stage 5). As
determined from the collection efficiencies for the surface coatings tested,
of the approximately 36% mass which penetrated past stage 5, 34% of this mass
penetrated on through to the final filter when the spray silicone coatings
were used, 42% when glass fiber surfaces were used, and 89% when the uncoated
aluminum was used as a collection surface. Observation of the stage 6 spray
silicone collection surface revealed uniform circular primary deposits (Figure
29). The stage 6 and 7 uncoated aluminum collection surfaces as well as the
stage 7 spray silicone surface exhibited secondary deposition about the area
of the primary deposit (Figure 29).
70
-------
99.9 _
99
N
•fi
tfl
"8
+->
a
o
rt
in
(A
(D
90
80
70
60
50
40
30
20
10
0.1
particle
bounce
a
\
_L
0.1
6 * fi
O - 1-0 cfm
- 0.5 cfm
- 0.25 cfm
Typical 95% confidence
interval for all flow rates
t
1.0 . 10.0
Aerodynamic Diameter (ym)
100.0
Figure 23. Effect of flow rate on measured size distribution for polydisperse uranine aerosol sampled onto
glass fiber collection surfaces (MK III University of Washington source test cascade impactor,
70°F, 30.29 "Hg, 0.3 mg total loading, fluorometric analysis).
-------
--a
99.9
99
o
V.
'O
o
IT
tc
90
80
20
10
* ft a
a
o
*M
0
•o
e
•H
c
CO
f.
in
i
^* ^f
70
60
50
40
30
particle
- bounce
- (
-
"
0.1
O- 1.0 cfra
0- 0.5 cfra
- 0.25 cfm
- Typical 95% conficence
interval for all flow rates
_t~
0.1
1.0
0.0
Figure 24.
Aerodynamic Diameter (ymj '
Effect of flow rate on measured size distribution for polydisperse oil aerosol sampled onto
spray silicone collection surface (MK III University of Washington source test cascade impactor,
70°F, 30.04 "Hg, 5.0 mg total loading, gravimetric analysis).
-------
99.9
99 -
o
K
(A
TJ
0>
-------
a) Stage 6 onto spray silicone at 0.5 cfm.
b) Stage 7 onto spray silicone at 1.0 cfm.
Figure 26. MK III University of Washington source test cascade
impactor deposition of oil aerosol.
-------
c) Stage 7 onto spray silicone at 0.5 cfm.
I-'igure 26. (cont.)
-------
o\
N
•H
OS
o
tn
cs
£
99.9
99
90
80
C t>U
'1-1
r n
I 40
T <*
t/l Ow
irt
O
20
10
0.1
particle
bounce
* 66
a—i—1_
0.1
- 1.0 cfm
0-0.5 cfm
D
-0.25 cfm
- Typical 95% confidence interval
for all flow rates
_1 lv_
100.0
Aerodynamic Diameter (urn)'
Figure 27 . Effect of flow rate on measured size distribution for polydisperse oil aerosol (increased mmd)
sampled onto spray silicone collection surfaces (MK III University of Washington source test
cascade impactor, 70 F, 30.00 "Hg, 30.0 mg total loading, gravimetric analysis).
-------
99.9
9S _
i /°T
JK.
i
o
N
•H
(fl
•M
a
u
•H
•o
c
•H
c
cij
V,
rH
(ft
rt
0\«
90
80
70
60
50
40
30
20
10
1
0.1
/
¥
D/
.O o- 1.0 cfm
r D»:
/ *- 0.5 cfm
0
m.
D' Q- 0.25 cfm
o
/
particle JT _
- bounce D T- Typical !
T / •*- for all j
\y
*
r- ...... L t i .1 . • l 1 - .1 J A L__l J__nA ...._»...
0.1
1.0 10.0
Aerodynamic Diameter (ym)
100.0
Figure 28.
Effect of flow rate on measured size distribution for polydisperse oil aerosol (increased mmd)
sampled onto glass fiber collection surfaces (MK III University of Washington source test
cascade impactor, 70°F, 30.00 "Hg, 30.0 mg total loading, gravimetric analysis).
-------
a) Stage 6 onto spray silicone at 0.5 cfm,
b) Stage 7 onto spray silicone at 0.5 cfm.
Figure 29. MK III University of Washington source test cascade
impactor deposition of polystyrene latex spheres.
-------
c) Stage 6 onto uncoated aluminum at 0.5 cfm.
o
%^<
d) Stage 7 onto uncoated aluminum at 0.5 cfm.
Figure 29. (cont.)
-------
A polydisperse uranine aerosol was also sampled at various flow rates
with the Andersen, Sierra and modified Brink impactors. Figures 30, 31 and
32 are the cumulative distributions obtained from each of these impactors when
employing glass fiber collection surfaces. In the case of both the Andersen
and the Sierra, particle bounce was not observed at the upper flow rate. The
Brink, however, showed visual signs of bounce off of the last stages at both
0.05 and 0.1 cfm flow rates.
As a point of interest, it was noted that the increase in mmd of the
distribution observed when using glass fiber collection surfaces was also
evident when sampling at higher or lower flow rates, as is evident in Figures
33 and 34.
E. STAGE LOADING
1. Effect of Stage Loading on Stage Collection Efficiency
The effect of a given stage loading upon that stage's collection
efficiency was observed using the MK III University of Washington source test
cascade impactor. This was accomplished by varying the stage loading, through
increased sampling times, and measuring the mass penetration past that stage.
Measurements were made at the design flow rate of 0.5 cfm using spray
silicone, glass fiber and uncoated aluminum collection surface coatings in an
attempt to characterize the collection characteristics of each type of coating
with increasing stage loadings.
Results are presented in Figure 35 for a polydisperse uranine aerosol.
In the case of both uncoated aluminum and spray silicone collection surfaces,
an initial effect was observed with the increased mass loading to stage 7.
Penetration decreased with increased loading up to approximately 0.5 mg, at
which time it leveled out. This effect corresponds to an increase in the
collection efficiency after some deposit has occurred upon the
collection surface. In contrast, the glass fiber collection surface showed a
gradual increase in penetration with increased loading. Observation of the
various stage 7 collection surfaces at 2 mg mass loading revealed heavy
uniform depositions, with their peaks standing almost to the back side of the
nozzle plate.
The results of a similar test using an oil aerosol (OOP) are presented
in Figure 36- Collection characteristics of the glass fiber surfaces were
observed to be relatively stable, with penetration gradually increasing with
increased loading. Penetration past the spray silicone collection surfaces
was also quite constant, as seen in the figure. Collection characteristics
of the uncoated aluminum surfaces were found to be quite variable, with no
trend evident.
Results of tests extending the stage loadings from 2-3 mg to 15-20 mg for
oil polydisperse aerosols (DOP and DNP) suggest no drastic changes in
collection efficiencies from the previously cited observations (Figure 37).
When the glass fiber collection surfaces were loaded to greater than 15-20 mg
with the oil aerosol (DNP and DOP), the oil was observed to penetrate through
80
-------
oo
99.9 r-
99
o
N
•irt
v>
2
u
•rt
c
•rt
e
4J
(A
I/)
V)
03
80
70
60
50
40
30
20
10
0.1
0.1
>
»fi
/c
a
/o
O - 1-0 scfm
• - C.5 scfm
Q - 0.25 scfm
J L.
X
_! L.
1.0 10.0
Aerodynamic Diameter (um)
Figure 3d Effect of flow rate on measured size distribution for polydisperse uranine aerosol collected on
glass fiber collection surface (Andersen MIC III stack sampler, 70 F, 30.00 "Hg, 15.0 rag total
loading, gravimetric analysis).
-------
00
to
99.9
99
a>
N
•H
in
-o
o
u
(A
rt
90
80
70
60
50
40
30
20
10
1 •-
0.1
T57T
^
I
a
O - 1.0 scfm
0 - 0.5 scfm
| - 0.125 scfm
n - 0.25 scfm
_* l_
-I L.
_L
1.0 10.0
Aerodynamic Diameter (pm)
-i 1.
J
100.0
Figure 31. Effect of flow rate on measured size distribution for polydisperse uranine aerosol collected
on glass fiber collection surface (Sierra Model 226 cascade impactor, 70°F, 30.04 "Hg, 15.0 mg
total loading, gravimetric analysis).
-------
99.9.-
00
99 -
o
N
•rt
eri
»w
O
•ft
•a
•H
C
R)
JS
*J
in
0.1
1.0 10.0
Aerodynamic Diameter (ym)
J
100.0
Figure32 • Effect of flow rate on measured size distribution for polydisperse uranine aerosol collected on
glass fiber collection surfaces (Modified Brink Model B cascade impactor, 70°F, 30.06 "Hg,
15.0 mg total loading, gravimetric analysis).
-------
O3
99.9
99
o
in
cd
u
•H
•o
c
I
CO
v>
0)
CO
to
cd
90
80
70
60
50
40
30
20
10
0.1
particle
bounce
O - glass fiber
I - spray silicone
I - Typical 95?0 confidence intervals
0.1
1.0 10.0
Aerodynamic Diameter (ym)
100.0
Figure 33. Effect of collection surface coating on measured size distribution of polydisperse uranine
aerosol sampled at 1.0 cfm (MK III University of Washington source test cascade impactor, 70°F,
30.12 "Hg, 15.0 rag total loading, gravimetric analysis).
-------
00
tn
99.9 r
99
90
80
70
60
g 50
£ 40
in
2 30
tf)
•o
ctf
o
•H
•o
(A
tf)
rt
€
20
10
0.1
0.1
Q - glass fiber
spray
silicone
- Typical 95% confidence intervals
_L
1.0 10.0
Aerodynamic Diameter (ym)
100.0
Figure 34.. Effect of collection surface coating on measured size distribution of polydisperse uranine aerosol
sampled at 0.25 cfm (MK III University of Washington source test cascade impactor, 70°F, 30.12 "Hg
15.0 mg total loading, gravimetric analysis).
-------
00
M-i
O
V,
60
60
03
IT,
•P
«n
ca
O
rt
-------
00
•-J
u
tr>
O
00
OJ
4-»
t/1
+J
c
o
•H
4J
A
s
Ci)
0.
0.2
0.1
0
O - Glass Fiber Filter
Spray Silicone
Uncoated Aluminum
- Typical 95% confidence interval for all
values
1.0 2.0
Stage 7 Loading (mg)
3.0
Figure 36. Effect of stage loading on penetration past stage 7 for polydisperse oil aerosol (MK III
University of Washington source test cascade impactor, 0.5 cfro, 70°F, 30.02 "Hg, fluorometric
analysis).
-------
50
00
OO
**->.
u
tn
CO
4->
tf)
in
o:
o
rt
g
Cu
40
30
20
10
o -
c
I
Glass Fiber Filter
Spray Silicone >
Uncoated Aluminum
Typical 95% confidence intervals
. I
1.0
10.0
Stage 3 Loading (mg)
100.0
Figure 37. Effect of stage loading on penetration past stage 3 of the MK III University of Washington
source test cascade impactor at 0.5 cfm (70 F, 30.36 "Hg, gravimetric analysis).
-------
the glass fiber mats. This suggests the possibility of mass loss to the
collection plate supporting the glass fiber filter and a subsequent error
in sizing data. No tests were run to quantitate such mass loss. When the
uncoated aluminum surfaces were loaded to greater than 10-15 mg with the oil
aerosol (DNP and OOP), there were visible signs of the oil deposits running,
due to the vertical alignment of the collection surface. This effect was not
observed when spray silicone collection surfaces were used.
At 10-15 mg loadings of a hygroscopic solid (sodium chloride), all of the
tested collection surfaces were observed to be overloaded. Individual
deposits were observed to be overlapping. In addition, there was a
significant increase in the visual losses to outer walls and undersides of the
nozzle plates. The penetration for these heavy loadings was highly variable
for all the surface coatings tested (Figure 38).
2. Effect of Stage Loading on the Measured Size Distribution
Further characterization of the surface coating's collection
characteristics with increased mass loading was made based upon a comparison
of cumulative particle size distributions obtained from impactor catches of
increasing total mass. Such a comparison also indicated the effect of stage
loading on the measured particle size distribution.
Figure 39 presents the measured size distributions of a polydisperse
uranine aerosol collected on spray silicone at 0.5 cfm, obtained after
sampling a total of: 1) 0.3 mg and 2) 10.0 mg mass. These total loadings
corresponded to the initial and final stage loadings for which the penetration
was measured past stage 7, as previously described (Figure 35). There is
evidence that with increasing stage loadings (total loading) the distribution
shifts toward a larger mmd and 6g, indicating decreased penetration. When
glass fiber collection surfaces were employed, the size distributions obtained
after total loadings of 0.3 mg and 10.0 mg were not significantly different
(Figure 40). The data point corresponding to the stage 7 collection showed
the only sign of any discrepancy, indicating a possible increase in
penetration. Referring back to the plot of penetration versus stage loading
for both spray silicone and glass fiber surfaces (Figure 35), it should be
noted that the effect of stage loading on the measured size distribution was
as expected from those results. In both tests, the glass fiber collection
surface displayed stable collection characteristics over the range of stage
loadings, while the spray silicone initially had relatively greater
penetration (decreased collection efficiency) resulting in the observed
difference in measured distributions.
Total loadings of 30 mg and 80 mg, corresponding to maximum stage
loadings of 8 mg and 16 mg, were sampled of a polydispersed oil aerosol (DNP)
at 0.5 cfm. As shown in Figure 41, the measured size distribution does not
show'any effect due to increased loading when collected on glass fiber
surfaces. Similar results were obtained using spray silicone collection
surfaces (Figure 42).
89
-------
<£>
O
1.0
u
(A
00
rt
in
2, 0.5
§
•H
+J
rt
0
Glass Fiber Filter
- Spray Silicone
Q - Uncoated Aluminum
T - Typical 95% confidence intervals
1
' i I I I 1 1
1
J 1 L
I t 1 IJ
0.1 1.0 10.0 100.0
Stage 7 loading (mg)
Figure 38. Effect of stage loading on penetration past stage 7 for polydisperse salt aerosol (MK III
University of Washington source test cascade impactor, 0.5 cfm, 30.13 "Hg, gravimetric analysis).
-------
99.9p
99 _
90
(fl
«3 80
a
X 70
3 60
c 50
| 40
2 30
-------
N)
99.9
99
a>
N
•H
90
80
70
60
50
40
w 30
(A
" 20
•o
o>
<->
rt
o
•H
ej
.e
(A
(A
rt
10
0.1
I
i
o
•i
/
- 10.0 mg total loading
0.5 mg total loading
- Typical 95% confidence intervals
oT
i—i i i
J-
i i i
JL
J
1.0.
• r>- , JO-0 100.0
Aerodynamic Diameter (umj
Figure 40. Effect of stage loading on measured size distribution for polydisperse uranine aerosol sampled
onto glass fiber collection surface (MK III University of Washington source test cascade impactor,
0.5 cfm, 70°F, 30.13 "Hg, fluorometric analysis).
-------
<£>
99.9r-
99
N
•o
90
OS
o
•H
c
•H
C
X
w
(0
^— <
^^^
in
in
Ctf
€
80
70
60
50
40
30
20
10
0.1
0.1
- 80.0 rag total loading
30.0 mg total loading
T - Typical 95% confidence interval
1.0
1 JO.
Aerodynamic Diameter (ym)
0
100.0
Figure 41. Effect of stage loading on measured size distribution for polydisperse oil aerosol sampled onto
glass fiber collection surfaces (MK III University of Washington source test cascade impactor,
0.5 cfm, 70°F, 30.00 "Hg, gravimetric analysis).
-------
99.9
99
Q>
N
•H
w
Q>
4-)
a
u
•H
"
I
OT
0)
Ift
rt
90
80
70
60
50
40
30
20
10
0.1
O - 30 mg total loading
- SO mg total loading
I
- Typical 95% confidence
intervals
JL
_L
0.1
100.0
1.0 10.0
Aerodynamic Diameter (urn)
Figure 42 Effect of loading on measured size distribution for an oil aerosol collected on spray silicone
collection surfaces (MK III University of Washington source test cascade impactor, 0.5 cfm,
70°F, 30.00 "Hg, gravimetric analysis).
-------
F. INTERSTAGE LOSSES
1- Interstage Losses as a Function of Particle Size
Monodisperse particles of uranine-methylene blue (90%:10% by weight) were
generated over a size range from 2-13 urn diameter and sampled with the MK III
University of Washington source test cascade impactor. The impactor was
operated at 0.5 cfm and collection was onto spray silicone collection surface
coatings. Total interstage losses were determined as previously described
(Chapter IV, Sections F and G).
In Figure 43, results are plotted as total interstage losses versus
particle diameter. Interstage losses accounted for as little as 2% of the
total collected mass for 2 urn diameter particles, to as much as 45% for 13 urn
diameter particles.
2. Effect of Interstage Losses on Size Distribution
Wall losses for the MK III University of Washington source test cascade
impactor were also evaluated for a polydisperse uranine aerosol. The impactor
was operated at its design flow rate of 0.5 cfm and equal volumes of the
aerosol stream were sampled except where noted.
Histograms of the interstage losses incurred when sampling different
total loadings of a polydisperse uranine aerosol collected onto both glass
fiber and spray silicone collection surfaces are shown in Figures 44 and 45.
At a total loading of 0.3 mg (Figure 44), the total interstage losses were 2%
and 4% of the total collected mass for spray silicone and glass fiber
collection surfaces, respectively. However, when the total loadings were
increased to 10 mg (Figure 45), interstage losses increased to 13% and 12% for
these same type collection surfaces. It was observed that at 10 mg total
loading, interstage losses associated with the uncoated aluminum collection
surfaces were significantly greater, at 20%. In all cases, the majority of
the losses were to the nozzle plates rather than the collection plates.
Observation of the nozzle plates revealed losses to: 1) the inside walls of
the nozzles, 2) the top center of the plates, directly beneath the opening
from the previous collection plate, and 3) the underside of the nozzle plates.
Losses to the collection plates were predominantly in the area of the outer
wall.
The effect of excluding these losses from the apparent measured size
distribution is shown in Figures 46 through 49, for both glass fiber and
spray silicone collection surfaces at total loadings of 0.3 mg and 10.0 mg.
In all cases, there was little change in the apparent distribution when the
recovered losses were excluded. Corresponding measured size distributions
Similar tests were conducted using the Andersen, Sierra and modified
Brink impactors, all operated at their suggested flow rate. Equal volumes of
the aerosol stream were sampled, resulting in total impactor catches of
approximately 15 mg. Histograms of the measured interstage losses are shown
in Figures 50 a through c. Table 3 summarizes the measured total interstage
losses for each of the impactors used. Corresponding measured size distri-
butions again show little effect when these losses are excluded (Figures 49-53)
95
-------
70 ,_
60
50
o
o\°
W
trt
q 30
CD
bo
ed
20
10
o
0.5
Figure 43.
i 1111
1.0
Particle Diameter (ym)
10.0
20
Total interstage loss versus particle diameter for the
University of Washington MK III source test cascade
impactor.
96
-------
Interstage loss
- Nozzle loss only
7
vi *•
cd
"3 *
Glass Fiber Filter
Total Interstage Loss =
I/)
10
V)
9)
DO
(4
+J
(/>
V.
4)
O
Spray Silicone
Total Interstage Loss = 2%
S-l
S-2
S-3
S-4
S-5
S-6
S-7
Figure 44. MK III University of Washington source test cascade impactor
measured interstage losses for a polydisperse uranine aerosol
of nund = 0.45 ym, at a total loading of 0.3 mg.
97
-------
v>
5
4
3
* 2
rt
4J
8 i
#> 0
^.x
tn
w 7
tn
.3
o> 6
bO
cfl
I 5
•M
tS 4
Glass Fiber Filter
Total Interstage Loss = 12%
^^E
Spray Silicone
Total Interstage Loss = 13%
S-l
S-2
S-3
S-4
S-5
S-6
S-7
Figure 45. MK III University of Washington source test cascade impactor
measured interstage losses for a polydisperse uranine
aerosol of mmd = 0.45 urn, at a total loading of 10.0 mg.
98
-------
ID
99.9 r-
99
Oi
N
•H
in
0>
rt
u
*H
13
c
•rt
1
**
(A
to
tt)
tn
in
cd
90
80
70
60
50
40
30
20
10
1
0.1
0.1
0 - includes interstage losses
- excludes interstage losses
- Typical 95% confidence intervals
. . t
j
1•°Aerodynamic Diameter (ymj°'°
100.0
Figure 46. Effect of interstage losses on measured size distribution of polydisperse uranine aerosol
sampled onto spray silicone at low stage loading (MK III University of Washington source test
cascade impactor, 0.5 cfm, 70°F, 29.93 "Hg, 0.3 rag total loading, fluorometric analysis).
-------
O
O
Figure 47.
99.9
99
M
•rt
(A
*o
o>
cti
u
•H
"O
•H
5
rt
5
(A
IA
0>
i-l
CO
(A
C8
£
80
70
60
50
40
30
20
10
0.1
0.1
- includes interstage losses
- excludes interstage losses
- Typical 95% confidence intervals
j—L
L
J
100.0
1•°Aerodynamic Diameter (urn)10'0
Effect of interstage losses on measured size distribution for polydisperse uranine aerosol
sampled onto spray silicone collection surfaces at high stage loadings (MK III University of
Washington source test cascade impactor, 0.5 cfm, 70 F, 30.20 "Hg, 10.0 mg total loading,
fluorometric analysis). -.
-------
99.9,-
99
•H
TJ
O
4-t
re
•j
• H
I
V)
Q>
V)
tn
90
80
70
60
50
40
30
20
10
0.1
I
^f
/
n
0- includes interstage losses
\- excludes-interstage losses
T- Typical 95% confidence intervals
. . I
. I
0.1
Aerodynamic Diameter (umj *
100.0
Figure 48. Effect of interstage losses on measured size distribution for polydisperse uranine aerosol
sampled onto glass fiber collection surface at low stage loading (MK III University of
Washington source test cascade impactor, 0.5 cfm, 70°F, 30.19 "Hg, 0.3 mg total loading,
fluorometric analysis).
-------
o
K)
N
•rl
in
•O
4)
4->
rt
o
•H
T3
I
10
in
VI
Figure
99.9
99
90
80
70
60
50
40
30
20
10
0.1
1 -
0.1
•
.5
1
«*
mm
/
- includes interstage losses
- excludes interstage losses
T- Typical 95% confidence intervals
I
1.0 10.0
Aerodynamic Diameter (vim)
100.0
49. Effect of interstage losses on measured size distribution for polydisperse uranine aerosol
sampled onto glass fiber collection surface at high stage loadings (HK III University of
Washington source test cascade impactor, 0.5 cfm, 70°F, 30.20 "Hg, 10.0 mg total loading,
fluorometric analysis).
-------
Total Interstage Loss =8.6%
P»
»
. , , 1
S-l
S-2 S-3
S-4
S-5
S-6
S-7
S-8
a.
Andersen MK III stack sampler, 0.5 cfm, glass fiber collection
surface, 15.0 mg total loading, fluorometric analysis.
(/>
i
nl
•M
O
4J
(4-1
O
in
3
OJ
00
rt
10
Total Interstage Loss - 4.2%
0
S-l S-2 S~3 S-4 S-f. S-6
b. Sierra Model 226 source cascade impactor, 0.25 cfm, glass fiber
collection surfaci, 15.0 mg total loaidug, fluorometric
analysis.
IS
10
Total Interstage Loss » 35.8%
0
c.
Figure 50 .
S-l S-2
Modified Brink Model B cascade impactor, 0.05 cfm, glass fiber
collection surface, 15.0 mg total loading, fluorometric
analysis.
Histograms of measured interstage losses for the Andersen,
Sierra and Brink impactors (polydisperse uranine aerosol
of mmd = 0.72 urn).
103
-------
•n
-------
o
C/l
99.9 r-
99
01
N
•H
tft
0)
•o
•H
CO
tn
0)
in
V)
90
80
70
60
50
40
30
20
10
0.1
0.1
^
excludes losses
includes losses
. t
1
100.0
1.0 10.0
Aerodynamic Diameter (urn)
Figure 52. Effect of interstage losses on measured size distribution for a polydisperse uranine aerosol
sampled using the Sierra Model 226 source cascade impactor (0.25 cfm, glass fiber collection
surface, 70 F, 30.04 "Hg, 15.0 mg total loading, gravimetric analysis).
-------
o
Figure
99.9
99
•M
0»
O
•H
I
4-*
to
tf)
90
80
70
60
SO
40
30
t/> 20
10
0.1
.1
includes losses
excludes losses
-L
1.0 10.0
Aerodynamic Diameter (ym)
100.0
Effect of interstage losses on measured size distribution of polydisperse uranine aerosol
sampled using the modified Brink Model B cascade impactor (0.05 cfm, glass-fitter collection surface,
70°F, 30.06 "Hg, 15.0 mg total loading, gravimetric analysis).
-------
REFERENCES ,
1. National Council of the Paper Industry for Air and Stream
Improvement, Inc., "Procedures for Source Emission Particle
Sizing Using Cascade Impactors and Microscopic Sizing Techniques,"
Atmospheric Quality Improvement Technical Bulletin No 71, May 1974.
2. Calvert, S., C. Lake and R. Parker, "Cascade Impactor Calibration
Guidelines," Environmental Protection Technology Series,
EPA-600/2-76-118, (NTIS No. PB 252 656), April 1976.
3. Gushing, K.M., G.E. Lacey, J.D. McCain and W.B. Smith, "Paniculate
Sizing Techniques for Control Device Evaluation: Cascade Impactor
Calibrations," Environmental Protection Technology Series,
EPA-600/2-76-280, U.S. Environmental Protection Agency, Research
Triangle Park, N.C., October 1976.
4. Harris, D.B., 'Tentative Procedures for Particle Sizing in
Process Streams: Cascade Iinpactors," Environmental Protection
Technology Series, EPA-600/2-76-023 (NTIS No. PB 250 375),
February 1976.
5. McCain, J.D. , K.M. dishing and A.N. Bird, Jr., "Field Measurements
of Particle Size Distribution with Inertial Sizing Devices,"
Environmental Protection Technology Series, EPA-650/2-73-035,
(NTIS No. PB 226 292), October 1973.
6. Harris, D.B., "Procedures for Cascade Impactor Calibration and
Operation in Process Streams," Environmental Protection
Technology Series, EPA-600/2-77-004, U.S. Environmental Protection
Agency, Research Triangle Park, N.C., January 1977.
7. Smith, W.B., K.M. Cushing, G E. Lacty and J.D. McCain,
"Particulate Sizing Techniques for Control Device Evaluation,
Environmental Protection Technology Series, IiPA-650/2-74-102-a,
U.S. Environmental Protection Agency, Research Triangle Park,
N.C., August 1975.
«. Smith, W.B., K.M. Gushing and G.E. Lacey, "Andersen Filter
Substrate Weight Loss," Environmental Protection Technology
Series,, EPA-650/2-75-022 (NTIS No. PB 240 720), February 1975.
Rao, A.K., An Experimental Stiidy_ojJneTtULggS£Sg"' D<^r
Thesis, UniveTslty of Minnesota; Minneapolis, Minn., June 1975.
107
-------
10. Willeke, K., "Performance of the Slotted fmpactor," Amer. Ind.
Hygiene Assoc. J., _36:683 (1975).
11. May, K.R., "The Cascade Impactor: An Instrument for Sampling
Coarse Aerosols," J. Sci. Instr., 22:187, 1945.
12. Marple, V.A., A Fundamental Study of Inert ial Impactors,
Doctoral Thesis, University of Minnesota, Minneapolis, Minn.,
December 1970.
13. Ranz, W.E. and J.B. Wong, "Impaction of Dust and Smoke Particles,"
Ind. and Eng. Chem.. 44_: 1371-81, (1952) t
14. Davies, C.N. and M. Aylward, "The Trajectories of Heavy, Solid
Particles in a Two Dimensional Jet of Ideal fluid Impinging
Normally Upon a Plate," Proc. Phys. Soc., 864:889 (1951).
15. Mercer, T.T. and H.Y. Chow, "Impaction from Rectangular Jets,"
J. Coll. and Interface Sci., 27_:75 (1968).
16. _ , and R.G. Stafford, "Impaction from Round Jets,"
Ann. Occup. Hyg., 12:41 (1969).
\
17. Wilcox, J.D., "Design of a New Five-Stage Cascade Impactor,"
A.M. A. Arch. Ind. Hyg. and Occup. Med., 7^376 (1953).
18. Fuchs, N.A., The Mechanics of Aerosols, Pergamon Press, New
York, (1964).
19. Cooper, D.W. and J.W. Davis, "Cascade Impactors for Aerosols:
Improved Data Analysis," Amer. Ind. Hygiene Assoc. J., 33:79
(1972).
20. Mercer, T.T., "The Interpretation of Cascade Impactor Data,"
Amer. Ind. Hygiene Assoc. J^., 26_:236 (1965).
21. Natusch, D.F.S. and J.R. Wallace, "Determination of Airborne
Particle Size Distributions: Calculation of Cross-Sensitivity
and Discreteness Effects in Cascade Impaction," Atmos. Environ.
JU):315 (1976). '
22. Jordan, D.W., "The Adhesion of Dust Particles," Brit. J. Appl.
Phys. Suppl.. 3_:sl94 (1954).
23. Dahneke, B., "The Capture of Aerosol Particles by Surfaces,"
J. Coll. and Interface Sci.. 3J7:342 (1971).
24. Loffler, F., "The Adhesion of Dust Particles to Fibrous and
Particulaje Surfaces," Staub-Reinhalt. Luft, 28:29 (1968).
108
-------
25- Bucchloz, H., "On the Separation of Airborne Matter by Inertia
Effect in the Submicron Range," Staub-Reinhalt. Luft. 30:15 (1970).
26. Berner, A., "Practical Experience with a 20-Stage Impactor,"
Staub-Reinhalt. Luft. 32_:1 (1972).
27. Lundgren, D.A., "An Aerosol Sampler for Determination of Particle
Concentration of Size and Time," J. Air Poll. Contr. Assoc.,
r7:225 (1967). ~~~
28- Loffler, F., "Blow-off of Particles Collected on Fiber Filters,"
Filtration and Separation. 9_:688 (1972j.
29- Winkler, P., "Relative Humidity and the Adhesion of Atmospheric
Particles to the Plates of Impactors," Aerosol Science, 5:235
(1974). ~
30. Willeke, K. and J.J. McFeters, "The Influence of Flow Entry and
Collecting Surface on the Impaction Efficiency of Inertial
Impactors,"'J. Coll. and Interface Sci., 53_:121 (1975). »
31. Felix, G., G.E. Clinard, G.E. Lacey and J.D. McCain, "Inertial
Cascade Impactor Substrate Media for Flue Gas Sampling,"
Interagency Energy-Environmental Research and Development Series,
EPA-600/7-77-060, U.S. Environmental Protection Agency, Research
Triangle Park, N.C., June 1977.
32. Picknett, R.G., "A New Method for Determining Aerosol Size
Distributions from Multistage Sampler Data," Aerosol Science,
3_:18S (1972).
33. Berglund, R.N. and B.Y.H. Liu, "Generation of Monodisperse Aerosol
Standards," Environ. Sci. and Tech., _7:147 (1973).
34. Green, H,L. and W.R. Lane, Pariiculate Clouds: Dusts, Smokes and
Mists, E. and F.N. Spon. Ltd., London (1957).
35 . Whitby, K.T., "Generator for Producing High Concentrations of
Small Ions," Rev. Sci. Inst., 32_:1351 (1961).
109
-------
APPENDIX A
GUIDE FOR THE USE OF IN-STACK CASCADE IMPACTORS
The guide presented below is not meant to be a complete set of
instructions for the field use of in-stack impactors. The information is
presented as supplementary material to previously published procedures for the
use of in-stack impactors (4,6).
I. Selection of collection surface
A. Spray silicone coating
1. For use at temperatures less than 400°F.
2. Apply the approximate thickness of the size particles which
will be caught.
3. Precondition for at least one hour at the temperature at which
sampling will occur.
B. Glass fiber
1. For use at temperatures less than approximately 1000°F.
2. Precondition to avoid SOa uptake (31,8).
3. Precondition for at least one hour at the temperature at which
sampling will occur.
4. Measured size distribution mmd could be on the order of 30%
greater than actual.
5. Avoid use when sampling "hard" aerosols due to increased bounce
related errors.
C. Uncoated aluminum
1. For use at temperatures less than approximately 900°F.
2. Precondition for at least one hour at the temperature at which
sampling will occur.
3. Avoid use when sampling oil aerosols or "hard" aerosols due to
unstable collection characteristics.
II. Gas sampling rate
A. MK III University of Washington source test cascade impactor
1. 0.25 cfm <_ Q < 1.0 cfm when sampling either oil or hygroscopic
type aerosols.
110
-------
2. 0.25 cfm <_ Q < 0.5 cfra when sampling "hard" aerosols.
B. Sierra Model 226 source test cascade impactor
1. 0.25 cfm £ Q <^ 1.0 cfm when sampling either oil or hygroscopic
type aerosols.
2. 0.25 cfm <^ Q <_ 0.5 cfm when sampling "hard" aerosols.
C. Andersen MK III stack sampler
1. 0.25 cfm <_ Q <^ 1.0 cfm when sampling either oil or hygroscopic
type aerosols.
2. 0.25 cfm <_ Q <_ 0.5 cfm when sampling "hard" aerosols.
D. Modified Brink Model B cascade impactor
1. Q < 0.05 cfm when sampling either oil or hygroscopic type
aerosols.
2. Q < 0.025 cfm when sampling "hard" aerosols.
E. General selection of flow rate
1. Maintain jet velocities < 75 m/sec when sampling either oil or
hygroscopic type aerosols.
2. Maintain jet velocities < 50 m/sec when sampling "hard"
aerosols.
3. Choose a flow rate which will provide sizing information over
the range of the expected mmd, within the above limits.
III. Sta'ge loadings
A. Hygroscopic type aerosols — 5-7 mg upper limit
B. Oil aerosols — 15 mg upper limit
C. General upper limit
1. Observe back side of nozzle for increased deposition.
2. Observe primary deposits for uniformity.
D. General lower limit
i
1. Collect at least 10X the error associated with the mass
measurement.
IV. Treatment of interstage losses
A. Suggest excluding them from calculations due to errors involved in
trying to recover these losses
B. Must include losses as total collected mass if calculating mass
loadings of the process stream
V. Treatment of sizing data
Consider that the error associated with a mass measurement is
inversely proportional to the mass collected, therefore when
constructing the distribution, give greater weight to the data
points representing the majority of the collected mass.
Ill
-------
4. TITLE AND SUBTI TLE
USE AND LIMITATIONS OF IN-STACK IMPACTORS
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing/
1. REPORT NO.
EPA-60Q/2-80-048
3. RECIPIENT'S ACCESSION NO.
5. REPORT DATE
February 1980
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
Dale A. Lundgren and W. David Balfour
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Department of Environmental Engineering Sciences
University of Florida
Gainesville, Florida 32611
12. SPONSORING AGENCY NAME AND ADDRESS
Environmental Sciences Research Laboratory - RTP, NC
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, N.C. 27711
10. PROGRAM ELEMENT NO.
1AD712 BA 28 (FY-77)
11. CONTRACT/GRANT NO.
Grant R-803692
13. TYPE OF REPORT AND PERIOD COVERED
Final
14. SPONSORING AGENCY CODE
EPA/600/09
15. SUPPLEMENTARY NOTES
16. ABSTRACT
A systematic evaluation of the operating parameters for four commercially available
in-stack cascade impactors was carried out with polydisperse test aerosols. Test
aerosols used were polystyrene latex spheres, uranine, sodium chloride, dioctyl
phthalate, or dinonyl phthalate. The effect upon the apparent measured size distri-
bution of each polydisperse test aerosol was noted for various gas sampling rates
(flow rates), types of impactor collection surfaces (glass fiber, uncoated aluminum,
and aluminum coated with silicone), stage loadings and interstage losses. Collection
surfaces were further characterized as to their weight loss during exposure to ele-
vated temperatures and their tendency to be blown off by an impinging jet of air.
Measurements revealed that interstage losses may amount to 30% of the total collected
mass; however, there is little effect upon the apparent measured size distribution
when these losses are ignored. The useful range of flow rates available for the
impactors was defined at the lower end by a loss of useful sizing data and at the
upper end by the presence of particle bounce off the latter stages. In general, the
impactors were found to give similar apparent measured size distributions when
operated.at various flow rates within this useful range.
Recommendations were made for: 1) optimum operation of the impactor when sampling
various types of aerosols, and 2) accounting for observed or known errors in the data.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
* Air pollution
* Evaluation
* Instruments
* Particle size
distribution
b.IDENTIFIERS/OPEN ENDED TERMS C. COSATI Field/Group
In-Stack Cascade
Impactor
13B
14B
18. DISTRIBUTION STATEMENT
RELEASE TO PUBLIC
19. SECURITY CLASS (This Report)
UNCLASSIFIED
21. NO. OF PAGES
126
20. SECURITY CLASS (Thispage)
UNCLASSIFIED
22. PRICE
EPA Form 2220-1 (Rev. 4-77)
PREVIOUS EDITION IS OBSOLETE
112
------- |